WIP #81

api/api.py

@@ -83,7 +83,8 @@ async def search_triplets(
         return TripletResult(
             triplets=triplets,
             references={
-                wid: ReferenceObject(doi=value)
+                # TODO
+                wid: ReferenceObject(doi=value, title="None", date="None", journal="None")
                 for wid, value in dm.get_dict_of_rid_to_reference_doi(
                     [reference_id for reference_id, _, _ in triplets]
                 ).items()
@@ -168,7 +169,9 @@ async def search_references(
         return ReferenceResult(
             ids=dict_items.keys(),
             objects={
-                rid: ReferenceObject(doi=value) for rid, value in dict_items.items()
+                # TODO
+                rid: ReferenceObject(doi=value, title="None", date="None", journal="None")
+                for rid, value in dict_items.items()
             },
             description="References matching the query",
             count=len(dict_items),

api/models.py

@@ -2,8 +2,10 @@
 
 from pydantic import BaseModel
 
-# class ReferenceOption(BaseModel):
-#     TODO
+class ReferenceOption(BaseModel):
+    date_min: Optional[str] = None
+    date_max: Optional[str] = None
+    journal: Optional[str] = None
 
 
 class StructureOption(BaseModel):
@@ -18,7 +20,8 @@ class TaxonOption(BaseModel):
 class ReferenceItem(BaseModel):
     wid: Optional[int] = None
     doi: Optional[str] = None
-    # option: ReferenceOption = ReferenceOption()
+    title: Optional[str] = None
+    option: ReferenceOption = ReferenceOption()
     # limit: Optional[int] = None
 
 
@@ -49,9 +52,12 @@ class Item(BaseModel):
                     "reference": {
                         "wid": 44488598,
                         "doi": "10.1080/1057563021000040466",
-                        # "option": {
-                        #     "TODO": True,
-                        # },
+                        "title": "Iridoids from Seeds of Gentiana Lutea",
+                        "option": {
+                            "date_min": "1999-12-31",
+                            "date_max": "2024-01-01",
+                            "journal": "Natural Product Research",
+                        },
                     },
                     "structure": {
                         "wid": 27151406,
@@ -76,6 +82,9 @@ class Item(BaseModel):
 
 class ReferenceObject(BaseModel):
     doi: str
+    title: str
+    date: str
+    journal: str
 
 
 class ReferenceResult(BaseModel):

api/queries.py

@@ -6,6 +6,7 @@
 from api.models import (  # ReferenceOption,
     Item,
     ReferenceItem,
+    ReferenceOption,
     StructureItem,
     StructureOption,
     TaxonItem,
@@ -20,21 +21,35 @@
 
 def references_from_reference_in_item(dm: DataModel, item: Item) -> set[int] | None:
     """Returns the WID of matching references."""
+    references = None
+
     wid = item.reference.wid
     doi = item.reference.doi
+    title = item.reference.title
+    date_min = item.reference.option.date_min
+    date_max = item.reference.option.date_max
+    journal = item.reference.option.journal
 
-    if doi and wid:
+    if len([param for param in [wid, doi, title] if param is not None]) >= 2:
         raise HTTPException(
             status_code=500,
-            detail=f"You cannot provide both 'doi' and 'wid'",
+            detail=f"Only one of ['wid', 'doi', 'title'] should be provided",
         )
-    elif wid is not None or doi is not None:
-        # This needs to be explained in the API doc
+    elif wid is not None or doi is not None or title is not None:
         if wid:
             return dm.get_reference_with_id(wid)
-        return dm.get_references_with_doi(doi)
-
-    return None
+        elif doi:
+            return dm.get_references_with_doi(doi)
+        # else:
+        #     return dm.get_references_with_title(title)
+        # if date_min:
+        #     TODO
+        # if date_max:
+        #     TODO
+        # if journal:
+        #     TODO
+    else:
+        return references
 
 
 def structures_from_structure_in_item(dm: DataModel, item: Item) -> set[int] | None:
@@ -46,10 +61,10 @@ def structures_from_structure_in_item(dm: DataModel, item: Item) -> set[int] | N
     sub = item.structure.option.substructure_search
     sim = item.structure.option.similarity_level
 
-    if molecule and wid:
+    if len([param for param in [wid, molecule] if param is not None]) >= 2:
         raise HTTPException(
             status_code=500,
-            detail=f"You cannot provide both 'molecule' and 'wid'",
+            detail=f"Only one of ['wid', 'molecule'] should be provided",
         )
     elif molecule is not None or wid is not None:
         # This needs to be explained in the API doc
@@ -88,10 +103,10 @@ def taxa_from_taxon_in_item(dm: DataModel, item: Item) -> set[int] | None:
     name = item.taxon.name
     children = item.taxon.option.taxon_children
 
-    if name and wid:
+    if len([param for param in [wid, name] if param is not None]) >= 2:
         raise HTTPException(
             status_code=500,
-            detail=f"You cannot provide both 'name' and 'wid'",
+            detail=f"Only one of ['wid', 'name'] should be provided",
         )
     if wid is not None or name is not None:
         # This needs to be explained in the API doc

doc/api/schemas/_index.yaml

@@ -8,8 +8,8 @@ referenceItem:
   $ref: "./referenceItem.yaml"
 referenceObject:
   $ref: "./referenceObject.yaml"
-# referenceOption:
-#   $ref: "./referenceOption.yaml"
+referenceOption:
+  $ref: "./referenceOption.yaml"
 referenceResult:
   $ref: "./referenceResult.yaml"
 structureId:

doc/api/schemas/referenceItem.yaml

@@ -13,9 +13,15 @@ properties:
     default: ~
     example: "10.1080/1057563021000040466"
     type: string
+  title:
+    description: |
+      The title of the reference.
+    default: ~
+    example: "Iridoids from Seeds of Gentiana Lutea"
+    type: string
   # external_id:
   #   $ref: "./referenceId.yaml"
-  # option:
-  #   $ref: "./referenceOption.yaml"
+  option:
+    $ref: "./referenceOption.yaml"
   # filter:
   #   $ref: "./referenceFilter.yaml"

doc/api/schemas/referenceObject.yaml

@@ -6,10 +6,20 @@ properties:
       DOI
     example: "10.1080/1057563021000040466"
     type: string
+  title:
+    description: |
+      The title of the reference.
+    example: "Iridoids from Seeds of Gentiana Lutea"
+    type: string
+  date:
+    description: |
+      The date of the reference.
+    example: "2003-08"
+    type: string
+  journal:
+    description: |
+      The journal of the reference.
+    example: "Natural Product Research"
+    type: string
   # external_ids:
   #   $ref: "./referenceId.yaml"
-  # title:
-  #   description: |
-  #     Title
-  #   type: string
-  #   example: "Iridoids from Seeds of Gentiana Lutea"

doc/api/schemas/referenceOption.yaml

@@ -0,0 +1,18 @@
+title: TaxonOption
+type: object
+properties:
+  date_min:
+    description: |
+      The oldest allowed date of the reference.
+    example: "1999-12-31"
+    type: string
+  date_max:
+    description: |
+      The most recent allowed date of the reference.
+    example: "2024-01-01"
+    type: string
+  journal:
+    description: |
+      The journal of the reference.
+    example: "Natural Product Research"
+    type: string

model/data_model.py

@@ -203,9 +203,13 @@ def get_reference_with_id(self, rid: int) -> set[int]:
 
     def get_dict_of_rid_to_reference_doi(self, rid: Iterable[int]) -> dict[int, str]:
         with self.storage.session() as session:
-            result = session.query(References.id, References.doi).filter(
-                References.id.in_(rid)
-            )
+            result = session.query(
+                References.id,
+                References.doi,
+                References.title,
+                References.date,
+                References.journal,
+            ).filter(References.id.in_(rid))
             return {row[0]: row[1] for row in result}
 
     def get_reference_doi_from_rid(self, rid: int) -> str | None:
@@ -222,6 +226,10 @@ def get_references_with_doi(self, doi: str) -> set[int]:
             )
             return {row[0] for row in result}
 
+    # TODO ref from title
+    # TODO ref from date
+    # TODO ref from journal
+
     ### Mixonomy
     # Todo, we probably want to still return that as a set
     def get_structures_of_taxon(self, tid: int, recursive: bool = True) -> set[int]:

storage/models/references.py

@@ -9,19 +9,17 @@ class References(Base):
 
     id: Mapped[int] = mapped_column(primary_key=True)
     doi: Mapped[str]
-    # TODO uncomment when ready
-    # title: Mapped[str]
-    # date: Mapped[str]
-    # journal: Mapped[int]
+    title: Mapped[str]
+    date: Mapped[str]
+    journal: Mapped[int]
 
     __table_args__ = (
         Index("reference_doi", "doi"),
         Index("reference_id", "id"),
-        # Index("reference_title", "title"),
-        # Index("reference_date", "date"),
-        # Index("reference_journal", "journal"),
+        Index("reference_title", "title"),
+        Index("reference_date", "date"),
+        Index("reference_journal", "journal"),
     )
 
     def __repr__(self):
-        return f"References(id={self.id}, doi={self.doi})"
-        # return f"References(id={self.id}, doi={self.doi}, title={self.title}, date={self.date}, journal={self.journal})"
+        return f"References(id={self.id}, doi={self.doi}, title={self.title}, date={self.date}, journal={self.journal})"

tests/test_references.py

@@ -2,5 +2,14 @@
 
 
 def test_references():
-    ref = References(id=1, doi="42.1/1")
-    assert ref.__repr__() == "References(id=1, doi=42.1/1)"
+    ref = References(
+        id=1,
+        doi="42.1/1",
+        title="Iridoids from Seeds of Gentiana Lutea",
+        date="2003-08",
+        journal="Natural Product Research",
+    )
+    assert (
+        ref.__repr__()
+        == "References(id=1, doi=42.1/1, title=Iridoids from Seeds of Gentiana Lutea, date=2003-08, journal=Natural Product Research)"
+    )