гих хук и сохранение изменений в контексте стори

src/rag_agent/cli.py

@@ -5,9 +5,15 @@ import hashlib
 from pathlib import Path
 
 from rag_agent.config import load_config
-from rag_agent.ingest.chunker import chunk_text
+from rag_agent.ingest.chunker import chunk_text_by_lines
 from rag_agent.ingest.file_loader import iter_text_files
-from rag_agent.ingest.git_watcher import filter_existing, filter_removed, get_changed_files
+from rag_agent.ingest.git_watcher import (
+    filter_existing,
+    filter_removed,
+    get_changed_files,
+    get_merge_base,
+    read_file_at_ref,
+)
 from rag_agent.index.embeddings import get_embedding_client
 from rag_agent.index.postgres import (
     connect,
@@ -16,6 +22,7 @@ from rag_agent.index.postgres import (
     get_or_create_story,
     get_story_id,
     replace_chunks,
+    update_story_indexed_range,
     upsert_document,
 )
 from rag_agent.agent.pipeline import StubLLMClient, answer_query
@@ -34,26 +41,66 @@ def cmd_index(args: argparse.Namespace) -> None:
     story_id = get_or_create_story(conn, args.story)
     embedding_client = get_embedding_client(config.embeddings_dim)
 
+    base_ref = args.base_ref.strip()
+    head_ref = args.head_ref.strip()
     if args.changed:
-        changed_files = get_changed_files(config.repo_path, args.base_ref, args.head_ref)
+        if base_ref.lower() == "auto":
+            base_ref = get_merge_base(
+                config.repo_path, args.default_branch, head_ref
+            ) or args.default_branch
+        changed_files = get_changed_files(
+            config.repo_path, base_ref, head_ref
+        )
         removed = filter_removed(changed_files)
         existing = filter_existing(changed_files)
     else:
         removed = []
-        existing = [path for path in Path(config.repo_path).rglob("*") if path.is_file()]
+        existing = [
+            p for p in Path(config.repo_path).rglob("*") if p.is_file()
+        ]
 
     for path in removed:
         delete_document(conn, story_id, str(path))
 
     for path, text in iter_text_files(existing, config.allowed_extensions):
-        chunks = chunk_text(text, config.chunk_size, config.chunk_overlap)
+        chunks = chunk_text_by_lines(
+            text,
+            config.chunk_size_lines,
+            config.chunk_overlap_lines,
+        )
         if not chunks:
             continue
-        embeddings = embedding_client.embed_texts([chunk.text for chunk in chunks])
+        base_chunks = None
+        if args.changed:
+            base_text = read_file_at_ref(config.repo_path, path, base_ref)
+            if base_text is not None:
+                base_chunks = chunk_text_by_lines(
+                    base_text,
+                    config.chunk_size_lines,
+                    config.chunk_overlap_lines,
+                )
+        embeddings = embedding_client.embed_texts(
+            [chunk.text for chunk in chunks]
+        )
         document_id = upsert_document(
             conn, story_id, str(path), _file_version(path)
         )
-        replace_chunks(conn, document_id, chunks, embeddings)
+        replace_chunks(
+            conn, document_id, chunks, embeddings, base_chunks=base_chunks
+        )
+
+    if args.changed:
+        update_story_indexed_range(conn, story_id, base_ref, head_ref)
+
+
+def cmd_serve(args: argparse.Namespace) -> None:
+    import uvicorn
+    uvicorn.run(
+        "rag_agent.webhook:app",
+        host=args.host,
+        port=args.port,
+        log_level="info",
+    )
 
 
 def cmd_ask(args: argparse.Namespace) -> None:
@@ -89,9 +136,26 @@ def build_parser() -> argparse.ArgumentParser:
         required=True,
         help="Story slug (e.g. branch name or story id); documents are tied to this story",
     )
-    index_parser.add_argument("--changed", action="store_true", help="Index only changed files")
-    index_parser.add_argument("--base-ref", default="HEAD~1", help="Base git ref")
-    index_parser.add_argument("--head-ref", default="HEAD", help="Head git ref")
+    index_parser.add_argument(
+        "--changed",
+        action="store_true",
+        help="Index only files changed in the story range (base-ref..head-ref); all commits in range belong to the story",
+    )
+    index_parser.add_argument(
+        "--base-ref",
+        default="main",
+        help="Start of story range (e.g. main). Use 'auto' for merge-base(default-branch, head-ref). All commits from base to head are the story.",
+    )
+    index_parser.add_argument(
+        "--head-ref",
+        default="HEAD",
+        help="End of story range (e.g. current branch tip)",
+    )
+    index_parser.add_argument(
+        "--default-branch",
+        default="main",
+        help="Default branch name for --base-ref auto",
+    )
     index_parser.set_defaults(func=cmd_index)
 
     ask_parser = sub.add_parser("ask", help="Ask a question")
@@ -104,6 +168,23 @@ def build_parser() -> argparse.ArgumentParser:
     ask_parser.add_argument("--top-k", type=int, default=5, help="Top K chunks to retrieve")
     ask_parser.set_defaults(func=cmd_ask)
 
+    serve_parser = sub.add_parser(
+        "serve",
+        help="Run webhook server: on push to remote repo, pull and index changes",
+    )
+    serve_parser.add_argument(
+        "--host",
+        default="0.0.0.0",
+        help="Bind host (default: 0.0.0.0)",
+    )
+    serve_parser.add_argument(
+        "--port",
+        type=int,
+        default=8000,
+        help="Bind port (default: 8000)",
+    )
+    serve_parser.set_defaults(func=cmd_serve)
+
     return parser