Make ChemicalMetadata __repr__

chemicals/identifiers.py

@@ -247,7 +247,46 @@ class ChemicalMetadata:
 
     __slots__ = ('pubchemid', 'formula', 'MW', 'smiles', 'InChI', 'InChI_key',
                  'iupac_name', 'common_name', 'synonyms', 'CAS', '_charge')
+
     def __repr__(self):
+        return (f"ChemicalMetadata(pubchemid={self.pubchemid!r}, "
+                f"CAS={self.CAS!r}, "
+                f"formula={self.formula!r}, "
+                f"MW={self.MW!r}, "
+                f"smiles={self.smiles!r}, "
+                f"InChI={self.InChI!r}, "
+                f"InChI_key={self.InChI_key!r}, "
+                f"iupac_name={self.iupac_name!r}, "
+                f"common_name={self.common_name!r}, "
+                f"synonyms={self.synonyms!r})")
+
+    def __eq__(self, other):
+        if self is other:
+            return True
+        return (self.pubchemid == other.pubchemid and
+                self.CAS == other.CAS and
+                self.formula == other.formula and
+                self.MW == other.MW and
+                self.smiles == other.smiles and
+                self.InChI == other.InChI and
+                self.InChI_key == other.InChI_key and
+                self.iupac_name == other.iupac_name and
+                self.common_name == other.common_name and
+                self.synonyms == other.synonyms)
+
+    def __hash__(self):
+        return hash((self.pubchemid,
+                    self.CAS,
+                    self.formula,
+                    self.MW,
+                    self.smiles, 
+                    self.InChI,
+                    self.InChI_key,
+                    self.iupac_name,
+                    self.common_name,
+                    tuple(self.synonyms)))
+
+    def __str__(self):
         return (f'<ChemicalMetadata, name={self.common_name}, formula={self.formula}, smiles={self.smiles}, MW={self.MW:g}>')
 
     @property
@@ -610,17 +649,17 @@ def search_chemical(ID, autoload=False, cache=True):
 
     Examples
     --------
-    >>> search_chemical('water')
+    >>> print(search_chemical('water'))
     <ChemicalMetadata, name=water, formula=H2O, smiles=O, MW=18.0153>
-    >>> search_chemical('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3')
+    >>> print(search_chemical('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3'))
     <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
-    >>> search_chemical('CCCCCCCCCC')
+    >>> print(search_chemical('CCCCCCCCCC'))
     <ChemicalMetadata, name=decane, formula=C10H22, smiles=CCCCCCCCCC, MW=142.282>
-    >>> search_chemical('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N')
+    >>> print(search_chemical('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N'))
     <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
-    >>> search_chemical('pubchem=702')
+    >>> print(search_chemical('pubchem=702'))
     <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
-    >>> search_chemical('O') # only elements can be specified by symbol
+    >>> print(search_chemical('O')) # only elements can be specified by symbol
     <ChemicalMetadata, name=oxygen, formula=O, smiles=[O], MW=15.9994>
     """
     if cache and ID in chemical_search_cache:

docs/tutorial.rst

@@ -152,17 +152,17 @@ inputs may be specified as
 
 Some sample queries illustrating the topic:
 
->>> search_chemical('water')
+>>> print(search_chemical('water'))
 <ChemicalMetadata, name=water, formula=H2O, smiles=O, MW=18.0153>
->>> search_chemical('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3')
+>>> print(search_chemical('InChI=1S/C2H6O/c1-2-3/h3H,2H2,1H3'))
 <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
->>> search_chemical('CCCCCCCCCC')
+>>> print(search_chemical('CCCCCCCCCC'))
 <ChemicalMetadata, name=decane, formula=C10H22, smiles=CCCCCCCCCC, MW=142.282>
->>> search_chemical('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N')
+>>> print(search_chemical('InChIKey=LFQSCWFLJHTTHZ-UHFFFAOYSA-N'))
 <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
->>> search_chemical('pubchem=702')
+>>> print(search_chemical('pubchem=702'))
 <ChemicalMetadata, name=ethanol, formula=C2H6O, smiles=CCO, MW=46.0684>
->>> search_chemical('O') # only elements can be specified by symbol
+>>> print(search_chemical('O')) # only elements can be specified by symbol
 <ChemicalMetadata, name=oxygen, formula=O, smiles=[O], MW=15.9994>
 
 

tests/test_identifiers.py

@@ -34,6 +34,7 @@
     CAS_from_any,
     CAS_to_int,
     ChemicalMetadataDB,
+    ChemicalMetadata,
     IDs_to_CASs,
     check_CAS,
     common_mixtures,
@@ -1448,6 +1449,73 @@ def test_sorted_CAS_key():
 
 
 
+def test_ChemicalMetadata_basics():
+    from copy import deepcopy
+    import pickle
+    obj = ChemicalMetadata(pubchemid=712, CAS=50000, formula='CH2O', MW=30.02598, smiles='C=O', InChI='CH2O/c1-2/h1H2', 
+                    InChI_key='WSFSSNUMVMOOMR-UHFFFAOYSA-N', iupac_name='methanal', common_name='formaldehyde', 
+                    synonyms=['methanal', 'formaldehyde', 'formalin', 'methanal', 'methylene oxide'])
+    assert eval(obj.__repr__()) == obj
+    assert hash(eval(obj.__repr__())) == hash(obj)
+
+    # Test equality with identical object
+    assert obj == obj
+
+    # Test hash consistency
+    assert hash(obj) == hash(obj)
+
+    # Test deepcopy
+    obj_copy = deepcopy(obj)
+    assert obj == obj_copy
+    assert hash(obj) == hash(obj_copy)
+
+    # Test pickling
+    obj_pickle = pickle.loads(pickle.dumps(obj))
+    assert obj_pickle == obj
+    assert hash(obj_pickle) == hash(obj)
+
+    # Test hash consistency after accessing computed property
+    hash_before = hash(obj)
+    _ = obj.charge
+    hash_after = hash(obj)
+    assert hash_before == hash_after
+
+
+def test_ChemicalMetadata_inequality():
+    base_obj = ChemicalMetadata(
+        pubchemid=712, 
+        CAS=50000, 
+        formula='CH2O', 
+        MW=30.02598, 
+        smiles='C=O', 
+        InChI='CH2O/c1-2/h1H2', 
+        InChI_key='WSFSSNUMVMOOMR-UHFFFAOYSA-N', 
+        iupac_name='methanal', 
+        common_name='formaldehyde', 
+        synonyms=['methanal', 'formaldehyde']
+    )
+
+    # Test inequality with modified attributes
+    modifications = {
+        'pubchemid': 713,
+        'CAS': 50001,
+        'formula': 'CH3O',
+        'MW': 30.02599,
+        'smiles': 'CO',
+        'InChI': 'different_inchi',
+        'InChI_key': 'different_key',
+        'iupac_name': 'different_name',
+        'common_name': 'different_common_name',
+        'synonyms': ['different', 'synonyms']
+    }
+
+    for attr, new_value in modifications.items():
+        kwargs = {k: getattr(base_obj, k) for k in base_obj.__slots__ if k != '_charge'}
+        kwargs[attr] = new_value
+        different_obj = ChemicalMetadata(**kwargs)
+        assert base_obj != different_obj
+        assert hash(base_obj) != hash(different_obj)
+
 def test_formula_search_exceptions():
     """
     Test that FORMULA_SEARCH_BEFORE_SMILES_EXCEPTIONS contains exactly the