Added APIs: Unique authors, Unique citations

Author: dashrath-chauhan

src/main.py

@@ -76,10 +76,8 @@ def main() -> FastAPI:
 
     db_conn = setup.DatabaseConnection(dsn)
     db_conn.connect()
-
     session = SessionLocal()
-
-    logger.info("Database connected via SQLAlchemy ORM")                           
+    logger.info("Database connected via SQLAlchemy ORM")                     
 
     # Fields setup
     record_fields = set(Record.model_fields.keys())

src/models/citation.py

@@ -31,4 +31,8 @@ class Reference(BaseModel):
     cited_order: int
     match: bool
 
-    model_config = {"from_attributes": True}
+    model_config = {"from_attributes": True}
+    
+class CitationList(BaseModel):
+    citations: list[Citation]
+    citation_count: int

src/repositories/epmc.py

@@ -288,10 +288,8 @@ def get_all_pmc_affiliations(self, limit: int = 100, skip: int = 0) -> list[PMCA
         return self.db.query(PMCAffiliation).offset(skip).limit(limit).all()
 
     def get_all_articles_ids(self) -> list[tuple[str | None, int]]:
-        return (
-            self.db.query(PMCArticle.pm_id, PMCArticle.record_id)
-            .all()
-        )
+        return  self.db.query(PMCArticle.pm_id, PMCArticle.record_id).all()
+        
 
     def _get_latest_version_subquery(self, entity_class: Type[Any]):
         """
@@ -409,13 +407,21 @@ def _get_latest_entities_by_column(self, entity_class: Type[Any], group_column,
         )
 
         # Query entities where row number is 1 (latest version per group)
-        return (
-            self.db.query(entity_class)
-            .join(version_subq, and_(entity_class.id == version_subq.c.id, version_subq.c.rn == 1))
-            .offset(skip)
-            .limit(limit)
-            .all()
-        )
+        if entity_class == Citation:
+            return (
+                self.db.query(entity_class)
+                .join(version_subq, and_(entity_class.id == version_subq.c.id, version_subq.c.rn == 1))
+                .offset(skip)
+                .all()
+            )
+        else:
+            return (
+                self.db.query(entity_class)
+                .join(version_subq, and_(entity_class.id == version_subq.c.id, version_subq.c.rn == 1))
+                .offset(skip)
+                .limit(limit)
+                .all()
+            )
 
 
     def _get_entities_by_column_value(self, entity_class: Type[Any], column_name: str, value: Any, limit: int = 100, skip: int = 0) -> list[Any]:
@@ -920,3 +926,22 @@ def get_unique_citations(self, limit: int = 100, skip: int = 0) -> list[Citation
             List of Citation entities, one per unique citation_id with highest version
         """
         return self._get_latest_entities_by_column(Citation, Citation.citation_id, limit=limit, skip=skip)
+    
+    def count_unique_authors(self) -> int:
+        return self.db.query(
+            func.count(
+                func.distinct(
+                    func.concat(
+                        func.lower(func.trim(PMCAuthor.firstname)),
+                        " ",
+                        func.lower(func.trim(PMCAuthor.lastname)),
+                        " ",
+                        func.lower(func.trim(PMCAuthor.initials))
+                    )
+                )
+            )
+        ).filter(
+            PMCAuthor.firstname.isnot(None),
+            PMCAuthor.lastname.isnot(None),
+            PMCAuthor.initials.isnot(None)
+        ).scalar()

src/routers/epmc.py

@@ -3,7 +3,7 @@
 
 from src.models.pmc_article import PMCArticle, PMCArticleFull
 from src.models.pmc_author import PMCAuthor
-from src.models.citation import Citation as CitationModel
+from src.models.citation import Citation as CitationModel, CitationList
 from src.services.epmc import EPMCService as EPMCService
 from src.repositories.epmc import EPMCRepo as EPMCRepo
 from src.services.grant import GrantService as Grant
@@ -138,11 +138,12 @@ async def get_affiliation_countries_count():
             repo = EPMCRepo(self.db)
             return repo.get_affiliation_countries_count()        
 
-        @self.router.get("/epmc/unique-citations", response_model=list[CitationModel])
+        @self.router.get("/epmc/unique-citations", response_model=CitationList)
         async def get_unique_citations(limit: int = 1000, skip: int = 0):
             repo = EPMCRepo(self.db)
             citations = repo.get_unique_citations(limit=limit, skip=skip)
-            return [CitationModel.model_validate(c) for c in citations]
+            return CitationList(citations=[CitationModel.model_validate(c) for c in citations], citation_count=len(citations))
+
 
         @self.router.get("/epmc/top-authors")
         async def get_top_authors(count: int = 15):
@@ -151,4 +152,12 @@ async def get_top_authors(count: int = 15):
             try:
                 return repo.get_top_authors(count=count)
             except Exception as e:
-                raise HTTPException(status_code=500, detail=f"Failed to fetch top authors: {e}")
+                raise HTTPException(status_code=500, detail=f"Failed to fetch top authors: {e}")
+            
+        @self.router.get("/epmc/unique-authors-count")
+        async def unique_authors_count():
+            repo = EPMCRepo(self.db)
+            service = EPMCService(repo)
+            
+            count = service.get_unique_authors_count()
+            return {"unique_authors": count}    

src/services/epmc.py

@@ -171,3 +171,6 @@ def insert_references(self, created_by: str):
                 counts["references"] += 1
 
         return counts
+    
+    def get_unique_authors_count(self) -> int:
+        return self.epmc_repo.count_unique_authors()