Refactored validator_factory interface

Added drop_empty_items

.clang-format

@@ -0,0 +1,22 @@
+BasedOnStyle: LLVM
+UseTab: AlignWithSpaces
+IndentWidth: 4
+TabWidth: 4
+BreakBeforeBraces: Allman
+ColumnLimit: 0
+NamespaceIndentation: Inner
+FixNamespaceComments: true
+AccessModifierOffset: -2
+AllowShortCaseLabelsOnASingleLine: true
+IndentCaseLabels: true
+BreakConstructorInitializers: BeforeComma
+BraceWrapping:
+  BeforeLambdaBody: false
+AlignAfterOpenBracket: DontAlign
+Cpp11BracedListStyle: false
+IncludeBlocks: Regroup
+LambdaBodyIndentation: Signature
+AllowShortLambdasOnASingleLine: Inline
+EmptyLineBeforeAccessModifier: LogicalBlock
+IndentPPDirectives: AfterHash
+PPIndentWidth: 1

.github/workflows/cmake-multi-platform.yml

@@ -33,13 +33,18 @@ jobs:
 
     - name: Install dependencies Ubuntu
       if: matrix.os == 'ubuntu-latest'
-      run: sudo apt-get update && sudo apt-get install mrc
+      run: sudo apt-get update && sudo apt-get install mrc catch2
 
     - name: Install dependencies Window
       if: matrix.os == 'windows-latest'
       run: ./tools/depends.cmd
       shell: cmd
 
+    - name: Install Catch2 macOS
+      if: matrix.os == 'macos-latest'
+      run: >
+        brew install catch2
+
     - name: Configure CMake
       run: >
         cmake -B ${{ steps.strings.outputs.build-output-dir }}

CMakeLists.txt

@@ -24,16 +24,18 @@
 
 cmake_minimum_required(VERSION 3.23)
 
+cmake_policy(SET CMP0135 NEW)
+
 if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR AND NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
-	set(CMAKE_BUILD_TYPE Release CACHE STRING "Build type" FORCE)
-	set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release" "MinSizeRel" "RelWithDebInfo")
+    set(CMAKE_BUILD_TYPE Release CACHE STRING "Build type" FORCE)
+    set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release" "MinSizeRel" "RelWithDebInfo")
 endif()
 
 # set the project name
 project(
-	libcifpp
-	VERSION 9.0.0
-	LANGUAGES CXX C)
+    libcifpp
+    VERSION 10.0.0
+    LANGUAGES CXX C)
 
 list(PREPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake")
 
@@ -43,19 +45,20 @@ include(GenerateExportHeader)
 include(CTest)
 include(ExternalProject)
 include(FetchContent)
+include(VersionString)
 
 # When building with ninja-multiconfig, build both debug and release by default
 if(CMAKE_GENERATOR STREQUAL "Ninja Multi-Config")
-	set(CMAKE_CROSS_CONFIGS "Debug;Release")
-	set(CMAKE_DEFAULT_CONFIGS "Debug;Release")
+    set(CMAKE_CROSS_CONFIGS "Debug;Release")
+    set(CMAKE_DEFAULT_CONFIGS "Debug;Release")
 endif()
 
 if("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU")
-	set(CMAKE_CXX_FLAGS
-		"${CMAKE_CXX_FLAGS} -Wall -Wextra -Wno-unused-parameter -Wno-missing-field-initializers"
-	)
+    set(CMAKE_CXX_FLAGS
+        "${CMAKE_CXX_FLAGS} -Wall -Wextra -Wno-unused-parameter -Wno-missing-field-initializers"
+    )
 elseif(MSVC)
-	set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /W4")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /W4")
 endif()
 
 # Build documentation?
@@ -64,126 +67,147 @@ set(BUILD_DOCUMENTATION OFF CACHE BOOL "Build the documentation")
 # Optionally build a version to be installed inside CCP4
 set(BUILD_FOR_CCP4 OFF CACHE BOOL "Build a version to be installed in CCP4")
 
+# Create the cql/sqlite interface
+set(BUILD_SQLITE_INTERFACE ON CACHE BOOL "Build the sqlite interface")
+
 # Building shared libraries?
-if(NOT(BUILD_FOR_CCP4 AND WIN32))
-	set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build a shared library instead of a static one")
+if(NOT (BUILD_FOR_CCP4 AND WIN32))
+    set(BUILD_SHARED_LIBS OFF CACHE BOOL "Build a shared library instead of a static one")
 endif()
 
 if(PROJECT_IS_TOP_LEVEL AND NOT BUILD_FOR_CCP4)
-	# Lots of code depend on the availability of the components.cif file
-	set(CIFPP_DOWNLOAD_CCD ON CACHE BOOL "Download the CCD file components.cif during installation")
+    # Lots of code depend on the availability of the components.cif file
+    set(CIFPP_DOWNLOAD_CCD ON CACHE BOOL "Download the CCD file components.cif during installation")
 
-	# An optional cron script can be installed to keep the data files up-to-date
-	if(UNIX AND NOT APPLE)
-		set(CIFPP_INSTALL_UPDATE_SCRIPT ON CACHE BOOL "Install the script to update CCD and dictionary files")
-	endif()
+    # An optional cron script can be installed to keep the data files up-to-date
+    if(UNIX AND NOT APPLE)
+        set(CIFPP_INSTALL_UPDATE_SCRIPT ON CACHE BOOL "Install the script to update CCD and dictionary files")
+    endif()
 else()
-	unset(CIFPP_DOWNLOAD_CCD)
-	unset(CIFPP_INSTALL_UPDATE_SCRIPT)
+    unset(CIFPP_DOWNLOAD_CCD)
+    unset(CIFPP_INSTALL_UPDATE_SCRIPT)
 endif()
 
 # When CCP4 is sourced in the environment, we can recreate the symmetry
 # operations table
 if(EXISTS "$ENV{CCP4}/lib/data/syminfo.lib")
-	set(CIFPP_RECREATE_SYMOP_DATA ON CACHE BOOL "Recreate SymOp data table in case it is out of date")
+    set(CIFPP_RECREATE_SYMOP_DATA ON CACHE BOOL "Recreate SymOp data table in case it is out of date")
 endif()
 
 # CCP4 build
 if(BUILD_FOR_CCP4)
-	if("$ENV{CCP4}" STREQUAL "" OR NOT EXISTS $ENV{CCP4})
-		message(FATAL_ERROR "cifpp: A CCP4 built was requested but CCP4 was not sourced")
-	else()
-		list(PREPEND CMAKE_MODULE_PATH "$ENV{CCP4}")
-		list(PREPEND CMAKE_PREFIX_PATH "$ENV{CCP4}")
-		set(CMAKE_INSTALL_PREFIX "$ENV{CCP4}")
+    if("$ENV{CCP4}" STREQUAL "" OR NOT EXISTS $ENV{CCP4})
+        message(FATAL_ERROR "cifpp: A CCP4 built was requested but CCP4 was not sourced")
+    else()
+        list(PREPEND CMAKE_MODULE_PATH "$ENV{CCP4}")
+        list(PREPEND CMAKE_PREFIX_PATH "$ENV{CCP4}")
+        set(CMAKE_INSTALL_PREFIX "$ENV{CCP4}")
 
-		if(WIN32)
-			set(BUILD_SHARED_LIBS ON)
-		endif()
-	endif()
+        if(WIN32)
+            set(BUILD_SHARED_LIBS ON)
+        endif()
+    endif()
 endif()
 
 # Now include the GNUInstallDirs module
 include(GNUInstallDirs)
 
 if(WIN32)
-	if(${CMAKE_SYSTEM_VERSION} GREATER_EQUAL 10) # Windows 10
-		add_definitions(-D _WIN32_WINNT=0x0A00)
-	elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.3) # Windows 8.1
-		add_definitions(-D _WIN32_WINNT=0x0603)
-	elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.2) # Windows 8
-		add_definitions(-D _WIN32_WINNT=0x0602)
-	elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.1) # Windows 7
-		add_definitions(-D _WIN32_WINNT=0x0601)
-	elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.0) # Windows Vista
-		add_definitions(-D _WIN32_WINNT=0x0600)
-	else() # Windows XP (5.1)
-		add_definitions(-D _WIN32_WINNT=0x0501)
-	endif()
+    if(${CMAKE_SYSTEM_VERSION} GREATER_EQUAL 10) # Windows 10
+        add_definitions(-D _WIN32_WINNT=0x0A00)
+    elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.3) # Windows 8.1
+        add_definitions(-D _WIN32_WINNT=0x0603)
+    elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.2) # Windows 8
+        add_definitions(-D _WIN32_WINNT=0x0602)
+    elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.1) # Windows 7
+        add_definitions(-D _WIN32_WINNT=0x0601)
+    elseif(${CMAKE_SYSTEM_VERSION} EQUAL 6.0) # Windows Vista
+        add_definitions(-D _WIN32_WINNT=0x0600)
+    else() # Windows XP (5.1)
+        add_definitions(-D _WIN32_WINNT=0x0501)
+    endif()
 
-	# We do not want to write an export file for all our symbols...
-	set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
+    # We do not want to write an export file for all our symbols...
+    set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
 endif()
 
 if(MSVC)
-	# make msvc standards compliant...
-	add_compile_options(/permissive- /bigobj)
-	add_link_options(/NODEFAULTLIB:library)
+    # make msvc standards compliant...
+    add_compile_options(/permissive- /bigobj)
+    add_link_options(/NODEFAULTLIB:library)
 
-	# This is dubious...
-	if(BUILD_SHARED_LIBS)
-		set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>DLL")
-	else()
-		set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>")
-	endif()
+    # This is dubious...
+    if(BUILD_SHARED_LIBS)
+        set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>DLL")
+    else()
+        set(CMAKE_MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>")
+    endif()
 endif()
 
 # Libraries
 
 if(MSVC)
-	# Avoid linking the shared library of zlib. Search ZLIB_ROOT first if it is
-	# set.
-	if(ZLIB_ROOT)
-		set(_ZLIB_SEARCH_ROOT PATHS ${ZLIB_ROOT} NO_DEFAULT_PATH)
-		list(APPEND _ZLIB_SEARCHES _ZLIB_SEARCH_ROOT)
-	endif()
+    # Avoid linking the shared library of zlib. Search ZLIB_ROOT first if it is
+    # set.
+    if(ZLIB_ROOT)
+        set(_ZLIB_SEARCH_ROOT PATHS ${ZLIB_ROOT} NO_DEFAULT_PATH)
+        list(APPEND _ZLIB_SEARCHES _ZLIB_SEARCH_ROOT)
+    endif()
 
-	# Normal search.
-	set(_ZLIB_x86 "(x86)")
-	set(_ZLIB_SEARCH_NORMAL
-		PATHS "[HKEY_LOCAL_MACHINE\\SOFTWARE\\GnuWin32\\Zlib;InstallPath]"
-		"$ENV{ProgramFiles}/zlib" "$ENV{ProgramFiles${_ZLIB_x86}}/zlib")
-	unset(_ZLIB_x86)
-	list(APPEND _ZLIB_SEARCHES _ZLIB_SEARCH_NORMAL)
+    # Normal search.
+    set(_ZLIB_x86 "(x86)")
+    set(_ZLIB_SEARCH_NORMAL
+        PATHS "[HKEY_LOCAL_MACHINE\\SOFTWARE\\GnuWin32\\Zlib;InstallPath]"
+        "$ENV{ProgramFiles}/zlib" "$ENV{ProgramFiles${_ZLIB_x86}}/zlib")
+    unset(_ZLIB_x86)
+    list(APPEND _ZLIB_SEARCHES _ZLIB_SEARCH_NORMAL)
 
-	if(BUILD_FOR_CCP4)
-		list(PREPEND _ZLIB_SEARCHES "$ENV{CCP4}/lib")
-	endif()
+    if(BUILD_FOR_CCP4)
+        list(PREPEND _ZLIB_SEARCHES "$ENV{CCP4}/lib")
+    endif()
 
-	foreach(search ${_ZLIB_SEARCHES})
-		find_library(
-			ZLIB_LIBRARY
-			NAMES zlibstatic NAMES_PER_DIR ${${search}}
-			PATH_SUFFIXES lib)
-	endforeach()
+    foreach(search ${_ZLIB_SEARCHES})
+        find_library(
+            ZLIB_LIBRARY
+            NAMES zlibstatic NAMES_PER_DIR ${${search}}
+            PATH_SUFFIXES lib)
+    endforeach()
 endif()
 
-# First check if <format> is available
-find_file(FMT NAME format)
+# Using fast_float for float parsing, but only if needed
+try_compile(STD_CHARCONV_COMPILING
+	SOURCES ${CMAKE_CURRENT_SOURCE_DIR}/cmake/test-charconv.cpp
+	CXX_STANDARD 20
+	CXX_STANDARD_REQUIRED ON)
 
-if(FMT EQUAL "FMT-NOTFOUND")
-	if(NOT (fmt_FOUND OR TARGET fmt))
-		find_package(fmt REQUIRED)
-		message(FATAL_ERROR "fmt not found, compiler too old, you're out of luck")
+if(NOT STD_CHARCONV_COMPILING)
+	message(NOTICE "libcifpp: Using fast_float for std::from_chars")
+	find_package(FastFloat 8.0 QUIET CONFIG)
+	if(NOT FastFloat_FOUND)
+		message(STATUS "FastFloat not found in system, fetching from GitHub")
+		FetchContent_Declare(fastfloat
+			GIT_REPOSITORY "https://github.com/fastfloat/fast_float"
+			GIT_TAG v8.0.2
+			EXCLUDE_FROM_ALL)
+		FetchContent_MakeAvailable(fastfloat)
 	endif()
 endif()
 
 find_package(Threads)
-include(FindPCRE2)
 find_package(ZLIB QUIET)
 
 if(NOT ZLIB_FOUND)
-	message(FATAL_ERROR "cifpp: The zlib development files were not found you this system, please install them and try again (hint: on debian/ubuntu use apt-get install zlib1g-dev)")
+    message(FATAL_ERROR "cifpp: The zlib development files were not found you this system, please install them and try again (hint: on debian/ubuntu use apt-get install zlib1g-dev)")
+endif()
+
+include(FindPkgConfig)
+
+if(PKG_CONFIG_FOUND)
+    pkg_check_modules(PCRE2 IMPORTED_TARGET libpcre2-8)
+endif()
+
+if(NOT PCRE2_FOUND)
+    add_subdirectory(pcre2-simple)
 endif()
 
 # Using Eigen3 is a bit of a thing. We don't want to build it completely since
@@ -192,352 +216,362 @@ endif()
 find_package(Eigen3 3.4 QUIET)
 
 if(Eigen3_FOUND AND TARGET Eigen3::Eigen)
-	get_target_property(EIGEN_INCLUDE_DIR Eigen3::Eigen
-		INTERFACE_INCLUDE_DIRECTORIES)
+    get_target_property(EIGEN_INCLUDE_DIR Eigen3::Eigen
+        INTERFACE_INCLUDE_DIRECTORIES)
 else()
-	# Use ExternalProject since FetchContent always tries to install the result...
-	ExternalProject_Add(my-eigen3
-		URL https://gitlab.com/libeigen/eigen/-/archive/3.4.0/eigen-3.4.0.zip
-		DOWNLOAD_EXTRACT_TIMESTAMP TRUE
-		CONFIGURE_COMMAND ""
-		BUILD_COMMAND ""
-		INSTALL_COMMAND "")
+    # Use ExternalProject since FetchContent always tries to install the result...
+    ExternalProject_Add(my-eigen3
+        URL https://gitlab.com/libeigen/eigen/-/archive/3.4.0/eigen-3.4.0.zip
+        DOWNLOAD_EXTRACT_TIMESTAMP TRUE
+        CONFIGURE_COMMAND ""
+        BUILD_COMMAND ""
+        INSTALL_COMMAND "")
 
-	ExternalProject_Get_Property(my-eigen3 SOURCE_DIR)
-	set(EIGEN_INCLUDE_DIR ${SOURCE_DIR})
+    ExternalProject_Get_Property(my-eigen3 SOURCE_DIR)
+    set(EIGEN_INCLUDE_DIR ${SOURCE_DIR})
 endif()
 
-message(STATUS "cifpp: Eigen include dir is ${EIGEN_INCLUDE_DIR}")
-
-# Create a revision file, containing the current git version info
-include(VersionString)
-write_version_header(${CMAKE_CURRENT_SOURCE_DIR}/src/ LIB_NAME "LibCIFPP")
-
 # SymOp data table
 if(CIFPP_RECREATE_SYMOP_DATA)
-	# The tool to create the table
-	add_executable(symop-map-generator
-		"${CMAKE_CURRENT_SOURCE_DIR}/src/symop-map-generator.cpp")
+    # The tool to create the table
+    add_executable(symop-map-generator
+        "${CMAKE_CURRENT_SOURCE_DIR}/src/symop-map-generator.cpp")
 
-	target_compile_features(symop-map-generator PUBLIC cxx_std_20)
+    target_compile_features(symop-map-generator PUBLIC cxx_std_20)
 
-	add_custom_command(
-		OUTPUT ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
-		COMMAND
-		$<TARGET_FILE:symop-map-generator> $ENV{CLIBD}/syminfo.lib
-		$ENV{CLIBD}/symop.lib ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp)
+    add_custom_command(
+        OUTPUT ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
+        COMMAND
+        $<TARGET_FILE:symop-map-generator> $ENV{CLIBD}/syminfo.lib
+        $ENV{CLIBD}/symop.lib ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp)
 
-	add_custom_target(
-		OUTPUT
-		${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
-		DEPENDS symop-map-generator "$ENV{CLIBD}/syminfo.lib"
-		"$ENV{CLIBD}/symop.lib")
+    add_custom_target(
+        OUTPUT
+        ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
+        DEPENDS symop-map-generator "$ENV{CLIBD}/syminfo.lib"
+        "$ENV{CLIBD}/symop.lib")
 endif()
 
+# Create a revision file, containing the current git version info
+write_version_header("${CMAKE_CURRENT_SOURCE_DIR}/src/" LIB_NAME "LibCIFPP")
+
 # Sources
-set(project_sources
-	${CMAKE_CURRENT_SOURCE_DIR}/src/category.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/condition.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/datablock.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/dictionary_parser.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/file.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/item.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/parser.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/row.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/validate.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/text.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/utilities.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/atom_type.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/compound.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/point.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/symmetry.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/model.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/cif2pdb.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb_record.hpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif_remark_3.hpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif_remark_3.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/reconstruct.cpp
-	${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/validate-pdbx.cpp
+list(APPEND project_sources
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/category.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/condition.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/datablock.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/dictionary_parser.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/file.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/item.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/parser.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/row.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/validate.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/text.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/utilities.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/atom_type.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/compound.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/point.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/symmetry.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/model.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/cql.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/cif2pdb.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb_record.hpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif_remark_3.hpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/pdb2cif_remark_3.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/reconstruct.cpp
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/pdb/validate-pdbx.cpp
 )
 
-set(project_headers
-	include/cif++.hpp
-	include/cif++/atom_type.hpp
-	include/cif++/category.hpp
-	include/cif++/compound.hpp
-	include/cif++/condition.hpp
-	include/cif++/datablock.hpp
-	include/cif++/dictionary_parser.hpp
-	include/cif++/exports.hpp
-	include/cif++/file.hpp
-	include/cif++/format.hpp
-	include/cif++/forward_decl.hpp
-	include/cif++/gzio.hpp
-	include/cif++/item.hpp
-	include/cif++/iterator.hpp
-	include/cif++/matrix.hpp
-	include/cif++/model.hpp
-	include/cif++/parser.hpp
-	include/cif++/pdb/cif2pdb.hpp
-	include/cif++/pdb.hpp
-	include/cif++/pdb/io.hpp
-	include/cif++/pdb/pdb2cif.hpp
-	include/cif++/pdb/tls.hpp
-	include/cif++/point.hpp
-	include/cif++/row.hpp
-	include/cif++/symmetry.hpp
-	include/cif++/text.hpp
-	include/cif++/utilities.hpp
-	include/cif++/validate.hpp
+if(BUILD_SQLITE_INTERFACE)
+    list(APPEND project_sources
+        ${CMAKE_CURRENT_SOURCE_DIR}/src/sqlite3/sqlite3.c)
+endif()
+
+list(APPEND project_headers
+    include/cif++.hpp
+    include/cif++/atom_type.hpp
+    include/cif++/category.hpp
+    include/cif++/compound.hpp
+    include/cif++/condition.hpp
+    include/cif++/datablock.hpp
+    include/cif++/dictionary_parser.hpp
+    include/cif++/exports.hpp
+    include/cif++/file.hpp
+    include/cif++/format.hpp
+    include/cif++/forward_decl.hpp
+    include/cif++/gzio.hpp
+    include/cif++/item.hpp
+    include/cif++/iterator.hpp
+    include/cif++/matrix.hpp
+    include/cif++/model.hpp
+    include/cif++/parser.hpp
+    include/cif++/pdb/cif2pdb.hpp
+    include/cif++/pdb.hpp
+    include/cif++/pdb/io.hpp
+    include/cif++/pdb/pdb2cif.hpp
+    include/cif++/pdb/tls.hpp
+    include/cif++/point.hpp
+    include/cif++/row.hpp
+    include/cif++/symmetry.hpp
+    include/cif++/text.hpp
+    include/cif++/cql.hpp
+    include/cif++/utilities.hpp
+    include/cif++/validate.hpp
 )
 
 add_library(cifpp)
 add_library(cifpp::cifpp ALIAS cifpp)
 
 if(TARGET my-eigen3)
-	add_dependencies(cifpp my-eigen3)
+    add_dependencies(cifpp my-eigen3)
 endif()
 
 target_sources(cifpp
-	PRIVATE ${project_sources}
-	${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
-	PUBLIC
-	FILE_SET cifpp_headers TYPE HEADERS
-	BASE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/include
-	FILES ${project_headers}
+    PRIVATE ${project_sources}
+    ${CMAKE_CURRENT_SOURCE_DIR}/src/symop_table_data.hpp
+    PUBLIC
+    FILE_SET cifpp_headers TYPE HEADERS
+    BASE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/include
+    FILES ${project_headers}
 )
 
 # The code now really requires C++20
 target_compile_features(cifpp PUBLIC cxx_std_20)
 
 generate_export_header(cifpp EXPORT_FILE_NAME
-	${CMAKE_CURRENT_SOURCE_DIR}/include/cif++/exports.hpp)
+    ${CMAKE_CURRENT_SOURCE_DIR}/include/cif++/exports.hpp)
 
 if(MSVC)
-	target_compile_definitions(cifpp PUBLIC NOMINMAX=1)
+    target_compile_definitions(cifpp PUBLIC NOMINMAX=1)
 endif()
 
 set_target_properties(cifpp PROPERTIES POSITION_INDEPENDENT_CODE ON)
 
 target_include_directories(
-	cifpp
-	PUBLIC "$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>"
-	"$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>"
-	PRIVATE "${EIGEN_INCLUDE_DIR}")
+    cifpp
+    PUBLIC "$<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>"
+    "$<INSTALL_INTERFACE:${CMAKE_INSTALL_INCLUDEDIR}>"
+    PRIVATE "${EIGEN_INCLUDE_DIR}")
 
 target_link_libraries(cifpp
-	PUBLIC Threads::Threads ZLIB::ZLIB $<$<TARGET_EXISTS:std::atomic>:std::atomic> $<BUILD_INTERFACE:pcre2-8>)
+    PUBLIC Threads::Threads ZLIB::ZLIB $<$<TARGET_EXISTS:std::atomic>:std::atomic>)
 
-if(fmt_FOUND)
-	target_link_libraries(cifpp PUBLIC fmt)
+if(PCRE2_FOUND)
+    target_include_directories(cifpp PRIVATE ${PCRE2_INCLUDE_DIRS})
+    target_link_libraries(cifpp PRIVATE ${PCRE2_LINK_LIBRARIES})
+else()
+    target_link_libraries(cifpp PRIVATE $<BUILD_INTERFACE:pcre2s>)
+endif()
+
+if(NOT STD_CHARCONV_COMPILING)
+    target_link_libraries(cifpp PRIVATE FastFloat::fast_float)
 endif()
 
 if(CMAKE_CXX_COMPILER_ID STREQUAL "AppleClang")
-	target_link_options(cifpp PRIVATE -undefined dynamic_lookup)
+    target_link_options(cifpp PRIVATE -undefined dynamic_lookup)
 endif(CMAKE_CXX_COMPILER_ID STREQUAL "AppleClang")
 
 if(CIFPP_DOWNLOAD_CCD)
-	# download the components.cif file from CCD
-	set(COMPONENTS_CIF ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/components.cif)
+    # download the components.cif file from CCD
+    set(COMPONENTS_CIF ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/components.cif)
 
-	if(EXISTS ${COMPONENTS_CIF})
-		file(SIZE ${COMPONENTS_CIF} CCD_FILE_SIZE)
+    if(EXISTS ${COMPONENTS_CIF})
+        file(SIZE ${COMPONENTS_CIF} CCD_FILE_SIZE)
 
-		if(CCD_FILE_SIZE EQUAL 0)
-			message(STATUS "cifpp: Removing empty ${COMPONENTS_CIF} file")
-			file(REMOVE "${COMPONENTS_CIF}")
-		endif()
-	endif()
+        if(CCD_FILE_SIZE EQUAL 0)
+            message(STATUS "cifpp: Removing empty ${COMPONENTS_CIF} file")
+            file(REMOVE "${COMPONENTS_CIF}")
+        endif()
+    endif()
 
-	if(NOT EXISTS ${COMPONENTS_CIF})
-		# Since the file(DOWNLOAD) command in cmake does not use compression, we try
-		# to download the gzipped version and decompress it ourselves.
-		find_program(GUNZIP gunzip)
+    if(NOT EXISTS ${COMPONENTS_CIF})
+        # Since the file(DOWNLOAD) command in cmake does not use compression, we try
+        # to download the gzipped version and decompress it ourselves.
+        find_program(GUNZIP gunzip)
 
-		if(WIN32 OR GUNZIP STREQUAL "GUNZIP-NOTFOUND")
-			file(
-				DOWNLOAD https://files.wwpdb.org/pub/pdb/data/monomers/components.cif
-				${COMPONENTS_CIF}
-				SHOW_PROGRESS
-				STATUS CCD_FETCH_STATUS)
-		else()
-			if(NOT EXISTS "${COMPONENTS_CIF}.gz")
-				file(
-					DOWNLOAD
-					https://files.wwpdb.org/pub/pdb/data/monomers/components.cif.gz
-					${COMPONENTS_CIF}.gz
-					SHOW_PROGRESS
-					STATUS CCD_FETCH_STATUS)
-			endif()
+        if(WIN32 OR GUNZIP STREQUAL "GUNZIP-NOTFOUND")
+            file(
+                DOWNLOAD https://files.wwpdb.org/pub/pdb/data/monomers/components.cif
+                ${COMPONENTS_CIF}
+                SHOW_PROGRESS
+                STATUS CCD_FETCH_STATUS)
+        else()
+            if(NOT EXISTS "${COMPONENTS_CIF}.gz")
+                file(
+                    DOWNLOAD
+                    https://files.wwpdb.org/pub/pdb/data/monomers/components.cif.gz
+                    ${COMPONENTS_CIF}.gz
+                    SHOW_PROGRESS
+                    STATUS CCD_FETCH_STATUS)
+            endif()
 
-			add_custom_command(
-				OUTPUT ${COMPONENTS_CIF}
-				COMMAND "${GUNZIP}" ${COMPONENTS_CIF}.gz
-				WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/)
+            add_custom_command(
+                OUTPUT ${COMPONENTS_CIF}
+                COMMAND "${GUNZIP}" ${COMPONENTS_CIF}.gz
+                WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/)
 
-			add_custom_target(COMPONENTS ALL DEPENDS ${COMPONENTS_CIF})
-		endif()
+            add_custom_target(COMPONENTS ALL DEPENDS ${COMPONENTS_CIF})
+        endif()
 
-		# Do not continue if downloading went wrong
-		list(POP_FRONT CCD_FETCH_STATUS CCD_FETCH_STATUS_CODE)
+        # Do not continue if downloading went wrong
+        list(POP_FRONT CCD_FETCH_STATUS CCD_FETCH_STATUS_CODE)
 
-		if(CCD_FETCH_STATUS_CODE)
-			message(
-				FATAL_ERROR "cifpp: Error trying to download CCD file: ${CCD_FETCH_STATUS}")
-		endif()
-	endif()
+        if(CCD_FETCH_STATUS_CODE)
+            message(
+                FATAL_ERROR "cifpp: Error trying to download CCD file: ${CCD_FETCH_STATUS}")
+        endif()
+    endif()
 endif()
 
 # Installation directories
 if(BUILD_FOR_CCP4)
-	set(CIFPP_DATA_DIR
-		"$ENV{CCP4}/share/libcifpp"
-		CACHE PATH "Directory where dictionary and other static data is stored")
+    set(CIFPP_DATA_DIR
+        "$ENV{CCP4}/share/libcifpp"
+        CACHE PATH "Directory where dictionary and other static data is stored")
 else()
-	set(CIFPP_DATA_DIR
-		"${CMAKE_INSTALL_FULL_DATADIR}/libcifpp"
-		CACHE PATH "Directory where dictionary and other static data is stored")
+    set(CIFPP_DATA_DIR
+        "${CMAKE_INSTALL_FULL_DATADIR}/libcifpp"
+        CACHE PATH "Directory where dictionary and other static data is stored")
 endif()
 
 if(CIFPP_DATA_DIR)
-	target_compile_definitions(cifpp PUBLIC DATA_DIR="${CIFPP_DATA_DIR}")
-	set_target_properties(cifpp PROPERTIES CIFPP_DATA_DIR ${CIFPP_DATA_DIR})
+    target_compile_definitions(cifpp PUBLIC DATA_DIR="${CIFPP_DATA_DIR}")
+    set_target_properties(cifpp PROPERTIES CIFPP_DATA_DIR ${CIFPP_DATA_DIR})
 endif()
 
 if(NOT PROJECT_IS_TOP_LEVEL)
-	set(CIFPP_SHARE_DIR ${CIFPP_DATA_DIR} PARENT_SCOPE)
+    set(CIFPP_SHARE_DIR ${CIFPP_DATA_DIR} PARENT_SCOPE)
 endif()
 
 if(UNIX AND NOT BUILD_FOR_CCP4)
-	if("${CMAKE_INSTALL_PREFIX}" STREQUAL "/usr/local")
-		set(CIFPP_CACHE_DIR
-			"/var/cache/libcifpp"
-			CACHE PATH "The directory where downloaded data files are stored")
-	else()
-		set(CIFPP_CACHE_DIR
-			"${CMAKE_INSTALL_FULL_LOCALSTATEDIR}/cache/libcifpp"
-			CACHE PATH "The directory where downloaded data files are stored")
-	endif()
+    if("${CMAKE_INSTALL_PREFIX}" STREQUAL "/usr/local")
+        set(CIFPP_CACHE_DIR
+            "/var/cache/libcifpp"
+            CACHE PATH "The directory where downloaded data files are stored")
+    else()
+        set(CIFPP_CACHE_DIR
+            "${CMAKE_INSTALL_FULL_LOCALSTATEDIR}/cache/libcifpp"
+            CACHE PATH "The directory where downloaded data files are stored")
+    endif()
 
-	target_compile_definitions(cifpp PUBLIC CACHE_DIR="${CIFPP_CACHE_DIR}")
+    target_compile_definitions(cifpp PUBLIC CACHE_DIR="${CIFPP_CACHE_DIR}")
 
-	set(CIFPP_ETC_DIR
-		"${CMAKE_INSTALL_FULL_SYSCONFDIR}"
-		CACHE PATH "The directory where the update configuration file is stored")
+    set(CIFPP_ETC_DIR
+        "${CMAKE_INSTALL_FULL_SYSCONFDIR}"
+        CACHE PATH "The directory where the update configuration file is stored")
 else()
-	unset(CIFPP_CACHE_DIR)
+    unset(CIFPP_CACHE_DIR)
 endif()
 
 # Install rules
 install(TARGETS cifpp
-	EXPORT cifpp
-	FILE_SET cifpp_headers DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+    EXPORT cifpp
+    FILE_SET cifpp_headers DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 
 if(MSVC AND BUILD_SHARED_LIBS)
-	install(
-		FILES $<TARGET_PDB_FILE:cifpp>
-		DESTINATION ${CMAKE_INSTALL_LIBDIR}
-		OPTIONAL)
+    install(
+        FILES $<TARGET_PDB_FILE:cifpp>
+        DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        OPTIONAL)
 endif()
 
 # Clean up old config files (with old names)
 file(GLOB OLD_CONFIG_FILES
-	${CMAKE_INSTALL_FULL_LIBDIR}/cmake/cifpp/cifppConfig*.cmake
-	${CMAKE_INSTALL_FULL_LIBDIR}/cmake/cifpp/cifppTargets*.cmake)
+    ${CMAKE_INSTALL_FULL_LIBDIR}/cmake/cifpp/cifppConfig*.cmake
+    ${CMAKE_INSTALL_FULL_LIBDIR}/cmake/cifpp/cifppTargets*.cmake)
 
 if(OLD_CONFIG_FILES)
-	message(
-		STATUS "cifpp: Installation will remove old config files: ${OLD_CONFIG_FILES}")
-	install(CODE "file(REMOVE ${OLD_CONFIG_FILES})")
+    message(
+        STATUS "cifpp: Installation will remove old config files: ${OLD_CONFIG_FILES}")
+    install(CODE "file(REMOVE ${OLD_CONFIG_FILES})")
 endif()
 
 install(EXPORT cifpp
-	NAMESPACE cifpp::
-	FILE "cifpp-targets.cmake"
-	DESTINATION lib/cmake/cifpp)
+    NAMESPACE cifpp::
+    FILE "cifpp-targets.cmake"
+    DESTINATION lib/cmake/cifpp)
 
 install(
-	FILES ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_ddl.dic
-	${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_pdbx.dic
-	${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_ma.dic
-	DESTINATION ${CMAKE_INSTALL_DATADIR}/libcifpp)
+    FILES ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_ddl.dic
+    ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_pdbx.dic
+    ${CMAKE_CURRENT_SOURCE_DIR}/rsrc/mmcif_ma.dic
+    DESTINATION ${CMAKE_INSTALL_DATADIR}/libcifpp)
 
 if(CIFPP_DATA_DIR AND CIFPP_DOWNLOAD_CCD)
-	install(FILES ${COMPONENTS_CIF}
-		DESTINATION ${CMAKE_INSTALL_DATADIR}/libcifpp)
+    install(FILES ${COMPONENTS_CIF}
+        DESTINATION ${CMAKE_INSTALL_DATADIR}/libcifpp)
 endif()
 
 set(CONFIG_TEMPLATE_FILE ${CMAKE_CURRENT_SOURCE_DIR}/cmake/cifpp-config.cmake.in)
-set(REQUIRE_FMT ${fmt_FOUND})
 
 configure_package_config_file(
-	${CONFIG_TEMPLATE_FILE} ${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config.cmake
-	INSTALL_DESTINATION lib/cmake/cifpp
-	PATH_VARS CIFPP_DATA_DIR)
+    ${CONFIG_TEMPLATE_FILE} ${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config.cmake
+    INSTALL_DESTINATION lib/cmake/cifpp
+    PATH_VARS CIFPP_DATA_DIR)
 
 install(
-	FILES "${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config.cmake"
-	"${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config-version.cmake"
-	DESTINATION lib/cmake/cifpp)
+    FILES "${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config.cmake"
+    "${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config-version.cmake"
+    DESTINATION lib/cmake/cifpp)
 
 set_target_properties(
-	cifpp
-	PROPERTIES VERSION ${PROJECT_VERSION}
-	SOVERSION "${PROJECT_VERSION_MAJOR}.${PROJECT_VERSION_MINOR}"
-	INTERFACE_cifpp_MAJOR_VERSION ${PROJECT_VERSION_MAJOR})
+    cifpp
+    PROPERTIES VERSION ${PROJECT_VERSION}
+    SOVERSION "${PROJECT_VERSION_MAJOR}.${PROJECT_VERSION_MINOR}"
+    INTERFACE_cifpp_MAJOR_VERSION ${PROJECT_VERSION_MAJOR})
 
 set_property(
-	TARGET cifpp
-	APPEND
-	PROPERTY COMPATIBLE_INTERFACE_STRING cifpp_MAJOR_VERSION)
+    TARGET cifpp
+    APPEND
+    PROPERTY COMPATIBLE_INTERFACE_STRING cifpp_MAJOR_VERSION)
 
 write_basic_package_version_file(
-	"${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config-version.cmake"
-	VERSION ${PROJECT_VERSION}
-	COMPATIBILITY AnyNewerVersion)
+    "${CMAKE_CURRENT_BINARY_DIR}/cifpp/cifpp-config-version.cmake"
+    VERSION ${PROJECT_VERSION}
+    COMPATIBILITY AnyNewerVersion)
 
 if(BUILD_TESTING AND PROJECT_IS_TOP_LEVEL)
-	add_subdirectory(test)
+    add_subdirectory(test)
 endif()
 
 # Optionally install the update scripts for CCD and dictionary files
 if(CIFPP_INSTALL_UPDATE_SCRIPT)
-	configure_file(${CMAKE_CURRENT_SOURCE_DIR}/tools/update-libcifpp-data.in
-		update-libcifpp-data @ONLY)
+    configure_file(${CMAKE_CURRENT_SOURCE_DIR}/tools/update-libcifpp-data.in
+        update-libcifpp-data @ONLY)
 
-	if(${CMAKE_SYSTEM_NAME} STREQUAL "Linux" OR
-		${CMAKE_SYSTEM_NAME} STREQUAL "GNU" OR
-		${CMAKE_SYSTEM_NAME} STREQUAL "FreeBSD")
-		install(
-			FILES ${CMAKE_CURRENT_BINARY_DIR}/update-libcifpp-data
-			DESTINATION ${CMAKE_INSTALL_SYSCONFDIR}/cron.weekly
-			PERMISSIONS OWNER_EXECUTE OWNER_READ GROUP_EXECUTE GROUP_READ WORLD_EXECUTE
-			WORLD_READ)
-	else()
-		message(FATAL_ERROR "cifpp: Don't know where to install the update script")
-	endif()
+    if(${CMAKE_SYSTEM_NAME} STREQUAL "Linux" OR
+        ${CMAKE_SYSTEM_NAME} STREQUAL "GNU" OR
+        ${CMAKE_SYSTEM_NAME} STREQUAL "FreeBSD")
+        install(
+            FILES ${CMAKE_CURRENT_BINARY_DIR}/update-libcifpp-data
+            DESTINATION ${CMAKE_INSTALL_SYSCONFDIR}/cron.weekly
+            PERMISSIONS OWNER_EXECUTE OWNER_READ GROUP_EXECUTE GROUP_READ WORLD_EXECUTE
+            WORLD_READ)
+    else()
+        message(FATAL_ERROR "cifpp: Don't know where to install the update script")
+    endif()
 
-	# a config file, to make it complete
-	# install(DIRECTORY DESTINATION "${CMAKE_INSTALL_LOCALSTATEDIR}/libcifpp")
-	if(NOT EXISTS "${CMAKE_INSTALL_SYSCONFDIR}/libcifpp.conf")
-		file(
-			WRITE ${CMAKE_CURRENT_BINARY_DIR}/libcifpp.conf
-			[[# Uncomment the next line to enable automatic updates
+    # a config file, to make it complete
+    # install(DIRECTORY DESTINATION "${CMAKE_INSTALL_LOCALSTATEDIR}/libcifpp")
+    if(NOT EXISTS "${CMAKE_INSTALL_SYSCONFDIR}/libcifpp.conf")
+        file(
+            WRITE ${CMAKE_CURRENT_BINARY_DIR}/libcifpp.conf
+            [[# Uncomment the next line to enable automatic updates
 # update=true
 ]])
-		install(FILES ${CMAKE_CURRENT_BINARY_DIR}/libcifpp.conf
-			DESTINATION ${CMAKE_INSTALL_SYSCONFDIR})
-		install(
-			CODE "message(\"cifpp: A configuration file has been written to ${CIFPP_ETC_DIR}/libcifpp.conf, please edit this file to enable automatic updates\")"
-		)
+        install(FILES ${CMAKE_CURRENT_BINARY_DIR}/libcifpp.conf
+            DESTINATION ${CMAKE_INSTALL_SYSCONFDIR})
+        install(
+            CODE "message(\"cifpp: A configuration file has been written to ${CIFPP_ETC_DIR}/libcifpp.conf, please edit this file to enable automatic updates\")"
+        )
 
-		install(DIRECTORY DESTINATION ${CMAKE_INSTALL_SYSCONFDIR}/libcifpp/cache-update.d)
-	endif()
+        install(DIRECTORY DESTINATION ${CMAKE_INSTALL_SYSCONFDIR}/libcifpp/cache-update.d)
+    endif()
 
-	target_compile_definitions(cifpp PUBLIC CACHE_DIR="${CIFPP_CACHE_DIR}")
+    target_compile_definitions(cifpp PUBLIC CACHE_DIR="${CIFPP_CACHE_DIR}")
 endif()
 
 if(BUILD_DOCUMENTATION)
-	add_subdirectory(docs)
+    add_subdirectory(docs)
 endif()

changelog

@@ -1,3 +1,26 @@
+Version 10.0.0
+- Added a SQLite interface.
+
+Version 9.0.5
+- Added exists to compound_factory
+- Added sub_matrix, fix and extend determinant calculation
+- Added yet another structure::create_non_poly
+- Remove revision.hpp file in make clean (new VersionString.cmake)
+
+Version 9.0.4
+- Fix various stopping and reconstruction errors
+
+Version 9.0.3
+- Reconstruction fixed when some entity ids are missing
+
+Version 9.0.2
+- Fix code that reconstructs sequences, could throw a map::at
+- Many optimisations in validation and reconstruction code.
+
+Version 9.0.1
+- Use pcre2 from pkg-config if available, if not
+  build a version from the original code.
+
 Version 9.0.0
 - Rename fields of cif::mm::polymer to match the naming
   in mmcif_pdbx.dic. Also, related, fix building mm::structure

cmake/FindPCRE2.cmake

@@ -1,27 +1,12 @@
-set(PCRE2_USE_STATIC_LIBS ON)
+# The problem is, find_package(PCRE2) does not work
+# and using pkg-config results in linking to a shared library
+# causing all kinds of trouble later on
 
-# The cmake config files for pcre2 are broken
+find_path(PCRE2_INCLUDEDIR NAMES pcre2.h HINTS "C:/Program Files (x86)/PCRE2/include" REQUIRED)
+find_library(PCRE2_LIBRARY NAMES pcre2-8-static libpcre2-8.a HINTS "C:/Program Files (x86)/PCRE2/lib" REQUIRED)
 
-include(FindPkgConfig)
-
-if(PKG_CONFIG_FOUND)
-	pkg_check_modules(PCRE2 IMPORTED_TARGET libpcre2-8)
-
-	if(PCRE2_FOUND)
-		message(STATUS "Using pcre2 found using pkg-config")
-
-		add_library(pcre2-8 ALIAS PkgConfig::PCRE2)
-	endif()
-endif()
-
-if(NOT PCRE2_FOUND)
-	find_path(PCRE2_INCLUDEDIR NAMES pcre2.h HINTS "C:/Program Files (x86)/PCRE2/include" REQUIRED)
-	find_library(PCRE2_LIBRARY NAMES pcre2-8-static HINTS "C:/Program Files (x86)/PCRE2/lib" REQUIRED)
-
-	add_library(pcre2-8 IMPORTED STATIC)
-	target_include_directories(pcre2-8 INTERFACE ${PCRE2_INCLUDEDIR})
-	target_compile_definitions(pcre2-8 INTERFACE PCRE2_STATIC)
-	# target_link_libraries(pcre2-8 INTERFACE ${PCRE2_LIBRARY})
-	set_target_properties(pcre2-8 PROPERTIES IMPORTED_LOCATION ${PCRE2_LIBRARY})
-	set_target_properties(pcre2-8 PROPERTIES IMPORTED_IMPLIB ${PCRE2_LIBRARY})
-endif()
+add_library(pcre2-8 IMPORTED STATIC)
+target_include_directories(pcre2-8 INTERFACE ${PCRE2_INCLUDEDIR})
+target_compile_definitions(pcre2-8 INTERFACE PCRE2_STATIC)
+set_target_properties(pcre2-8 PROPERTIES IMPORTED_LOCATION ${PCRE2_LIBRARY})
+set_target_properties(pcre2-8 PROPERTIES IMPORTED_IMPLIB ${PCRE2_LIBRARY})

cmake/VersionString.cmake

@@ -238,7 +238,7 @@ function(write_version_header dir)
 		if(res EQUAL 0)
 			set(REVISION_STRING "${out}")
 		else()
-			message(STATUS "Git hash not found, does this project has a 'build' tag?")
+			message(STATUS "Git hash not found, does this project have a 'build' tag?")
 		endif()
 	else()
 		message(STATUS "Git hash not found")

cmake/cifpp-config.cmake.in

@@ -8,8 +8,5 @@ include(CMakeFindDependencyMacro)
 
 find_dependency(Threads)
 find_dependency(ZLIB REQUIRED)
-if(@REQUIRE_FMT@)
-    find_dependency(fmt REQUIRED)
-endif()
 
 check_required_components(cifpp)

cmake/test-charconv.cpp

@@ -0,0 +1,17 @@
+#include <charconv>
+#include <cassert>
+#include <cstring>
+
+int main()
+{
+	float v;
+	char s[] = "1.0";
+
+	auto r = std::from_chars(s, s + strlen(s), v);
+
+	assert(r.ec == std::errc{});
+	assert(r.ptr = s + strlen(s));
+	assert(v == 1.0f);
+
+	return 0;
+}

include/cif++/category.hpp

@@ -26,15 +26,12 @@
 
 #pragma once
 
-#include "cif++/forward_decl.hpp"
-
 #include "cif++/condition.hpp"
+#include "cif++/forward_decl.hpp"
 #include "cif++/iterator.hpp"
 #include "cif++/row.hpp"
 #include "cif++/text.hpp"
 
-#include <array>
-
 /** \file category.hpp
  * Documentation for the cif::category class
  *
@@ -181,6 +178,27 @@ class category
 
 	const std::string &name() const { return m_name; } ///< Returns the name of the category
 
+	/// \brief Rename category to @a new_name
+	void name(std::string_view new_name)
+	{
+		m_name = new_name;
+		m_dirty = true;
+	}
+
+	/// \brief Return true if the category has been modified since last open/save
+	constexpr bool is_dirty() const
+	{
+		return m_dirty;
+	}
+
+	/// \brief Mark the category as modified according to @a dirty
+	void set_dirty(bool dirty)
+	{
+		m_dirty = dirty;
+	}
+
+	// --------------------------------------------------------------------
+	
 	[[deprecated("use key_items instead")]] iset key_fields() const; ///< Returns the cif::iset of key item names. Retrieved from the @ref category_validator for this category
 
 	iset key_items() const; ///< Returns the cif::iset of key item names. Retrieved from the @ref category_validator for this category
@@ -1055,60 +1073,6 @@ class category
 			{ return value; });
 	}
 
-	// --------------------------------------------------------------------
-	// Naming used to be very inconsistent. For backward compatibility,
-	// the old function names are here as deprecated variants.
-
-	/// \brief Return the index number for \a column_name
-	[[deprecated("Use get_item_ix instead")]] uint16_t get_column_ix(std::string_view column_name) const
-	{
-		return get_item_ix(column_name);
-	}
-
-	/// @brief Return the name for column with index @a ix
-	/// @param ix The index number
-	/// @return The name of the column
-	[[deprecated("use get_item_name instead")]] std::string_view get_column_name(uint16_t ix) const
-	{
-		return get_item_name(ix);
-	}
-
-	/// @brief Make sure a item with name @a item_name is known and return its index number
-	/// @param item_name The name of the item
-	/// @return The index number of the item
-	[[deprecated("use add_item instead")]] uint16_t add_column(std::string_view item_name)
-	{
-		return add_item(item_name);
-	}
-
-	/** @brief Remove column name @a colum_name
-	 * @param column_name The column to be removed
-	 */
-	[[deprecated("use remove_item instead")]] void remove_column(std::string_view column_name)
-	{
-		remove_item(column_name);
-	}
-
-	/** @brief Rename column @a from_name to @a to_name */
-	[[deprecated("use rename_item instead")]] void rename_column(std::string_view from_name, std::string_view to_name)
-	{
-		rename_item(from_name, to_name);
-	}
-
-	/// @brief Return whether a column with name @a name exists in this category
-	/// @param name The name of the column
-	/// @return True if the column exists
-	[[deprecated("use has_item instead")]] bool has_column(std::string_view name) const
-	{
-		return has_item(name);
-	}
-
-	/// @brief Return the cif::iset of columns in this category
-	[[deprecated("use get_items instead")]] iset get_columns() const
-	{
-		return get_items();
-	}
-
 	// --------------------------------------------------------------------
 	/// \brief Return the index number for \a item_name
 
@@ -1135,6 +1099,9 @@ class category
 	 */
 	void remove_item(std::string_view item_name);
 
+	/// \brief Drop items in this category that contain empty values in all rows.
+	void drop_empty_items();
+
 	/** @brief Rename item @a from_name to @a to_name */
 	void rename_item(std::string_view from_name, std::string_view to_name);
 
@@ -1146,8 +1113,14 @@ class category
 		return get_item_ix(name) < m_items.size();
 	}
 
-	/// @brief Return the cif::iset of items in this category
-	iset get_items() const;
+	/// @brief Return the items in this category
+	std::vector<std::string> get_items() const;
+
+	/// @brief Return the number of items (colums) in this category
+	size_t get_item_count() const noexcept
+	{
+		return m_items.size();
+	}
 
 	// --------------------------------------------------------------------
 
@@ -1177,16 +1150,48 @@ class category
 	/// Write the contents of the category to the std::ostream @a os
 	void write(std::ostream &os) const;
 
+	/// \brief Various supported output formats
+	enum class output_format
+	{
+		cif,	  // Output in mmCIF format
+		csv,      // comma separated values
+		tsv,      // tab separated values
+		list,     // values delimited by a '|' character
+		column,   // output in columns
+		markdown, //
+		table,	  // ascii art table
+		box,      // table with unicode line characters
+	};
+
+	/// @brief
+	/// @brief Write the contents of the category to the std::ostream @a os and
+	/// use @a order as the order of the items. If @a addMissingItems is
+	/// false, items that do not contain any value will be suppressed. Use this version
+	/// to write out
+	/// @param os The std::ostream to write to
+	/// @param fmt The format to use
+	/// @param order The order in which the items should appear
+	/// @param addMissingItems When false, empty items are suppressed from the output
+	void write(std::ostream &os, output_format fmt,
+		const std::vector<std::string> &order, bool addMissingItems = true);
+
 	/// @brief Write the contents of the category to the std::ostream @a os and
 	/// use @a order as the order of the items. If @a addMissingItems is
 	/// false, items that do not contain any value will be suppressed
 	/// @param os The std::ostream to write to
 	/// @param order The order in which the items should appear
 	/// @param addMissingItems When false, empty items are suppressed from the output
-	void write(std::ostream &os, const std::vector<std::string> &order, bool addMissingItems = true);
+	void write(std::ostream &os, const std::vector<std::string> &order, bool addMissingItems = true)
+	{
+		write(os, output_format::cif, order, addMissingItems);
+	}
 
   private:
-	void write(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const;
+	void write_cif(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const;
+	void write_delimited(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems,
+		std::string_view delimiter, bool aligned, bool header) const;
+	void write_markdown(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const;
+	void write_table(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems, bool ascii) const;
 
   public:
 	/// friend function to make it possible to do:
@@ -1286,6 +1291,8 @@ class category
 	uint32_t m_last_unique_num = 0;
 	class category_index *m_index = nullptr;
 	row *m_head = nullptr, *m_tail = nullptr;
+
+	bool m_dirty = false;	// Keep track of modifications
 };
 
 } // namespace cif

include/cif++/compound.hpp

@@ -179,7 +179,7 @@ class compound
 	friend class compound_factory_impl;
 	friend class local_compound_factory_impl;
 
-	compound(cif::datablock &db);
+	compound(datablock &db);
 
 	std::string m_id;
 	std::string m_name;
@@ -270,11 +270,15 @@ class compound_factory
 		return is_std_base(res_name) or is_std_peptide(res_name);
 	}
 
+	/// Return whether @a res_name is water
 	bool is_water(std::string_view res_name) const
 	{
 		return res_name == "HOH" or res_name == "H2O" or res_name == "WAT";
 	}
 
+	/// Return whether @a res_name already exists, without creating it.
+	bool exists(std::string_view res_name) const;
+
 	/// \brief Create the compound object for \a id
 	///
 	/// This will create the compound instance for \a id if it doesn't exist already.
@@ -331,14 +335,14 @@ class compound_factory
 class compound_source
 {
   public:
-	compound_source(const cif::file &file)
+	compound_source(const file &file)
 	{
-		cif::compound_factory::instance().push_dictionary(file);
+		compound_factory::instance().push_dictionary(file);
 	}
 
 	~compound_source()
 	{
-		cif::compound_factory::instance().pop_dictionary();
+		compound_factory::instance().pop_dictionary();
 	}
 };
 

include/cif++/condition.hpp

@@ -26,6 +26,7 @@
 
 #pragma once
 
+#include "cif++/text.hpp"
 #include "cif++/row.hpp"
 #include "cif++/format.hpp"
 
@@ -1076,6 +1077,16 @@ condition operator==(const key &key, const T &v)
 	return condition(new detail::key_equals_number_condition_impl(key.m_item_name, static_cast<double>(v)));
 }
 
+/**
+ * @brief Operator to create an not-equals condition based on a key @a key and a numeric value @a v
+ */
+template <Numeric T>
+condition operator!=(const key &key, const T &v)
+{
+	// TODO: change key_equals_etc... to use std::variant<double,int64_t> or something
+	return condition(new detail::not_condition_impl(key == v));
+}
+
 /**
  * @brief Operator to create an equals condition based on a key @a key and a value @a value
  */

include/cif++/cql.hpp

@@ -0,0 +1,444 @@
+/*-
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright (c) 2025  NKI/AVL, Netherlands Cancer Institute
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ *    list of conditions and the following disclaimer
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include "cif++/category.hpp"
+#include "cif++/datablock.hpp"
+#include "cif++/item.hpp"
+#include "cif++/iterator.hpp"
+#include "cif++/row.hpp"
+#include "cif++/validate.hpp"
+
+#include <iterator>
+#include <memory>
+#include <stdexcept>
+#include <string>
+
+// --------------------------------------------------------------------
+
+namespace cif::cql
+{
+
+class result;
+class row;
+class transaction;
+class connection;
+
+struct result_impl;
+
+// --------------------------------------------------------------------
+
+class field_ref final
+{
+  public:
+	std::string_view name() const &
+	{
+		return m_row.get_category().get_item_name(m_index);
+	}
+
+	constexpr size_t num() const noexcept
+	{
+		return m_index;
+	}
+
+	std::string_view text() const &
+	{
+		return m_row[m_index].text();
+	}
+
+	/** Return the contents of this item as type @tparam T */
+	template <typename T = std::string>
+	auto as() const -> T
+	{
+		return m_row[m_index].as<T>();
+	}
+
+	/** Return the contents of this item as type @tparam T or, if not
+	 * set, use @a dv as the default value.
+	 */
+	template <typename T>
+	auto value_or(const T &dv) const
+	{
+		return m_row[m_index].value_or(dv);
+	}
+
+	field_ref(row_handle rh, int col, std::shared_ptr<result_impl> result_impl)
+		: m_row(rh)
+		, m_index(col)
+		, m_result_impl(result_impl)
+	{
+	}
+
+	field_ref(const field_ref &) = default;
+	field_ref(field_ref &&) = default;
+
+	field_ref &operator=(const field_ref &) = default;
+	field_ref &operator=(field_ref &&) = default;
+
+  private:
+	row_handle m_row;
+	int m_index;
+
+	std::shared_ptr<result_impl> m_result_impl;
+};
+
+// --------------------------------------------------------------------
+
+class row_ref final
+{
+  public:
+	class const_field_iterator
+	{
+	  public:
+		friend class result;
+
+		using iterator_category = std::forward_iterator_tag;
+		using value_type = const field_ref;
+		using difference_type = std::ptrdiff_t;
+		using pointer = value_type *;
+		using reference = value_type &;
+
+		const_field_iterator(const const_field_iterator &) = default;
+		const_field_iterator(const_field_iterator &&) = default;
+
+		const_field_iterator &operator=(const const_field_iterator &) = default;
+		const_field_iterator &operator=(const_field_iterator &&) = default;
+
+		reference operator*()
+		{
+			return m_current;
+		}
+
+		pointer operator->()
+		{
+			return &m_current;
+		}
+
+		const_field_iterator &operator++()
+		{
+			if (m_row)
+			{
+				++m_col;
+				m_current = field_ref(m_row, m_col, m_result_impl);
+			}
+
+			return *this;
+		}
+
+		const_field_iterator operator++(int)
+		{
+			const_field_iterator result(*this);
+			this->operator++();
+			return result;
+		}
+
+		bool operator==(const const_field_iterator &rhs) const
+		{
+			return m_row == rhs.m_row and m_col == rhs.m_col;
+		}
+
+		bool operator!=(const const_field_iterator &rhs) const
+		{
+			return m_row != rhs.m_row or m_col != rhs.m_col;
+		}
+
+	  private:
+		friend class row_ref;
+
+		const_field_iterator(row_handle row, int column, std::shared_ptr<result_impl> result_impl)
+			: m_row(row)
+			, m_col(column)
+			, m_current(m_row, m_col, result_impl)
+			, m_result_impl(result_impl)
+		{
+		}
+
+		row_handle m_row;
+		int m_col;
+		field_ref m_current;
+
+		std::shared_ptr<result_impl> m_result_impl;
+	};
+
+	// --------------------------------------------------------------------
+
+	row_ref() = default;
+
+	row_ref(row_handle rh, std::shared_ptr<result_impl> result_impl)
+		: m_row(rh)
+		, m_result_impl(result_impl)
+	{
+	}
+
+	row_ref(const row_ref &) = default;
+	row_ref &operator=(const row_ref &) = default;
+
+	// --------------------------------------------------------------------
+
+	const_field_iterator cbegin() const noexcept { return const_field_iterator(m_row, 0, m_result_impl); }
+	const_field_iterator begin() const noexcept { return const_field_iterator(m_row, 0, m_result_impl); }
+	const_field_iterator cend() const noexcept { return const_field_iterator(m_row, size(), m_result_impl); }
+	const_field_iterator end() const noexcept { return const_field_iterator(m_row, size(), m_result_impl); }
+
+	field_ref front() const noexcept { return field_ref(m_row, 0, m_result_impl); }
+	field_ref back() const noexcept { return field_ref(m_row, size() - 1, m_result_impl); }
+
+	size_t size() const noexcept;
+	bool empty() const noexcept { return size() == 0; }
+
+	field_ref operator[](size_t index) const noexcept { return field_ref(m_row, index, m_result_impl); }
+	field_ref operator[](std::string_view name) const;
+
+	// --------------------------------------------------------------------
+
+	bool operator==(const row_ref &rhs) const { return m_row == rhs.m_row; }
+	bool operator!=(const row_ref &rhs) const { return m_row != rhs.m_row; }
+
+  private:
+	row_handle m_row;
+	std::shared_ptr<result_impl> m_result_impl;
+};
+
+// --------------------------------------------------------------------
+
+class result
+{
+  public:
+	// --------------------------------------------------------------------
+
+	class iterator
+	{
+	  public:
+		friend class view;
+
+		using iterator_category = std::forward_iterator_tag;
+		using value_type = const row_ref;
+		using difference_type = std::ptrdiff_t;
+		using pointer = value_type *;
+		using reference = value_type &;
+
+		// const_row_iterator() = default;
+
+		iterator(std::shared_ptr<result_impl> result_impl, category::iterator cat_iter)
+			: m_iter(cat_iter)
+			, m_current(*m_iter, result_impl)
+			, m_result_impl(result_impl)
+		{
+		}
+
+		iterator(const iterator &) = default;
+		iterator(iterator &&) = default;
+
+		// const_row_iterator &operator=(const const_row_iterator &) = default;
+		// const_row_iterator &operator=(const_row_iterator &&) = default;
+
+		reference operator*()
+		{
+			return m_current;
+		}
+
+		pointer operator->()
+		{
+			return &m_current;
+		}
+
+		iterator &operator++()
+		{
+			++m_iter;
+			m_current = { *m_iter, m_result_impl };
+			return *this;
+		}
+
+		iterator operator++(int)
+		{
+			iterator result(*this);
+			this->operator++();
+			return result;
+		}
+
+		bool operator==(const iterator &rhs) const
+		{
+			return m_result_impl == rhs.m_result_impl and m_iter == rhs.m_iter;
+		}
+
+		bool operator!=(const iterator &rhs) const
+		{
+			return m_result_impl != rhs.m_result_impl or m_iter != rhs.m_iter;
+		}
+
+	  private:
+		category::iterator m_iter;
+		row_ref m_current;
+		std::shared_ptr<result_impl> m_result_impl;
+	};
+
+	// --------------------------------------------------------------------
+
+	result() = delete;
+	result(result const &rhs) noexcept = default;
+	result(result &&rhs) noexcept = default;
+	result &operator=(result const &rhs) noexcept = default;
+	result &operator=(result &&rhs) noexcept = default;
+
+	result(category &&data, const std::string &query = "");
+
+	~result() = default;
+
+	row_ref one_row() const
+	{
+		if (size() != 1)
+			throw std::runtime_error("Expected one row");
+		return front();
+	}
+
+	field_ref one_field() const
+	{
+		expect_columns(1);
+
+		if (size() != 1)
+			throw std::runtime_error("Expected one row");
+
+		return one_row().front();
+	}
+
+	// --------------------------------------------------------------------
+
+	iterator begin() const noexcept;
+	iterator cbegin() const noexcept;
+
+	iterator end() const noexcept;
+	iterator cend() const noexcept;
+
+	row_ref front() const;
+	row_ref back() const;
+
+	size_t size() const noexcept;
+	bool empty() const noexcept { return size() == 0; }
+
+	size_t column_count() const;
+
+	category &get_category() const;
+
+	result expect_columns(size_t cols) const
+	{
+		if (auto actual = column_count(); cols != actual)
+			throw std::runtime_error("Unexpected number of columns");
+		return *this;
+	}
+
+	// --------------------------------------------------------------------
+	
+	friend std::ostream &operator<<(std::ostream &os, const result &r)
+	{
+		os << r.get_category();
+		return os;
+	}
+
+  private:
+	friend class transaction;
+	friend class SelectStatement;
+
+	std::shared_ptr<result_impl> m_impl;
+};
+
+// --------------------------------------------------------------------
+
+template <typename... Ts>
+class cql_iterator_proxy : public cif::iterator_proxy<category, Ts...>
+{
+  public:
+	cql_iterator_proxy(result &&res)
+		: cif::iterator_proxy<category, Ts...>(res.get_category())
+		, m_result(std::forward<result>(res))
+	{
+		m_result.expect_columns(cif::iterator_proxy<category, Ts...>::N);
+	}
+
+  private:
+	result m_result;
+};
+
+// --------------------------------------------------------------------
+
+class transaction final
+{
+  public:
+	transaction(connection &conn);
+	~transaction();
+
+	transaction(const transaction &) = delete;
+	transaction &operator=(const transaction &) = delete;
+
+	/// \brief Execute the sql in @a query returning an iterable result
+	result exec(std::string query);
+
+	/// \brief Execute the sql in @a query returning an iterable result.
+	/// Updates @a tail with what remains after the first statement in @a query
+	result exec(std::string query, std::string &tail);
+
+	template<typename... Ts>
+	cql_iterator_proxy<Ts...> stream(const std::string &sql)
+	{
+		return cql_iterator_proxy<Ts...>{ exec(sql) };
+	}
+
+	void commit();
+	void rollback();
+
+  private:
+	connection &m_conn;
+	bool m_transaction_active = false;
+};
+
+// --------------------------------------------------------------------
+
+class connection final
+{
+  public:
+	connection(datablock &db);
+	~connection();
+
+	friend class transaction;
+
+	/// \brief Return true if the string @a sql contains a complete statement.
+	bool is_complete_statement(const std::string &sql) const;
+
+	/// \brief Execute the sql in @a query returning an iterable result
+	result exec(std::string query);
+
+	/// \brief Execute the sql in @a query returning an iterable result.
+	/// Updates @a tail with what remains after the first statement in @a query
+	result exec(std::string query, std::string &tail);
+
+	/// \brief Return true if the underlying data was modified by any query.
+	bool is_modified() const;
+
+  private:
+	struct connection_impl *m_impl;
+};
+
+} // namespace cif::cql

include/cif++/datablock.hpp

@@ -43,7 +43,7 @@ namespace cif
 
 /**
  * @brief A datablock is a list of category objects with some additional features
- * 
+ *
  */
 
 class datablock : public std::list<category>
@@ -53,7 +53,7 @@ class datablock : public std::list<category>
 
 	/**
 	 * @brief Construct a new datablock object with name @a name
-	 * 
+	 *
 	 * @param name The name for the new datablock
 	 */
 	datablock(std::string_view name)
@@ -80,7 +80,7 @@ class datablock : public std::list<category>
 	{
 		std::swap(a.m_name, b.m_name);
 		std::swap(a.m_validator, b.m_validator);
-		std::swap(static_cast<std::list<category>&>(a), static_cast<std::list<category>&>(b));
+		std::swap(static_cast<std::list<category> &>(a), static_cast<std::list<category> &>(b));
 	}
 
 	// --------------------------------------------------------------------
@@ -92,7 +92,7 @@ class datablock : public std::list<category>
 
 	/**
 	 * @brief Set the name of this datablock to @a name
-	 * 
+	 *
 	 * @param name The new name
 	 */
 	void set_name(std::string_view name)
@@ -102,27 +102,33 @@ class datablock : public std::list<category>
 
 	/**
 	 * @brief Attempt to load the dictionary specified in audit_conform category
-	 * 
+	 *
 	 */
 	void load_dictionary();
 
+	/**
+	 * @brief Attempt to load the dictionary @a dict
+	 *
+	 */
+	void load_dictionary(std::string_view dict);
+
 	/**
 	 * @brief Set the validator object to @a v
-	 * 
+	 *
 	 * @param v The new validator object, may be null
 	 */
 	void set_validator(const validator *v);
 
 	/**
 	 * @brief Get the validator object
-	 * 
+	 *
 	 * @return const validator* The validator or nullptr if there is none
 	 */
 	const validator *get_validator() const;
 
 	/**
 	 * @brief Validates the content of this datablock and all its content
-	 * 
+	 *
 	 * @return true If the content is valid
 	 * @return false If the content is not valid
 	 */
@@ -131,7 +137,7 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Validates all contained data for valid links between parents and children
 	 * as defined in the validator
-	 * 
+	 *
 	 * @return true If all links are valid
 	 * @return false If all links are not valid
 	 */
@@ -140,7 +146,7 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Strip removes all categories and items that are invalid according
 	 * to the assigned validator. Will also add a valid audit_conform block.
-	 * 
+	 *
 	 * @return true if the remaining datablock is valid
 	 */
 	bool strip();
@@ -150,7 +156,7 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Return the category named @a name, will create a new and empty
 	 * category named @a name if it does not exist.
-	 * 
+	 *
 	 * @param name The name of the category to return
 	 * @return category& Reference to the named category
 	 */
@@ -159,7 +165,7 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Return the const category named @a name, will return a reference
 	 * to a static empty category if it was not found.
-	 * 
+	 *
 	 * @param name The name of the category to return
 	 * @return category& Reference to the named category
 	 */
@@ -168,7 +174,7 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Return a pointer to the category named @a name or nullptr if
 	 * it does not exist.
-	 * 
+	 *
 	 * @param name The name of the category
 	 * @return category* Pointer to the category found or nullptr
 	 */
@@ -177,13 +183,12 @@ class datablock : public std::list<category>
 	/**
 	 * @brief Return a pointer to the category named @a name or nullptr if
 	 * it does not exist.
-	 * 
+	 *
 	 * @param name The name of the category
 	 * @return category* Pointer to the category found or nullptr
 	 */
 	const category *get(std::string_view name) const;
 
-
 	/**
 	 * @brief Return true if this datablock contains a non-empty category
 	 */
@@ -197,7 +202,7 @@ class datablock : public std::list<category>
 	 * new one if it is not found. The result is a tuple of an iterator
 	 * pointing to the category and a boolean indicating whether the category
 	 * was created or not.
-	 * 
+	 *
 	 * @param name The name for the category
 	 * @return std::tuple<iterator, bool> A tuple containing an iterator pointing
 	 * at the category and a boolean indicating whether the category was newly

include/cif++/item.hpp

@@ -37,7 +37,6 @@
 #include <iomanip>
 #include <iostream>
 #include <limits>
-#include <memory>
 #include <optional>
 #include <utility>
 
@@ -53,12 +52,12 @@ namespace cif
 // --------------------------------------------------------------------
 /** @brief item is a transient class that is used to pass data into rows
  * but it also takes care of formatting data.
- * 
- * 
- * 
+ *
+ *
+ *
  * The class cif::item is often used implicitly when creating a row in a category
  * using the emplace function.
- * 
+ *
  * @code{.cpp}
  * cif::category cat("my-cat");
  * cat.emplace({
@@ -68,12 +67,12 @@ namespace cif
  *   { "item-4", std::make_optional<int>(42) },   // <- stores an item with value 42
  *   { "item-5" }                                 // <- stores an item with value .
  * });
- * 
+ *
  * std::cout << cat << '\n';
  * @endcode
- * 
+ *
  * Will result in:
- * 
+ *
  * @code{.txt}
  * _my-cat.item-1 1
  * _my-cat.item-2 1.00
@@ -176,7 +175,7 @@ class item
 
 	/// \brief constructor for an item with name \a name and as
 	/// content value \a value
-	template<typename T, std::enable_if_t<std::is_same_v<T, std::string>, int> = 0>
+	template <typename T, std::enable_if_t<std::is_same_v<T, std::string>, int> = 0>
 	item(const std::string_view name, T &&value)
 		: m_name(name)
 		, m_value(std::move(value))
@@ -221,15 +220,15 @@ class item
 	item &operator=(item &&rhs) noexcept = default;
 	/** @endcond */
 
-	std::string_view name() const { return m_name; }   ///< Return the name of the item
-	std::string_view value() const & { return m_value; } ///< Return the value of the item
+	std::string_view name() const { return m_name; }            ///< Return the name of the item
+	std::string_view value() const & { return m_value; }        ///< Return the value of the item
 	std::string value() const && { return std::move(m_value); } ///< Return the value of the item
 
 	/// \brief replace the content of the stored value with \a v
 	void value(std::string_view v) { m_value = v; }
 
 	/// \brief empty means either null or unknown
-	bool empty() const { return m_value.empty(); }
+	bool empty() const { return is_null() or is_unknown() or m_value.empty(); }
 
 	/// \brief returns true if the item contains '.'
 	bool is_null() const { return m_value == "."; }
@@ -260,8 +259,8 @@ class item
 // --------------------------------------------------------------------
 /// \brief the internal storage for items in a category
 ///
-/// Internal storage, strictly forward linked list with minimal space
-/// requirements. Strings of size 7 or shorter are stored internally.
+/// Internal storage, with minimal space requirements. Strings of
+/// size 7 or shorter are stored internally.
 /// Typically, more than 99% of the strings in an mmCIF file are less
 /// than 8 bytes in length.
 
@@ -338,7 +337,8 @@ struct item_value
 	/** Return the content of the item as a std::string_view */
 	constexpr inline std::string_view text() const
 	{
-		return { m_length >= kBufferSize ? m_data : m_local_data, m_length };
+		const char *ptr = m_length >= kBufferSize ? m_data : m_local_data;
+		return (m_length == 1 and *ptr == '?') ? std::string_view{} : std::string_view{ ptr, m_length };
 	}
 };
 
@@ -560,7 +560,9 @@ struct item_handle::item_value_as<T, std::enable_if_t<std::is_arithmetic_v<T> an
 			auto b = txt.data();
 			auto e = txt.data() + txt.size();
 
-			std::from_chars_result r = (b + 1 < e and *b == '+' and std::isdigit(b[1])) ? selected_charconv<value_type>::from_chars(b + 1, e, result) : selected_charconv<value_type>::from_chars(b, e, result);
+			std::from_chars_result r = (b + 1 < e and *b == '+' and std::isdigit(b[1])) //
+			                               ? from_chars(b + 1, e, result)
+			                               : from_chars(b, e, result);
 
 			if ((bool)r.ec or r.ptr != e)
 			{
@@ -595,7 +597,9 @@ struct item_handle::item_value_as<T, std::enable_if_t<std::is_arithmetic_v<T> an
 			auto b = txt.data();
 			auto e = txt.data() + txt.size();
 
-			std::from_chars_result r = (b + 1 < e and *b == '+' and std::isdigit(b[1])) ? selected_charconv<value_type>::from_chars(b + 1, e, v) : selected_charconv<value_type>::from_chars(b, e, v);
+			std::from_chars_result r = (b + 1 < e and *b == '+' and std::isdigit(b[1]))
+			                               ? from_chars(b + 1, e, v)
+			                               : from_chars(b, e, v);
 
 			if ((bool)r.ec or r.ptr != e)
 			{

include/cif++/iterator.hpp

@@ -26,9 +26,12 @@
 
 #pragma once
 
+#include "cif++/condition.hpp"
 #include "cif++/row.hpp"
 
 #include <array>
+#include <cstdint>
+#include <numeric>
 
 /**
  * @file iterator.hpp
@@ -262,6 +265,11 @@ class iterator_impl<Category>
 		return m_current;
 	}
 
+	int64_t row_id() const
+	{
+		return reinterpret_cast<int64_t>(m_current.m_row);
+	}
+
 	iterator_impl &operator++()
 	{
 		if (m_current)
@@ -489,6 +497,9 @@ class iterator_proxy
 		std::swap(m_item_ix, rhs.m_item_ix);
 	}
 
+  protected:
+	iterator_proxy(category_type &cat);
+
   private:
 	category_type *m_category;
 	row_iterator m_begin, m_end;
@@ -530,6 +541,7 @@ class conditional_iterator_proxy
 		using pointer = value_type *;
 		using reference = value_type;
 
+		conditional_iterator_impl() = default;
 		conditional_iterator_impl(CategoryType &cat, row_iterator pos, const condition &cond, const std::array<uint16_t, N> &cix);
 		conditional_iterator_impl(const conditional_iterator_impl &i) = default;
 		conditional_iterator_impl &operator=(const conditional_iterator_impl &i) = default;
@@ -649,6 +661,15 @@ iterator_proxy<Category, Ts...>::iterator_proxy(Category &cat, row_iterator pos,
 		m_item_ix[i++] = m_category->get_item_ix(item);
 }
 
+template <typename Category, typename... Ts>
+iterator_proxy<Category, Ts...>::iterator_proxy(Category &cat)
+	: m_category(&cat)
+	, m_begin(cat.begin())
+	, m_end(cat.end())
+{
+	std::iota(m_item_ix.begin(), m_item_ix.end(), 0);
+}
+
 // --------------------------------------------------------------------
 
 template <typename Category, typename... Ts>
@@ -661,6 +682,8 @@ conditional_iterator_proxy<Category, Ts...>::conditional_iterator_impl::conditio
 {
 	if (m_condition == nullptr or m_condition->empty())
 		m_begin = m_end;
+	else
+		m_current = *m_begin;
 }
 
 template <typename Category, typename... Ts>

include/cif++/matrix.hpp

@@ -124,6 +124,23 @@ class matrix_expression
 
 		return os;
 	}
+
+	template <typename M2>
+	constexpr bool operator==(const matrix_expression<M2> &m) const
+	{
+		bool same = false;
+		if (dim_m() == m.dim_m() and dim_n() == m.dim_n())
+		{
+			same = true;
+			for (std::size_t i = 0; same and i < m.dim_m(); ++i)
+			{
+				for (std::size_t j = 0; same and j < m.dim_n(); ++j)
+					same = operator()(i, j) == m(i, j);
+			}
+		}
+
+		return same;
+	}
 };
 
 // --------------------------------------------------------------------
@@ -594,6 +611,35 @@ auto operator*(const matrix_expression<M1> &m1, const matrix_expression<M2> &m2)
 
 // --------------------------------------------------------------------
 
+template <typename M2>
+class sub_matrix : public matrix_expression<sub_matrix<M2>>
+{
+  public:
+	sub_matrix(const M2 &m, int i, int j)
+		: m_m(m)
+		, m_i(i)
+		, m_j(j)
+	{
+	}
+
+	constexpr std::size_t dim_m() const { return m_m.dim_m() - 1; } ///< Return dimension m
+	constexpr std::size_t dim_n() const { return m_m.dim_n() - 1; } ///< Return dimension n
+
+	/** Access to the value of element [ @a i, @a j ] */
+	constexpr auto operator()(std::size_t i, std::size_t j) const
+	{
+		return m_m(
+			i >= m_i ? i + 1 : i,
+			j >= m_j ? j + 1 : j);
+	}
+
+  private:
+	const M2 &m_m;
+	std::size_t m_i, m_j;
+};
+
+// --------------------------------------------------------------------
+
 /** Generic routine to calculate the determinant of a matrix
  * 
  * @note This is currently only implemented for fixed matrices of size 3x3
@@ -605,11 +651,23 @@ auto determinant(const M &m);
 template <typename F = float>
 auto determinant(const matrix3x3<F> &m)
 {
-	return (m(0, 0) * (m(1, 1) * m(2, 2) - m(1, 2) * m(2, 1)) +
-			m(0, 1) * (m(1, 2) * m(2, 0) - m(1, 0) * m(2, 2)) +
-			m(0, 2) * (m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0)));
+	return (m(0, 0) * ((m(1, 1) * m(2, 2) - m(1, 2) * m(2, 1))) +
+			m(0, 1) * ((m(1, 2) * m(2, 0) - m(1, 0) * m(2, 2))) +
+			m(0, 2) * ((m(1, 0) * m(2, 1) - m(1, 1) * m(2, 0))));
 }
 
+/** Implementation of the determinant function for fixed size matrices of size 4x4 */
+template <typename F = float>
+F determinant(const matrix4x4<F> &m)
+{
+	return m(0, 0) * determinant(matrix3x3<F>(sub_matrix<decltype(m)>(m, 0, 0))) -
+	       m(0, 1) * determinant(matrix3x3<F>(sub_matrix<decltype(m)>(m, 0, 1))) +
+	       m(0, 2) * determinant(matrix3x3<F>(sub_matrix<decltype(m)>(m, 0, 2))) -
+	       m(0, 3) * determinant(matrix3x3<F>(sub_matrix<decltype(m)>(m, 0, 3)));
+}
+
+// --------------------------------------------------------------------
+
 /** Generic routine to calculate the inverse of a matrix
  * 
  * @note This is currently only implemented for fixed matrices of size 3x3

include/cif++/model.hpp

@@ -29,6 +29,7 @@
 #include "cif++/atom_type.hpp"
 #include "cif++/datablock.hpp"
 #include "cif++/point.hpp"
+#include "cif++/row.hpp"
 
 #include <memory>
 #include <numeric>
@@ -134,14 +135,20 @@ class atom
 
 		row_handle row_aniso()
 		{
+			row_handle result{};
 			auto cat = m_db.get("atom_site_anisotrop");
-			return cat ? cat->operator[]({ { "id", m_id } }) : row_handle{};
+			if (cat)
+				result = cat->operator[]({ { "id", m_id } });
+			return result;
 		}
 
 		const row_handle row_aniso() const
 		{
+			row_handle result{};
 			auto cat = m_db.get("atom_site_anisotrop");
-			return cat ? cat->operator[]({ { "id", m_id } }) : row_handle{};
+			if (cat)
+				result = cat->operator[]({ { "id", m_id } });
+			return result;
 		}
 
 		const datablock &m_db;
@@ -1059,12 +1066,30 @@ class structure
 	/// \return				The newly create asym ID
 	std::string create_non_poly(const std::string &entity_id, std::vector<row_initializer> atoms);
 
+	/// \brief Create a new NonPolymer struct_asym for a compound of type \a compound_id, returns asym_id.
+	/// This method creates new atom records filled with info from the CCD compound info.
+	///
+	/// \param compound_id	 The compound ID of the new nonpoly
+	/// \param skip_hydrogen Do not create hydrogen atoms when true
+	/// \return				 The newly create asym ID
+	std::string create_non_poly(const std::string &compound_id, bool skip_hydrogen);
+
 	/// \brief Create a new water with atom constructed from info in \a atom_info
 	/// This method creates a new atom record filled with info from the info.
 	///
 	/// \param atom			The set of item data containing the data for the atoms.
 	void create_water(row_initializer atom);
 
+	/// \brief Create a link, a struct_conn record for two atoms.
+	///
+	/// \param a1			Atom 1
+	/// \param a2			Atom 2
+	/// \param link_type	The struct_conn_type ID for the link
+	/// \param role			The pdbx_role field value
+	/// \return 			The ID of the struct_conn record created
+
+	std::string create_link(atom a1, atom a2, const std::string &link_type, const std::string &role);
+
 	/// \brief Create a new and empty (sugar) branch
 	branch &create_branch();
 

include/cif++/pdb.hpp

@@ -149,6 +149,23 @@ void fixup_pdbx(file &pdbx_file, const validator &v);
 
 bool reconstruct_pdbx(file &pdbx_file, const validator &v);
 
+/** \brief This is an extension to cif::validator, use the logic in common
+ * PDBx files to see if the file is internally consistent.
+ *
+ * This function for now checks if the following categories are consistent:
+ *
+ * atom_site -> pdbx_poly_seq_scheme -> entity_poly_seq -> entity_poly -> entity
+ *
+ * Use the common \ref cif::VERBOSE flag to turn on diagnostic messages.
+ * 
+ * This function throws a std::system_error in case of an error
+ *
+ * \param pdbx_file The input file
+ * \result Returns true if the file was valid and consistent
+ */
+
+bool is_valid_pdbx_file(const file &pdbx_file);
+
 /** \brief This is an extension to cif::validator, use the logic in common
  * PDBx files to see if the file is internally consistent.
  *
@@ -165,8 +182,7 @@ bool reconstruct_pdbx(file &pdbx_file, const validator &v);
  * \result Returns true if the file was valid and consistent
  */
 
-bool is_valid_pdbx_file(const file &pdbx_file,
-	const validator &v = validator_factory::instance().get("mmcif_pdbx.dic"));
+bool is_valid_pdbx_file(const file &pdbx_file, const validator &v);
 
 /** \brief This is an extension to cif::validator, use the logic in common
  * PDBx files to see if the file is internally consistent.

include/cif++/pdb/cif2pdb.hpp

@@ -1,33 +0,0 @@
-/*-
- * SPDX-License-Identifier: BSD-2-Clause
- *
- * Copyright (c) 2020 NKI/AVL, Netherlands Cancer Institute
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- * 1. Redistributions of source code must retain the above copyright notice, this
- *    list of conditions and the following disclaimer
- * 2. Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
- * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#pragma once
-
-/// \file cif2pdb.hpp
-/// \deprecated This file is no longer used. Please use "cif++/pdb.hpp" instead
-
-#warning "Use of this file is deprecated, please use "cif++/pdb.hpp"
-

include/cif++/pdb/io.hpp

@@ -1,32 +0,0 @@
-/*-
- * SPDX-License-Identifier: BSD-2-Clause
- * 
- * Copyright (c) 2022 NKI/AVL, Netherlands Cancer Institute
- * 
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- * 
- * 1. Redistributions of source code must retain the above copyright notice, this
- *    list of conditions and the following disclaimer
- * 2. Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- * 
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
- * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#pragma once
-
-/// \file io.hpp
-/// \deprecated This file is no longer used. Please use "cif++/pdb.hpp" instead
-
-#warning "Use of this file is deprecated, please use "cif++/pdb.hpp"

include/cif++/pdb/pdb2cif.hpp

@@ -1,32 +0,0 @@
-/*-
- * SPDX-License-Identifier: BSD-2-Clause
- *
- * Copyright (c) 2020 NKI/AVL, Netherlands Cancer Institute
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- * 1. Redistributions of source code must retain the above copyright notice, this
- *    list of conditions and the following disclaimer
- * 2. Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
- * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#pragma once
-
-/// \file pdb2cif.hpp
-/// \deprecated This file is no longer used. Please use "cif++/pdb.hpp" instead
-
-#warning "Use of this file is deprecated, please use "cif++/pdb.hpp"

include/cif++/pdb/tls.hpp

@@ -1,32 +0,0 @@
-/*-
- * SPDX-License-Identifier: BSD-2-Clause
- *
- * Copyright (c) 2020 NKI/AVL, Netherlands Cancer Institute
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions are met:
- *
- * 1. Redistributions of source code must retain the above copyright notice, this
- *    list of conditions and the following disclaimer
- * 2. Redistributions in binary form must reproduce the above copyright notice,
- *    this list of conditions and the following disclaimer in the documentation
- *    and/or other materials provided with the distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
- * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
- * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
- * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
- * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
- * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#pragma once
-
-/// \file tls.hpp
-/// \deprecated This code has been moved to libpdb-redo
-
-#warning "This code has been moved to libpdb-redo"

include/cif++/point.hpp

@@ -30,7 +30,9 @@
 #include <cmath>
 #include <complex>
 #include <cstdint>
+#include <format>
 #include <functional>
+#include <optional>
 #include <valarray>
 
 #if __has_include(<clipper/core/coords.h>)
@@ -365,11 +367,18 @@ class quaternion_type
 	}
 
 	/// \brief test for all zero values
-	constexpr operator bool() const
+	constexpr explicit operator bool() const
 	{
 		return a != 0 or b != 0 or c != 0 or d != 0;
 	}
 
+	/// \brief for debugging e.g.
+	friend std::ostream &operator<<(std::ostream &os, const quaternion_type &rhs)
+	{
+		os << std::format("{{ a: {}, b: {}, c: {}, d: {} }}", rhs.a, rhs.b, rhs.c, rhs.d);
+		return os;
+	}
+
   private:
 	value_type a, b, c, d;
 };
@@ -743,6 +752,55 @@ inline constexpr auto cross_product(const point_type<F1> &a, const point_type<F2
 		a.m_x * b.m_y - b.m_x * a.m_y);
 }
 
+/// \brief return the squared norm of point @a p
+template <typename F>
+constexpr F norm_squared(const point_type<F> &p)
+{
+	return p.m_x * p.m_x + p.m_y * p.m_y + p.m_z * p.m_z;
+}
+
+/// \brief return the norm of point @a p
+template <typename F>
+constexpr point_type<F> norm(const point_type<F> &p)
+{
+	return std::sqrt(norm_squared(p));
+}
+
+/// \brief return the point where two lines intersect, or an empty value if they don't intersect at all
+template <typename F>
+std::optional<cif::point> line_line_intersection(const point_type<F> &p1,
+	const point_type<F> &p2, const point_type<F> &p3, const point_type<F> &p4)
+{
+	auto p13 = p1 - p3;
+	auto p43 = p4 - p3;
+	if (std::abs(p43.m_x) < std::numeric_limits<F>::epsilon() and std::abs(p43.m_y) < std::numeric_limits<F>::epsilon() and std::abs(p43.m_z) < std::numeric_limits<F>::epsilon())
+		return {};
+
+	auto p21 = p2 - p1;
+	if (std::abs(p21.m_x) < std::numeric_limits<F>::epsilon() and std::abs(p21.m_y) < std::numeric_limits<F>::epsilon() and std::abs(p21.m_z) < std::numeric_limits<F>::epsilon())
+		return {};
+
+	auto d1343 = cif::dot_product(p43, p13);
+	auto d4321 = cif::dot_product(p43, p21);
+	auto d1321 = cif::dot_product(p13, p21);
+	auto d4343 = cif::dot_product(p43, p43);
+	auto d2121 = cif::dot_product(p21, p21);
+
+	auto denom = d2121 * d4343 - d4321 * d4321;
+	if (std::abs(denom) < std::numeric_limits<F>::epsilon())
+		return {};
+
+	auto numer = d1343 * d4321 - d1321 * d4343;
+
+	auto mua = numer / denom;
+	auto mub = (d1343 + d4321 * mua) / d4343;
+
+	auto pa = p1 + mua * p21;
+	auto pb = p3 + mub * p43;
+
+	return { (pa + pb) / 2 };
+}
+
 /// \brief return the angle in degrees between the vectors from point @a p2 to @a p1 and @a p2 to @a p3
 template <typename F>
 constexpr auto angle(const point_type<F> &p1, const point_type<F> &p2, const point_type<F> &p3)
@@ -806,6 +864,9 @@ constexpr auto distance_point_to_line(const point_type<F> &l1, const point_type<
 	return cross.length() / line.length();
 }
 
+/// \brief return the smallest sphere around the points in @a pts
+std::tuple<point, float> smallest_sphere_around_points(std::vector<point> pts);
+
 // --------------------------------------------------------------------
 /**
  * @brief For e.g. simulated annealing, returns a new point that is moved in

include/cif++/row.hpp

@@ -29,55 +29,61 @@
 #include "cif++/item.hpp"
 
 #include <array>
+#include <cstdint>
 
 /**
  * @file row.hpp
- * 
+ *
  * The class cif::row should be an opaque type. It is used to store the
  * internal data per row in a category. You should use cif::row_handle
  * to get access to the contents in a row.
- * 
+ *
  * One could think of rows as vectors of cif::item. But internally
  * that's not the case.
- * 
+ *
  * You can access the values of stored items by name or index.
  * The return value of operator[] is an cif::item_handle object.
- * 
+ *
  * @code {.cpp}
  * cif::category &atom_site = my_db["atom_site"];
  * cif::row_handle rh = atom_site.front();
- * 
+ *
  * // by name:
  * std::string name = rh["label_atom_id"].as<std::string>();
- * 
+ *
  * // by index:
  * uint16_t ix = atom_site.get_item_ix("label_atom_id");
  * assert(rh[ix].as<std::string() == name);
  * @endcode
- * 
+ *
  * There some template magic here to allow easy extracting of data
  * from rows. This can be done using cif::tie e.g.:
- * 
+ *
  * @code {.cpp}
  * std::string name;
  * float x, y, z;
- * 
+ *
  * cif::tie(name, x, y, z) = rh.get("label_atom_id", "cartn_x", "cartn_y", "cartn_z");
  * @endcode
- * 
+ *
  * However, a more modern way uses structured binding:
- * 
+ *
  * @code {.cpp}
  * const auto &[name, x, y, z] = rh.get<std::string,float,float,float>("label_atom_id", "cartn_x", "cartn_y", "cartn_z");
  * @endcode
- * 
- * 
- * 
+ *
+ *
+ *
  */
 
 namespace cif
 {
 
+namespace cql
+{
+	struct connection_impl;
+}
+
 namespace detail
 {
 
@@ -141,7 +147,7 @@ namespace detail
 
 } // namespace detail
 
-/// \brief similar to std::tie, assign values to each element in @a v from the 
+/// \brief similar to std::tie, assign values to each element in @a v from the
 /// result of a get on a row_handle.
 template <typename... Ts>
 auto tie(Ts &...v)
@@ -160,7 +166,7 @@ class row : public std::vector<item_value>
 	/**
 	 * @brief Return the item_value pointer for item at index @a ix
 	 */
-	item_value* get(uint16_t ix)
+	item_value *get(uint16_t ix)
 	{
 		return ix < size() ? &data()[ix] : nullptr;
 	}
@@ -168,7 +174,7 @@ class row : public std::vector<item_value>
 	/**
 	 * @brief Return the const item_value pointer for item at index @a ix
 	 */
-	const item_value* get(uint16_t ix) const
+	const item_value *get(uint16_t ix) const
 	{
 		return ix < size() ? &data()[ix] : nullptr;
 	}
@@ -184,7 +190,7 @@ class row : public std::vector<item_value>
 	{
 		if (ix >= size())
 			resize(ix + 1);
-		
+
 		at(ix) = std::move(iv);
 	}
 
@@ -208,7 +214,8 @@ class row_handle
 	friend class category;
 	friend class category_index;
 	friend class row_initializer;
-	template <typename, typename...> friend class iterator_impl;
+	template <typename, typename...>
+	friend class iterator_impl;
 
 	row_handle() = default;
 
@@ -233,6 +240,12 @@ class row_handle
 		return *m_category;
 	}
 
+	/// \brief return the row ID
+	int64_t row_id() const
+	{
+		return reinterpret_cast<int64_t>(m_row);
+	}
+
 	/// \brief Return true if the row is empty or uninitialised
 	bool empty() const
 	{
@@ -299,19 +312,19 @@ class row_handle
 	}
 
 	/// \brief assign each of the items named in @a values to their respective value
-	void assign(const std::vector<item> &values)
+	void assign(const std::vector<item> &values, bool updateLinked = true)
 	{
 		for (auto &value : values)
-			assign(value, true);
+			assign(value, updateLinked);
 	}
 
-	/** \brief assign the value @a value to the item named @a name 
-	 * 
+	/** \brief assign the value @a value to the item named @a name
+	 *
 	 * If updateLinked it true, linked records are updated as well.
 	 * That means that if item @a name is part of the link definition
 	 * and the link results in a linked record in another category
 	 * this record in the linked category is updated as well.
-	 * 
+	 *
 	 * If validate is true, which is default, the assigned value is
 	 * checked to see if it conforms to the rules defined in the dictionary
 	 */
@@ -322,12 +335,12 @@ class row_handle
 	}
 
 	/** \brief assign the value @a value to item at index @a item
-	 * 
+	 *
 	 * If updateLinked it true, linked records are updated as well.
 	 * That means that if item @a item is part of the link definition
 	 * and the link results in a linked record in another category
 	 * this record in the linked category is updated as well.
-	 * 
+	 *
 	 * If validate is true, which is default, the assigned value is
 	 * checked to see if it conforms to the rules defined in the dictionary
 	 */
@@ -346,6 +359,8 @@ class row_handle
 
 	uint16_t add_item(std::string_view name);
 
+	friend cql::connection_impl;
+
 	row *get_row()
 	{
 		return m_row;
@@ -371,7 +386,7 @@ class row_handle
 
 /**
  * @brief The class row_initializer is a list of cif::item's.
- * 
+ *
  * This class is used to construct new rows, it allows to
  * group a list of item name and value pairs and pass it
  * in one go to the constructing function.
@@ -406,7 +421,6 @@ class row_initializer : public std::vector<item>
 	/// \brief constructor taking the values of an existing row
 	row_initializer(row_handle rh);
 
-
 	/// \brief set the value for item name @a name to @a value
 	void set_value(std::string_view name, std::string_view value);
 

include/cif++/text.hpp

@@ -355,279 +355,35 @@ std::string cif_id_for_number(int number);
 std::vector<std::string> word_wrap(const std::string &text, std::size_t width);
 
 // --------------------------------------------------------------------
-/// \brief std::from_chars for floating point types.
-///
-/// These are optional, there's a selected_charconv class below that selects
-/// the best option to use based on support by the stl library.
-///
-/// I.e. that in case of GNU < 12 (or something) the cif implementation will
-/// be used, all other cases will use the stl version.
 
-template <typename FloatType, std::enable_if_t<std::is_floating_point_v<FloatType>, int> = 0>
-std::from_chars_result from_chars(const char *first, const char *last, FloatType &value)
-{
-	std::from_chars_result result{ first, {} };
-
-	enum State
-	{
-		IntegerSign,
-		Integer,
-		Fraction,
-		ExponentSign,
-		Exponent
-	} state = IntegerSign;
-	int sign = 1;
-	unsigned long long vi = 0;
-	int fl = 0, tz = 0;
-	int exponent_sign = 1;
-	int exponent = 0;
-	bool done = false;
-
-	while (not done and not (bool)result.ec)
-	{
-		char ch = result.ptr != last ? *result.ptr : 0;
-		++result.ptr;
-
-		switch (state)
-		{
-			case IntegerSign:
-				if (ch == '-')
-				{
-					sign = -1;
-					state = Integer;
-				}
-				else if (ch == '+')
-					state = Integer;
-				else if (ch >= '0' and ch <= '9')
-				{
-					vi = ch - '0';
-					state = Integer;
-				}
-				else if (ch == '.')
-					state = Fraction;
-				else
-					result.ec = std::errc::invalid_argument;
-				break;
-
-			case Integer:
-				if (ch >= '0' and ch <= '9')
-					vi = 10 * vi + (ch - '0');
-				else if (ch == 'e' or ch == 'E')
-					state = ExponentSign;
-				else if (ch == '.')
-					state = Fraction;
-				else
-				{
-					done = true;
-					--result.ptr;
-				}
-				break;
-
-			case Fraction:
-				if (ch >= '0' and ch <= '9')
-				{
-					vi = 10 * vi + (ch - '0');
-
-					if (ch == '0')
-						tz += 1;
-					else
-					{
-						fl += tz + 1;
-						tz = 0;
-					}
-				}
-				else if (ch == 'e' or ch == 'E')
-					state = ExponentSign;
-				else
-				{
-					done = true;
-					--result.ptr;
-				}
-				break;
-
-			case ExponentSign:
-				if (ch == '-')
-				{
-					exponent_sign = -1;
-					state = Exponent;
-				}
-				else if (ch == '+')
-					state = Exponent;
-				else if (ch >= '0' and ch <= '9')
-				{
-					exponent = ch - '0';
-					state = Exponent;
-				}
-				else
-					result.ec = std::errc::invalid_argument;
-				break;
-
-			case Exponent:
-				if (ch >= '0' and ch <= '9')
-					exponent = 10 * exponent + (ch - '0');
-				else
-				{
-					done = true;
-					--result.ptr;
-				}
-				break;
-		}
-	}
-
-	if (not (bool)result.ec)
-	{
-		while (tz-- > 0)
-			vi /= 10;
-
-		long double v = std::pow(10, -fl) * vi * sign;
-		if (exponent != 0)
-			v *= std::pow(10, exponent * exponent_sign);
-
-		if (std::isnan(v))
-			result.ec = std::errc::invalid_argument;
-		else if (std::abs(v) > std::numeric_limits<FloatType>::max())
-			result.ec = std::errc::result_out_of_range;
-
-		value = static_cast<FloatType>(v);
-	}
-
-	return result;
-}
-
-/// \brief duplication of std::chars_format for deficient STL implementations
-enum class chars_format
-{
-	scientific = 1,
-	fixed = 2,
-	// hex,
-	general = fixed | scientific
-};
-
-/// \brief a simplistic implementation of std::to_chars for old STL implementations
-template <typename FloatType, std::enable_if_t<std::is_floating_point_v<FloatType>, int> = 0>
-std::to_chars_result to_chars(char *first, char *last, FloatType &value, chars_format fmt)
-{
-	int size = static_cast<int>(last - first);
-	int r = 0;
-
-	switch (fmt)
-	{
-		case chars_format::scientific:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%le", value);
-			else
-				r = snprintf(first, last - first, "%e", value);
-			break;
-
-		case chars_format::fixed:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%lf", value);
-			else
-				r = snprintf(first, last - first, "%f", value);
-			break;
-
-		case chars_format::general:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%lg", value);
-			else
-				r = snprintf(first, last - first, "%g", value);
-			break;
-	}
-
-	std::to_chars_result result;
-	if (r < 0 or r >= size)
-		result = { first, std::errc::value_too_large };
-	else
-		result = { first + r, std::errc() };
-
-	return result;
-}
-
-/// \brief a simplistic implementation of std::to_chars for old STL implementations
-template <typename FloatType, std::enable_if_t<std::is_floating_point_v<FloatType>, int> = 0>
-std::to_chars_result to_chars(char *first, char *last, FloatType &value, chars_format fmt, int precision)
-{
-	int size = static_cast<int>(last - first);
-	int r = 0;
-
-	switch (fmt)
-	{
-		case chars_format::scientific:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%.*le", precision, value);
-			else
-				r = snprintf(first, last - first, "%.*e", precision, value);
-			break;
-
-		case chars_format::fixed:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%.*lf", precision, value);
-			else
-				r = snprintf(first, last - first, "%.*f", precision, value);
-			break;
-
-		case chars_format::general:
-			if constexpr (std::is_same_v<FloatType, long double>)
-				r = snprintf(first, last - first, "%.*lg", precision, value);
-			else
-				r = snprintf(first, last - first, "%.*g", precision, value);
-			break;
-	}
-
-	std::to_chars_result result;
-	if (r < 0 or r >= size)
-		result = { first, std::errc::value_too_large };
-	else
-		result = { first + r, std::errc() };
-
-	return result;
-}
-
-/// \brief class that uses our implementation of std::from_chars and std::to_chars
 template <typename T>
-struct my_charconv
-{
-	/// @brief Simply call our version of std::from_chars
-	static std::from_chars_result from_chars(const char *a, const char *b, T &d)
-	{
-		return cif::from_chars(a, b, d);
-	}
+using from_chars_function = decltype(std::from_chars(std::declval<const char *>(), std::declval<const char *>(), std::declval<T &>()));
 
-	/// @brief Simply call our version of std::to_chars
-	static std::to_chars_result to_chars(char *first, char *last, T &value, chars_format fmt)
-	{
-		return cif::to_chars(first, last, value, fmt);
-	}
-};
-
-/// \brief class that uses the STL implementation of std::from_chars and std::to_chars
 template <typename T>
 struct std_charconv
 {
-	/// @brief Simply call std::from_chars
 	static std::from_chars_result from_chars(const char *a, const char *b, T &d)
 	{
 		return std::from_chars(a, b, d);
 	}
-
-	/// @brief Simply call std::to_chars
-	static std::to_chars_result to_chars(char *first, char *last, T &value, chars_format fmt)
-	{
-		return std::to_chars(first, last, value, fmt);
-	}
 };
 
-/// \brief helper to find a from_chars function
-template <typename T>
-using from_chars_function = decltype(std::from_chars(std::declval<const char *>(), std::declval<const char *>(), std::declval<T &>()));
+template <typename T, typename = void>
+struct ff_charconv;
 
-/**
- * @brief Helper to select the best implementation of charconv based on availability of the
- * function in the std:: namespace
- *
- * @tparam T The type for which we want to find a from_chars/to_chars function
- */
 template <typename T>
-using selected_charconv = typename std::conditional_t<std_experimental::is_detected_v<from_chars_function, T>, std_charconv<T>, my_charconv<T>>;
+struct ff_charconv<T, typename std::enable_if_t<std::is_floating_point_v<T>>>
+{
+	static std::from_chars_result from_chars(const char *a, const char *b, T &v);
+};
+
+template <typename T>
+using charconv = typename std::conditional_t<std_experimental::is_detected_v<from_chars_function, T>, std_charconv<T>, ff_charconv<T>>;
+
+template <typename T>
+constexpr auto from_chars(const char *s, const char *e, T &v)
+{
+	return charconv<T>::from_chars(s, e, v);
+}
 
 } // namespace cif

include/cif++/utilities.hpp

@@ -296,6 +296,11 @@ class progress_bar
 	 */
 	void message(const std::string &inMessage);
 
+	/**
+	 * @brief Flush the progress bar to the output stream
+	 */
+	void flush();
+
   private:
 	progress_bar(const progress_bar &) = delete;
 	progress_bar &operator=(const progress_bar &) = delete;

include/cif++/validate.hpp

@@ -30,7 +30,6 @@
 #include "cif++/text.hpp"
 
 #include <cassert>
-#include <filesystem>
 #include <list>
 #include <mutex>
 #include <optional>
@@ -343,11 +342,11 @@ struct item_validator
  */
 struct category_validator
 {
-	std::string m_name;                         ///< The name of the category
-	std::vector<std::string> m_keys;            ///< The list of items that make up the key
-	cif::iset m_groups;                         ///< The category groups this category belongs to
-	cif::iset m_mandatory_items;                ///< The mandatory items for this category
-	std::set<item_validator> m_item_validators; ///< The item validators for the items in this category
+	std::string m_name;                            ///< The name of the category
+	std::vector<std::string> m_keys;               ///< The list of items that make up the key
+	cif::iset m_groups;                            ///< The category groups this category belongs to
+	cif::iset m_mandatory_items;                   ///< The mandatory items for this category
+	std::vector<item_validator> m_item_validators; ///< The item validators for the items in this category
 
 	/// @brief return true if this category sorts before @a rhs
 	bool operator<(const category_validator &rhs) const
@@ -520,10 +519,18 @@ class validator_factory
 	static validator_factory &instance();
 
 	/// @brief Return validator with info recorded in @a audit_conform
-	const validator &get(const category &audit_conform);
+	const validator *get(const category &audit_conform);
 
 	/// @brief Return the single-file validator with name @a dictionary_name
-	const validator &get(std::string_view dictionary_name);
+	/// and the dictionary name may be a set of dictionaries separated by comma
+	const validator *get(std::string_view dictionary_name);
+
+	/// @brief Return validator with info recorded in @a audit_conform
+	const validator &operator[](const category &audit_conform);
+
+	/// @brief Return the single-file validator with name @a dictionary_name
+	/// and the dictionary name may be a set of dictionaries separated by comma
+	const validator &operator[](std::string_view dictionary_name);
 
 	/// @brief Return true if the version @a found is equal or higher than @a expected for dictionary @a name
 	static bool check_version(std::string_view name, std::string_view expected, std::string_view found);
@@ -535,6 +542,21 @@ class validator_factory
 		return m_validators.emplace_back(std::move(v));
 	}
 
+#if __cplusplus >= 202302L
+	/// @brief Return validator with info recorded in @a audit_conform
+	static validator &operator[](const category &audit_conform)
+	{
+		return instance()[audit_conform];
+	}
+
+	/// @brief Return the single-file validator with name @a dictionary_name
+	/// and the dictionary name may be a set of dictionaries separated by comma
+	static validator &operator[](std::string_view dict)
+	{
+		return instance()[dict];
+	}
+#endif
+
   private:
 	validator_factory() = default;
 

pcre2-simple/CMakeLists.txt

@@ -0,0 +1,316 @@
+# SPDX-License-Identifier: BSD-2-Clause
+# 
+# Copyright (c) 2025 Maarten L. Hekkelman
+# 
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+# 
+# 1. Redistributions of source code must retain the above copyright notice, this
+#    list of conditions and the following disclaimer
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+#    this list of conditions and the following disclaimer in the documentation
+#    and/or other materials provided with the distribution.
+# 
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+# ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# A simplified wrapper CMakeLists.txt file for PCRE2
+#
+# This will generate an OBJECT library so it can be linked into another library
+
+cmake_minimum_required(VERSION 3.25)
+
+include(FetchContent)
+
+project(pcre2s VERSION 1.0.0 LANGUAGES C CXX)
+
+# The original code:
+
+file(DOWNLOAD https://github.com/PCRE2Project/pcre2/releases/download/pcre2-10.46/pcre2-10.46.tar.gz
+    ${CMAKE_CURRENT_BINARY_DIR}/pcre2-code.tgz
+    EXPECTED_HASH SHA256=8d28d7f2c3b970c3a4bf3776bcbb5adfc923183ce74bc8df1ebaad8c1985bd07)
+file(ARCHIVE_EXTRACT INPUT ${CMAKE_CURRENT_BINARY_DIR}/pcre2-code.tgz
+    DESTINATION ${CMAKE_CURRENT_BINARY_DIR})
+set(PCRE2_SOURCE_DIR ${CMAKE_CURRENT_BINARY_DIR}/pcre2-10.46)
+set(PCRE2_MAJOR 10)
+set(PCRE2_MINOR 46)
+set(PCRE2_VERSION "${PCRE2_MAJOR}.${PCRE2_MINOR}")
+set(PCRE2_DATE "2024-06-09")
+
+# Some needed configuration options
+
+# option(PCRE2_BUILD_PCRE2_8 "Build 8 bit PCRE2 library" ON)
+# option(PCRE2_BUILD_PCRE2_16 "Build 16 bit PCRE2 library" OFF)
+# option(PCRE2_BUILD_PCRE2_32 "Build 32 bit PCRE2 library" OFF)
+
+option(PCRE2_STATIC_PIC "Build the static library with the option position independent code enabled." OFF)
+
+set(PCRE2_NEWLINE "LF" CACHE STRING "What to recognize as a newline (one of CR, LF, CRLF, ANY, ANYCRLF, NUL)." FORCE)
+set_property(CACHE PCRE2_NEWLINE PROPERTY STRINGS "CR" "LF" "CRLF" "ANY" "ANYCRLF" "NUL")
+
+set(PCRE2_LINK_SIZE "2" CACHE STRING "Internal link size (2, 3 or 4 allowed). See LINK_SIZE in config.h.in for details.")
+set_property(CACHE PCRE2_LINK_SIZE PROPERTY STRINGS "2" "3" "4")
+
+set(PCRE2_PARENS_NEST_LIMIT "250" CACHE STRING "Default nested parentheses limit. See PARENS_NEST_LIMIT in config.h.in for details.")
+set(PCRE2_HEAP_LIMIT "20000000" CACHE STRING "Default limit on heap memory (kibibytes). See HEAP_LIMIT in config.h.in for details.")
+set(PCRE2_MAX_VARLOOKBEHIND "255" CACHE STRING "Default limit on variable lookbehinds.")
+set(PCRE2_MATCH_LIMIT "10000000" CACHE STRING "Default limit on internal looping. See MATCH_LIMIT in config.h.in for details.")
+set(PCRE2_MATCH_LIMIT_DEPTH "MATCH_LIMIT" CACHE STRING "Default limit on internal depth of search. See MATCH_LIMIT_DEPTH in config.h.in for details.")
+set(PCRE2GREP_BUFSIZE "20480" CACHE STRING "Buffer starting size parameter for pcre2grep. See PCRE2GREP_BUFSIZE in config.h.in for details.")
+set(PCRE2GREP_MAX_BUFSIZE "1048576" CACHE STRING "Buffer maximum size parameter for pcre2grep. See PCRE2GREP_MAX_BUFSIZE in config.h.in for details.")
+set(PCRE2_SUPPORT_JIT OFF CACHE BOOL "Enable support for Just-in-time compiling.")
+
+if(${CMAKE_SYSTEM_NAME} MATCHES Linux|NetBSD)
+    set(PCRE2_SUPPORT_JIT_SEALLOC OFF CACHE BOOL "Enable SELinux compatible execmem allocator in JIT (experimental).")
+else()
+    set(PCRE2_SUPPORT_JIT_SEALLOC IGNORE)
+endif()
+
+set(PCRE2GREP_SUPPORT_JIT ON CACHE BOOL "Enable use of Just-in-time compiling in pcre2grep.")
+set(PCRE2GREP_SUPPORT_CALLOUT ON CACHE BOOL "Enable callout string support in pcre2grep.")
+set(PCRE2GREP_SUPPORT_CALLOUT_FORK ON CACHE BOOL "Enable callout string fork support in pcre2grep.")
+set(PCRE2_SUPPORT_UNICODE ON CACHE BOOL "Enable support for Unicode and UTF-8/UTF-16/UTF-32 encoding.")
+set(PCRE2_SUPPORT_BSR_ANYCRLF OFF CACHE BOOL "ON=Backslash-R matches only LF CR and CRLF, OFF=Backslash-R matches all Unicode Linebreaks")
+set(PCRE2_NEVER_BACKSLASH_C OFF CACHE BOOL "If ON, backslash-C (upper case C) is locked out.")
+set(PCRE2_SUPPORT_VALGRIND OFF CACHE BOOL "Enable Valgrind support.")
+
+if(MINGW)
+    option(NON_STANDARD_LIB_PREFIX "ON=Shared libraries built in mingw will be named pcre2.dll, etc., instead of libpcre2.dll, etc." OFF)
+    option(NON_STANDARD_LIB_SUFFIX "ON=Shared libraries built in mingw will be named libpcre2-0.dll, etc., instead of libpcre2.dll, etc." OFF)
+endif()
+
+# 
+
+set(NEWLINE_DEFAULT "")
+
+if(PCRE2_NEWLINE STREQUAL "CR")
+    set(NEWLINE_DEFAULT "1")
+elseif(PCRE2_NEWLINE STREQUAL "LF")
+    set(NEWLINE_DEFAULT "2")
+elseif(PCRE2_NEWLINE STREQUAL "CRLF")
+    set(NEWLINE_DEFAULT "3")
+elseif(PCRE2_NEWLINE STREQUAL "ANY")
+    set(NEWLINE_DEFAULT "4")
+elseif(PCRE2_NEWLINE STREQUAL "ANYCRLF")
+    set(NEWLINE_DEFAULT "5")
+elseif(PCRE2_NEWLINE STREQUAL "NUL")
+    set(NEWLINE_DEFAULT "6")
+else()
+    message(FATAL_ERROR "The PCRE2_NEWLINE variable must be set to one of the following values: \"LF\", \"CR\", \"CRLF\", \"ANY\", \"ANYCRLF\".")
+endif()
+
+# Some tests
+
+include(CheckCSourceCompiles)
+include(CheckFunctionExists)
+include(CheckSymbolExists)
+include(CheckIncludeFile)
+
+check_include_file(assert.h HAVE_ASSERT_H)
+check_include_file(dirent.h HAVE_DIRENT_H)
+check_include_file(sys/stat.h HAVE_SYS_STAT_H)
+check_include_file(sys/types.h HAVE_SYS_TYPES_H)
+check_include_file(unistd.h HAVE_UNISTD_H)
+check_include_file(windows.h HAVE_WINDOWS_H)
+
+check_symbol_exists(bcopy "strings.h" HAVE_BCOPY)
+check_symbol_exists(memfd_create "sys/mman.h" HAVE_MEMFD_CREATE)
+check_symbol_exists(memmove "string.h" HAVE_MEMMOVE)
+check_symbol_exists(secure_getenv "stdlib.h" HAVE_SECURE_GETENV)
+check_symbol_exists(strerror "string.h" HAVE_STRERROR)
+
+check_c_source_compiles(
+  "int main(void) { char buf[128] __attribute__((uninitialized)); (void)buf; return 0; }"
+  HAVE_ATTRIBUTE_UNINITIALIZED
+)
+
+check_c_source_compiles(
+  [=[
+  extern __attribute__ ((visibility ("default"))) int f(void);
+  int main(void) { return f(); }
+  int f(void) { return 42; }
+  ]=]
+  HAVE_VISIBILITY
+)
+
+if(HAVE_VISIBILITY)
+  set(PCRE2_EXPORT [=[__attribute__ ((visibility ("default")))]=])
+else()
+  set(PCRE2_EXPORT)
+endif()
+
+check_c_source_compiles("int main(void) { __assume(1); return 0; }" HAVE_BUILTIN_ASSUME)
+
+check_c_source_compiles(
+  [=[
+  #include <stddef.h>
+  int main(void) { int a,b; size_t m; __builtin_mul_overflow(a,b,&m); return 0; }
+  ]=]
+  HAVE_BUILTIN_MUL_OVERFLOW
+)
+
+check_c_source_compiles(
+  "int main(int c, char *v[]) { if (c) __builtin_unreachable(); return (int)(*v[0]); }"
+  HAVE_BUILTIN_UNREACHABLE
+)
+
+# # Check whether Intel CET is enabled, and if so, adjust compiler flags. This
+# # code was written by PH, trying to imitate the logic from the autotools
+# # configuration.
+
+# check_c_source_compiles(
+#   [=[
+#   #ifndef __CET__
+#   #error CET is not enabled
+#   #endif
+#   int main() { return 0; }
+#   ]=]
+#   INTEL_CET_ENABLED
+# )
+
+# if(INTEL_CET_ENABLED)
+#   set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mshstk")
+# endif()
+
+# Set up some dependencies first
+
+configure_file(
+    ${PCRE2_SOURCE_DIR}/src/pcre2_chartables.c.dist
+    ${CMAKE_CURRENT_BINARY_DIR}/pcre2_chartables.c
+    COPYONLY
+)
+
+configure_file(
+    ${PCRE2_SOURCE_DIR}/config-cmake.h.in
+    ${CMAKE_CURRENT_BINARY_DIR}/interface/config.h
+    @ONLY
+)
+
+configure_file(
+    ${PCRE2_SOURCE_DIR}/src/pcre2.h.in
+    ${CMAKE_CURRENT_BINARY_DIR}/interface/pcre2.h
+    @ONLY
+)
+
+# Define our library
+
+list(APPEND PCRE2_HEADERS
+    ${CMAKE_CURRENT_BINARY_DIR}/interface/pcre2.h)
+
+list(APPEND PCRE2_SOURCES
+    ${PCRE2_SOURCE_DIR}/src/pcre2_auto_possess.c
+    ${CMAKE_CURRENT_BINARY_DIR}/pcre2_chartables.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_chkdint.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_compile.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_compile_class.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_config.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_context.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_convert.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_dfa_match.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_error.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_extuni.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_find_bracket.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_jit_compile.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_maketables.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_match.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_match_data.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_newline.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_ord2utf.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_pattern_info.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_script_run.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_serialize.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_string_utils.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_study.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_substitute.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_substring.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_tables.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_ucd.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_valid_utf.c
+    ${PCRE2_SOURCE_DIR}/src/pcre2_xclass.c
+)
+
+add_library(pcre2s OBJECT)
+
+target_sources(pcre2s
+    PRIVATE ${PCRE2_SOURCES}
+    PUBLIC
+    FILE_SET pcre2_headers TYPE HEADERS
+    BASE_DIRS ${PCRE2_SOURCE_DIR}/include ${CMAKE_CURRENT_BINARY_DIR}/interface
+    FILES ${PCRE2_HEADERS}
+)
+
+target_compile_definitions(pcre2s PUBLIC PCRE2_CODE_UNIT_WIDTH=8 HAVE_CONFIG_H)
+if(NOT BUILD_SHARED_LIBS)
+    target_compile_definitions(pcre2s PUBLIC PCRE2_STATIC)
+endif()
+
+target_include_directories(pcre2s PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/interface ${PCRE2_SOURCE_DIR}/src)
+
+if(PCRE2_STATIC_PIC)
+    set_target_properties(pcre2s PROPERTIES POSITION_INDEPENDENT_CODE 1)
+endif()
+
+# # Installation and config files
+
+# include(CMakePackageConfigHelpers)
+# include(GenerateExportHeader)
+
+# # Install rules
+# install(TARGETS pcre2s
+#     EXPORT pcre2s
+#     FILE_SET pcre2_headers DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+
+# if(MSVC AND BUILD_SHARED_LIBS)
+#     install(
+#         FILES $<TARGET_PDB_FILE:pcre2s>
+#         DESTINATION ${CMAKE_INSTALL_LIBDIR}
+#         OPTIONAL)
+# endif()
+
+# install(EXPORT pcre2s
+#     NAMESPACE pcre2s::
+#     FILE "pcre2s-targets.cmake"
+#     DESTINATION lib/cmake/pcre2s)
+
+# configure_package_config_file(
+#     ${CMAKE_CURRENT_SOURCE_DIR}/pcre2s-config.cmake.in ${CMAKE_CURRENT_BINARY_DIR}/pcre2s/pcre2s-config.cmake
+#     INSTALL_DESTINATION lib/cmake/pcre2s)
+
+# install(
+#     FILES "${CMAKE_CURRENT_BINARY_DIR}/pcre2s/pcre2s-config.cmake"
+#     "${CMAKE_CURRENT_BINARY_DIR}/pcre2s/pcre2s-config-version.cmake"
+#     DESTINATION lib/cmake/pcre2s)
+
+# set_target_properties(
+#     pcre2s
+#     PROPERTIES VERSION ${PCRE2_VERSION}
+#     SOVERSION ${PCRE2_VERSION}
+#     INTERFACE_pcre2s_MAJOR_VERSION ${PCRE2_MAJOR})
+
+# set_property(
+#     TARGET pcre2s
+#     APPEND
+#     PROPERTY COMPATIBLE_INTERFACE_STRING pcre2s_MAJOR_VERSION)
+
+# write_basic_package_version_file(
+#     "${CMAKE_CURRENT_BINARY_DIR}/pcre2s/pcre2s-config-version.cmake"
+#     VERSION "${PCRE2_VERSION}"
+#     COMPATIBILITY AnyNewerVersion)
+
+# # Testing
+
+# if(PROJECT_IS_TOP_LEVEL)
+#     include(CTest)
+
+#     if(BUILD_TESTING)
+#         add_subdirectory(test)
+#     endif()
+# endif()

src/category.cpp

@@ -25,6 +25,7 @@
  */
 
 #include "cif++/category.hpp"
+
 #include "cif++/datablock.hpp"
 #include "cif++/parser.hpp"
 #include "cif++/utilities.hpp"
@@ -32,6 +33,8 @@
 
 #include <numeric>
 #include <stack>
+#include <utility>
+#include <vector>
 
 // TODO: Find out what the rules are exactly for linked items, the current implementation
 // is inconsistent. It all depends whether a link is satified if a item taking part in the
@@ -108,7 +111,7 @@ class row_comparator
 			std::string_view ka = ai->value;
 			std::string_view kb = rhb[k].text();
 
-			if (not (ai->may_be_null and rhb[k].empty()))
+			if (not(ai->may_be_null and rhb[k].empty()))
 				d = f(ka, kb);
 
 			if (d != 0)
@@ -535,6 +538,7 @@ void swap(category &a, category &b) noexcept
 	std::swap(a.m_index, b.m_index);
 	std::swap(a.m_head, b.m_head);
 	std::swap(a.m_tail, b.m_tail);
+	std::swap(a.m_dirty, b.m_dirty);
 }
 
 category::~category()
@@ -606,6 +610,30 @@ void category::remove_item(std::string_view item_name)
 	}
 }
 
+void category::drop_empty_items()
+{
+	std::vector<bool> is_empty(m_items.size(), true);
+	
+	for (auto &row : *this)
+	{
+		for (size_t ix = 0; ix < m_items.size(); ++ix)
+		{
+			if (is_empty[ix] and not row[ix].empty())
+				is_empty[ix] = false;
+		}
+	}
+
+	std::vector<std::string> items;
+	for (size_t ix = 0; ix < m_items.size(); ++ix)
+	{
+		if (is_empty[ix])
+			items.push_back(m_items[ix].m_name);
+	}
+
+	for (auto &item : items)
+		remove_item(item);
+}
+
 void category::rename_item(std::string_view from_name, std::string_view to_name)
 {
 	for (std::size_t ix = 0; ix < m_items.size(); ++ix)
@@ -620,12 +648,12 @@ void category::rename_item(std::string_view from_name, std::string_view to_name)
 	}
 }
 
-iset category::get_items() const
+std::vector<std::string> category::get_items() const
 {
-	iset result;
+	std::vector<std::string> result;
 
 	for (auto &col : m_items)
-		result.insert(col.m_name);
+		result.emplace_back(col.m_name);
 
 	return result;
 }
@@ -1248,6 +1276,7 @@ void category::clear()
 
 	delete m_index;
 	m_index = nullptr;
+	m_dirty = true;
 }
 
 void category::erase_orphans(condition &&cond, category &parent)
@@ -1496,6 +1525,8 @@ void category::update_value(row *row, uint16_t item, std::string_view value, boo
 	if (value == oldValue) // no need to update
 		return;
 
+	m_dirty = true;
+
 	std::string oldStrValue{ oldValue };
 
 	// check the value
@@ -1637,6 +1668,8 @@ void category::delete_row(row *r)
 		row_allocator_type ra(get_allocator());
 		row_allocator_traits::destroy(ra, r);
 		row_allocator_traits::deallocate(ra, r, 1);
+
+		m_dirty = true;
 	}
 }
 
@@ -1737,6 +1770,8 @@ category::iterator category::insert_impl(const_iterator pos, row *n)
 				n = n->m_next = m_head->m_next;
 		}
 
+		m_dirty = true;
+
 		return iterator(*this, n);
 	}
 	catch (const std::exception &e)
@@ -1759,6 +1794,8 @@ void category::swap_item(uint16_t item_ix, row_handle &a, row_handle &b)
 	auto &ra = *a.m_row;
 	auto &rb = *b.m_row;
 
+	m_dirty = true;
+
 	while (ra.size() <= item_ix)
 		ra.emplace_back("");
 
@@ -1773,6 +1810,8 @@ void category::sort(std::function<int(row_handle, row_handle)> f)
 	if (m_head == nullptr)
 		return;
 
+	m_dirty = true;
+
 	std::vector<row_handle> rows;
 	for (auto itemRow = m_head; itemRow != nullptr; itemRow = itemRow->m_next)
 		rows.emplace_back(*this, *itemRow);
@@ -1912,10 +1951,10 @@ void category::write(std::ostream &os) const
 {
 	std::vector<uint16_t> order(m_items.size());
 	iota(order.begin(), order.end(), static_cast<uint16_t>(0));
-	write(os, order, false);
+	write_cif(os, order, false);
 }
 
-void category::write(std::ostream &os, const std::vector<std::string> &items, bool addMissingItems)
+void category::write(std::ostream &os, output_format fmt, const std::vector<std::string> &items, bool addMissingItems)
 {
 	// make sure all items are present
 	for (auto &c : items)
@@ -1936,10 +1975,43 @@ void category::write(std::ostream &os, const std::vector<std::string> &items, bo
 		}
 	}
 
-	write(os, order, true);
+	switch (fmt)
+	{
+		case output_format::cif:
+			write_cif(os, order, addMissingItems);
+			break;
+
+		case output_format::csv:
+			write_delimited(os, order, addMissingItems, ",", false, true);
+			break;
+
+		case output_format::tsv:
+			write_delimited(os, order, addMissingItems, "\t", false, true);
+			break;
+
+		case output_format::list:
+			write_delimited(os, order, addMissingItems, "|", false, false);
+			break;
+
+		case output_format::column:
+			write_delimited(os, order, addMissingItems, "  ", true, true);
+			break;
+
+		case output_format::markdown:
+			write_markdown(os, order, addMissingItems);
+			break;
+
+		case output_format::table:
+			write_table(os, order, addMissingItems, true);
+			break;
+
+		case output_format::box:
+			write_table(os, order, addMissingItems, false);
+			break;
+	}
 }
 
-void category::write(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const
+void category::write_cif(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const
 {
 	if (empty())
 		return;
@@ -2112,6 +2184,423 @@ void category::write(std::ostream &os, const std::vector<uint16_t> &order, bool
 	os << "# \n";
 }
 
+void category::write_delimited(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems,
+	std::string_view delimiter, bool aligned, bool header) const
+{
+	if (empty())
+		return;
+
+	std::vector<bool> right_aligned(m_items.size(), false);
+
+	if (aligned and m_cat_validator != nullptr)
+	{
+		for (auto cix : order)
+		{
+			auto &col = m_items[cix];
+			right_aligned[cix] = col.m_validator != nullptr and
+			                     col.m_validator->m_type != nullptr and
+			                     col.m_validator->m_type->m_primitive_type == cif::DDL_PrimitiveType::Numb;
+		}
+	}
+
+	std::vector<std::size_t> itemWidths(m_items.size());
+	auto get_line = [delimiter](std::string_view s) -> std::string
+	{
+		if (delimiter == ",")
+		{
+			if (s.find_first_of("\",") == std::string::npos)
+				return std::string{ s };
+
+			std::string r{ '"' };
+			r.reserve(s.length() + 2);
+			for (auto ch : s)
+			{
+				if (ch == '"')
+					r.append("\"\"");
+				else
+					r.push_back(ch);
+			}
+			r.push_back('"');
+			return r;
+		}
+		else if (delimiter == "\t")
+		{
+			std::string r;
+			r.reserve(s.length());
+			for (auto ch : s)
+			{
+				if (ch == '\r' or ch == '\n' or ch == '\t' or ch == '\\')
+					r.push_back('\\');
+				r.push_back(ch);
+			}
+			return r;
+		}
+		else if (delimiter == "|" or delimiter == "  ")
+			return std::string{ s };
+		else
+		{
+			assert(false);
+			return std::string{ s };
+		}
+	};
+
+	if (aligned)
+	{
+		if (header)
+		{
+			for (auto cix : order)
+			{
+				auto &col = m_items[cix];
+				itemWidths[cix] = col.m_name.length();
+			}
+		}
+
+		for (auto r = m_head; r != nullptr; r = r->m_next)
+		{
+			for (uint16_t ix = 0; ix < r->size(); ++ix)
+			{
+				auto v = r->get(ix);
+				if (v == nullptr)
+					continue;
+
+				size_t l = get_line(v->text()).length();
+				if (itemWidths[ix] < l)
+					itemWidths[ix] = l;
+			}
+		}
+	}
+
+	if (header)
+	{
+		for (bool first = true; uint16_t cix : order)
+		{
+			if (not std::exchange(first, false))
+				os << delimiter;
+
+			std::size_t w = itemWidths[cix];
+			std::string_view s = m_items[cix].m_name;
+
+			if (s.length() < w)
+			{
+				if (delimiter == "  ")
+				{
+					int l = (w - s.length()) / 2;
+					int r = w - s.length() - l;
+					os << std::string(l, ' ') << s << std::string(r, ' ');
+				}
+				else
+				{
+					if (right_aligned[cix])
+						os << std::string(w - s.length(), ' ');
+					os << s;
+					if (not right_aligned[cix])
+						os << std::string(w - s.length(), ' ');
+				}
+			}
+			else
+				os << s;
+		}
+
+		os << '\n';
+
+		if (delimiter == "  ")
+		{
+			for (bool first = true; uint16_t cix : order)
+			{
+				if (not std::exchange(first, false))
+					os << delimiter;
+				os << std::string(itemWidths[cix], '-');
+			}
+
+			os << '\n';
+		}
+	}
+
+	for (auto r = m_head; r != nullptr; r = r->m_next) // loop over rows
+	{
+		for (bool first = true; uint16_t cix : order)
+		{
+			if (not std::exchange(first, false))
+				os << delimiter;
+
+			std::size_t w = itemWidths[cix];
+
+			std::string_view s;
+			auto iv = r->get(cix);
+
+			if (iv != nullptr)
+				s = iv->text();
+
+			if (s == "?" or s == ".")
+				s = "";
+
+			if (s.length() < w)
+			{
+				if (right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+				os << s;
+				if (not right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+			}
+			else
+				os << s;
+		}
+
+		os << '\n';
+	}
+}
+
+void category::write_markdown(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems) const
+{
+	if (empty())
+		return;
+
+	std::vector<bool> right_aligned(m_items.size(), false);
+
+	if (m_cat_validator != nullptr)
+	{
+		for (auto cix : order)
+		{
+			auto &col = m_items[cix];
+			right_aligned[cix] = col.m_validator != nullptr and
+			                     col.m_validator->m_type != nullptr and
+			                     col.m_validator->m_type->m_primitive_type == cif::DDL_PrimitiveType::Numb;
+		}
+	}
+
+	std::vector<std::size_t> itemWidths(m_items.size());
+
+	for (auto cix : order)
+	{
+		auto &col = m_items[cix];
+		itemWidths[cix] = col.m_name.length();
+	}
+
+	for (auto r = m_head; r != nullptr; r = r->m_next)
+	{
+		for (uint16_t ix = 0; ix < r->size(); ++ix)
+		{
+			auto v = r->get(ix);
+			if (v == nullptr)
+				continue;
+
+			size_t l = v->text().length();
+			if (itemWidths[ix] < l)
+				itemWidths[ix] = l;
+		}
+	}
+
+	os << "| ";
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << " | ";
+
+		std::size_t w = itemWidths[cix];
+		std::string_view s = m_items[cix].m_name;
+
+		if (s.length() < w)
+		{
+			int l = (w - s.length()) / 2;
+			int r = w - s.length() - l;
+			os << std::string(l, ' ') << s << std::string(r, ' ');
+		}
+		else
+			os << s;
+	}
+	os << " |\n";
+
+	os << "| ";
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << " | ";
+		if (not right_aligned[cix])
+			os << ':';
+		os << std::string(itemWidths[cix] - 1, '-');
+		if (right_aligned[cix])
+			os << ':';
+	}
+	os << " |\n";
+
+	for (auto r = m_head; r != nullptr; r = r->m_next) // loop over rows
+	{
+		os << "| ";
+		for (bool first = true; uint16_t cix : order)
+		{
+			if (not std::exchange(first, false))
+				os << " | ";
+
+			std::size_t w = itemWidths[cix];
+
+			std::string_view s;
+			auto iv = r->get(cix);
+
+			if (iv != nullptr)
+				s = iv->text();
+
+			if (s == "?" or s == ".")
+				s = "";
+
+			if (s.length() < w)
+			{
+				if (right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+				os << s;
+				if (not right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+			}
+			else
+				os << s;
+		}
+
+		os << " |\n";
+	}
+}
+
+void category::write_table(std::ostream &os, const std::vector<uint16_t> &order, bool includeEmptyItems, bool ascii) const
+{
+	static constexpr const std::string_view
+		kUnicodeBox[13] = {
+			"┌─", "─┬─", "─┐\n",
+			"├─", "─┼─", "─┤\n",
+			"└─", "─┴─", "─┘\n",
+			"│ ", " │ ", " │\n",
+			"─"
+		},
+		kAsciiBox[13] = {        //
+			"+-", "-+-", "-+\n", //
+			"+-", "-+-", "-+\n", //
+			"+-", "-+-", "-+\n", //
+			"| ", " | ", " |\n", //
+			"-"
+		};
+
+	if (empty())
+		return;
+
+	auto box = ascii ? kAsciiBox : kUnicodeBox;
+
+	std::vector<bool> right_aligned(m_items.size(), false);
+
+	if (m_cat_validator != nullptr)
+	{
+		for (auto cix : order)
+		{
+			auto &col = m_items[cix];
+			right_aligned[cix] = col.m_validator != nullptr and
+			                     col.m_validator->m_type != nullptr and
+			                     col.m_validator->m_type->m_primitive_type == cif::DDL_PrimitiveType::Numb;
+		}
+	}
+
+	std::vector<std::size_t> itemWidths(m_items.size());
+
+	for (auto cix : order)
+	{
+		auto &col = m_items[cix];
+		itemWidths[cix] = col.m_name.length();
+	}
+
+	for (auto r = m_head; r != nullptr; r = r->m_next)
+	{
+		for (uint16_t ix = 0; ix < r->size(); ++ix)
+		{
+			auto v = r->get(ix);
+			if (v == nullptr)
+				continue;
+
+			size_t l = v->text().length();
+			if (itemWidths[ix] < l)
+				itemWidths[ix] = l;
+		}
+	}
+
+	os << box[0];
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << box[1];
+		for (size_t i = 0; i < itemWidths[cix]; ++i)
+			os << box[12];
+	}
+	os << box[2];
+
+	os << box[9];
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << box[10];
+
+		std::size_t w = itemWidths[cix];
+		std::string_view s = m_items[cix].m_name;
+
+		if (s.length() < w)
+		{
+			int l = (w - s.length()) / 2;
+			int r = w - s.length() - l;
+			os << std::string(l, ' ') << s << std::string(r, ' ');
+		}
+		else
+			os << s;
+	}
+	os << box[11];
+
+	os << box[3];
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << box[4];
+		for (size_t i = 0; i < itemWidths[cix]; ++i)
+			os << box[12];
+	}
+	os << box[5];
+
+	for (auto r = m_head; r != nullptr; r = r->m_next) // loop over rows
+	{
+		os << box[9];
+		for (bool first = true; uint16_t cix : order)
+		{
+			if (not std::exchange(first, false))
+				os << box[10];
+
+			std::size_t w = itemWidths[cix];
+
+			std::string_view s;
+			auto iv = r->get(cix);
+
+			if (iv != nullptr)
+				s = iv->text();
+
+			if (s == "?" or s == ".")
+				s = "";
+
+			if (s.length() < w)
+			{
+				if (right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+				os << s;
+				if (not right_aligned[cix])
+					os << std::string(w - s.length(), ' ');
+			}
+			else
+				os << s;
+		}
+
+		os << box[11];
+	}
+
+	os << box[6];
+	for (bool first = true; uint16_t cix : order)
+	{
+		if (not std::exchange(first, false))
+			os << box[7];
+		for (size_t i = 0; i < itemWidths[cix]; ++i)
+			os << box[12];
+	}
+	os << box[8];
+}
+
 bool category::operator==(const category &rhs) const
 {
 	// shortcut

src/compound.cpp

@@ -24,14 +24,38 @@
  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
-#include "cif++.hpp"
+#include "cif++/compound.hpp" // for compound_atom, compound_bond, compoun...
 
-#include <filesystem>
-#include <fstream>
-#include <map>
-#include <mutex>
-#include <numeric>
-#include <shared_mutex>
+#include "cif++/atom_type.hpp" // for atom_type_traits
+#include "cif++/category.hpp"  // for category
+#include "cif++/datablock.hpp" // for datablock
+#include "cif++/file.hpp"      // for file
+#include "cif++/item.hpp"      // for item
+#include "cif++/iterator.hpp"  // for iterator_proxy
+#include "cif++/parser.hpp"    // for parser
+#include "cif++/point.hpp"     // for distance, point
+#include "cif++/row.hpp"       // for tie, row_initializer, tie_wrap
+#include "cif++/text.hpp"      // for iequals, replace_all, iset
+#include "cif++/utilities.hpp" // for load_resource, VERBOSE, colour_type
+
+#include <algorithm>    // for find_if
+#include <cstddef>      // for size_t
+#include <exception>    // for exception, throw_with_nested
+#include <filesystem>   // for path, exists
+#include <fstream>      // for char_traits, basic_ostream, operator<<
+#include <iomanip>      // for operator<<, quoted
+#include <iostream>     // for clog, cout, cerr
+#include <limits>       // for numeric_limits
+#include <list>         // for _List_iterator
+#include <map>          // for allocator, map, _Rb_tree_iterator
+#include <memory>       // for shared_ptr, unique_ptr, __shared_ptr_...
+#include <optional>     // for optional
+#include <shared_mutex> // for shared_lock, shared_timed_mutex
+#include <stdexcept>    // for runtime_error, invalid_argument, out_...
+#include <string>       // for basic_string, string, operator==, ope...
+#include <string_view>  // for string_view, basic_string_view
+#include <utility>      // for pair, exchange, move
+#include <vector>       // for vector
 
 namespace fs = std::filesystem;
 
@@ -140,7 +164,7 @@ compound::compound(cif::datablock &db)
 
 	cif::tie(m_id, m_name, m_type, m_formula, m_formula_weight, m_formal_charge, one_letter_code, m_parent_id) =
 		chemComp.front().get("id", "name", "type", "formula", "formula_weight", "pdbx_formal_charge", "one_letter_code", "mon_nstd_parent_comp_id");
-	
+
 	if (one_letter_code.length() == 1)
 		m_one_letter_code = one_letter_code.front();
 
@@ -159,7 +183,7 @@ compound::compound(cif::datablock &db)
 		if (stereo_config.empty())
 			atom.stereo_config = stereo_config_type::N;
 		else
-		atom.stereo_config = parse_stereo_config_from_string(stereo_config);
+			atom.stereo_config = parse_stereo_config_from_string(stereo_config);
 		m_atoms.push_back(std::move(atom));
 	}
 
@@ -172,7 +196,7 @@ compound::compound(cif::datablock &db)
 		if (valueOrder.empty())
 			bond.type = bond_type::sing;
 		else
-		bond.type = parse_bond_type_from_string(valueOrder);
+			bond.type = parse_bond_type_from_string(valueOrder);
 		m_bonds.push_back(std::move(bond));
 	}
 }
@@ -231,12 +255,12 @@ float compound::bond_length(const std::string &atomId_1, const std::string &atom
 
 bool compound::is_peptide() const
 {
-	return iequals(m_type, "l-peptide linking")	or iequals(m_type, "peptide linking");
+	return iequals(m_type, "l-peptide linking") or iequals(m_type, "peptide linking");
 }
 
 bool compound::is_base() const
 {
-	return iequals(m_type, "dna linking")	or iequals(m_type, "rna linking");
+	return iequals(m_type, "dna linking") or iequals(m_type, "rna linking");
 }
 
 // --------------------------------------------------------------------
@@ -294,12 +318,31 @@ class compound_factory_impl : public std::enable_shared_from_this<compound_facto
 			delete c;
 	}
 
+	virtual bool exists_self(const std::string &id) const
+	{
+		if (m_missing.contains(id))
+			return false;
+
+		if (std::find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c)
+				{ return c->id() == id; }) != m_compounds.end())
+			return true;
+
+		return m_next and m_next->exists_self(id);
+	}
+
+	bool exists(std::string_view id)
+	{
+		std::shared_lock lock(mMutex);
+
+		return exists_self(std::string{ id });
+	}
+
 	compound *get(std::string id)
 	{
 		std::shared_lock lock(mMutex);
 
 		compound *result = nullptr;
-		
+
 		for (auto impl = shared_from_this(); impl; impl = impl->m_next)
 		{
 			result = impl->create(id);
@@ -363,7 +406,9 @@ compound *compound_factory_impl::create(const std::string &id)
 	if (m_missing.contains(id))
 		return nullptr;
 
-	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c) { return c->id() == id; }); i != m_compounds.end())
+	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c)
+			{ return c->id() == id; });
+		i != m_compounds.end())
 		return *i;
 
 	compound *result = nullptr;
@@ -454,7 +499,6 @@ class local_compound_factory_impl : public compound_factory_impl
 	compound *create(const std::string &id) override;
 
   private:
-
 	compound *construct_compound(const datablock &db, const std::string &id, const std::string &name, const std::string &three_letter_code, const std::string &group);
 
 	cif::file m_local_file;
@@ -465,7 +509,9 @@ compound *local_compound_factory_impl::create(const std::string &id)
 	if (m_missing.contains(id))
 		return nullptr;
 
-	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c) { return c->id() == id; }); i != m_compounds.end())
+	if (auto i = find_if(m_compounds.begin(), m_compounds.end(), [id](compound *c)
+			{ return c->id() == id; });
+		i != m_compounds.end())
 		return *i;
 
 	compound *result = nullptr;
@@ -510,12 +556,10 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 
 	float formula_weight = 0;
 	int formal_charge = 0;
-	std::map<std::string,std::size_t> formula_data;
+	std::map<std::string, std::size_t> formula_data;
 
 	for (std::size_t ord = 1; const auto &[atom_id, type_symbol, type, charge, x, y, z, xi, yi, zi] :
-		rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int,
-			std::optional<float>, std::optional<float>, std::optional<float>,
-			std::optional<float>, std::optional<float>, std::optional<float>>(
+		rdb["chem_comp_atom"].rows<std::string, std::string, std::string, int, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>, std::optional<float>>(
 			"atom_id", "type_symbol", "type", "charge",
 			"model_Cartn_x", "model_Cartn_y", "model_Cartn_z",
 			"pdbx_model_Cartn_x_ideal", "pdbx_model_Cartn_y_ideal", "pdbx_model_Cartn_z_ideal"))
@@ -525,16 +569,14 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 
 		formula_data[type_symbol] += 1;
 
-		db["chem_comp_atom"].emplace({
-			{ "comp_id", id },
+		db["chem_comp_atom"].emplace({ { "comp_id", id },
 			{ "atom_id", atom_id },
 			{ "type_symbol", type_symbol },
 			{ "charge", charge },
-			{ "model_Cartn_x",  x.has_value() ? x : xi, 3 },
-			{ "model_Cartn_y",  y.has_value() ? y : yi, 3 },
-			{ "model_Cartn_z",  z.has_value() ? z : zi, 3 },
-			{ "pdbx_ordinal", ord++ }
-		});
+			{ "model_Cartn_x", x.has_value() ? x : xi, 3 },
+			{ "model_Cartn_y", y.has_value() ? y : yi, 3 },
+			{ "model_Cartn_z", z.has_value() ? z : zi, 3 },
+			{ "pdbx_ordinal", ord++ } });
 
 		formal_charge += charge;
 	}
@@ -551,21 +593,19 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 		else if (cif::iequals(type, "triple") or cif::iequals(type, "trip"))
 			value_order = "TRIP";
 
-		db["chem_comp_bond"].emplace({
-			{ "comp_id", id },
+		db["chem_comp_bond"].emplace({ { "comp_id", id },
 			{ "atom_id_1", atom_id_1 },
 			{ "atom_id_2", atom_id_2 },
 			{ "value_order", value_order },
 			{ "pdbx_aromatic_flag", aromatic },
 			// TODO: fetch stereo_config info from chem_comp_chir
-			{ "pdbx_ordinal", ord++ }
-		});
+			{ "pdbx_ordinal", ord++ } });
 	}
 
 	db.emplace_back(rdb["pdbx_chem_comp_descriptor"]);
 
 	std::string formula;
-	for (bool first = true; const auto &[symbol, count]: formula_data)
+	for (bool first = true; const auto &[symbol, count] : formula_data)
 	{
 		if (std::exchange(first, false))
 			formula += ' ';
@@ -584,15 +624,13 @@ compound *local_compound_factory_impl::construct_compound(const datablock &rdb,
 	else
 		type = "NON-POLYMER";
 
-	db["chem_comp"].emplace({
-		{ "id", id },
+	db["chem_comp"].emplace({ { "id", id },
 		{ "name", name },
 		{ "type", type },
 		{ "formula", formula },
 		{ "pdbx_formal_charge", formal_charge },
 		{ "formula_weight", formula_weight },
-		{ "three_letter_code", three_letter_code }
-	});
+		{ "three_letter_code", three_letter_code } });
 
 	std::shared_lock lock(mMutex);
 
@@ -698,6 +736,11 @@ void compound_factory::pop_dictionary()
 		m_impl = m_impl->next();
 }
 
+bool compound_factory::exists(std::string_view id) const
+{
+	return m_impl and m_impl->exists(id);
+}
+
 const compound *compound_factory::create(std::string_view id)
 {
 	auto result = m_impl ? m_impl->get(std::string{ id }) : nullptr;
@@ -722,7 +765,7 @@ bool compound_factory::is_peptide(std::string_view res_name) const
 	bool result = is_std_peptide(res_name);
 	if (not result and m_impl)
 	{
-		auto compound = const_cast<compound_factory&>(*this).create(res_name);
+		auto compound = const_cast<compound_factory &>(*this).create(res_name);
 		result = compound != nullptr and compound->is_peptide();
 	}
 	return result;
@@ -734,7 +777,7 @@ bool compound_factory::is_base(std::string_view res_name) const
 	bool result = is_std_base(res_name);
 	if (not result and m_impl)
 	{
-		auto compound = const_cast<compound_factory&>(*this).create(res_name);
+		auto compound = const_cast<compound_factory &>(*this).create(res_name);
 		result = compound != nullptr and compound->is_base();
 	}
 	return result;
@@ -784,7 +827,7 @@ void compound_factory::report_missing_compound(std::string_view compound_id)
 				  << "in /var/cache/libcifpp using the following commands:\n\n"
 				  << "curl -o " << CACHE_DIR << "/components.cif https://files.wwpdb.org/pub/pdb/data/monomers/components.cif\n"
 				  << "curl -o " << CACHE_DIR << "/mmcif_pdbx.dic https://mmcif.wwpdb.org/dictionaries/ascii/mmcif_pdbx_v50.dic\n"
-				  << "curl -o " << CACHE_DIR << "/mmcif_ma.dic https://github.com/ihmwg/ModelCIF/raw/master/dist/mmcif_ma.dic\n\n";
+				  << "curl -o " << CACHE_DIR << "/mmcif_ma.dic https://mmcif.wwpdb.org/dictionaries/ascii/mmcif_ma.dic\n\n";
 #endif
 
 		if (m_impl)

src/datablock.cpp

@@ -25,8 +25,11 @@
  */
 
 #include "cif++/datablock.hpp"
+
 #include "cif++/validate.hpp"
 
+#include <exception>
+
 namespace cif
 {
 
@@ -42,7 +45,28 @@ datablock::datablock(const datablock &db)
 void datablock::load_dictionary()
 {
 	if (auto *audit_conform = get("audit_conform"); audit_conform and not audit_conform->empty())
-		set_validator(&validator_factory::instance().get(*audit_conform));
+	{
+		try
+		{
+			set_validator(validator_factory::instance().get(*audit_conform));
+		}
+		catch (const std::exception &ex)
+		{
+			std::clog << ex.what() << '\n';
+		}
+	}
+}
+
+void datablock::load_dictionary(std::string_view dict)
+{
+	try
+	{
+		set_validator(validator_factory::instance().get(dict));
+	}
+	catch (const std::exception &ex)
+	{
+		std::clog << ex.what() << '\n';
+	}
 }
 
 void datablock::set_validator(const validator *v)
@@ -96,7 +120,8 @@ bool datablock::strip()
 	bool result = true;
 
 	// remove all categories that have no validator
-	erase(std::remove_if(begin(), end(), [](category &c) {
+	erase(std::remove_if(begin(), end(), [](category &c)
+			  {
 		bool result = false;
 		if (c.get_cat_validator() == nullptr)
 		{
@@ -104,8 +129,8 @@ bool datablock::strip()
 				std::clog << "Dropping category " << c.name() << '\n';
 			result = true;
 		}
-		return result;
-	}), end());
+		return result; }),
+		end());
 
 	// then strip the remaining categories
 	for (auto &cat : *this)

src/dictionary_parser.cpp

@@ -28,6 +28,9 @@
 #include "cif++/dictionary_parser.hpp"
 #include "cif++/file.hpp"
 #include "cif++/parser.hpp"
+#include <exception>
+#include <iomanip>
+#include <stdexcept>
 
 namespace cif
 {
@@ -46,7 +49,7 @@ class dictionary_parser : public parser
 	void load_dictionary()
 	{
 		std::unique_ptr<datablock> dict;
-		auto savedDatablock = m_datablock;
+		auto savedDatablock = std::exchange(m_datablock, nullptr);
 
 		try
 		{
@@ -75,6 +78,9 @@ class dictionary_parser : public parser
 			error(ex.what());
 		}
 
+		if (m_datablock == nullptr)
+			throw std::runtime_error("Dictionary file is empty?");
+
 		// store all validators
 		for (auto &ic : mCategoryValidators)
 			m_validator.add_category_validator(std::move(ic));

src/model.cpp

@@ -24,13 +24,17 @@
  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include "cif++/model.hpp"
 #include "cif++.hpp"
+#include "cif++/point.hpp"
 
 #include <filesystem>
 #include <fstream>
+#include <initializer_list>
 #include <iomanip>
 #include <numeric>
 #include <stack>
+#include <stdexcept>
 
 namespace fs = std::filesystem;
 
@@ -348,17 +352,7 @@ std::tuple<point, float> residue::center_and_radius() const
 	for (auto &a : m_atoms)
 		pts.push_back(a.get_location());
 
-	auto center = centroid(pts);
-	float radius = 0;
-
-	for (auto &pt : pts)
-	{
-		float d = static_cast<float>(distance(pt, center));
-		if (radius < d)
-			radius = d;
-	}
-
-	return std::make_tuple(center, radius);
+	return smallest_sphere_around_points(pts);
 }
 
 bool residue::has_alternate_atoms() const
@@ -1275,28 +1269,28 @@ void structure::load_atoms_for_model(structure_open_options options)
 	else
 	{
 		std::vector<cif::mm::atom> atoms;
-		std::map<std::tuple<std::string,int>, std::map<std::string, float>> alts;
-	
+		std::map<std::tuple<std::string, int>, std::map<std::string, float>> alts;
+
 		for (auto id : atom_site.find<std::string>(std::move(c), "id"))
 		{
 			auto a = atoms.emplace_back(std::make_shared<atom::atom_impl>(m_db, id));
-	
+
 			if (a.is_alternate())
 			{
 				auto key = std::make_tuple(a.get_label_asym_id(), a.get_label_seq_id());
 				auto alt_id = a.get_label_alt_id();
-	
+
 				if (auto i = alts.find(key); i != alts.end())
 					i->second[alt_id] += a.get_occupancy();
 				else
 					alts[key][alt_id] = a.get_occupancy();
 			}
 		}
-	
+
 		for (auto &&[key, value] : alts)
 		{
 			// const auto &[asym_id, seq_id] = key;
-	
+
 			// select highest occupancy for this residue's alternates
 			std::string alt_id;
 			float occupancy = options.occupancy_mode == occupancy_policy::MAX ? 0.f : std::numeric_limits<float>::max();
@@ -1319,11 +1313,11 @@ void structure::load_atoms_for_model(structure_open_options options)
 					}
 				}
 			}
-	
+
 			value.clear();
 			value.emplace(alt_id, occupancy);
 		}
-	
+
 		for (auto a : atoms)
 		{
 			if (a.is_alternate())
@@ -1335,10 +1329,8 @@ void structure::load_atoms_for_model(structure_open_options options)
 			}
 			else
 				emplace_atom(a);
-	
 		}
 	}
-
 }
 
 void structure::load_data()
@@ -1898,13 +1890,12 @@ void structure::swap_atoms(atom a1, atom a2)
 		auto r1 = atomSites.find1(key("id") == a1.id());
 		auto r2 = atomSites.find1(key("id") == a2.id());
 
-		auto l1 = r1["label_atom_id"];
-		auto l2 = r2["label_atom_id"];
-		l1.swap(l2);
-
-		auto l3 = r1["auth_atom_id"];
-		auto l4 = r2["auth_atom_id"];
-		l3.swap(l4);
+		for (std::string fld : std::initializer_list<std::string>{ "label_atom_id", "auth_atom_id", "type_symbol" })
+		{
+			auto l1 = r1[fld];
+			auto l2 = r2[fld];
+			l1.swap(l2);
+		}
 	}
 	catch (const std::exception &ex)
 	{
@@ -2348,6 +2339,36 @@ std::string structure::create_non_poly(const std::string &entity_id, std::vector
 	return asym_id;
 }
 
+std::string structure::create_non_poly(const std::string &compound_id, bool skip_hydrogen)
+{
+	auto compound = cif::compound_factory::instance().create(compound_id);
+	if (compound == nullptr)
+		throw std::runtime_error(std::format("{} is not a known compound", compound_id));
+	
+	std::vector<cif::row_initializer> atoms;
+	for (auto a : compound->atoms())
+	{
+		// We skip H-atoms, as fitting without H-atoms works better and we avoid conflicts in protonation states between CCD and MONLIB
+		if (skip_hydrogen and cif::atom_type_traits(a.type_symbol).symbol() == "H")
+			continue;
+
+		auto ax = a.get_location().get_x();
+		auto ay = a.get_location().get_y();
+		auto az = a.get_location().get_z();
+
+		atoms.emplace_back(cif::row_initializer{
+			{ "type_symbol", cif::atom_type_traits(a.type_symbol).symbol() },
+			{ "label_atom_id", a.id },
+			{ "auth_atom_id", a.id },
+			{ "Cartn_x", ax },
+			{ "Cartn_y", ay },
+			{ "Cartn_z", az },
+			{ "B_iso_or_equiv", 30.00 } });
+	}
+
+	return create_non_poly(create_non_poly_entity(compound_id), atoms);
+}
+
 void structure::create_water(row_initializer atom)
 {
 	using namespace literals;
@@ -2412,6 +2433,61 @@ void structure::create_water(row_initializer atom)
 	});
 }
 
+std::string structure::create_link(atom a1, atom a2, const std::string &link_type, const std::string &role)
+{
+	using namespace literals;
+
+	auto &struct_conn = m_db["struct_conn"];
+	auto &struct_conn_type = m_db["struct_conn_type"];
+
+	// This will validate link_type :-)
+	if (not struct_conn_type.contains("id"_key == link_type))
+		struct_conn_type.emplace({ { "id", link_type } });
+
+	std::string link_id = struct_conn.get_unique_id(link_type + '_');
+
+	item label_seq_id_1("ptnr1_label_seq_id");
+	if (int nr = a1.get_label_seq_id(); nr != 0)
+		label_seq_id_1.value(std::to_string(nr));
+
+	item label_seq_id_2("ptnr2_label_seq_id");
+	if (int nr = a2.get_label_seq_id(); nr != 0)
+		label_seq_id_2.value(std::to_string(nr));
+
+	struct_conn.emplace(
+		{ //
+			{ "id", link_id },
+			{ "conn_type_id", link_type },
+			{ "pdbx_leaving_atom_flag", "one" },
+
+			{ "ptnr1_label_asym_id", a1.get_label_asym_id() },
+			{ "ptnr1_label_comp_id", a1.get_label_comp_id() },
+			label_seq_id_1,
+			{ "ptnr1_label_atom_id", a1.get_label_atom_id() },
+			{ "pdbx_ptnr1_label_alt_id", a1.get_label_alt_id() },
+			{ "pdbx_ptnr1_PDB_ins_code", a1.get_pdb_ins_code() },
+			{ "ptnr1_auth_asym_id", a1.get_auth_asym_id() },
+			{ "ptnr1_auth_comp_id", a1.get_auth_comp_id() },
+			{ "ptnr1_auth_seq_id", a1.get_auth_seq_id() },
+			{ "ptnr1_symmetry", a1.symmetry() },
+
+			{ "ptnr2_label_asym_id", a2.get_label_asym_id() },
+			{ "ptnr2_label_comp_id", a2.get_label_comp_id() },
+			label_seq_id_2,
+			{ "ptnr2_label_atom_id", a2.get_label_atom_id() },
+			{ "pdbx_ptnr2_label_alt_id", a2.get_label_alt_id() },
+			{ "pdbx_ptnr2_PDB_ins_code", a2.get_pdb_ins_code() },
+			{ "ptnr2_auth_asym_id", a2.get_auth_asym_id() },
+			{ "ptnr2_auth_comp_id", a2.get_auth_comp_id() },
+			{ "ptnr2_auth_seq_id", a2.get_auth_seq_id() },
+			{ "ptnr2_symmetry", a2.symmetry() },
+
+			{ "pdbx_dist_value", distance(a1.get_location(), a2.get_location()), 3 },
+			{ "pdbx_role", role } });
+
+	return link_id;
+}
+
 branch &structure::create_branch()
 {
 	auto &entity = m_db["entity"];
@@ -2845,8 +2921,8 @@ static int compare_numbers(std::string_view a, std::string_view b)
 
 	std::from_chars_result ra, rb;
 
-	ra = selected_charconv<double>::from_chars(a.data(), a.data() + a.length(), da);
-	rb = selected_charconv<double>::from_chars(b.data(), b.data() + b.length(), db);
+	ra = from_chars(a.data(), a.data() + a.length(), da);
+	rb = from_chars(b.data(), b.data() + b.length(), db);
 
 	if (not(bool) ra.ec and not(bool) rb.ec)
 	{

src/parser.cpp

@@ -345,6 +345,7 @@ sac_parser::CIFToken sac_parser::get_next_token()
 				{
 					retract();
 					result = CIFToken::VALUE;
+					m_token_value = "?";
 				}
 				else
 					state = State::Value;

src/pdb/pdb2cif.cpp

@@ -26,21 +26,17 @@
 
 #include "pdb2cif_remark_3.hpp"
 
-#include "cif++.hpp"
+#include <cif++/compound.hpp>
+#include <cif++/gzio.hpp>
+#include <cif++/model.hpp>
+#include <cif++/pdb.hpp>
+#include <cif++/symmetry.hpp>
 
-#include <iomanip>
 #include <map>
 #include <set>
 #include <stack>
 #include <stdexcept>
 
-using cif::category;
-using cif::datablock;
-using cif::iequals;
-using cif::key;
-using cif::to_lower;
-using cif::to_lower_copy;
-
 // --------------------------------------------------------------------
 // attempt to come up with better error handling
 
@@ -3142,7 +3138,6 @@ void PDBFileParser::ParseRemark350()
 	std::map<std::string, std::string> values;
 	std::vector<std::string> asymIdList;
 	std::smatch m;
-	cif::row_handle genR;
 
 	std::vector<double> mat, vec;
 
@@ -6375,7 +6370,7 @@ void read_pdb_file(std::istream &pdbFile, cif::file &cifFile)
 	{
 		cifFile.front().load_dictionary();
 		if (cifFile.front().get_validator() == nullptr)
-			cifFile.front().set_validator(&validator_factory::instance().get("mmcif_pdbx.dic"));
+			cifFile.front().set_validator(validator_factory::instance().get("mmcif_pdbx.dic"));
 
 		if (not cifFile.is_valid() and cif::VERBOSE >= 0)
 			std::cerr << "Resulting mmCIF file is not valid!\n";
@@ -6445,7 +6440,7 @@ file read(std::istream &is)
 	{
 		auto &db = result.front();
 		if (db.get_validator() == nullptr)
-			db.set_validator(&validator_factory::instance().get("mmcif_pdbx.dic"));
+			db.set_validator(validator_factory::instance().get("mmcif_pdbx.dic"));
 		if (db.is_valid())
 			db.get_validator()->fill_audit_conform(db["audit_conform"]);
 	}

src/pdb/pdb2cif_remark_3.cpp

@@ -1478,7 +1478,7 @@ bool Remark3Parser::parse(const std::string &expMethod, PDBRecord *r, cif::datab
 
 		best.parser->fixup();
 
-		auto &validator = cif::validator_factory::instance().get("mmcif_pdbx.dic");
+		auto &validator = cif::validator_factory::instance()["mmcif_pdbx.dic"];
 
 		for (auto &cat1 : best.parser->mDb)
 		{

src/pdb/reconstruct.cpp

@@ -25,8 +25,12 @@
  */
 
 #include "cif++.hpp"
+#include "cif++/compound.hpp"
 #include "cif++/row.hpp"
 
+#include <stdexcept>
+#include <string>
+
 // --------------------------------------------------------------------
 
 namespace cif::pdb
@@ -186,6 +190,8 @@ void checkEntities(datablock &db)
 
 void createEntityIDs(datablock &db)
 {
+	using namespace literals;
+
 	// Suppose the file does not have entity ID's. We have to make up some
 
 	// walk the atoms. For each auth_asym_id we have a new struct_asym.
@@ -197,28 +203,44 @@ void createEntityIDs(datablock &db)
 	// that should cover it
 
 	auto &atom_site = db["atom_site"];
+	auto &entity = db["entity"];
 	auto &cf = compound_factory::instance();
 
-	std::vector<std::vector<residue_key_type>> entities;
+	std::vector<std::vector<row_handle>> entities;
 
 	std::string lastAsymID;
 	int lastSeqID = -1;
-	std::vector<residue_key_type> waters;
+	std::vector<row_handle> waters;
 
-	for (residue_key_type k : atom_site.rows<std::optional<std::string>,
-			 std::optional<int>,
-			 std::optional<std::string>,
-			 std::optional<std::string>,
-			 std::optional<int>,
-			 std::optional<std::string>>(
-			 "auth_asym_id", "auth_seq_id", "auth_comp_id",
-			 "label_asym_id", "label_seq_id", "label_comp_id"))
+	int nextEntityID;
+	try
 	{
+		if (entity.empty())
+			nextEntityID = 1;
+		else
+			nextEntityID = entity.find_max<int>("id") + 1;
+	}
+	catch (...)
+	{
+		nextEntityID = 1;
+	}
+
+	for (auto rh : atom_site.find("label_entity_id"_key == cif::null))
+	{
+		residue_key_type k = rh.get<std::optional<std::string>,
+			std::optional<int>,
+			std::optional<std::string>,
+			std::optional<std::string>,
+			std::optional<int>,
+			std::optional<std::string>>(
+			"auth_asym_id", "auth_seq_id", "auth_comp_id",
+			"label_asym_id", "label_seq_id", "label_comp_id");
+
 		std::string comp_id = get_comp_id(k);
 
 		if (cf.is_water(comp_id))
 		{
-			waters.emplace_back(k);
+			waters.emplace_back(rh);
 			continue;
 		}
 
@@ -227,19 +249,20 @@ void createEntityIDs(datablock &db)
 
 		bool is_monomer = cf.is_monomer(comp_id);
 
-		if (lastAsymID == asym_id and lastSeqID == seq_id and not is_monomer)
-			continue;
+		// if (lastAsymID == asym_id and lastSeqID == seq_id and not is_monomer)
+		// 	continue;
 
 		if (asym_id != lastAsymID or (not is_monomer and lastSeqID != seq_id))
 			entities.push_back({});
 
-		entities.back().emplace_back(k);
+		entities.back().emplace_back(rh);
 
 		lastAsymID = asym_id;
 		lastSeqID = seq_id;
 	}
 
 	std::map<std::size_t, std::string> entity_ids;
+	std::map<std::string, std::string> newEntitiesForCompound;
 
 	atom_site.add_item("label_entity_id");
 
@@ -248,7 +271,39 @@ void createEntityIDs(datablock &db)
 		if (entity_ids.contains(i))
 			continue;
 
-		auto entity_id = std::to_string(i + 1);
+		residue_key_type k = entities[i].front().get<std::optional<std::string>, std::optional<int>, std::optional<std::string>, std::optional<std::string>, std::optional<int>, std::optional<std::string>>(
+			"auth_asym_id", "auth_seq_id", "auth_comp_id",
+			"label_asym_id", "label_seq_id", "label_comp_id");
+
+		std::string comp_id = get_comp_id(k);
+
+		std::string entity_id;
+		if (auto v = db["pdbx_entity_nonpoly"].find_first("comp_id"_key == comp_id); v)
+			entity_id = v.get<std::string>("entity_id");
+		else if (auto i2 = newEntitiesForCompound.find(comp_id); i2 != newEntitiesForCompound.end())
+			entity_id = i2->second;
+		else
+		{
+			entity_id = std::to_string(nextEntityID++);
+
+			if (cf.is_monomer(comp_id))
+				entity.emplace({ //
+					{ "id", entity_id },
+					{ "type", "polymer" } });
+			else if (cf.is_water(comp_id))
+				entity.emplace({ //
+					{ "id", entity_id },
+					{ "type", "water" } });
+			else
+			{
+				entity.emplace({ //
+					{ "id", entity_id },
+					{ "type", "non-polymer" } });
+
+				newEntitiesForCompound[comp_id] = entity_id;
+			}
+		}
+
 		entity_ids[i] = entity_id;
 
 		for (std::size_t j = i + 1; j < entities.size(); ++j)
@@ -260,20 +315,17 @@ void createEntityIDs(datablock &db)
 
 	for (std::size_t ix = 0; auto &e : entities)
 	{
-		auto k = e.front();
 		const auto &entity_id = entity_ids[ix++];
 
-		std::string comp_id = get_comp_id(k);
-
-		for (auto &k2 : e)
-			atom_site.update_value(get_condition(k2), "label_entity_id", entity_id);
+		for (auto rh : e)
+			rh["label_entity_id"] = entity_id;
 	}
 
 	if (not waters.empty())
 	{
 		std::string waterEntityID = std::to_string(entities.size() + 1);
-		for (auto &k : waters)
-			atom_site.update_value(get_condition(k), "label_entity_id", waterEntityID);
+		for (auto rh : waters)
+			rh["label_entity_id"] = waterEntityID;
 	}
 }
 
@@ -440,9 +492,38 @@ void checkAtomRecords(datablock &db)
 	if (atom_site.contains(key("label_seq_id") < 0))
 		fixNegativeSeqID(atom_site);
 
-	std::set<int> polymer_entities;
-	for (int id : db["entity"].find<int>("type"_key == "polymer", "id"))
-		polymer_entities.insert(id);
+	std::set<std::string> polymer_entities;
+	if (db["entity"].empty())
+	{
+		// No entity, so we have to guess the types based on the content of atom_site
+
+		std::string last_entity_id;
+		std::optional<int> last_label_seq_id, last_auth_seq_id;
+
+		std::set<std::string> entityIDs;
+		for (auto &[entity_id, label_comp_id, label_seq_id, auth_comp_id, auth_seq_id] :
+			atom_site.rows<std::string, std::string, std::optional<int>, std::string, std::optional<int>>(
+				"label_entity_id", "label_comp_id", "label_seq_id", "auth_comp_id", "auth_seq_id"))
+		{
+			if (cf.is_water(label_comp_id) or cf.is_water(auth_comp_id))
+				continue;
+
+			if (polymer_entities.contains(entity_id))
+				continue;
+
+			if (last_entity_id == entity_id and (last_label_seq_id != label_seq_id or last_auth_seq_id != auth_seq_id))
+				polymer_entities.emplace(entity_id);
+
+			last_entity_id = entity_id;
+			last_label_seq_id = label_seq_id;
+			last_auth_seq_id = auth_seq_id;
+		}
+	}
+	else
+	{
+		for (std::string id : db["entity"].find<std::string>("type"_key == "polymer", "id"))
+			polymer_entities.insert(id);
+	}
 
 	std::set<std::string> missingCompounds;
 
@@ -479,7 +560,7 @@ void checkAtomRecords(datablock &db)
 		if (missingCompounds.contains(comp_id))
 			continue;
 
-		bool is_polymer = polymer_entities.contains(row["label_entity_id"].as<int>());
+		bool is_polymer = polymer_entities.contains(row["label_entity_id"].as<std::string>());
 		auto compound = cf.create(comp_id);
 
 		if (not compound)
@@ -533,14 +614,20 @@ void checkAtomRecords(datablock &db)
 		if (is_polymer and row["label_seq_id"].empty() and cf.is_monomer(comp_id))
 			row["label_seq_id"] = std::to_string(seq_id);
 
-		if (row["label_atom_id"].empty())
-			row["label_atom_id"] = row["auth_atom_id"].text();
 		if (row["label_asym_id"].empty())
 			row["label_asym_id"] = row["auth_asym_id"].text();
+		else if (row["auth_asym_id"].empty())
+			row["auth_asym_id"] = row["label_asym_id"].text();
+
 		if (row["label_comp_id"].empty())
 			row["label_comp_id"] = row["auth_comp_id"].text();
+		else if (row["auth_comp_id"].empty())
+			row["auth_comp_id"] = row["label_comp_id"].text();
+
 		if (row["label_atom_id"].empty())
 			row["label_atom_id"] = row["auth_atom_id"].text();
+		else if (row["auth_atom_id"].empty())
+			row["auth_atom_id"] = row["label_atom_id"].text();
 
 		// Rewrite the coordinates and other items that look better in a fixed format
 		// Be careful not to nuke invalidly formatted data here
@@ -563,7 +650,7 @@ void checkAtomRecords(datablock &db)
 			{
 				char b[12];
 
-				if (auto [ptr, ec] = cif::to_chars(b, b + sizeof(b), v, cif::chars_format::fixed, prec); ec == std::errc{})
+				if (auto [ptr, ec] = std::to_chars(b, b + sizeof(b), v, std::chars_format::fixed, prec); ec == std::errc{})
 					row.assign(item_name, { b, static_cast<std::string::size_type>(ptr - b) }, false, false);
 			}
 		}
@@ -605,19 +692,24 @@ void checkAtomAnisotropRecords(datablock &db)
 
 	std::vector<row_handle> to_be_deleted;
 
+	std::map<int, row_handle> atoms;
+	for (auto rh : atom_site)
+		atoms[rh.get<int>("id")] = rh;
+
 	bool warnReplaceTypeSymbol = true;
 	for (auto row : atom_site_anisotrop)
 	{
-		auto parents = atom_site_anisotrop.get_parents(row, atom_site);
-		if (parents.size() != 1)
+		auto ai = atoms.find(row.get<int>("id"));
+
+		if (ai == atoms.end())
 		{
 			to_be_deleted.emplace_back(row);
 			continue;
 		}
 
-		// this happens sometimes (Phenix):
+		auto parent = ai->second;
 
-		auto parent = parents.front();
+		// this happens sometimes (Phenix):
 
 		if (row["type_symbol"].empty())
 			row["type_symbol"] = parent["type_symbol"].text();
@@ -638,8 +730,6 @@ void checkAtomAnisotropRecords(datablock &db)
 			row["pdbx_label_atom_id"] = parent["label_atom_id"].text();
 		if (row["pdbx_label_comp_id"].empty() and not parent["label_comp_id"].empty())
 			row["pdbx_label_comp_id"] = parent["label_comp_id"].text();
-		// if (row["pdbx_PDB_model_num"].empty() and not parent["pdbx_PDB_model_num"].empty())
-		// 	row["pdbx_PDB_model_num"] = parent["pdbx_PDB_model_num"].text();
 	}
 
 	if (not to_be_deleted.empty())
@@ -652,23 +742,53 @@ void checkAtomAnisotropRecords(datablock &db)
 	}
 }
 
-void createStructAsym(datablock &db)
+void checkStructAsym(datablock &db)
 {
 	auto &atom_site = db["atom_site"];
 	auto &struct_asym = db["struct_asym"];
 
-	for (const auto &[label_asym_id, entity_id] : atom_site.rows<std::string, std::string>("label_asym_id", "label_entity_id"))
+	if (struct_asym.empty())
 	{
-		if (label_asym_id.empty())
-			throw std::runtime_error("File contains atom_site records without a label_asym_id");
-		if (struct_asym.count(key("id") == label_asym_id) == 0)
+		for (const auto &[label_asym_id, entity_id] : atom_site.rows<std::string, std::string>("label_asym_id", "label_entity_id"))
 		{
-			struct_asym.emplace({
-				// clang-format off
-				{ "id", label_asym_id },
-				{ "entity_id", entity_id }
-				//clang-format on
-			});
+			if (label_asym_id.empty())
+				throw std::runtime_error("File contains atom_site records without a label_asym_id");
+			if (struct_asym.count(key("id") == label_asym_id) == 0)
+			{
+				struct_asym.emplace({
+					// clang-format off
+					{ "id", label_asym_id },
+					{ "entity_id", entity_id }
+					//clang-format on
+				});
+			}
+		}
+	}
+	else
+	{
+		for (const auto &[label_asym_id, entity_id] :
+			atom_site.rows<std::string, std::string>("label_asym_id", "label_entity_id"))
+		{
+			if (label_asym_id.empty())
+				throw std::runtime_error("File contains atom_site records without a label_asym_id");
+			
+			auto sa = struct_asym.find_first(key("id") == label_asym_id);
+			if (sa)
+			{
+				if (sa["entity_id"].empty())
+					sa.assign("entity_id", entity_id, false, true);
+				else if (sa.get<std::string>("entity_id") != entity_id)
+					throw std::runtime_error("Inconsistent entity ID's in struct_asym");
+			}
+			else
+			{
+				struct_asym.emplace({
+					// clang-format off
+					{ "id", label_asym_id },
+					{ "entity_id", entity_id }
+					//clang-format on
+				});
+			}
 		}
 	}
 }
@@ -739,8 +859,11 @@ void createEntity(datablock &db)
 			auto c = cf.create(first_comp_id);
 
 			type = "non-polymer";
-			desc = c->name();
-			weight = c->formula_weight();
+			if (c)
+			{
+				desc = c->name();
+				weight = c->formula_weight();
+			}
 		}
 		else
 		{
@@ -811,28 +934,28 @@ void createEntityPoly(datablock &db)
 			if (type != "other")
 			{
 				std::string c_type;
-				if (cf.is_base(comp_id))
+				if (auto i = compound_factory::kBaseMap.find(comp_id); i != compound_factory::kBaseMap.end())
 				{
 					c_type = "polydeoxyribonucleotide";
-					letter_can = compound_factory::kBaseMap.at(comp_id);
+
+					letter_can = i->second;
+
 					if (comp_id.length() == 1)
 						letter = letter_can;
 					else
-						letter = '(' + letter_can + ')';
+						letter = '(' + comp_id + ')';
 				}
-				else if (cf.is_peptide(comp_id))
+				else if (auto i2 = compound_factory::kAAMap.find(comp_id); i2 != compound_factory::kAAMap.end())
 				{
 					c_type = "polypeptide(L)";
-					letter = letter_can = compound_factory::kAAMap.at(comp_id);
+
+					letter = letter_can = i2->second;
 				}
 				else if (iequals(c->type(), "D-PEPTIDE LINKING"))
 				{
 					c_type = "polypeptide(D)";
 
 					letter_can = c->one_letter_code();
-					if (letter_can == 0)
-						letter_can = 'X';
-
 					letter = '(' + comp_id + ')';
 
 					non_std_linkage = true;
@@ -843,9 +966,6 @@ void createEntityPoly(datablock &db)
 					c_type = "polypeptide(L)";
 
 					letter_can = c->one_letter_code();
-					if (letter_can == 0)
-						letter_can = 'X';
-
 					letter = '(' + comp_id + ')';
 
 					non_std_monomer = true;
@@ -855,9 +975,6 @@ void createEntityPoly(datablock &db)
 					// c_type = "other";
 
 					letter_can = c->one_letter_code();
-					if (letter_can == 0)
-						letter_can = 'X';
-
 					letter = '(' + comp_id + ')';
 
 					non_std_monomer = true;
@@ -870,7 +987,7 @@ void createEntityPoly(datablock &db)
 			}
 
 			seq[auth_asym_id] += letter;
-			seq_can[auth_asym_id] += letter_can;
+			seq_can[auth_asym_id] += letter_can ? letter_can : 'X';
 
 			if (find(pdb_strand_ids.begin(), pdb_strand_ids.end(), auth_asym_id) == pdb_strand_ids.end())
 				pdb_strand_ids.emplace_back(auth_asym_id);
@@ -1347,9 +1464,9 @@ bool reconstruct_pdbx(file &file)
 	auto &db = file.front();
 
 	if (auto ac = db.get("audit_conform"); ac != nullptr)
-		return reconstruct_pdbx(file, validator_factory::instance().get(*ac));
+		return reconstruct_pdbx(file, validator_factory::instance()[*ac]);
 	else
-		return reconstruct_pdbx(file, validator_factory::instance().get("mmcif_pdbx.dic"));
+		return reconstruct_pdbx(file, validator_factory::instance()["mmcif_pdbx.dic"]);
 }
 
 bool reconstruct_pdbx(file &file, const validator &validator)
@@ -1390,7 +1507,7 @@ bool reconstruct_pdbx(file &file, const validator &validator)
 	checkChemCompRecords(db);
 
 	// If the data is really horrible, it might not contain entities
-	if (not db["atom_site"].find_first(key("label_entity_id") != null))
+	if (db["atom_site"].find_first(key("label_entity_id") == null))
 		createEntityIDs(db);
 
 	// Now see if atom records make sense at all
@@ -1534,9 +1651,8 @@ bool reconstruct_pdbx(file &file, const validator &validator)
 		checkAtomAnisotropRecords(db);
 
 	// Now create any missing categories
-	// Next make sure we have struct_asym records
-	if (auto cat = db.get("struct_asym"); cat == nullptr or cat->empty())
-		createStructAsym(db);
+	// Next make sure we have good struct_asym records
+	checkStructAsym(db);
 
 	if (auto cat = db.get("entity"); cat == nullptr or cat->empty())
 		createEntity(db);
@@ -1573,9 +1689,9 @@ void fixup_pdbx(file &file)
 	auto &db = file.front();
 
 	if (auto ac = db.get("audit_conform"); ac != nullptr)
-		fixup_pdbx(file, validator_factory::instance().get(*ac));
+		fixup_pdbx(file, validator_factory::instance()[*ac]);
 	else
-		fixup_pdbx(file, validator_factory::instance().get("mmcif_pdbx.dic"));
+		fixup_pdbx(file, validator_factory::instance()["mmcif_pdbx.dic"]);
 }
 
 void fixup_pdbx(file &file, const validator &validator)
@@ -1645,9 +1761,8 @@ void fixup_pdbx(file &file, const validator &validator)
 	db.set_validator(&validator);
 
 	// Now create any missing categories
-	// Next make sure we have struct_asym records
-	if (auto cat = db.get("struct_asym"); cat == nullptr or cat->empty())
-		createStructAsym(db);
+	// Next make sure we have good struct_asym records
+	checkStructAsym(db);
 
 	if (auto cat = db.get("entity"); cat == nullptr or cat->empty())
 		createEntity(db);

src/pdb/validate-pdbx.cpp

@@ -25,6 +25,7 @@
  */
 
 #include "cif++.hpp"
+#include "cif++/validate.hpp"
 
 namespace cif::pdb
 {
@@ -65,11 +66,18 @@ condition get_parents_condition(const validator &validator, row_handle rh, const
 	return result;
 }
 
+bool is_valid_pdbx_file(const file &file)
+{
+	std::error_code ec;
+	bool result = is_valid_pdbx_file(file, validator_factory::instance()["mmcif_pdbx.dic"], ec);
+	return result and not(bool) ec;
+}
+
 bool is_valid_pdbx_file(const file &file, const validator &v)
 {
 	std::error_code ec;
 	bool result = is_valid_pdbx_file(file, v, ec);
-	return result and not (bool)ec;
+	return result and not(bool) ec;
 }
 
 bool is_valid_pdbx_file(const file &file, std::error_code &ec)
@@ -79,10 +87,10 @@ bool is_valid_pdbx_file(const file &file, std::error_code &ec)
 	if (file.empty())
 		ec = make_error_code(validation_error::empty_file);
 	else if (auto ac = file.front().get("audit_conform"); ac != nullptr)
-		result = is_valid_pdbx_file(file, validator_factory::instance().get(*ac), ec);
+		result = is_valid_pdbx_file(file, validator_factory::instance()[*ac], ec);
 	else
-		result = is_valid_pdbx_file(file, validator_factory::instance().get("mmcif_pdbx.dic"), ec);
-	
+		result = is_valid_pdbx_file(file, validator_factory::instance()["mmcif_pdbx.dic"], ec);
+
 	return result;
 }
 
@@ -167,7 +175,7 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 
 			const auto entity_poly_type = entity_poly.find1<std::string>("entity_id"_key == entity_id, "type");
 
-			std::map<int,std::set<std::string>> mon_per_seq_id;
+			std::map<int, std::set<std::string>> mon_per_seq_id;
 
 			for (const auto &[num, mon_id, hetero] : entity_poly_seq.find<int, std::string, bool>("entity_id"_key == entity_id, "num", "mon_id", "hetero"))
 			{
@@ -202,28 +210,37 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 					throw std::runtime_error("Mismatch between the hetero flag in the poly seq schemes and the number residues per seq_id");
 			}
 
-			for (const auto &[seq_id, mon_ids] : mon_per_seq_id)
-			{
-				for (auto asym_id : struct_asym.find<std::string>("entity_id"_key == entity_id, "id"))
-				{
-					condition cond;
-					
-					for (auto mon_id : mon_ids)
-						cond = std::move(cond) or "label_comp_id"_key == mon_id;
+			// This code proved to take too much time ...
 
-					cond = "label_entity_id"_key == entity_id and
-						"label_asym_id"_key == asym_id and
-						"label_seq_id"_key == seq_id and not std::move(cond);
-					
-					if (atom_site.contains(std::move(cond)))
-						throw std::runtime_error("An atom_site record exists that has no parent in the poly seq scheme categories");
-				}
+			// for (const auto &[seq_id, mon_ids] : mon_per_seq_id)
+			// {
+			// 	for (auto asym_id : struct_asym.find<std::string>("entity_id"_key == entity_id, "id"))
+			// 	{
+			// 		condition cond;
+
+			// 		for (auto mon_id : mon_ids)
+			// 			cond = std::move(cond) or "label_comp_id"_key == mon_id;
+
+			// 		cond = "label_entity_id"_key == entity_id and
+			// 			"label_asym_id"_key == asym_id and
+			// 			"label_seq_id"_key == seq_id and not std::move(cond);
+
+			// 		if (atom_site.contains(std::move(cond)))
+			// 			throw std::runtime_error("An atom_site record exists that has no parent in the poly seq scheme categories");
+			// 	}
+			// }
+
+			// ... so we're using this instead, should be almost the same...
+
+			for (const auto &[comp_id, seq_id] :
+				atom_site.find<std::string, int>("label_entity_id"_key == entity_id, "label_comp_id", "label_seq_id"))
+			{
+				if (not mon_per_seq_id[seq_id].contains(comp_id))
+					throw std::runtime_error("An atom_site record exists that has no parent in the poly seq scheme categories");
 			}
 
 			auto &&[seq, seq_can] = entity_poly.find1<std::optional<std::string>, std::optional<std::string>>("entity_id"_key == entity_id,
 				"pdbx_seq_one_letter_code", "pdbx_seq_one_letter_code_can");
-			
-			std::string::const_iterator si, sci, se, sce;
 
 			auto seq_match = [&](bool can, std::string::const_iterator si, std::string::const_iterator se)
 			{
@@ -260,8 +277,8 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 							else
 								letter = '(' + comp_id + ')';
 						}
-						
-						if (iequals(std::string{si, si + letter.length()}, letter))
+
+						if (iequals(std::string{ si, si + letter.length() }, letter))
 						{
 							match = true;
 							si += letter.length();
@@ -285,7 +302,9 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 			}
 			else
 			{
-				seq->erase(std::remove_if(seq->begin(), seq->end(), [](char ch) { return std::isspace(ch); }), seq->end());
+				seq->erase(std::remove_if(seq->begin(), seq->end(), [](char ch)
+							   { return std::isspace(ch); }),
+					seq->end());
 
 				if (not seq_match(false, seq->begin(), seq->end()))
 					throw std::runtime_error("Sequences do not match for entity " + entity_id);
@@ -298,7 +317,9 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 			}
 			else
 			{
-				seq_can->erase(std::remove_if(seq_can->begin(), seq_can->end(), [](char ch) { return std::isspace(ch); }), seq_can->end());
+				seq_can->erase(std::remove_if(seq_can->begin(), seq_can->end(), [](char ch)
+								   { return std::isspace(ch); }),
+					seq_can->end());
 
 				if (not seq_match(true, seq_can->begin(), seq_can->end()))
 					throw std::runtime_error("Canonical sequences do not match for entity " + entity_id);
@@ -315,11 +336,10 @@ bool is_valid_pdbx_file(const file &file, const validator &validator, std::error
 		ec = make_error_code(validation_error::not_valid_pdbx);
 	}
 
-	if (not result and (bool)ec)
+	if (not result and (bool) ec)
 		ec = make_error_code(validation_error::not_valid_pdbx);
 
 	return result;
 }
 
 } // namespace cif::pdb
-  

src/point.cpp

@@ -25,17 +25,19 @@
  */
 
 #include "cif++/point.hpp"
-#include "cif++/matrix.hpp"
 
-#include <cassert>
-#include <random>
+#include "cif++/matrix.hpp" // for matrix_subtraction, matrix_cofactors
+
+#include <initializer_list>
+#include <random> // for uniform_real_distribution, normal_distri...
+#include <stdexcept>
 
 namespace cif
 {
 
 // --------------------------------------------------------------------
 
-template<typename T>
+template <typename T>
 quaternion_type<T> normalize(quaternion_type<T> q)
 {
 	std::valarray<double> t(4);
@@ -126,10 +128,9 @@ quaternion construct_for_dihedral_angle(point p1, point p2, point p3, point p4,
 	p4 -= p3;
 	p3 -= p3;
 
-	quaternion q;
 	auto axis = -p2;
-
 	float dh = dihedral_angle(p1, p2, p3, p4);
+
 	return construct_from_angle_axis(angle - dh, axis);
 }
 
@@ -293,9 +294,9 @@ quaternion align_points(const std::vector<point> &pa, const std::vector<point> &
 	}
 
 	quaternion q(
-		static_cast<float>(cf(maxR, 0)), 
-		static_cast<float>(cf(maxR, 1)), 
-		static_cast<float>(cf(maxR, 2)), 
+		static_cast<float>(cf(maxR, 0)),
+		static_cast<float>(cf(maxR, 1)),
+		static_cast<float>(cf(maxR, 2)),
 		static_cast<float>(cf(maxR, 3)));
 	q = normalize(q);
 
@@ -327,4 +328,251 @@ point nudge(point p, float offset)
 	return p + r;
 }
 
+// --------------------------------------------------------------------
+
+std::tuple<point, float> smallest_sphere_around_2_points(std::array<cif::point, 2> pts)
+{
+	return { (pts[0] + pts[1]) / 2, distance(pts[0], pts[1]) / 2 };
+}
+
+std::tuple<point, float> smallest_sphere_around_3_points(std::array<cif::point, 3> pts)
+{
+	// Find two bisectors
+	auto vz = cross_product(pts[1] - pts[0], pts[2] - pts[0]);
+
+	auto bs1 = cross_product(vz, pts[1] - pts[0]);
+	bs1.normalize();
+
+	auto v1 = (pts[1] - pts[0]);
+	v1.normalize();
+
+	auto s1 = pts[0] + (distance(pts[1], pts[0]) / 2) * v1;
+
+	auto bs2 = cross_product(vz, pts[2] - pts[0]);
+	bs2.normalize();
+
+	auto v2 = (pts[2] - pts[0]);
+	v2.normalize();
+
+	auto s2 = pts[0] + (distance(pts[2], pts[0]) / 2) * v2;
+
+	auto c = line_line_intersection(s1, s1 + bs1, s2, s2 + bs2);
+	if (c)
+		return { *c, distance(*c, pts[0]) };
+
+	// Colinear points I guess, try something else
+	auto l1 = distance_squared(pts[0], pts[1]);
+	auto l2 = distance_squared(pts[0], pts[2]);
+	auto l3 = distance_squared(pts[1], pts[2]);
+
+	if (l1 > l2 and l1 > l3)
+		return smallest_sphere_around_2_points({ pts[0], pts[1] });
+	else if (l2 > l1 and l2 > l3)
+		return smallest_sphere_around_2_points({ pts[0], pts[2] });
+	else
+		return smallest_sphere_around_2_points({ pts[1], pts[2] });
+}
+
+std::tuple<point, float> smallest_sphere_around_4_points(std::array<cif::point, 4> pts)
+{
+	auto t0 = -norm_squared(pts[0]);
+	auto t1 = -norm_squared(pts[1]);
+	auto t2 = -norm_squared(pts[2]);
+	auto t3 = -norm_squared(pts[3]);
+
+	// clang-format off
+	matrix4x4<float> Tm({
+		pts[0].m_x, pts[0].m_y, pts[0].m_z, 1,
+		pts[1].m_x, pts[1].m_y, pts[1].m_z, 1,
+		pts[2].m_x, pts[2].m_y, pts[2].m_z, 1,
+		pts[3].m_x, pts[3].m_y, pts[3].m_z, 1
+	});
+	auto T = determinant(Tm);
+
+	if (T != 0)
+	{
+		matrix4x4<float> Dm({
+			t0, pts[0].m_y, pts[0].m_z, 1,
+			t1, pts[1].m_y, pts[1].m_z, 1,
+			t2, pts[2].m_y, pts[2].m_z, 1,
+			t3, pts[3].m_y, pts[3].m_z, 1
+		});
+		auto D = determinant(Dm) / T;
+		
+		matrix4x4<float> Em({
+			pts[0].m_x, t0, pts[0].m_z, 1,
+			pts[1].m_x, t1, pts[1].m_z, 1,
+			pts[2].m_x, t2, pts[2].m_z, 1,
+			pts[3].m_x, t3, pts[3].m_z, 1
+		});
+		auto E = determinant(Em) / T;
+		
+		matrix4x4<float> Fm({
+			pts[0].m_x, pts[0].m_y, t0, 1,
+			pts[1].m_x, pts[1].m_y, t1, 1,
+			pts[2].m_x, pts[2].m_y, t2, 1,
+			pts[3].m_x, pts[3].m_y, t3, 1
+		});
+		
+		auto F = determinant(Fm) / T;
+		
+		matrix4x4<float> Gm({
+			pts[0].m_x, pts[0].m_y, pts[0].m_z, t0,
+			pts[1].m_x, pts[1].m_y, pts[1].m_z, t1,
+			pts[2].m_x, pts[2].m_y, pts[2].m_z, t2,
+			pts[3].m_x, pts[3].m_y, pts[3].m_z, t3
+		});
+		auto G = determinant(Gm) / T;
+		
+		point center{ -D / 2, -E / 2, -F / 2 };
+		float radius = std::sqrt(D * D + E * E + F * F - 4 * G) / 2;
+
+		// clang-format on
+
+		return { center, radius };
+	}
+
+	// Perhaps some colinear points, try something else:
+
+	for (auto ix : std::initializer_list<std::array<size_t, 4>>{
+			 { 1, 2, 3, 0 },
+			 { 0, 2, 3, 1 },
+			 { 0, 1, 3, 2 },
+			 { 0, 1, 2, 3 },
+		 })
+	{
+		auto [center, radius] =
+			smallest_sphere_around_3_points({ pts[ix[0]], pts[ix[1]], pts[ix[2]] });
+
+		if (distance(pts[ix[3]], center) <= radius)
+			return { center, radius };
+	}
+
+	assert(false);
+	exit(1);
+}
+
+std::tuple<point, float> smallest_sphere_around_all_points(std::vector<point> P, std::vector<point> R)
+{
+	if (P.empty() or R.size() == 4)
+	{
+		switch (R.size())
+		{
+			case 1:
+				return { R[0], 0 };
+
+			case 2:
+				return smallest_sphere_around_2_points({ R[0], R[1] });
+
+			case 3:
+				return smallest_sphere_around_3_points({ R[0], R[1], R[2] });
+
+			case 4:
+				return smallest_sphere_around_4_points({ R[0], R[1], R[2], R[3] });
+
+			default:
+				assert(false);
+		}
+	}
+
+	auto p = P.back();
+	P.pop_back();
+
+	auto [c, r] = smallest_sphere_around_all_points(P, R);
+	assert(not std::isnan(r));
+	if (distance(c, p) <= r)
+		return { c, r };
+
+	R.emplace_back(p);
+	return smallest_sphere_around_all_points(P, R);
+}
+
+bool point_in_circle(point p, std::vector<point> c)
+{
+	switch (c.size())
+	{
+		case 0:
+			return false;
+
+		case 1:
+			return p == c.front();
+
+		case 2:
+		{
+			auto [center, radius] = smallest_sphere_around_2_points({ c[0], c[1] });
+			return cif::distance_squared(p, center) <= radius * radius;
+		}
+
+		case 3:
+		{
+			auto [center, radius] = smallest_sphere_around_3_points({ c[0], c[1], c[2] });
+			return cif::distance_squared(p, center) <= radius * radius;
+		}
+
+		case 4:
+		{
+			auto [center, radius] = smallest_sphere_around_4_points({ c[0], c[1], c[2], c[3] });
+			return cif::distance_squared(p, center) <= radius * radius;
+		}
+
+		default:
+			assert(false);
+			throw std::runtime_error("Error finding smallest sphere");
+	}
+}
+
+std::tuple<point, float> smallest_sphere_around_points(std::vector<point> pts)
+{
+	std::random_device rd;
+	std::mt19937 g(rd());
+
+	std::shuffle(pts.begin(), pts.end(), g);
+
+	std::vector<size_t> cix;
+
+	auto cirle_points = [&]()
+	{
+		std::vector<point> result;
+		for (auto ix : cix)
+			result.emplace_back(pts[ix]);
+		return result;
+	};
+
+	size_t i = 0;
+	while (i < pts.size())
+	{
+		if (std::find(cix.begin(), cix.end(), i) != cix.end() or
+			point_in_circle(pts[i], cirle_points()))
+		{
+			++i;
+		}
+		else
+		{
+			cix.erase(std::remove_if(cix.begin(), cix.end(), [i](size_t j)
+						  { return j < i; }),
+				cix.end());
+			cix.push_back(i);
+			if (cix.size() < 4)
+				i = 0;
+			else
+				++i;
+		}
+	}
+
+	switch (cix.size())
+	{
+		case 1:
+			return { pts[cix[0]], 0 };
+		case 2:
+			return smallest_sphere_around_2_points({ pts[cix[0]], pts[cix[1]] });
+		case 3:
+			return smallest_sphere_around_3_points({ pts[cix[0]], pts[cix[1]], pts[cix[2]] });
+		case 4:
+			return smallest_sphere_around_4_points({ pts[cix[0]], pts[cix[1]], pts[cix[2]], pts[cix[3]] });
+		default:
+			assert(false);
+			throw std::runtime_error("Error finding smallest sphere");
+	}
+}
+
 } // namespace cif

src/text.cpp

@@ -28,6 +28,11 @@
 
 #include <algorithm>
 #include <cassert>
+#include <charconv>
+
+#if __has_include("fast_float/fast_float.h")
+#include "fast_float/fast_float.h"
+#endif
 
 namespace cif
 {
@@ -512,4 +517,32 @@ std::vector<std::string> word_wrap(const std::string &text, std::size_t width)
 	return result;
 }
 
+#if __has_include("fast_float/fast_float.h")
+
+template <typename T>
+std::from_chars_result ff_charconv<T, typename std::enable_if_t<std::is_floating_point_v<T>>>::from_chars(const char *a, const char *b, T &v)
+{
+	auto r = fast_float::from_chars(a, b, v);
+	return { r.ptr, r.ec };
+}
+
+template struct ff_charconv<float>;
+template struct ff_charconv<double>;
+// template struct ff_charconv<long double>;
+
+#ifdef __STDCPP_FLOAT64_T__
+template struct ff_charconv<std::float64_t>;
+#endif
+#ifdef __STDCPP_FLOAT32_T__
+template struct ff_charconv<std::float32_t>;
+#endif
+#ifdef __STDCPP_FLOAT16_T__
+template struct ff_charconv<std::float16_t>;
+#endif
+#ifdef __STDCPP_BFLOAT16_T__
+template struct ff_charconv<std::bfloat16_t>;
+#endif
+
+#endif
+
 } // namespace cif

src/utilities.cpp

@@ -34,14 +34,20 @@
 #include <condition_variable>
 #include <cstring>
 #include <deque>
+#include <format>
 #include <fstream>
-#include <functional>
 #include <iomanip>
 #include <iostream>
 #include <map>
 #include <mutex>
 #include <sstream>
+#include <string>
 #include <thread>
+#include <utility>
+
+#if __cpp_lib_jthread >= 201911L
+#include <stop_token>
+#endif
 
 namespace fs = std::filesystem;
 
@@ -65,27 +71,50 @@ std::string get_version_nr()
 
 #if defined(_WIN32) or defined(__MINGW32__)
 }
-#include <windows.h>
-#include <libloaderapi.h>
-#include <wincon.h>
-
-#include <codecvt>
+// clang-format off
+# include <windows.h>
+# include <libloaderapi.h>
+# include <wincon.h>
+// clang-format on
 
 namespace cif
 {
 
 uint32_t get_terminal_width()
 {
-    CONSOLE_SCREEN_BUFFER_INFO csbi;
-    ::GetConsoleScreenBufferInfo(::GetStdHandle(STD_OUTPUT_HANDLE), &csbi);
-    return csbi.srWindow.Right - csbi.srWindow.Left + 1;
+	CONSOLE_SCREEN_BUFFER_INFO csbi;
+	return ::GetConsoleScreenBufferInfo(::GetStdHandle(STD_OUTPUT_HANDLE), &csbi)
+	           ? csbi.srWindow.Right - csbi.srWindow.Left + 1
+	           : 80;
+}
+
+void write_to_console(const std::string &s)
+{
+	auto h = ::GetStdHandle(STD_OUTPUT_HANDLE);
+	CONSOLE_SCREEN_BUFFER_INFO csbi;
+
+	if (auto l = ::MultiByteToWideChar(CP_UTF8, 0, s.data(), s.length(), nullptr, 0);
+		l > 0 and ::GetConsoleScreenBufferInfo(::GetStdHandle(STD_OUTPUT_HANDLE), &csbi))
+	{
+		std::u16string ws(l, 0);
+
+		::MultiByteToWideChar(CP_UTF8, 0, s.data(), s.length(), (LPWSTR)ws.data(), l);
+
+		DWORD w;
+		::WriteConsoleW(h, ws.data(), ws.length(), &w, nullptr);
+	}
+	else
+	{
+		std::cout.write(s.data(), s.length());
+		std::cout.flush();
+	}
 }
 
 #else
 
-#include <sys/ioctl.h>
-#include <termios.h>
-#include <limits.h>
+# include <limits.h>
+# include <sys/ioctl.h>
+# include <termios.h>
 
 uint32_t get_terminal_width()
 {
@@ -100,59 +129,220 @@ uint32_t get_terminal_width()
 	return result;
 }
 
+inline void write_to_console(const std::string &s)
+{
+	std::cout << s << std::flush;
+}
+
 #endif
 
 // --------------------------------------------------------------------
 
 struct progress_bar_impl
 {
-	progress_bar_impl(int64_t inMax, const std::string &inAction)
-		: m_max_value(inMax)
+	progress_bar_impl(uint64_t max_value, const std::string &message)
+		: m_max_value(max_value)
 		, m_consumed(0)
-		, m_action(inAction)
-		, m_message(inAction)
-		, m_thread(std::bind(&progress_bar_impl::run, this))
+		, m_action(message)
+		, m_message(message)
 	{
 	}
 
-	progress_bar_impl(const progress_bar_impl&) = delete;
-	progress_bar_impl &operator=(const progress_bar_impl &) = delete;
+	virtual ~progress_bar_impl() {}
 
-	~progress_bar_impl();
-
-	void run();
-
-	void consumed(int64_t n);
-	void progress(int64_t p);
-	void message(const std::string &msg);
-
-	void print_progress();
-	void print_done();
+	virtual void consumed(uint64_t n);
+	virtual void progress(uint64_t p);
+	virtual void message(const std::string &msg);
+	virtual void print_done();
 
 	using time_point = std::chrono::time_point<std::chrono::system_clock>;
 
-	int64_t m_max_value;
-	std::atomic<int64_t> m_consumed;
-	int64_t m_last_consumed = 0;
-	int m_spinner_index = 0;
+	uint64_t m_max_value;
+	std::atomic<uint64_t> m_consumed;
 	std::string m_action, m_message;
-	std::mutex m_mutex;
-	std::thread m_thread;
 	time_point m_start = std::chrono::system_clock::now();
-	time_point m_last = std::chrono::system_clock::now();
-	bool m_stop = false;
 };
 
-progress_bar_impl::~progress_bar_impl()
+void progress_bar_impl::consumed(uint64_t n)
+{
+	m_consumed += n;
+}
+
+void progress_bar_impl::progress(uint64_t p)
+{
+	m_consumed = p;
+}
+
+void progress_bar_impl::message(const std::string &msg)
+{
+	m_message = msg;
+}
+
+void progress_bar_impl::print_done()
+{
+	std::chrono::duration<double> elapsed = std::chrono::system_clock::now() - m_start;
+	std::string days, hours, minutes, seconds;
+
+	uint64_t s = static_cast<uint64_t>(std::trunc(elapsed.count()));
+	if (s > 24 * 60 * 60)
+	{
+		days = std::format("{:d}d ", s / (24 * 60 * 60));
+		s %= 24 * 60 * 60;
+	}
+
+	if (s > 60 * 60)
+	{
+		hours = std::format("{:2d}h ", s / (60 * 60));
+		s %= 60 * 60;
+	}
+
+	if (s > 60)
+	{
+		minutes = std::format("{:2d}m ", s / 60);
+		s %= 60;
+	}
+
+	std::string msg = std::format("{} done in {}{}{}{:.1f}s", m_action, days, hours, minutes, s + 1e-6 * (elapsed.count() - s));
+
+	uint32_t width = get_terminal_width();
+
+	if (msg.length() < width)
+		msg += std::string(width - msg.length(), ' ');
+
+	write_to_console(msg += '\n');
+}
+
+// --------------------------------------------------------------------
+
+struct simple_progress_bar_impl : public progress_bar_impl
+{
+	simple_progress_bar_impl(uint64_t max_value, const std::string &message)
+		: progress_bar_impl(max_value, message)
+	{
+	}
+
+	~simple_progress_bar_impl()
+	{
+		if (m_printed_any)
+			print_done();
+	}
+
+	void consumed(uint64_t n) override
+	{
+		using namespace std::literals;
+
+		progress_bar_impl::consumed(n);
+
+		// print at most 10 steps, but only if it took long enough
+
+		int percentile = static_cast<int>(std::floor(10.f * m_consumed / m_max_value));
+		if (percentile > m_last_percentile and (m_printed_any or std::chrono::system_clock::now() - m_start >= 1s))
+		{
+			if (not std::exchange(m_printed_any, true))
+				write_to_console(m_action + ": ");
+
+			write_to_console(std::format("...{:d}0%", percentile));
+			m_last_percentile = percentile;
+		}
+	}
+
+	void progress(uint64_t p) override
+	{
+		consumed(p - m_consumed);
+	}
+
+	void print_done() override
+	{
+		if (m_printed_any)
+		{
+			write_to_console("\n");
+			progress_bar_impl::print_done();
+		}
+	}
+
+	bool m_printed_any = false;
+	int m_last_percentile = 0;
+};
+
+// --------------------------------------------------------------------
+
+struct fancy_progress_bar_impl : public progress_bar_impl
+{
+	fancy_progress_bar_impl(uint64_t max_value, const std::string &message)
+		: progress_bar_impl(max_value, message)
+		, m_thread(
+#if __cpp_lib_jthread >= 201911L
+			  [this](std::stop_token stoken)
+			  { this->run(stoken); }
+#else
+			  [this]()
+			  { this->run(); }
+#endif
+		  )
+	{
+	}
+
+	~fancy_progress_bar_impl();
+
+#if __cpp_lib_jthread >= 201911L
+	void run(std::stop_token stoken);
+#else
+	void run();
+#endif
+
+	void consumed(uint64_t n) override;
+	void progress(uint64_t p) override;
+	void message(const std::string &msg) override;
+
+	void print_progress();
+
+	std::mutex m_mutex;
+	std::condition_variable m_cv;
+
+	float m_progress;
+	uint32_t m_width, m_bar_width;
+	uint32_t m_steps, m_last_steps = 0;
+	uint64_t m_last_consumed = 0;
+#if __cpp_lib_jthread >= 201911L
+	std::jthread m_thread;
+#else
+	std::thread m_thread;
+	bool m_stop = false;
+#endif
+};
+
+const char *kBlocks[] = {
+	" ",
+	"▏",
+	"▎",
+	"▍",
+	"▌",
+	"▋",
+	"▊",
+	"▉",
+	"█",
+};
+
+const size_t kBlockCount = sizeof(kBlocks) / sizeof(void *) - 1;
+
+fancy_progress_bar_impl::~fancy_progress_bar_impl()
 {
 	using namespace std::literals;
 	assert(m_thread.joinable());
 
+#if __cpp_lib_jthread >= 201911L
+	m_thread.request_stop();
+#else
 	m_stop = true;
+#endif
 	m_thread.join();
 }
 
-void progress_bar_impl::run()
+#if __cpp_lib_jthread >= 201911L
+void fancy_progress_bar_impl::run(std::stop_token stoken)
+#else
+void fancy_progress_bar_impl::run()
+#endif
 {
 	using namespace std::literals;
 
@@ -160,25 +350,44 @@ void progress_bar_impl::run()
 
 	try
 	{
-		while (not m_stop)
+		for (;;)
 		{
+			std::unique_lock lock(m_mutex);
+
+			m_cv.wait_for(lock, 25ms);
+
+#if __cpp_lib_jthread >= 201911L
+			if (stoken.stop_requested())
+				break;
+#else
+			if (m_stop)
+				break;
+#endif
+
 			auto now = std::chrono::system_clock::now();
 
-			if (now - m_start < 2s or now - m_last < 100ms)
-			{
-				std::this_thread::sleep_for(10ms);
+			if (m_consumed == m_last_consumed or now - m_start < 1s)
 				continue;
-			}
 
-			std::lock_guard lock(m_mutex);
+			m_last_consumed = m_consumed;
 
-			if (not printedAny and isatty(STDOUT_FILENO))
-				std::cout << "\x1b[?25l";
+			// See if we need to do work
+			m_width = get_terminal_width();
+			m_progress = static_cast<float>(m_consumed) / m_max_value;
+			m_bar_width = 7 * m_width / 10; // 70% of the width of the terminal
+			m_steps = static_cast<uint32_t>(std::ceil(m_progress * m_bar_width * kBlockCount));
+
+			if (m_steps == m_last_steps)
+				continue;
+
+			m_last_steps = m_steps;
+
+			if (not printedAny)
+				write_to_console("\x1b[?25l");
 
 			print_progress();
 
 			printedAny = true;
-			m_last = std::chrono::system_clock::now();
 		}
 	}
 	catch (...)
@@ -187,161 +396,94 @@ void progress_bar_impl::run()
 
 	if (printedAny)
 	{
+		write_to_console("\r\x1b[?25h");
 		print_done();
-		if (isatty(STDOUT_FILENO))
-			std::cout << "\x1b[?25h";
 	}
 }
 
-void progress_bar_impl::consumed(int64_t n)
+void fancy_progress_bar_impl::consumed(uint64_t n)
 {
-	m_consumed += n;
+	progress_bar_impl::consumed(n);
+	// m_cv.notify_one();
 }
 
-void progress_bar_impl::progress(int64_t p)
+void fancy_progress_bar_impl::progress(uint64_t p)
 {
-	m_consumed = p;
+	progress_bar_impl::progress(p);
+	// m_cv.notify_one();
 }
 
-void progress_bar_impl::message(const std::string &msg)
+void fancy_progress_bar_impl::message(const std::string &msg)
 {
 	std::unique_lock lock(m_mutex);
-	m_message = msg;
+	progress_bar_impl::message(msg);
+	// m_cv.notify_one();
 }
 
-const char* kSpinner[] = {
-	// ".", "o", "O", "0", "O", "o", ".", " "
-	// "⢄", "⢂", "⢁", "⡁", "⡈", "⡐", "⡠"
-	 ".", "o", "O", "0", "@", "*", " "
-};
-
-const std::size_t kSpinnerCount = sizeof(kSpinner) / sizeof(char*);
-
-const int kSpinnerTimeInterval = 100;
-
 const uint32_t kMinBarWidth = 40, kMinMsgWidth = 12;
 
-void progress_bar_impl::print_progress()
+void fancy_progress_bar_impl::print_progress()
 {
-	const char *kBlocks[] = {
-		// "▯", // 0
-		// "▮", // 1
-		"=",
-		"-"
-	};
+	const uint32_t pct_width = 5;
+	uint32_t msg_width = m_width - m_bar_width - pct_width - 1;
 
-	uint32_t width = get_terminal_width();
-
-	float progress = static_cast<float>(m_consumed) / m_max_value;
-	
-	if (width < kMinBarWidth)
-		std::cout << (100 * progress) << "%\n";
-	else
+	if (msg_width < kMinMsgWidth)
 	{
-		uint32_t bar_width = 7 * width / 10;
-		uint32_t pct_width = 7;
-		uint32_t msg_width = width - bar_width - pct_width - 1;
+		m_bar_width += kMinMsgWidth - msg_width;
+		msg_width = kMinMsgWidth;
+	}
 
-		if (msg_width < kMinMsgWidth)
-		{
-			bar_width += kMinMsgWidth - msg_width;
-			msg_width = kMinMsgWidth;
-		}
+	std::string bar;
+	bar.reserve(m_bar_width * 4);
 
-		std::ostringstream msg;
-
-		if (m_message.length() <= msg_width)
-		{
-			msg << m_message;
-			if (m_message.length() < msg_width)
-				msg << std::string(msg_width - m_message.length(), ' ');
-		}
+	for (uint32_t i = 0; i < m_bar_width; ++i)
+	{
+		if (i * kBlockCount <= m_steps)
+			bar += kBlocks[kBlockCount];
+		else if (i * kBlockCount > m_steps + kBlockCount)
+			bar += kBlocks[0];
 		else
-			msg << m_message.substr(0, msg_width - 3) << "...";
-
-		msg << ' ';
-
-		uint32_t pi = static_cast<uint32_t>(std::ceil(progress * bar_width));
-
-		for (uint32_t i = 0; i < bar_width; ++i)
-			msg << kBlocks[i > pi ? 1 : 0];
-
-		msg << ' ';
-
-		msg << std::setw(3) << static_cast<int>(std::ceil(progress * 100)) << "% ";
-
-		auto now = std::chrono::system_clock::now();
-		m_spinner_index = (std::chrono::duration_cast<std::chrono::milliseconds>(now - m_start).count() / kSpinnerTimeInterval) % kSpinnerCount;
-
-		msg << kSpinner[m_spinner_index];
-
-		std::cout << '\r' << msg.str();
-		std::cout.flush();
+			bar += kBlocks[1 + m_steps % kBlockCount];
 	}
-}
 
-namespace
-{
-
-	std::ostream &operator<<(std::ostream &os, const std::chrono::duration<double> &t)
+	// make the bar more colorfull
+	struct color_type
 	{
-		uint64_t s = static_cast<uint64_t>(std::trunc(t.count()));
-		if (s > 24 * 60 * 60)
-		{
-			auto days = s / (24 * 60 * 60);
-			os << days << "d ";
-			s %= 24 * 60 * 60;
-		}
+		uint8_t r, g, b;
+	} fg{ 0, 3, 5 }, bg{ 0, 1, 2 };
 
-		if (s > 60 * 60)
-		{
-			auto hours = s / (60 * 60);
-			os << hours << "h ";
-			s %= 60 * 60;
-		}
+	auto esc_1 = std::format("\x1b[38;5;{}m\x1b[48;5;{}m",
+		16 + (fg.r * 36) + (fg.g * 6) + fg.b,
+		16 + (bg.r * 36) + (bg.g * 6) + bg.b);
+	std::string esc_2("\x1b[0m");
 
-		if (s > 60)
-		{
-			auto minutes = s / 60;
-			os << minutes << "m ";
-			s %= 60;
-		}
+	bar = esc_1 + bar + esc_2;
 
-		double ss = s + 1e-6 * (t.count() - s);
+	std::string msg = m_message.length() <= msg_width
+	                      ? m_message
+	                      : m_message.substr(0, msg_width - 3) + "...";
 
-		os << std::fixed << std::setprecision(1) << ss << 's';
-
-		return os;
-	}
-
-} // namespace
-
-void progress_bar_impl::print_done()
-{
-	std::chrono::duration<double> elapsed = std::chrono::system_clock::now() - m_start;
-
-	std::ostringstream msgstr;
-	msgstr << m_action << " done in " << elapsed << " seconds";
-	auto msg = msgstr.str();
-
-	uint32_t width = get_terminal_width();
-
-	if (msg.length() < width)
-		msg += std::string(width - msg.length(), ' ');
-
-	std::cout << '\r' << msg << '\n';
+	write_to_console(std::format("{:{}} {} {:3d}%\r", msg, msg_width, bar,
+		static_cast<int>(std::ceil(m_progress * 100))));
 }
 
-progress_bar::progress_bar(int64_t inMax, const std::string &inAction)
+// --------------------------------------------------------------------
+
+progress_bar::progress_bar(int64_t max_value, const std::string &message)
 	: m_impl(nullptr)
 {
-	if (isatty(STDOUT_FILENO) and VERBOSE >= 0)
-		m_impl = new progress_bar_impl(inMax, inAction);
+	if (VERBOSE >= 0)
+	{
+		if (isatty(STDOUT_FILENO) and get_terminal_width() > kMinBarWidth)
+			m_impl = new fancy_progress_bar_impl(max_value, message);
+		else
+			m_impl = new simple_progress_bar_impl(max_value, message);
+	}
 }
 
 progress_bar::~progress_bar()
 {
-	delete m_impl;
+	flush();
 }
 
 void progress_bar::consumed(int64_t inConsumed)
@@ -350,16 +492,25 @@ void progress_bar::consumed(int64_t inConsumed)
 		m_impl->consumed(inConsumed);
 }
 
-void progress_bar::progress(int64_t inProgress)
+void progress_bar::progress(int64_t value)
 {
 	if (m_impl != nullptr)
-		m_impl->progress(inProgress);
+		m_impl->progress(value);
 }
 
-void progress_bar::message(const std::string &inMessage)
+void progress_bar::message(const std::string &message)
 {
 	if (m_impl != nullptr)
-		m_impl->message(inMessage);
+		m_impl->message(message);
+}
+
+void progress_bar::flush()
+{
+	if (m_impl)
+	{
+		delete m_impl;
+		m_impl = nullptr;
+	}
 }
 
 } // namespace cif
@@ -387,13 +538,13 @@ struct rsrc_imp
 
 #if _WIN32
 
-#if __MINGW32__
+# if __MINGW32__
 
 extern "C" __attribute__((weak, alias("gResourceIndexDefault"))) const mrsrc::rsrc_imp gResourceIndex[];
 extern "C" __attribute__((weak, alias("gResourceDataDefault"))) const char gResourceData[];
 extern "C" __attribute__((weak, alias("gResourceNameDefault"))) const char gResourceName[];
 
-#else
+# else
 
 extern "C" const mrsrc::rsrc_imp *gResourceIndexDefault[1] = {};
 extern "C" const char *gResourceDataDefault[1] = {};
@@ -403,11 +554,11 @@ extern "C" const mrsrc::rsrc_imp gResourceIndex[];
 extern "C" const char gResourceData[];
 extern "C" const char gResourceName[];
 
-#pragma comment(linker, "/alternatename:gResourceIndex=gResourceIndexDefault")
-#pragma comment(linker, "/alternatename:gResourceData=gResourceDataDefault")
-#pragma comment(linker, "/alternatename:gResourceName=gResourceNameDefault")
+#  pragma comment(linker, "/alternatename:gResourceIndex=gResourceIndexDefault")
+#  pragma comment(linker, "/alternatename:gResourceData=gResourceDataDefault")
+#  pragma comment(linker, "/alternatename:gResourceName=gResourceNameDefault")
 
-#endif
+# endif
 
 #else
 extern const __attribute__((weak)) mrsrc::rsrc_imp gResourceIndex[];

src/validate.cpp

@@ -25,14 +25,17 @@
  */
 
 #include "cif++/validate.hpp"
+
 #include "cif++/category.hpp"
 #include "cif++/dictionary_parser.hpp"
-#include "cif++/gzio.hpp"
 #include "cif++/utilities.hpp"
 
 #include <cassert>
-#include <fstream>
+#include <format>
+#include <iomanip>
 #include <iostream>
+#include <mutex>
+#include <stdexcept>
 
 // The validator depends on regular expressions. Unfortunately,
 // the implementation of std::regex in g++ is buggy and crashes
@@ -62,14 +65,6 @@ validation_exception::validation_exception(std::error_code ec, std::string_view
 
 // --------------------------------------------------------------------
 
-// struct regex_impl : public regex
-// {
-// 	regex_impl(std::string_view rx)
-// 		: regex(rx.begin(), rx.end(), regex::extended | regex::optimize)
-// 	{
-// 	}
-// };
-
 struct regex_impl
 {
 	regex_impl(std::string_view rx);
@@ -81,9 +76,9 @@ struct regex_impl
 	bool match(std::string_view v) const;
 
   private:
-
 	pcre2_code *m_rx = nullptr;
 	pcre2_match_data *m_data = nullptr;
+	mutable std::mutex m_mutex;
 };
 
 regex_impl::regex_impl(std::string_view rx)
@@ -104,6 +99,8 @@ regex_impl::regex_impl(std::string_view rx)
 
 regex_impl::~regex_impl()
 {
+	std::unique_lock lock(m_mutex);
+
 	if (m_data)
 		pcre2_match_data_free(m_data);
 
@@ -113,6 +110,8 @@ regex_impl::~regex_impl()
 
 bool regex_impl::match(std::string_view v) const
 {
+	std::unique_lock lock(m_mutex);
+
 	bool result = false;
 
 	if (int rc = pcre2_match(m_rx, (PCRE2_SPTR)v.data(), v.length(), 0, 0, m_data, nullptr); rc >= 0)
@@ -190,8 +189,8 @@ int type_validator::compare(std::string_view a, std::string_view b) const
 
 				std::from_chars_result ra, rb;
 
-				ra = selected_charconv<double>::from_chars(a.data(), a.data() + a.length(), da);
-				rb = selected_charconv<double>::from_chars(b.data(), b.data() + b.length(), db);
+				ra = from_chars(a.data(), a.data() + a.length(), da);
+				rb = from_chars(b.data(), b.data() + b.length(), db);
 
 				if (not(bool) ra.ec and not(bool) rb.ec)
 				{
@@ -272,7 +271,7 @@ void item_validator::operator()(std::string_view value) const
 {
 	std::error_code ec;
 	if (not validate_value(value, ec))
-		throw std::system_error(ec, std::string{ value } + " does not match rx for " + m_item_name);
+		throw std::system_error(ec, std::format("'{}' is not a valid value for {}", value, m_item_name));
 }
 
 bool item_validator::validate_value(std::string_view value, std::error_code &ec) const noexcept
@@ -299,15 +298,20 @@ void category_validator::add_item_validator(item_validator &&v)
 
 	v.m_category = m_name;
 
-	auto r = m_item_validators.insert(std::move(v));
-	if (not r.second and VERBOSE >= 4)
-		std::cout << "Could not add validator for item " << v.m_item_name << " to category " << m_name << '\n';
+	auto i = std::find(m_item_validators.begin(), m_item_validators.end(), v);
+	if (i != m_item_validators.end())
+	{
+		if (VERBOSE >= 4)
+			std::cout << "Could not add validator for item " << v.m_item_name << " to category " << m_name << '\n';
+	}
+	else
+		m_item_validators.emplace_back(std::move(v));
 }
 
 const item_validator *category_validator::get_validator_for_item(std::string_view item_name) const
 {
 	const item_validator *result = nullptr;
-	auto i = m_item_validators.find(item_validator{ std::string(item_name) });
+	auto i = std::find(m_item_validators.begin(), m_item_validators.end(), item_validator{ std::string(item_name) });
 	if (i != m_item_validators.end())
 		result = &*i;
 	else if (VERBOSE > 4)
@@ -544,19 +548,18 @@ validator_factory &validator_factory::instance()
 	return s_instance;
 }
 
-const validator &validator_factory::get(std::string_view dictionary_name)
+const validator *validator_factory::get(std::string_view dictionary_name)
 {
 	category audit_conform("audit_conform");
-	for (auto part : cif::split(dictionary_name, ";", true))
+	for (auto part : cif::split(dictionary_name, ";,", true))
 		audit_conform.emplace({ { "dict_name", part } });
 
 	return get(audit_conform);
 }
 
-const validator &validator_factory::get(const category &audit_conform)
+const validator *validator_factory::get(const category &audit_conform)
 {
-	if (audit_conform.empty())
-		throw std::runtime_error("Empty audit_conform category, cannot create a validator");
+	const validator *result = nullptr;
 
 	std::lock_guard lock(m_mutex);
 
@@ -564,37 +567,60 @@ const validator &validator_factory::get(const category &audit_conform)
 	for (auto &v : m_validators)
 	{
 		if (v.matches_audit_conform(audit_conform))
-			return v;
+			result = &v;
 	}
 
 	// If the audit conform contains only one record, this is easy
-	if (audit_conform.size() == 1)
+	if (result == nullptr and audit_conform.size() == 1)
 	{
-		const auto &[name, version] = audit_conform.front().get<std::string, std::optional<std::string>>("dict_name", "dict_version");
-		return m_validators.emplace_back(construct_validator(name, version));
+		const auto &[name, version] =
+			audit_conform.front().get<std::string, std::optional<std::string>>("dict_name", "dict_version");
+		if (not name.empty())
+			result = &m_validators.emplace_back(construct_validator(name, version));
 	}
 
-	// A new, merged dictionary
-
-	std::optional<validator> v;
-	for (const auto &[name, version] : audit_conform.rows<std::string, std::optional<std::string>>("dict_name", "dict_version"))
+	if (result == nullptr)
 	{
-		if (not v)
-			v = construct_validator(name, version);
-		else
+		// A new, merged dictionary
+		std::optional<validator> v;
+		for (const auto &[name, version] : audit_conform.rows<std::string, std::optional<std::string>>("dict_name", "dict_version"))
 		{
-			auto data = load_resource(name);
-			if (not data)
-				throw std::runtime_error("Could not load dictionary " + std::string{ name });
+			if (name.empty())
+				continue;
 
-			v->parse(*data);
+			if (not v) // first dict
+				v = construct_validator(name, version);
+			else // additional/extending dict
+			{
+				auto data = load_resource(name);
+				if (not data)
+					throw std::runtime_error("Could not load dictionary " + std::string{ name });
+
+				v->parse(*data);
+			}
 		}
+
+		if (v)
+			result = &m_validators.emplace_back(std::move(*v));
 	}
 
-	if (not v)
-		throw std::runtime_error("Missing dictionary information?");
+	return result;
+}
 
-	return m_validators.emplace_back(std::move(*v));
+const validator &validator_factory::operator[](const category &audit_conform)
+{
+	auto v = get(audit_conform);
+	if (v == nullptr)
+		throw std::runtime_error("Could not load dictionary for audit_conform");
+	return *v;
+}
+
+const validator &validator_factory::operator[](std::string_view dictionary_name)
+{
+	auto v = get(dictionary_name);
+	if (v == nullptr)
+		throw std::runtime_error("Could not load dictionary for " + std::string{ dictionary_name });
+	return *v;
 }
 
 validator validator_factory::construct_validator(std::string_view name, std::optional<std::string> version)
@@ -613,7 +639,7 @@ validator validator_factory::construct_validator(std::string_view name, std::opt
 		not v.matches_audit_conform(category{ "audit_conform", //
 			{ { "dict_name", name }, { "dict_version", version } } }))
 	{
-		std::clog << "Invalid dictionary?\n";
+		std::clog << "Loaded dictionary does not match name=" << name << " and version=" << version.value_or("''") << "\n";
 	}
 
 	return v;

test/CMakeLists.txt

@@ -1,19 +1,39 @@
-# We're using the older version 2 of Catch2
+# SPDX-License-Identifier: BSD-2-Clause
+# 
+# Copyright (c) 2025  NKI/AVL, Netherlands Cancer Institute
+# 
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+# 
+# 1. Redistributions of source code must retain the above copyright notice, this
+#    list of conditions and the following disclaimer
+# 2. Redistributions in binary form must reproduce the above copyright notice,
+#    this list of conditions and the following disclaimer in the documentation
+#    and/or other materials provided with the distribution.
+# 
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+# ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+# (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+# LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-if(NOT(Catch2_FOUND OR TARGET Catch2))
-	find_package(Catch2 QUIET)
+if(NOT (Catch2_FOUND OR TARGET Catch2))
+	find_package(Catch2 3 QUIET)
 
 	if(NOT Catch2_FOUND)
-		include(FetchContent)
-
 		FetchContent_Declare(
 			Catch2
 			GIT_REPOSITORY https://github.com/catchorg/Catch2.git
-			GIT_TAG v2.13.9)
+			GIT_TAG v3.4.0)
 
 		FetchContent_MakeAvailable(Catch2)
 
-		set(Catch2_VERSION "2.13.9")
+		target_compile_features(Catch2 PRIVATE cxx_std_20)
 	endif()
 endif()
 
@@ -27,20 +47,16 @@ list(
 	rename-compound
 	sugar
 	spinner
-	# reconstruction
+	reconstruction
 	validate-pdbx
-	)
+	cql
+	matrix
+)
 
 add_library(test-main OBJECT "${CMAKE_CURRENT_SOURCE_DIR}/test-main.cpp")
 
 target_link_libraries(test-main cifpp::cifpp Catch2::Catch2)
 
-if("${Catch2_VERSION}" VERSION_LESS 3.0.0)
-	target_compile_definitions(test-main PUBLIC CATCH22=1)
-else()
-	target_compile_definitions(test-main PUBLIC CATCH22=0)
-endif()
-
 foreach(CIFPP_TEST IN LISTS CIFPP_tests)
 	set(CIFPP_TEST "${CIFPP_TEST}-test")
 	set(CIFPP_TEST_SOURCE "${CMAKE_CURRENT_SOURCE_DIR}/${CIFPP_TEST}.cpp")
@@ -48,12 +64,6 @@ foreach(CIFPP_TEST IN LISTS CIFPP_tests)
 	add_executable(
 		${CIFPP_TEST} ${CIFPP_TEST_SOURCE} $<TARGET_OBJECTS:test-main>)
 
-	if(${Catch2_VERSION} VERSION_GREATER_EQUAL 3.0.0)
-		target_compile_definitions(${CIFPP_TEST} PUBLIC CATCH22=0)
-	else()
-		target_compile_definitions(${CIFPP_TEST} PUBLIC CATCH22=1)
-	endif()
-
 	target_link_libraries(${CIFPP_TEST} PRIVATE cifpp::cifpp Catch2::Catch2)
 	target_include_directories(${CIFPP_TEST} PRIVATE "${EIGEN_INCLUDE_DIR}")
 
@@ -62,15 +72,8 @@ foreach(CIFPP_TEST IN LISTS CIFPP_tests)
 		target_compile_options(${CIFPP_TEST} PRIVATE /EHsc)
 	endif()
 
-	add_custom_target(
-		"run-${CIFPP_TEST}"
-		DEPENDS ${CMAKE_CURRENT_BINARY_DIR}/Run${CIFPP_TEST}.touch ${CIFPP_TEST})
-
-	add_custom_command(
-		OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/Run${CIFPP_TEST}.touch
-		COMMAND $<TARGET_FILE:${CIFPP_TEST}> --data-dir
-		${CMAKE_CURRENT_SOURCE_DIR})
-
-	add_test(NAME ${CIFPP_TEST} COMMAND $<TARGET_FILE:${CIFPP_TEST}> --data-dir
-		${CMAKE_CURRENT_SOURCE_DIR})
-endforeach()
+	if(NOT (CIFPP_TEST STREQUAL "spinner-test"))
+		add_test(NAME ${CIFPP_TEST}
+			COMMAND $<TARGET_FILE:${CIFPP_TEST}> --data-dir ${CMAKE_CURRENT_SOURCE_DIR})
+	endif()
+endforeach()

test/cql-test.cpp

@@ -0,0 +1,537 @@
+/*-
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright (c) 2025 NKI/AVL, Netherlands Cancer Institute
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ *    list of conditions and the following disclaimer
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "test-main.hpp"
+
+#include <catch2/catch_test_macros.hpp>
+#include <cif++.hpp>
+#include <cif++/cql.hpp>
+
+// --------------------------------------------------------------------
+
+cif::file operator""_cf(const char *text, std::size_t length)
+{
+	struct membuf : public std::streambuf
+	{
+		membuf(char *text, std::size_t length)
+		{
+			this->setg(text, text, text + length);
+		}
+	} buffer(const_cast<char *>(text), length);
+
+	std::istream is(&buffer);
+	return cif::file(is);
+}
+
+// --------------------------------------------------------------------
+
+const char *kAuthors[] = {
+	"Kleywegt, G.J.",
+	"Bergfors, T.",
+	"Senn, H.",
+	"Le Motte, P.",
+	"Gsell, B.",
+	"Shudo, K.",
+	"Jones, T.A.",
+
+	"Banaszak, L.",
+	"Winter, N.",
+	"Xu, Z.",
+	"Bernlohr, D.A.",
+	"Cowan, S.W.",
+	"Jones, T.A.",
+	"Bergfors, T.",
+	"Kleywegt, G.J.",
+	"Jones, T.A.",
+	"Cowan, S.W.",
+	"Newcomer, M.E.",
+	"Jones, T.A.",
+	"Jones, T.A.",
+	"Bergfors, T.",
+	"Sedzik, J.",
+	"Unge, T."
+};
+
+// Test simple SELECT
+TEST_CASE("cql-1")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT name, ordinal FROM citation_author WHERE citation_id = 'primary';");
+	CHECK(r.size() == 7);
+
+	for (size_t ix = 0; auto row : r)
+	{
+		REQUIRE(ix < (sizeof(kAuthors) / sizeof(char *)));
+
+		CHECK(row[0].as<std::string>() == kAuthors[ix]);
+		CHECK(row[1].as<size_t>() == ix + 1);
+
+		CHECK(row["name"].as<std::string>() == kAuthors[ix]);
+		CHECK(row["ordinal"].as<size_t>() == ix + 1);
+
+		++ix;
+	}
+
+	r = tx.exec("SELECT ordinal, name FROM citation_author WHERE citation_id = 'primary';");
+	CHECK(r.size() == 7);
+
+	for (size_t ix = 0; auto row : r)
+	{
+		REQUIRE(ix < (sizeof(kAuthors) / sizeof(char *)));
+
+		CHECK(row[1].as<std::string>() == kAuthors[ix]);
+		CHECK(row[0].as<size_t>() == ix + 1);
+
+		CHECK(row["name"].as<std::string>() == kAuthors[ix]);
+		CHECK(row["ordinal"].as<size_t>() == ix + 1);
+
+		++ix;
+	}
+
+	r = tx.exec("SELECT * FROM citation_author WHERE citation_id = 'primary';");
+	CHECK(r.size() == 7);
+
+	for (int ix = 0; auto row : r)
+	{
+		REQUIRE(static_cast<size_t>(ix) < (sizeof(kAuthors) / sizeof(char *)));
+
+		for (auto fld : row)
+		{
+			switch (fld.num())
+			{
+				case 0:
+					CHECK(fld.name() == "citation_id");
+					CHECK(fld.as<std::string>() == "primary");
+					break;
+				case 1:
+					CHECK(fld.name() == "name");
+					CHECK(fld.as<std::string>() == kAuthors[ix]);
+					break;
+				case 2:
+					CHECK(fld.name() == "ordinal");
+					CHECK(fld.as<int>() == ix + 1);
+					break;
+				default:
+					REQUIRE(false);
+					break;
+			}
+		}
+
+		CHECK(row["name"].as<std::string>() == kAuthors[ix]);
+		CHECK(row["ordinal"].as<int>() == ix + 1);
+		CHECK(row["citation_id"].as<std::string>() == "primary");
+
+		++ix;
+	}
+}
+
+// Test SELECT AS
+TEST_CASE("cql-2")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT name AS v1, ordinal AS v2 FROM citation_author WHERE citation_id = 'primary';");
+	CHECK(r.size() == 7);
+
+	for (size_t ix = 0; auto row : r)
+	{
+		REQUIRE(ix < (sizeof(kAuthors) / sizeof(char *)));
+
+		CHECK(row[0].as<std::string>() == kAuthors[ix]);
+		CHECK(row[1].as<size_t>() == ix + 1);
+
+		CHECK(row["v1"].as<std::string>() == kAuthors[ix]);
+		CHECK(row["v2"].as<size_t>() == ix + 1);
+
+		++ix;
+	}
+}
+
+TEST_CASE("cql-3")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT name FROM citation_author WHERE ordinal = 10").one_field();
+	CHECK(r.as<std::string>() == kAuthors[9]);
+}
+
+TEST_CASE("cql-4")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT name FROM citation_author WHERE ordinal BETWEEN 10 AND 15");
+	REQUIRE(r.size() == 6);
+}
+
+TEST_CASE("cql-5")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT (SELECT year FROM citation WHERE id = citation_id) AS jaar FROM citation_author WHERE ordinal IS 23").one_field();
+	CHECK(r.name() == "jaar");
+	CHECK(r.as<int>() == 1988);
+}
+
+TEST_CASE("cql-6")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto r = tx.exec("SELECT COUNT(*) FROM citation WHERE page_last IS NULL").one_field();
+	CHECK(r.as<int>() == 4);
+
+	r = tx.exec("SELECT COUNT(*) FROM citation WHERE page_last IS NOT NULL").one_field();
+	CHECK(r.as<int>() == 1);
+}
+
+TEST_CASE("cql-stream-1")
+{
+	cif::file f(gTestDir / ".." / "examples" / "1cbs.cif.gz");
+	auto &db = f.front();
+	db.load_dictionary("mmcif_pdbx.dic");
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	for (size_t ix = 0;
+		const auto &[name, ordinal] : tx.stream<std::string, size_t>(
+			"SELECT name, ordinal FROM citation_author WHERE citation_id = 'primary';"))
+	{
+		REQUIRE(ix < (sizeof(kAuthors) / sizeof(char *)));
+
+		CHECK(name == kAuthors[ix]);
+		CHECK(ordinal == (ix + 1));
+
+		++ix;
+	}
+}
+
+// --------------------------------------------------------------------
+
+TEST_CASE("cql-insert-1")
+{
+	auto f1 = R"(
+data_T1
+loop_
+_table1.id
+_table1.name
+1 aap
+2 noot)"_cf;
+	auto f0 = f1;
+
+	auto &db = f1.front();
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	auto count = tx.exec("SELECT COUNT(*) FROM table1;").one_field().as<int>();
+	CHECK(count == 2);
+
+	auto r = tx.exec("INSERT INTO table1 (id, name) VALUES (3, 'mies')");
+
+	count = tx.exec("SELECT COUNT(*) FROM table1").one_field().as<int>();
+	CHECK(count == 3);
+
+	(void)tx.exec("DELETE FROM table1 WHERE CAST(id AS INTEGER) = 1;");
+
+	count = tx.exec("SELECT COUNT(*) FROM table1;").one_field().as<int>();
+	CHECK(count == 2);
+
+	(void)tx.exec("UPDATE table1 SET name = 'amandel' WHERE CAST(id AS INTEGER) = 2");
+
+	auto f2 = R"(
+data_T1
+loop_
+_table1.id
+_table1.name
+2 amandel
+3 mies)"_cf;
+
+	CHECK(f1 == f2);
+
+	tx.rollback();
+
+	CHECK(f1 == f0);
+}
+
+// --------------------------------------------------------------------
+
+TEST_CASE("cql-rename")
+{
+	auto f1 = R"(
+data_T1
+loop_
+_table1.id
+_table1.name
+1 aap
+2 noot)"_cf;
+
+	auto &db = f1.front();
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	(void)tx.exec("ALTER TABLE table1 RENAME TO 'table2'");
+
+	auto f2 = R"(
+data_T1
+loop_
+_table2.id
+_table2.name
+1 aap
+2 noot)"_cf;
+
+	CHECK(f1 == f2);
+}
+
+// --------------------------------------------------------------------
+
+TEST_CASE("cql-foreign-keys-1")
+{
+	const char dict[] = R"(
+data_test_dict.dic
+    _datablock.id	test_dict.dic
+    _datablock.description
+;
+    A test dictionary
+;
+    _dictionary.title           test_dict.dic
+    _dictionary.datablock_id    test_dict.dic
+    _dictionary.version         1.0
+
+     loop_
+    _item_type_list.code
+    _item_type_list.primitive_code
+    _item_type_list.construct
+    _item_type_list.detail
+               code      char
+               '[][_,.;:"&<>()/\{}'`~!@#$%A-Za-z0-9*|+-]*'
+;              code item types/single words ...
+;
+               text      char
+               '[][ \n\t()_,.;:"&<>/\{}'`~!@#$%?+=*A-Za-z0-9|^-]*'
+;              text item types / multi-line text ...
+;
+               int       numb
+               '[+-]?[0-9]+'
+;              int item types are the subset of numbers that are the negative
+               or positive integers.
+;
+
+save_cat_1
+    _category.description     'A simple test category'
+    _category.id              cat_1
+    _category.mandatory_code  no
+    _category_key.name        '_cat_1.id'
+
+    save_
+
+save__cat_1.id
+    _item.name                '_cat_1.id'
+    _item.category_id         cat_1
+    _item.mandatory_code      yes
+    _item_aliases.dictionary  cif_core.dic
+    _item_aliases.version     2.0.1
+    _item_linked.child_name   '_cat_2.parent_id'
+    _item_linked.parent_name  '_cat_1.id'
+    _item_type.code           code
+    save_
+
+save__cat_1.name
+    _item.name                '_cat_1.name'
+    _item.category_id         cat_1
+    _item.mandatory_code      yes
+    _item_aliases.dictionary  cif_core.dic
+    _item_aliases.version     2.0.1
+    _item_type.code           text
+    save_
+
+save_cat_2
+    _category.description     'A second simple test category'
+    _category.id              cat_2
+    _category.mandatory_code  no
+    _category_key.name        '_cat_2.id'
+    save_
+
+save__cat_2.id
+    _item.name                '_cat_2.id'
+    _item.category_id         cat_2
+    _item.mandatory_code      yes
+    _item_aliases.dictionary  cif_core.dic
+    _item_aliases.version     2.0.1
+    _item_type.code           int
+    save_
+
+save__cat_2.parent_id
+    _item.name                '_cat_2.parent_id'
+    _item.category_id         cat_2
+    _item.mandatory_code      yes
+    _item_aliases.dictionary  cif_core.dic
+    _item_aliases.version     2.0.1
+    _item_type.code           code
+    save_
+
+save__cat_2.desc
+    _item.name                '_cat_2.desc'
+    _item.category_id         cat_2
+    _item.mandatory_code      yes
+    _item_aliases.dictionary  cif_core.dic
+    _item_aliases.version     2.0.1
+    _item_type.code           text
+    save_
+    )";
+
+	struct membuf : public std::streambuf
+	{
+		membuf(char *text, std::size_t length)
+		{
+			this->setg(text, text, text + length);
+		}
+	} buffer(const_cast<char *>(dict), sizeof(dict) - 1);
+
+	std::istream is_dict(&buffer);
+
+	cif::validator validator(is_dict);
+
+	cif::file f;
+
+	// --------------------------------------------------------------------
+
+	const char data[] = R"(
+data_test
+loop_
+_cat_1.id
+_cat_1.name
+1 Aap
+2 Noot
+3 Mies
+
+loop_
+_cat_2.id
+_cat_2.parent_id
+_cat_2.desc
+1 1 'Een dier'
+2 1 'Een andere aap'
+3 2 'walnoot bijvoorbeeld'
+    )";
+
+	struct data_membuf : public std::streambuf
+	{
+		data_membuf(char *text, std::size_t length)
+		{
+			this->setg(text, text, text + length);
+		}
+	} data_buffer(const_cast<char *>(data), sizeof(data) - 1);
+
+	std::istream is_data(&data_buffer);
+	f.load(is_data, validator);
+	
+	auto &db = f.front();
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	for (const auto &desc : tx.stream<std::string>(R"(SELECT b.desc FROM cat_1 a, cat_2 b WHERE a.id = b.parent_id AND a.name = 'Noot')"))
+	{
+		CHECK(desc == "walnoot bijvoorbeeld");
+	}
+
+	// Check cascading delete
+	tx.exec("DELETE FROM cat_1 WHERE id = 1");
+	CHECK(db["cat_1"].size() == 2);
+	CHECK(db["cat_2"].size() == 1);
+
+	tx.rollback();
+	CHECK(db["cat_1"].size() == 3);
+	CHECK(db["cat_2"].size() == 3);
+
+}
+
+// --------------------------------------------------------------------
+
+TEST_CASE("drop-table")
+{
+	auto f1 = R"(
+data_T1
+loop_
+_table1.id
+_table1.name
+1 aap
+2 noot)"_cf;
+
+	auto &db = f1.front();
+
+	cif::cql::connection connection(db);
+	cif::cql::transaction tx(connection);
+
+	SECTION("commit")
+	{
+		(void)tx.exec("DROP TABLE table1;");
+		tx.commit();
+	
+		CHECK(db.empty());
+	}
+
+	// Ah, too bad: this doesn't work
+	// SECTION("rollback")
+	// {
+	// 	(void)tx.exec("DROP TABLE table1;");
+	// 	tx.rollback();
+	
+	// 	CHECK(not db.empty());
+	// 	CHECK(db["table1"].size() == 2);
+	// }
+}

test/matrix-test.cpp

@@ -0,0 +1,103 @@
+/*-
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright (c) 2025 NKI/AVL, Netherlands Cancer Institute
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright notice, this
+ *    list of conditions and the following disclaimer
+ * 2. Redistributions in binary form must reproduce the above copyright notice,
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
+ * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+ * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
+ * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "cif++/matrix.hpp"
+#include "test-main.hpp"
+
+#include <catch2/catch_test_macros.hpp>
+#include <cif++.hpp>
+
+TEST_CASE("m1")
+{
+	cif::matrix3x3<int> m = cif::identity_matrix<int>(3);
+
+	CHECK(cif::determinant(m) == 1);
+}
+
+TEST_CASE("m2")
+{
+	cif::matrix4x4<int> m = cif::identity_matrix<int>(4);
+
+	cif::sub_matrix<cif::matrix4x4<int>> ms(m, 1, 1);
+	CHECK(ms == cif::identity_matrix<int>(3));
+}
+
+TEST_CASE("m3")
+{
+	cif::matrix4x4<int> m{
+		{ 1, 2, 3, 4,      //
+			5, 6, 7, 8,    //
+			9, 10, 11, 12, //
+			13, 14, 15, 16 }
+	};
+	cif::sub_matrix<cif::matrix4x4<int>> ms(m, 1, 1);
+
+	cif::matrix3x3<int> t{
+		{ 1, 3, 4, 9, 11, 12, 13, 15, 16 }
+	};
+
+	CHECK(ms == t);
+}
+
+TEST_CASE("m4")
+{
+	cif::matrix4x4<int> m{
+		{
+			-2,
+			3,
+			1,
+			0,
+			4,
+			1,
+			-3,
+			2,
+			0,
+			-1,
+			2,
+			5,
+			3,
+			2,
+			0,
+			-4,
+		}
+	};
+
+	// std::cout << m << "\n\n";
+
+	// std::cout << cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 0)) << "\n\n";
+	// std::cout << cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 1)) << "\n\n";
+	// std::cout << cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 2)) << "\n\n";
+	// std::cout << cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 3)) << "\n\n";
+
+	// std::cout << cif::determinant(cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 0))) << "\n\n";
+	// std::cout << cif::determinant(cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 1))) << "\n\n";
+	// std::cout << cif::determinant(cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 2))) << "\n\n";
+	// std::cout << cif::determinant(cif::matrix3x3<int>(cif::sub_matrix<decltype(m)>(m, 0, 3))) << "\n\n";
+
+	CHECK(cif::determinant(m) == 332);
+}
+
+

test/model-test.cpp

@@ -54,7 +54,7 @@ TEST_CASE("create_nonpoly_1")
 
 	cif::file file;
 	auto &&[dbi, ignore] = file.emplace("TEST"); // create a datablock
-	dbi->set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	dbi->load_dictionary("mmcif_pdbx.dic");
 
 	cif::mm::structure structure(file);
 
@@ -82,7 +82,7 @@ _atom_site.pdbx_formal_charge
 # that's enough to test with
 )"_cf;
 
-	atoms.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	atoms.front().load_dictionary("mmcif_pdbx.dic");
 
 	auto &hem_data = atoms["HEM"];
 	auto &atom_site = hem_data["atom_site"];
@@ -159,7 +159,7 @@ _struct_asym.details                       ?
 _atom_type.symbol   C
 )"_cf;
 
-	expected.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	expected.front().load_dictionary("mmcif_pdbx.dic");
 
 	if (not(expected.front() == structure.get_datablock()))
 	{
@@ -178,7 +178,7 @@ TEST_CASE("create_nonpoly_2")
 
 	cif::file file;
 	auto &&[dbi, ignore] = file.emplace("TEST"); // create a datablock
-	dbi->set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	dbi->load_dictionary("mmcif_pdbx.dic");
 
 	cif::mm::structure structure(file);
 
@@ -270,7 +270,7 @@ _struct_asym.details                       ?
 _atom_type.symbol   C
 )"_cf;
 
-	expected.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	expected.front().load_dictionary("mmcif_pdbx.dic");
 
 	REQUIRE(expected.front() == structure.get_datablock());
 
@@ -354,7 +354,7 @@ _struct_asym.details                       ?
 #
 )"_cf;
 
-	data.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	data.front().load_dictionary("mmcif_pdbx.dic");
 
 	cif::mm::structure s(data);
 
@@ -566,7 +566,7 @@ _struct_asym.details                       ?
 #
 )"_cf;
 
-	data.front().set_validator(&cif::validator_factory::instance().get("mmcif_pdbx.dic"));
+	data.front().load_dictionary("mmcif_pdbx.dic");
 
 	SECTION("max")
 	{

test/query-test.cpp

@@ -49,6 +49,4 @@ TEST_CASE("q-1")
 
 		CHECK(pdbx_poly_seq_scheme.count("asym_id"_key == "A" and "entity_id"_key == 1 and "seq_id"_key == 1 and "mon_id"_key == "PRO" and "hetero"_key == false) == 1);
 	}
-	
-
 }

test/reconstruction-test.cpp

@@ -28,6 +28,7 @@
 
 #include <cif++.hpp>
 
+#include <filesystem>
 #include <iostream>
 #include <fstream>
 

test/test-main.cpp

@@ -11,12 +11,7 @@ int main(int argc, char *argv[])
 	Catch::Session session; // There must be exactly one instance
 
 	// Build a new parser on top of Catch2's
-#if CATCH22
-	using namespace Catch::clara;
-#else
-	// Build a new parser on top of Catch2's
 	using namespace Catch::Clara;
-#endif
 
 	auto cli = session.cli()                               // Get Catch2's command line parser
 	           | Opt(gTestDir, "data-dir")                 // bind variable to a new option, with a hint string

test/test-main.hpp

@@ -1,17 +1,17 @@
 /*-
  * SPDX-License-Identifier: BSD-2-Clause
- * 
+ *
  * Copyright (c) 2024 NKI/AVL, Netherlands Cancer Institute
- * 
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions are met:
- * 
+ *
  * 1. Redistributions of source code must retain the above copyright notice, this
  *    list of conditions and the following disclaimer
  * 2. Redistributions in binary form must reproduce the above copyright notice,
  *    this list of conditions and the following disclaimer in the documentation
  *    and/or other materials provided with the distribution.
- * 
+ *
  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
@@ -26,11 +26,7 @@
 
 #pragma once
 
-#if CATCH22
-#include <catch2/catch.hpp>
-#else
 #include <catch2/catch_all.hpp>
-#endif
 
 #include <filesystem>
 

test/unit-3d-test.cpp

@@ -24,15 +24,17 @@
  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  */
 
+#include "cif++/point.hpp"
 #include "test-main.hpp"
 
+#include <catch2/catch_test_macros.hpp>
+#include <catch2/matchers/catch_matchers_floating_point.hpp>
+#include <cif++.hpp>
 #include <stdexcept>
 
-#include <cif++.hpp>
-
 #if defined(_MSC_VER)
-#pragma warning (disable : 5054)	// warning C5054: operator '&': deprecated between enumerations of different types
-#pragma warning (disable : 4127)	// conditional expression is constant
+# pragma warning(disable : 5054) // warning C5054: operator '&': deprecated between enumerations of different types
+# pragma warning(disable : 4127) // conditional expression is constant
 #endif
 
 #include <Eigen/Eigenvalues>
@@ -315,8 +317,7 @@ TEST_CASE("m2q_0a")
 		cif::point p2 = p1;
 		p2.rotate(q);
 
-		cif::matrix3x3<float> rot_c({
-			static_cast<float>(d[0]),
+		cif::matrix3x3<float> rot_c({ static_cast<float>(d[0]),
 			static_cast<float>(d[1]),
 			static_cast<float>(d[2]),
 			static_cast<float>(d[3]),
@@ -324,8 +325,7 @@ TEST_CASE("m2q_0a")
 			static_cast<float>(d[5]),
 			static_cast<float>(d[6]),
 			static_cast<float>(d[7]),
-			static_cast<float>(d[8])
-		});
+			static_cast<float>(d[8]) });
 
 		cif::point p3 = rot_c * p1;
 
@@ -610,3 +610,40 @@ TEST_CASE("volume_3bwh_1")
 
 	CHECK_THAT(c.get_cell().get_volume(), Catch::Matchers::WithinRel(741009.625f, 0.01f));
 }
+
+// --------------------------------------------------------------------
+
+TEST_CASE("smallest_sphere-1")
+{
+	std::vector<cif::point> pts{
+		{ 0.9295, 4.9006, 46.9706 },
+		{ -0.1215, 5.5936, 46.0726 },
+		{ -0.7975, 4.7046, 45.0796 },
+		{ -1.4875, 3.5486, 45.7196 },
+		{ -0.6535, 2.8816, 46.8186 },
+		{ 0.3825, 3.5156, 47.4496 },
+		{ 1.1995, 2.9206, 48.5286 },
+		{ 0.8255, 2.0466, 49.4716 },
+		{ 1.6625, 1.5036, 50.5176 },
+		{ 1.1165, 0.6056, 51.3626 },
+		{ 1.8325, -0.0064, 52.4656 },
+		{ 1.1945, -0.9044, 53.2216 },
+		{ 1.8135, -1.5534, 54.3566 },
+		{ 1.0925, -2.4574, 55.0656 },
+		{ 1.5205, -3.2204, 56.2476 },
+		{ 1.1955, 5.8066, 48.1796 },
+		{ 2.2495, 4.6896, 46.1796 },
+		{ -1.2515, 1.5186, 47.1786 },
+		{ 3.1385, 1.9106, 50.6166 },
+		{ 3.2605, -1.1834, 54.7206 },
+		{ 2.5975, -3.8554, 56.2096 },
+		{ 0.7975, -3.2184, 57.2686 }
+	};
+
+	for (int i = 0; i < 1000; ++i)
+	{
+		auto [c, r] = cif::smallest_sphere_around_points(pts);
+		CHECK_THAT(cif::distance(c, cif::point{ 0, 0.743099928, 51.1741028 }), Catch::Matchers::WithinAbs(0.f, 0.01f));
+		CHECK_THAT(r, Catch::Matchers::WithinAbs(7.31248331f, 0.01f));
+	}
+}

tools/depends.cmd

@@ -1,6 +1,5 @@
 @ECHO OFF
 SET ZLIB_VERSION=1.3
-SET PCRE2_VERSION=10.45
 
 IF NOT EXIST build_ci\libs (
   MKDIR build_ci\libs
@@ -22,19 +21,4 @@ cmake --build build --target ALL_BUILD --config Release || EXIT /b 1
 cmake --build build --target RUN_TESTS --config Release || EXIT /b 1
 cmake --build build --target INSTALL --config Release || EXIT /b 1
 
-@REM Install PCRE2
-IF NOT EXIST pcre2-%PCRE2_VERSION%.zip (
-  ECHO Downloading https://github.com/PCRE2Project/pcre2/releases/download/pcre2-%PCRE2_VERSION%/pcre2-%PCRE2_VERSION%.zip
-  curl -L -o pcre2-%PCRE2_VERSION%.zip https://github.com/PCRE2Project/pcre2/releases/download/pcre2-%PCRE2_VERSION%/pcre2-%PCRE2_VERSION%.zip || EXIT /b 1
-)
-IF NOT EXIST pcre2-%PCRE2_VERSION% (
-  ECHO Unpacking pcre2-%PCRE2_VERSION%.zip
-  C:\windows\system32\tar.exe -x -f pcre2-%PCRE2_VERSION%.zip || EXIT /b 1
-)
-CD pcre2-%PCRE2_VERSION%
-cmake -B build -DPCRE2_BUILD_PCRE2GREP=OFF || EXIT /b 1
-cmake --build build --target ALL_BUILD --config Release || EXIT /b 1
-@REM cmake --build build --target RUN_TESTS --config Release || EXIT /b 1
-cmake --build build --target INSTALL --config Release || EXIT /b 1
-
 @EXIT /b 0

tools/update-libcifpp-data.in

@@ -63,7 +63,7 @@ update_dictionary() {
 
 update_dictionary "@CIFPP_CACHE_DIR@/components.cif" "https://files.wwpdb.org/pub/pdb/data/monomers/components.cif.gz"
 update_dictionary "@CIFPP_CACHE_DIR@/mmcif_pdbx.dic" "https://mmcif.wwpdb.org/dictionaries/ascii/mmcif_pdbx_v50.dic.gz"
-update_dictionary "@CIFPP_CACHE_DIR@/mmcif_ma.dic" "https://github.com/ihmwg/ModelCIF/raw/master/dist/mmcif_ma.dic"
+update_dictionary "@CIFPP_CACHE_DIR@/mmcif_ma.dic" "https://mmcif.wwpdb.org/dictionaries/ascii/mmcif_ma.dic"
 
 # notify subscribers, using find instead of run-parts to make it work on FreeBSD as well