Store.groupby enhancements

src/maggma/core/store.py

@@ -111,7 +111,7 @@ def query(
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -161,7 +161,7 @@ def groupby(
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -190,10 +190,12 @@ def query_one(
         Queries the Store for a single document.
 
         Args:
-            criteria: PyMongo filter for documents to search
-            properties: properties to return in the document
+            criteria: PyMongo filter for documents to search in
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
+            skip: number documents to skip
+            limit: limit on total number of documents returned
         """
         return next(self.query(criteria=criteria, properties=properties, sort=sort), None)
 

src/maggma/stores/advanced_stores.py

@@ -243,7 +243,7 @@
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -292,7 +292,7 @@
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -317,7 +317,14 @@
 
         lazy_substitute(criteria, self.reverse_aliases)
 
-        return self.store.groupby(keys=keys, properties=properties, criteria=criteria, skip=skip, limit=limit)
+        return self.store.groupby(
+            keys=keys,
+            properties=properties,
+            criteria=criteria,
+            skip=skip,
+            limit=limit,
+            sort=sort,
+        )
 
     def update(self, docs: Union[List[Dict], Dict], key: Union[List, str, None] = None):
         """
@@ -443,7 +450,7 @@
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -468,7 +475,7 @@
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -479,7 +486,14 @@
         """
         criteria = dict(**criteria, **self.sbx_criteria) if criteria else self.sbx_criteria
 
-        return self.store.groupby(keys=keys, properties=properties, criteria=criteria, skip=skip, limit=limit)
+        return self.store.groupby(
+            keys=keys,
+            properties=properties,
+            criteria=criteria,
+            skip=skip,
+            limit=limit,
+            sort=sort,
+        )
 
     def update(self, docs: Union[List[Dict], Dict], key: Union[List, str, None] = None):
         """

src/maggma/stores/aws.py

@@ -172,12 +172,12 @@ def query(
         Queries the Store for a set of documents.
 
         Args:
-            criteria: PyMongo filter for documents to search in.
-            properties: properties to return in grouped documents.
-            sort: Dictionary of sort order for fields. Keys are field names and values
-                are 1 for ascending or -1 for descending.
-            skip: number documents to skip.
-            limit: limit on total number of documents returned.
+            criteria: PyMongo filter for documents to search in
+            properties: fields to include in returned documents. By default, all fields are returned.
+            sort: Dictionary of sort order for fields. Keys are field names and
+                values are 1 for ascending or -1 for descending.
+            skip: number documents to skip
+            limit: limit on total number of documents returned
 
         """
         prop_keys = set()
@@ -265,13 +265,13 @@ def groupby(
         Simple grouping function that will group documents by keys.
 
         Args:
-            keys: fields to group documents.
-            criteria: PyMongo filter for documents to search in.
-            properties: properties to return in grouped documents.
-            sort: Dictionary of sort order for fields. Keys are field names and values
-            are 1 for ascending or -1 for descending.
-            skip: number documents to skip.
-            limit: limit on total number of documents returned.
+            keys: fields to group documents
+            criteria: PyMongo filter for documents to search in
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
+            sort: Dictionary of sort order for fields. Keys are field names and
+                values are 1 for ascending or -1 for descending.
+            skip: number documents to skip
+            limit: limit on total number of documents returned
 
         Returns:
             generator returning tuples of (dict, list of docs)

src/maggma/stores/azure.py

@@ -192,7 +192,7 @@ def query(
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -266,7 +266,7 @@ def groupby(
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip

src/maggma/stores/compound_stores.py

@@ -240,6 +240,7 @@ def query(
         agg = self._collection.aggregate(pipeline)
         yield from agg
 
+    # TODO - sort kwarg is not passed anywhere
     def groupby(
         self,
         keys: Union[List[str], str],
@@ -267,8 +268,8 @@ def query_one(self, criteria=None, properties=None, **kwargs):
         Get one document.
 
         Args:
-            properties: properties to return in query
-            criteria: filter for matching
+            criteria: PyMongo filter for documents to search in
+            properties: fields to include in returned documents. By default, all fields are returned.
             kwargs: kwargs for collection.aggregate
 
         Returns:
@@ -436,7 +437,7 @@ def query(
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -463,7 +464,7 @@ def groupby(
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip

src/maggma/stores/file_store.py

@@ -388,7 +388,7 @@ def query(  # type: ignore
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             hint: Dictionary of indexes to use as hints for query optimizer.

src/maggma/stores/gridfs.py

@@ -224,7 +224,7 @@ def query(
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -289,6 +289,7 @@ def distinct(self, field: str, criteria: Optional[Dict] = None, all_exist: bool
 
         return self._files_store.distinct(field=field, criteria=criteria)
 
+    # TODO - sort, skip, limit, and properties are not functional
     def groupby(
         self,
         keys: Union[List[str], str],
@@ -306,7 +307,7 @@ def groupby(
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip

src/maggma/stores/mongolike.py

@@ -189,6 +189,7 @@
 
         return distinct_vals if distinct_vals is not None else []
 
+    # TODO - sort, skip, and limit are not functional
     def groupby(
         self,
         keys: Union[List[str], str],
@@ -205,7 +206,7 @@
         Args:
             keys: fields to group documents
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             skip: number documents to skip
@@ -308,7 +309,7 @@
 
         Args:
             criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
+            properties: fields to include in returned documents. By default, all fields are returned.
             sort: Dictionary of sort order for fields. Keys are field names and
                 values are 1 for ascending or -1 for descending.
             hint: Dictionary of indexes to use as hints for query optimizer.
@@ -553,51 +554,6 @@
         """Hash for the store."""
         return hash((self.name, self.last_updated_field))
 
-    def groupby(
-        self,
-        keys: Union[List[str], str],
-        criteria: Optional[Dict] = None,
-        properties: Union[Dict, List, None] = None,
-        sort: Optional[Dict[str, Union[Sort, int]]] = None,
-        skip: int = 0,
-        limit: int = 0,
-    ) -> Iterator[Tuple[Dict, List[Dict]]]:
-        """
-        Simple grouping function that will group documents
-        by keys.
-
-        Args:
-            keys: fields to group documents
-            criteria: PyMongo filter for documents to search in
-            properties: properties to return in grouped documents
-            sort: Dictionary of sort order for fields. Keys are field names and
-                values are 1 for ascending or -1 for descending.
-            skip: number documents to skip
-            limit: limit on total number of documents returned
-
-        Returns:
-            generator returning tuples of (key, list of elements)
-        """
-        keys = keys if isinstance(keys, list) else [keys]
-
-        if properties is None:
-            properties = []
-        if isinstance(properties, dict):
-            properties = list(properties.keys())
-
-        data = [
-            doc for doc in self.query(properties=keys + properties, criteria=criteria) if all(has(doc, k) for k in keys)
-        ]
-
-        def grouping_keys(doc):
-            return tuple(get(doc, k) for k in keys)
-
-        for vals, group in groupby(sorted(data, key=grouping_keys), key=grouping_keys):
-            doc = {}  # type: ignore
-            for k, v in zip(keys, vals):
-                set_(doc, k, v)
-            yield doc, list(group)
-
     def __eq__(self, other: object) -> bool:
         """
         Check equality for MemoryStore
@@ -944,3 +900,49 @@
                 search_doc = {k: d[k] for k in key} if isinstance(key, list) else {key: d[key]}
 
                 self._collection.replace_one(search_doc, d, upsert=True)
+
+    # Moved this from MemoryStore b/c MontyDB does not implement aggregate() as
+    # of May 2022. See https://github.com/davidlatwe/montydb/issues/66
+    def groupby(
+        self,
+        keys: Union[List[str], str],
+        criteria: Optional[Dict] = None,
+        properties: Union[Dict, List, None] = None,
+        sort: Optional[Dict[str, Union[Sort, int]]] = None,
+        skip: int = 0,
+        limit: int = 0,
+    ) -> Iterator[Tuple[Dict, List[Dict]]]:
+        """
+        Simple grouping function that will group documents
+        by keys.
+
+        Args:
+            keys: fields to group documents
+            criteria: PyMongo filter for documents to search in
+            properties: fields to include in grouped documents. By default, only the 'id' field is returned.
+            sort: Dictionary of sort order for fields. Keys are field names and
+                values are 1 for ascending or -1 for descending.
+            skip: number documents to skip
+            limit: limit on total number of documents returned
+        Returns:
+            generator returning tuples of (key, list of elements)
+        """
+        keys = keys if isinstance(keys, list) else [keys]
+
+        if properties is None:
+            properties = []
+        if isinstance(properties, dict):
+            properties = list(properties.keys())
+
+        data = [
+            doc for doc in self.query(properties=keys + properties, criteria=criteria) if all(has(doc, k) for k in keys)
+        ]
+
+        def grouping_keys(doc):
+            return tuple(get(doc, k) for k in keys)
+
+        for vals, group in groupby(sorted(data, key=grouping_keys), key=grouping_keys):
+            doc = {}  # type: Dict[Any,Any]
+            for k, v in zip(keys, vals):
+                set_(doc, k, v)
+            yield doc, list(group)