exclusion von subsections entfernt

math-tutor/backend/app/deterministic_services/vector_store.py

@@ -302,8 +302,6 @@ def retrieve(
     source_type_filter: Optional[List[str]] = None,
     expand_links: bool = True,
     neighbor_expand: int = 0,
-    exclude_subsection_refs: Optional[List["SubsectionRef"]] = None,
-    exclude_child_subsection_refs: Optional[List["SubsectionRef"]] = None,
 ) -> List[Source]:
     qvec = Vector(embed_query(embedder, query))
 
@@ -322,18 +320,6 @@ def retrieve(
         where.append("source_type = ANY(%(source_type_filter)s)")
         params["source_type_filter"] = source_type_filter
 
-    excluded_child_subsections = _normalize_subsection_refs(exclude_child_subsection_refs)
-    if excluded_child_subsections:
-        sec_arr = [sec for sec, _ in excluded_child_subsections]
-        sub_arr = [sub for _, sub in excluded_child_subsections]
-        where.append(
-            "NOT EXISTS ("
-            "SELECT 1 FROM unnest(%(exclude_sec_arr)s::int[], %(exclude_sub_arr)s::int[]) AS ex(sec, sub) "
-            "WHERE ex.sec = section_index AND ex.sub = subsection_index)"
-        )
-        params["exclude_sec_arr"] = sec_arr
-        params["exclude_sub_arr"] = sub_arr
-
     where_sql = " AND ".join(where)
 
     sql = f"""
@@ -355,7 +341,6 @@ def retrieve(
     sections_docs: List[Retrieved] = []
     children_direct: List[Retrieved] = []
     children_expanded: List[Retrieved] = []
-    excluded_subsections = set(_normalize_subsection_refs(exclude_subsection_refs))
 
     with psycopg.connect(pg_url, row_factory=dict_row) as conn:
         register_vector(conn)
@@ -379,25 +364,25 @@ def retrieve(
                         sec_sub_counts.items(), key=lambda x: x[1])[0]
                     most_common_sec, most_common_sub = most_common_sec_sub
 
-                    if (most_common_sec, most_common_sub) not in excluded_subsections:
-                        cur.execute(
-                            """
-                            SELECT
-                              d.uid, d.doc_type,
-                              d.section_index, d.subsection_index, d.child_index,
-                              d.section_title, d.subsection_title, d.title, d.source_type,
-                              d.path, d.markdown,
-                              1.0 AS score
-                            FROM docs d
-                            WHERE d.doc_type = ANY(%(sub_doc_types)s)
-                              AND d.section_index = %(sec)s
-                              AND d.subsection_index = %(sub)s
-                            """,
-                            {"sec": most_common_sec, "sub": most_common_sub,
-                                "sub_doc_types": ["subsection", "chapter"]},
-                        )
-                        subsections = [_row_to_retrieved(
-                            row, source_type="subsection") for row in cur.fetchall()]
+                    cur.execute(
+                        """
+                        SELECT
+                          d.uid, d.doc_type,
+                          d.section_index, d.subsection_index, d.child_index,
+                          d.section_title, d.subsection_title, d.title, d.source_type,
+                          d.path, d.markdown,
+                          1.0 AS score
+                        FROM docs d
+                        WHERE d.doc_type = ANY(%(sub_doc_types)s)
+                          AND d.section_index = %(sec)s
+                          AND d.subsection_index = %(sub)s
+                        """,
+                        {"sec": most_common_sec, "sub": most_common_sub,
+                            "sub_doc_types": ["subsection", "chapter"]},
+                    )
+                    subsections = [_row_to_retrieved(
+                        row, source_type="subsection") for row in cur.fetchall()]
+
 
                     cur.execute(
                         """
@@ -592,8 +577,6 @@ def retrieve_with_subsections(
         source_type_filter=source_type_filter,
         expand_links=expand_links,
         neighbor_expand=neighbor_expand,
-        exclude_subsection_refs=subsection_refs,
-        exclude_child_subsection_refs=subsection_refs,
     )
     vector_sources = vector_sources[:k]
     subsection_children = load_children_for_subsections(

math-tutor/backend/app/deterministic_services/vector_store_subsection.py

@@ -29,7 +29,6 @@ def retrieve(
     source_type_filter: Optional[List[str]] = None,
     expand_links: bool = False,
     neighbor_expand: int = 0,
-    exclude_subsection_refs: Optional[List[SubsectionRef]] = None,
 ) -> List[Source]:
     # Parameters kept for drop-in compatibility with child-level retrieve.
     _ = expand_links
@@ -56,19 +55,6 @@ def retrieve(
         where.append("source_type = ANY(%(source_type_filter)s)")
         params["source_type_filter"] = source_type_filter
 
-    excluded = set((int(sec), int(sub)) for sec, sub in (exclude_subsection_refs or []))
-    if excluded:
-        exclude_clauses: List[str] = []
-        for idx, (sec, sub) in enumerate(sorted(excluded)):
-            sec_key = f"exclude_sec_{idx}"
-            sub_key = f"exclude_sub_{idx}"
-            exclude_clauses.append(
-                f"(section_index = %({sec_key})s AND subsection_index = %({sub_key})s)"
-            )
-            params[sec_key] = sec
-            params[sub_key] = sub
-        where.append(f"NOT ({' OR '.join(exclude_clauses)})")
-
     where_sql = " AND ".join(where)
     sql = f"""
     SELECT
@@ -117,7 +103,6 @@ def retrieve_with_subsections(
         source_type_filter=source_type_filter,
         expand_links=expand_links,
         neighbor_expand=neighbor_expand,
-        exclude_subsection_refs=subsection_refs,
     )
     vector_sources = vector_sources[:k]
     subsection_children = load_children_for_subsections(