Fix drop_empty_items

include/cif++/model.hpp

@@ -122,7 +122,7 @@ class atom
 
 		// const compound *compound() const;
 
-		[[nodiscard]] const item_value &get_property(std::string_view name) const;
+		[[nodiscard]] const item_handle get_property(std::string_view name) const;
 		void set_property(const std::string_view name, item_value value);
 
 		row_handle row()
@@ -227,7 +227,7 @@ class atom
 	explicit operator bool() const { return m_impl.operator bool(); }
 
 	/// \brief Return the item named @a name in the _atom_site category for this atom
-	[[nodiscard]] const item_value &get_property_value(std::string_view name) const
+	[[nodiscard]] const item_handle get_property_value(std::string_view name) const
 	{
 		if (not m_impl)
 			throw std::logic_error("Error trying to fetch a property from an uninitialized atom");

src/category.cpp

@@ -614,11 +614,11 @@ void category::drop_empty_items()
 {
 	std::vector<bool> is_empty(get_item_count(), true);
 
-	for (auto row : *this)
+	for (size_t ix = 0; ix < get_item_count(); ++ix)
 	{
-		for (size_t ix = 0; ix < get_item_count(); ++ix)
+		for (auto row : *this)
 		{
-			if (is_empty[ix] and not row[ix].empty())
+			if (not row[ix].empty())
 			{
 				is_empty[ix] = false;
 				break;
@@ -1612,7 +1612,6 @@ void category::update_value(row *row, uint16_t item, item_value value, bool upda
 		}
 		else if (ec)
 			throw validation_exception(ec, m_name, m_items[item].m_name);
-
 	}
 
 	// If the item is part of the Key for this category, remove it from the index
@@ -2200,7 +2199,7 @@ void category::write_cif(std::ostream &os, const std::vector<uint16_t> &order, b
 					offset = 0;
 				}
 
-				offset = detail::write_value(os, s, offset, w, right_aligned[cix]/*  iv->is_number() */);
+				offset = detail::write_value(os, s, offset, w, right_aligned[cix] /*  iv->is_number() */);
 
 				if (offset > 132)
 				{
@@ -2722,70 +2721,20 @@ bool category::operator==(const category &rhs) const
 
 	using namespace std::placeholders;
 
-	//	set<std::string> item_namesA(a.items()), item_namesB(b.items());
-	//
-	//	if (item_namesA != item_namesB)
-	//		std::cout << "Unequal number of items\n";
+	std::set<std::string> keys;
 
-	const category_validator *catValidator = nullptr;
-
-	auto validator = a.get_validator();
-	if (validator != nullptr)
-		catValidator = validator->get_validator_for_category(a.name());
-
-	using compType = std::function<int(std::string_view, std::string_view)>;
-	std::vector<std::tuple<std::string, compType>> item_names;
-	std::vector<std::string> keys;
-	std::vector<std::size_t> keyIx;
-
-	if (catValidator == nullptr)
+	for (const auto &items : { a.get_items(), b.get_items()})
 	{
-		for (auto &item_name : a.get_items())
-		{
-			item_names.emplace_back(item_name, [](std::string_view va, std::string_view vb)
-				{ return va.compare(vb); });
-			keyIx.push_back(keys.size());
-			keys.push_back(item_name);
-		}
+		for (auto &item_name : items)
+			keys.insert(item_name);
 	}
-	else
-	{
-		keys = catValidator->m_keys;
-
-		for (auto &item_name : a.key_items())
-		{
-			auto iv = catValidator->get_validator_for_item(item_name);
-			if (iv == nullptr)
-				throw std::runtime_error("missing item validator");
-			auto tv = iv->m_type;
-			if (tv == nullptr)
-				throw std::runtime_error("missing type validator");
-			item_names.emplace_back(item_name, [tv](auto &&a1, auto &&a2)
-				{ return tv->compare(std::forward<decltype(a1)>(a1), std::forward<decltype(a2)>(a2)); });
-
-			auto pred = [item_name](const std::string &s) -> bool
-			{
-				return cif::iequals(item_name, s) == 0;
-			};
-			if (std::ranges::find_if(keys, pred) == keys.end())
-				keyIx.push_back(item_names.size() - 1);
-		}
-	}
-
-	// a.reorderByIndex();
-	// b.reorderByIndex();
 
 	auto rowEqual = [&](const_row_handle &a, const_row_handle &b)
 	{
 		int d = 0;
 
-		for (auto kix : keyIx)
+		for (const auto &item_name : keys)
 		{
-			std::string item_name;
-			compType compare;
-
-			std::tie(item_name, compare) = item_names[kix];
-
 			d = a[item_name].compare(b[item_name]);
 
 			if (d != 0)
@@ -2806,28 +2755,6 @@ bool category::operator==(const category &rhs) const
 		if (not rowEqual(ra, rb))
 			return false;
 
-		std::vector<std::string> missingA, missingB, different;
-
-		for (auto &tt : item_names)
-		{
-			std::string item_name;
-			compType compare;
-
-			std::tie(item_name, compare) = tt;
-
-			// make it an option to compare unapplicable to empty or something
-
-			// auto ta = ra[item_name].text();
-			// if (ta == "." or ta == "?")
-			// 	ta = "";
-			// auto tb = rb[item_name].text();
-			// if (tb == "." or tb == "?")
-			// 	tb = "";
-
-			if (ra[item_name].compare(rb[item_name]) != 0)
-				return false;
-		}
-
 		++ai;
 		++bi;
 	}

src/model.cpp

@@ -26,6 +26,7 @@
 
 #include "cif++/cif++.hpp"
 #include "cif++/item.hpp"
+#include "cif++/row.hpp"
 
 #include <algorithm>
 #include <cassert>
@@ -76,10 +77,10 @@ void atom::atom_impl::moveTo(const point &p)
 
 // const compound *compound() const;
 
-const item_value &atom::atom_impl::get_property(std::string_view name) const
+const item_handle atom::atom_impl::get_property(std::string_view name) const
 {
 	if (auto rh = row(); rh)
-		return rh[name].value();
+		return rh[name];
 	throw std::runtime_error(std::format("Missing property {} for atom", name));
 }
 
@@ -1911,7 +1912,13 @@ void structure::change_residue(residue &res, const std::string &newCompound,
 	for (const auto &[a1, a2] : remappedAtoms)
 	{
 		auto i = std::ranges::find_if(atoms, [id = a1](const atom &a)
-			{ return a.get_label_atom_id() == id; });
+			{
+				if (a.get_row())
+				{
+					auto ih = a.get_property_value("label_atom_id");
+					return not ih.empty() and ih.get<std::string>() == id;
+				}
+				return false; });
 		if (i == atoms.end())
 		{
 			if (VERBOSE >= 0)
@@ -2170,29 +2177,50 @@ std::string structure::create_non_poly(const std::string &entity_id, const std::
 	{
 		auto atom_id = atom_site.get_unique_id("");
 
-		auto row = atom_site.emplace({ { "group_PDB", atom.get_property("group_PDB") },
-			{ "id", atom_id },
-			{ "type_symbol", atom.get_property_value("type_symbol") },
-			{ "label_atom_id", atom.get_property_value("label_atom_id") },
-			{ "label_alt_id", atom.get_property_value("label_alt_id") },
-			{ "label_comp_id", comp_id },
-			{ "label_asym_id", asym_id },
-			{ "label_entity_id", entity_id },
-			{ "label_seq_id", cif::item_value_type::INAPPLICABLE },
-			{ "pdbx_PDB_ins_code", cif::item_value_type::MISSING },
-			{ "Cartn_x", atom.get_property_value("Cartn_x") },
-			{ "Cartn_y", atom.get_property_value("Cartn_y") },
-			{ "Cartn_z", atom.get_property_value("Cartn_z") },
-			{ "occupancy", atom.get_property_value("occupancy") },
-			{ "B_iso_or_equiv", atom.get_property_value("B_iso_or_equiv") },
-			{ "pdbx_formal_charge", atom.get_property_value("pdbx_formal_charge") },
-			{ "auth_seq_id", "1" },
-			{ "auth_comp_id", comp_id },
-			{ "auth_asym_id", asym_id },
-			{ "auth_atom_id", atom.get_property_value("label_atom_id") },
-			{ "pdbx_PDB_model_num", m_model_nr } });
+		cif::row_initializer data //
+			{
+				{ "id", atom_id },
+				{ "label_comp_id", comp_id },
+				{ "label_asym_id", asym_id },
+				{ "label_entity_id", entity_id },
+				{ "label_seq_id", cif::item_value_type::INAPPLICABLE },
+				{ "pdbx_PDB_ins_code", cif::item_value_type::MISSING },
+				{ "auth_seq_id", "1" },
+				{ "auth_comp_id", comp_id },
+				{ "auth_asym_id", asym_id },
+				{ "pdbx_PDB_model_num", m_model_nr } //
+			};
+
+		for (auto item : std::initializer_list<std::string>{
+				 // clang-format off
+				 "group_PDB",
+				 "type_symbol",
+				 "label_atom_id",
+				 "label_alt_id",
+				 "auth_atom_id",
+				 "Cartn_x",
+				 "Cartn_y",
+				 "Cartn_z",
+				 "occupancy",
+				 "B_iso_or_equiv",
+				 "pdbx_formal_charge"
+				 // clang-format on
+			 })
+		{
+			auto v = atom.get_property_value(item);
+			if (not v.empty())
+				data.push_back({ item, v.value() });
+			else
+				data.push_back({ item, cif::item_value_type::INAPPLICABLE });
+		}
+
+		auto row = atom_site.emplace(std::move(data));
 
 		auto &newAtom = emplace_atom(std::make_shared<atom::atom_impl>(m_db, atom_id));
+
+		if (newAtom.get_property_value("auth_atom_id").empty() and not newAtom.get_property_value("label_atom_id").empty())
+			newAtom.set_property("auth_atom_id", newAtom.get_property("label_atom_id"));
+
 		res.add_atom(newAtom);
 	}