гих хук и сохранение изменений в контексте стори

src/rag_agent/ingest/chunker.py

@@ -10,6 +10,8 @@ class TextChunk:
     index: int
     text: str
     hash: str
+    start_line: int | None = None
+    end_line: int | None = None
 
 
 def _hash_text(text: str) -> str:
@@ -27,7 +29,15 @@ def chunk_text(text: str, chunk_size: int, overlap: int) -> list[TextChunk]:
     while start < len(tokens):
         end = min(start + chunk_size, len(tokens))
         chunk_text_value = " ".join(tokens[start:end])
-        chunks.append(TextChunk(index=index, text=chunk_text_value, hash=_hash_text(chunk_text_value)))
+        chunks.append(
+            TextChunk(
+                index=index,
+                text=chunk_text_value,
+                hash=_hash_text(chunk_text_value),
+                start_line=None,
+                end_line=None,
+            )
+        )
         index += 1
         if end == len(tokens):
             break
@@ -40,3 +50,34 @@ def iter_chunks(
 ) -> Iterator[list[TextChunk]]:
     for text in texts:
         yield chunk_text(text, chunk_size, overlap)
+
+
+def chunk_text_by_lines(
+    text: str, max_lines: int, overlap_lines: int
+) -> list[TextChunk]:
+    """Chunk by consecutive lines; each chunk has start_line/end_line (1-based)."""
+    lines = text.splitlines()
+    if not lines:
+        return []
+
+    chunks: list[TextChunk] = []
+    index = 0
+    start = 0
+    while start < len(lines):
+        end = min(start + max_lines, len(lines))
+        chunk_lines = lines[start:end]
+        chunk_text_value = "\n".join(chunk_lines)
+        chunks.append(
+            TextChunk(
+                index=index,
+                text=chunk_text_value,
+                hash=_hash_text(chunk_text_value),
+                start_line=start + 1,
+                end_line=end,
+            )
+        )
+        index += 1
+        if end == len(lines):
+            break
+        start = max(end - overlap_lines, 0)
+    return chunks

src/rag_agent/ingest/git_watcher.py

@@ -40,3 +40,53 @@ def filter_existing(paths: Iterable[Path]) -> list[Path]:
 
 def filter_removed(paths: Iterable[Path]) -> list[Path]:
     return [path for path in paths if not path.exists()]
+
+
+def get_merge_base(
+    repo_path: str, ref1: str, ref2: str = "HEAD"
+) -> str | None:
+    """Return merge-base commit of ref1 and ref2 (start of story range). None on error."""
+    args = [
+        "git",
+        "-C",
+        repo_path,
+        "merge-base",
+        ref1,
+        ref2,
+    ]
+    try:
+        result = subprocess.run(
+            args,
+            check=True,
+            capture_output=True,
+            text=True,
+        )
+        return result.stdout.strip() or None
+    except subprocess.CalledProcessError:
+        return None
+
+
+def read_file_at_ref(
+    repo_path: str, path: Path, ref: str
+) -> str | None:
+    """Read file content at a git ref. Returns None if path is missing at ref."""
+    repo = Path(repo_path)
+    rel = path.relative_to(repo) if path.is_absolute() else path
+    rel_str = rel.as_posix()
+    args = [
+        "git",
+        "-C",
+        repo_path,
+        "show",
+        f"{ref}:{rel_str}",
+    ]
+    try:
+        result = subprocess.run(
+            args,
+            check=True,
+            capture_output=True,
+            text=True,
+        )
+        return result.stdout
+    except subprocess.CalledProcessError:
+        return None