* reader stub. Navigates the file successfullly
* works
* works
* read out all metadata, not just ours
* first pass at reading out molecules
* support mol blocks too
* add python wrapper for parser
* add direct writer
* get rid of multiple definitions in PNParser.h
* update from code review
* robustification
* handle reading compressed metadata
* support compressing metadata too
* reorder arguments to make this more consistent
* add writers to python wrappers
* forgotten file
* add pickle support
* explicit zlib dependency
* get windows builds working
at least with conda boost+zlib
* switch to using boost::iostreams to do the compression/decompression
* switch to using a vector of string pairs to store the metadata
need this so that we can contain "duplicate" keys
* add metadata output to MolDraw2DCairo
still need python test
* add a python test
* initial work at reading/writing reactions from PNGs
refactor the ReactionParser.h header a bit
* cleanup debug messages
* reaction PNG support -> python
* ReactionPickler no longer includes all ProductTemplate props
* handle metadata at the MolDraw2D level
Currently only supported by MolDraw2DCairo, it's worth extending this to SVG too
* support reading multiple molecules from a png
* support multiple molecules from files and in python
* stop duplicating tags with multiple molecules
* update to get windll builds working
this change should be propagated to more cmakelists.txt files
* make sure the metadata ends up in the notebook
* make sure PNGs in the notebook also have metadata
efficiency improvements for some notebook bits (i.e stop going PNG->Image->PNG)
* no need to pretend that we might be using PIL anymore
* documentation
* update docs to show new functionality
* not sure why the doctests failed on linux
* still trying to diagnose those failures
* protect doctests in case python interpreters are being re-used
* switch tags
* the wrapped functions to read png data from files weren't working
* <sigh>... windows dlls
