M V30 COUNTS kw metadata parser added.

cimm-kzn · Jan 19, 2021 · 78ce928 · 78ce928
1 parent 23d720a
commit 78ce928
Show file tree

Hide file tree

Showing 9 changed files with 31 additions and 23 deletions.
diff --git a/CGRtools/files/MRVrw.py b/CGRtools/files/MRVrw.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2017-2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2017-2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -146,7 +146,7 @@ def __reader(self):
                     self._info(f'record consist errors:\n{format_exc()}')
                     yield parse_error(n, parsed, self._format_log(), meta)
                 else:
-                    record['meta'] = meta
+                    record['meta'].update(meta)
                     try:
                         container = self._convert_structure(record)
                     except ValueError:
@@ -302,7 +302,7 @@ def __parse_molecule(self, data):
                         self._info('incorrect bondStereo tag')
                 bonds.append((atom_map[a1], atom_map[a2], order))
 
-        mol = {'atoms': atoms, 'bonds': bonds, 'stereo': stereo}
+        mol = {'atoms': atoms, 'bonds': bonds, 'stereo': stereo, 'meta': {}}
         if '@title' in data:
             mol['title'] = data['@title']
         return mol

diff --git a/CGRtools/files/RDFrw.py b/CGRtools/files/RDFrw.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2014-2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2014-2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  Copyright 2019 Dinar Batyrshin <batyrshin-dinar@mail.ru>
 #  This file is part of CGRtools.
 #
@@ -164,7 +164,7 @@ def __reader(self):
                     self._flush_log()
             elif line.startswith('$RFMT'):
                 if record:
-                    record['meta'] = self._prepare_meta(meta)
+                    record['meta'].update(self._prepare_meta(meta))
                     if title:
                         record['title'] = title
                     try:
@@ -200,7 +200,7 @@ def __reader(self):
                 meta = defaultdict(list)
             elif line.startswith('$MFMT'):
                 if record:
-                    record['meta'] = self._prepare_meta(meta)
+                    record['meta'].update(self._prepare_meta(meta))
                     if title:
                         record['title'] = title
                     try:
@@ -270,7 +270,7 @@ def __reader(self):
                         self.__already_seeked = False
                     self._flush_log()
         if record:
-            record['meta'] = self._prepare_meta(meta)
+            record['meta'].update(self._prepare_meta(meta))
             if title:
                 record['title'] = title
             try:

diff --git a/CGRtools/files/SDFrw.py b/CGRtools/files/SDFrw.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2014-2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2014-2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -147,7 +147,7 @@ def __reader(self):
                     self._flush_log()
             elif line.startswith("$$$$"):
                 if record:
-                    record['meta'] = self._prepare_meta(meta)
+                    record['meta'].update(self._prepare_meta(meta))
                     if title:
                         record['title'] = title
                     try:
@@ -220,7 +220,7 @@ def __reader(self):
                     self._flush_log()
 
         if record:  # True for MOL file only.
-            record['meta'] = self._prepare_meta(meta)
+            record['meta'].update(self._prepare_meta(meta))
             if title:
                 record['title'] = title
             try:

diff --git a/CGRtools/files/SMILESrw.py b/CGRtools/files/SMILESrw.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2018-2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2018-2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  Copyright 2019 Artem Mukanov <nostro32@mail.ru>
 #  This file is part of CGRtools.
 #
@@ -254,7 +254,7 @@ def parse(self, smiles: str) -> Union[MoleculeContainer, CGRContainer, ReactionC
                 self._info(f'line: {smi}\nconsist errors:\n{format_exc()}')
                 return meta
 
-            record['meta'] = meta
+            record['meta'].update(meta)
             try:
                 container = self._convert_structure(record)
             except ValueError:
@@ -728,7 +728,7 @@ def _parse_tokens(self, tokens):
 
         stereo_bonds = {n: ms for n, ms in stereo_bonds.items() if len(ms) == 1 or len(ms) == set(ms.values())}
         mol = {'atoms': atoms, 'bonds': bonds, 'order': order,
-               'stereo_bonds': stereo_bonds, 'stereo_atoms': stereo_atoms, 'hydrogens': hydrogens}
+               'stereo_bonds': stereo_bonds, 'stereo_atoms': stereo_atoms, 'hydrogens': hydrogens, 'meta': {}}
         if cgr or any(x == 11 for x in atoms_types):
             mol['cgr'] = cgr
         return mol

diff --git a/CGRtools/files/_mdl/emol.py b/CGRtools/files/_mdl/emol.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2020, 2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -26,14 +26,15 @@ def __init__(self, log_buffer=None):
         self.__bonds = []
         self.__atom_map = {}
         self.__stereo = []
+        self.__meta = {}
         if log_buffer is None:
             log_buffer = []
         self.__log_buffer = log_buffer
 
     def getvalue(self):
         if self.__in_mol or self.__in_mol is None:
             raise ValueError('molecule not complete')
-        return {'atoms': self.__atoms, 'bonds': self.__bonds, 'stereo': self.__stereo}
+        return {'atoms': self.__atoms, 'bonds': self.__bonds, 'stereo': self.__stereo, 'meta': self.__meta}
 
     def __call__(self, line, lineu=None):
         if lineu is None:
@@ -73,12 +74,17 @@ def __call__(self, line, lineu=None):
                     raise ValueError('invalid CTAB')
 
             else:  # M  V30 COUNTS line expected
-                a, b, *_ = line[13:].split()
+                a, b, *meta = line[13:].split()
                 atom_count = int(a)
                 if not atom_count:
                     raise EmptyMolecule
                 self.__bonds_count = int(b)
                 self.__atoms_count = atom_count
+                for kv in meta:
+                    if '=' in kv:
+                        k, v = kv.split('=', 1)
+                        if k and v:
+                            self.__meta[k] = v
 
         elif self.__in_mol is not None:
             raise SyntaxError('invalid usage')

diff --git a/CGRtools/files/_mdl/erxn.py b/CGRtools/files/_mdl/erxn.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2020, 2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -90,7 +90,7 @@ def __call__(self, line):
 
     def getvalue(self):
         if self.__rend:
-            return {'reactants': self.__reactants, 'products': self.__products, 'reagents': self.__reagents}
+            return {'reactants': self.__reactants, 'products': self.__products, 'reagents': self.__reagents, 'meta': {}}
         raise ValueError('reaction not complete')
 
     __parser_group = __parser = None

diff --git a/CGRtools/files/_mdl/mol.py b/CGRtools/files/_mdl/mol.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2020, 2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -57,7 +57,7 @@ def __new__(cls, line, log_buffer=None):
 
     def getvalue(self):
         if self.__mend:
-            mol = {'atoms': self.__atoms, 'bonds': self.__bonds, 'stereo': self.__stereo}
+            mol = {'atoms': self.__atoms, 'bonds': self.__bonds, 'stereo': self.__stereo, 'meta': {}}
             if self.__cgr:
                 mol['cgr'] = self.__cgr
             if self.__query:

diff --git a/CGRtools/files/_mdl/parser.py b/CGRtools/files/_mdl/parser.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2014-2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2014-2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -117,6 +117,7 @@ def _convert_reaction(self, reaction):
                 remapped = {x: y for x, y in enumerate(tmp[shift: atom_len + shift])}
                 shift += atom_len
                 g = self.__prepare_structure(j, remapped)
+                g.meta.update(j['meta'])
                 rc[i].append(g)
         return ReactionContainer(meta=reaction['meta'], name=reaction.get('title'), **rc)
 

diff --git a/CGRtools/files/_mdl/rxn.py b/CGRtools/files/_mdl/rxn.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-#  Copyright 2020 Ramil Nugmanov <nougmanoff@protonmail.com>
+#  Copyright 2020, 2021 Ramil Nugmanov <nougmanoff@protonmail.com>
 #  This file is part of CGRtools.
 #
 #  CGRtools is free software; you can redistribute it and/or modify
@@ -83,7 +83,8 @@ def getvalue(self):
         if self.__rend:
             return {'reactants': self.__molecules[:self.__reactants_count],
                     'products': self.__molecules[self.__reactants_count:self.__products_count],
-                    'reagents': self.__molecules[self.__products_count:self.__reagents_count]}
+                    'reagents': self.__molecules[self.__products_count:self.__reagents_count],
+                    'meta': {}}
         raise ValueError('reaction not complete')
 
     __parser = None