weitere Überarbeitung des retrival, Ingest funktioniert jetzt wieder

math-tutor/README.md

@@ -32,21 +32,21 @@ Init DB schema:
 ```powershell
 cd math-tutor/backend
 .\.venv\Scripts\Activate.ps1
-python .\scripts\retrieval_cli.py init-db
+python -m scripts.retrieval_cli init-db
 ```
 
 Ingest markdown docs (expects `markdown/sections`, `markdown/subsections`, `markdown/childs`):
 ```powershell
 cd math-tutor/backend
 .\.venv\Scripts\Activate.ps1
-python .\scripts\retrieval_cli.py ingest --base markdown --clear
+python -m scripts.retrieval_cli ingest --base markdown --clear
 ```
 
 Query via CLI:
 ```powershell
 cd math-tutor/backend
 .\.venv\Scripts\Activate.ps1
-python .\scripts\retrieval_cli.py query --q "Was ist eine Teilmenge?" --k 8 --expand
+python -m scripts.retrieval_cli query --q "Was ist eine Teilmenge?" --k 8 --expand
 ```
 
 ## Configuration

math-tutor/backend/app/config.py

@@ -91,20 +91,6 @@ def get_ollama_settings() -> OllamaSettings:
         temperature=_read_float(os.getenv("OLLAMA_TEMPERATURE")),
     )
 
-
-def get_embedding_settings() -> EmbeddingSettings:
-    base_url = os.getenv("OPENAI_BASE_URL")
-    api_key = os.getenv("OPENAI_API_KEY")
-    model = os.getenv("OPENAI_EMBED_MODEL", "text-embedding-3-large")
-    if not base_url or not api_key:
-        raise ValueError("Missing OPENAI_BASE_URL or OPENAI_API_KEY")
-    return EmbeddingSettings(
-        base_url=base_url,
-        api_key=api_key,
-        model=model,
-        target_dim=1024,
-    )
-
 def get_openai_chat_settings() -> OpenAIChatSettings | None:
     model = os.getenv("OPENAI_CHAT_MODEL")
     if not model:

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

@@ -186,8 +186,8 @@ class EmbeddingFactory:
             ValueError: Wenn der Typ nicht unterstützt wird.
         """
         if config.embedding_type == EmbeddingType.OPENAI_LIKE:
-            return OpenAILikeEmbeddings(config=config.config)
+            return OpenAILikeEmbeddings(config=config)
         elif config.embedding_type == EmbeddingType.SENTENCE_TRANSFORMER:
-            return SentenceTransformerEmbeddings(config=config.config)
+            return SentenceTransformerEmbeddings(config=config)
         else:
             raise ValueError(f"Unsupported embedding type: {config.embedding_type}")
\ No newline at end of file

math-tutor/backend/scripts/retrieval_cli.py

@@ -59,24 +59,10 @@ def cli_query(args: argparse.Namespace) -> None:
         expand_links=args.expand,
         section_index=args.section_index,
         subsection_index=args.subsection_index,
-        type_filter=args.type_filter,
+        source_type_filter=args.source_type_filter,
     )
-    print("=== SECTIONS ===")
-    for item in result["sections"]:
-        meta = item.metadata
-        print(f"- [{item.score:.3f}] s{meta.get('section_index')} {meta.get('title') or meta.get('section_title')}")
-    print("=== SUBSECTIONS ===")
-    for item in result["subsections"]:
-        meta = item.metadata
-        print(
-            f"- [{item.score:.3f}] s{meta.get('section_index')}, ss{meta.get('subsection_index')} {meta.get('title')}"
-        )
-    print("=== CHILDREN ===")
-    for item in result["children"]:
-        meta = item.metadata
-        print(
-            f"- [{item.score:.3f}] s{meta.get('section_index')}, ss{meta.get('subsection_index')}, c{meta.get('child_index')}"
-        )
+    for item in result:
+        print(f"- [{item.score:.3f}] {item.source_id.title} ({item.source_type})")
 
 
 def main() -> int:
@@ -101,7 +87,7 @@ def main() -> int:
     ap_q.add_argument("--expand", action="store_true")
     ap_q.add_argument("--section-index", type=int, default=None)
     ap_q.add_argument("--subsection-index", type=int, default=None)
-    ap_q.add_argument("--type-filter", nargs="*", default=None)
+    ap_q.add_argument("--source_type_filter", nargs="*", default=None)
     ap_q.set_defaults(func=cli_query)
 
     args = parser.parse_args()