Starting to fix references

api/api.py

@@ -5,7 +5,11 @@
 
 from api.models import (CoupleResult, Item, ReferenceInfo, ReferenceResult,
                         StructureInfo, StructureResult, TaxonInfo, TaxonResult)
-from api.queries import (get_matching_references_from_reference_in_item,
+from api.queries import (
+                         # get_matching_references_from_couple_in_item,
+                         get_matching_references_from_reference_in_item,
+                         get_matching_references_from_structure_in_item,
+                         get_matching_references_from_taxon_in_item,
                          get_matching_structures_from_reference_in_item,
                          get_matching_structures_from_structure_in_item,
                          get_matching_structures_from_taxon_in_item,
@@ -183,9 +187,9 @@ async def search_references(item: Item) -> ReferenceResult:
     )
 
     # We want the set of all the references which have couples matching the query
-    matching_references_by_couple = get_matching_references_from_couple_in_item(
-        dm, item
-    )
+    # matching_references_by_couple = get_matching_references_from_couple_in_item(
+    #     dm, item
+    # )
 
     # We want the set of all the references which have structures matching the query
     matching_references_by_structure = get_matching_references_from_structure_in_item(
@@ -200,18 +204,18 @@ async def search_references(item: Item) -> ReferenceResult:
         matching_references_by_structure
         & matching_references_by_taxon
         & matching_references_by_reference
-        & matching_references_by_couple
+        # & matching_references_by_couple
         if matching_references_by_structure
         and matching_references_by_taxon
         and matching_references_by_reference
-        and matching_references_by_couple
+        # and matching_references_by_couple
         else matching_references_by_structure
         or matching_references_by_taxon
         or matching_references_by_reference
-        or matching_references_by_couple
+        # or matching_references_by_couple
     )
 
-    items = list(dm.get_dict_of_sid_to_smiles(matching_structures).items())
+    items = list(dm.get_dict_of_rid_to_ref_doi(matching_references).items())
 
     if item.limit == 0:
         items = items
@@ -220,7 +224,7 @@ async def search_references(item: Item) -> ReferenceResult:
 
     return ReferenceResult(
         ids=matching_references,
-        references={rid: ReferenceInfo(name=value) for rid, value in items},
+        references={rid: ReferenceInfo(doi=value) for rid, value in items},
         description="References matching the query",
         count=len(matching_references),
     )

api/queries.py

@@ -9,6 +9,28 @@
 )
 
 
+def get_matching_references_from_reference_in_item(
+    dm: DataModel, item: Item
+) -> set[int] | None:
+    """Returns the WID of matching references."""
+    if item.reference_wid is None and item.reference_doi is None:
+        return None
+    else:
+        # This needs to be explained in the API doc
+        if item.reference_wid:
+            if item.reference_wid in dm.get_references():
+                return {item.reference_wid}
+        else:
+            if item.reference_doi:
+                references = set(dm.get_references_with_doi(item.reference_doi))
+                if references is None:
+                    references = dm.get_references()
+            else:
+                references = dm.get_references()
+
+        return references
+
+
 def get_matching_structures_from_structure_in_item(
     dm: DataModel, item: Item
 ) -> set[int]:
@@ -75,39 +97,58 @@ def get_matching_taxa_from_taxon_in_item(dm: DataModel, item: Item) -> set[int]
         return taxa
 
 
-def get_matching_references_from_reference_in_item(
+# TODO WIP
+# def get_matching_references_from_couple_in_item(dm: DataModel, item: Item) -> set[int]:
+#     # We need to get all the matching taxa
+#     taxa = get_matching_taxa_from_taxon_in_item(dm, item)
+#     # We need to get all the matching structures
+#     structures = get_matching_structures_from_structure_in_item(dm, item)
+
+#     if taxa is None:
+#         return None
+
+#     if structures is None:
+#         return None
+
+#     tax = set()
+#     for taxon in taxa:
+#         tax.update(dm.get_structures_of_taxon(taxon))
+
+#     stru = set()
+#     for structure in structures:
+#         stru.update(dm.get_references_containing_structure(structure))
+
+#     # TODO get couples and intersect
+
+#     return out
+
+
+def get_matching_references_from_structure_in_item(
     dm: DataModel, item: Item
-) -> set[int] | None:
-    """Returns the WID of matching references."""
-    if item.reference_wid is None and item.reference_doi is None:
+) -> set[int]:
+    # We need to get all the matching structures
+    structures = get_matching_structures_from_structure_in_item(dm, item)
+
+    if structures is None:
         return None
-    else:
-        # This needs to be explained in the API doc
-        if item.reference_wid:
-            if item.reference_wid in dm.get_refs():
-                return {item.reference_wid}
-        else:
-            if item.reference_doi:
-                references = set(dm.get_references_with_doi(item.reference_doi))
-                if references is None:
-                    references = dm.get_refs()
-            else:
-                references = dm.get_refs()
 
-        return references
+    out = set()
+    for structure in structures:
+        out.update(dm.get_references_containing_structure(structure))
+
+    return out
 
 
-def get_matching_structures_from_taxon_in_item(dm: DataModel, item: Item) -> set[int]:
+def get_matching_references_from_taxon_in_item(dm: DataModel, item: Item) -> set[int]:
     # We need to get all the matching taxa
     taxa = get_matching_taxa_from_taxon_in_item(dm, item)
 
     if taxa is None:
         return None
 
-    # Set recursive=True to have all the structures from the parents too
     out = set()
     for taxon in taxa:
-        out.update(dm.get_structures_of_taxon(taxon))
+        out.update(dm.get_references_containing_taxon(taxon))
 
     return out
 
@@ -128,6 +169,21 @@ def get_matching_structures_from_reference_in_item(
     return out
 
 
+def get_matching_structures_from_taxon_in_item(dm: DataModel, item: Item) -> set[int]:
+    # We need to get all the matching taxa
+    taxa = get_matching_taxa_from_taxon_in_item(dm, item)
+
+    if taxa is None:
+        return None
+
+    # Set recursive=True to have all the structures from the parents too
+    out = set()
+    for taxon in taxa:
+        out.update(dm.get_structures_of_taxon(taxon))
+
+    return out
+
+
 def get_matching_taxa_from_structure_in_item(dm: DataModel, item: Item) -> set[int]:
     # We need to get all the matching structures
     structures = get_matching_structures_from_structure_in_item(dm, item)

model.py

@@ -242,7 +242,7 @@ def structure_get_tsv_from_scores(self, sids: list[int], scores) -> str:
 
     ### Biblionomy
     @functools.lru_cache(maxsize=None)
-    def get_refs(self) -> dict[int, str]:
+    def get_references(self) -> dict[int, str]:
         return self.db["reference_doi"]
 
     def get_ref_doi_from_list_of_rids(self, rids: list[int]) -> list[str]:
@@ -334,6 +334,7 @@ def get_number_of_references_containing_couple(self, sid: int, tid: int) -> int:
             return len(self.db["tc2r"][key])
         return 0
 
+    # TODO not working, fix
     def get_references_containing_structure(self, sid: int) -> list[int]:
         references = [
             rid
@@ -352,7 +353,8 @@ def get_number_of_references_containing_structure(self, sid: int) -> int:
         ]
         return len(references)
 
-    def get_references_containing_taxa(self, tid: int) -> list[int]:
+    # TODO not working, fix
+    def get_references_containing_taxon(self, tid: int) -> list[int]:
         references = [
             rid
             for (taxon_id, sid), rids in self.db["tc2r"].items()
@@ -361,7 +363,7 @@ def get_references_containing_taxa(self, tid: int) -> list[int]:
         ]
         return references
 
-    def get_number_of_references_containing_taxa(self, tid: int) -> int:
+    def get_number_of_references_containing_taxon(self, tid: int) -> int:
         references = [
             rid
             for (taxon_id, sid), rids in self.db["tc2r"].items()