Develop

.env.example

@@ -35,6 +35,14 @@ ACI_VECTOR_STORE_VECTOR_SIZE=1024
 ACI_INDEXING_MAX_WORKERS=8
 ACI_INDEXING_MAX_CHUNK_TOKENS=8192
 ACI_INDEXING_CHUNK_OVERLAP_LINES=2
+# Tokenizer strategy for chunk token counting.
+# Use "character" or "simple" for Ollama/BERT-based models (e.g. nomic-embed-text,
+# mxbai-embed-large) to avoid the tiktoken/WordPiece mismatch that causes
+# "input length exceeds context length" errors.
+#   tiktoken  - OpenAI BPE (default, accurate for OpenAI models)
+#   character - len(text)/4 estimate (conservative, works with any model)
+#   simple    - whitespace split (for generic non-BPE models)
+ACI_TOKENIZER=tiktoken
 # Optional comma-separated lists:
 # ACI_INDEXING_FILE_EXTENSIONS=.py,.js,.ts,.go
 # ACI_INDEXING_IGNORE_PATTERNS=__pycache__,*.pyc,.git,node_modules

src/aci/cli/__init__.py

@@ -22,7 +22,11 @@
 from rich.syntax import Syntax
 from rich.table import Table
 
-from aci.core.path_utils import get_collection_name_for_path, validate_indexable_path
+from aci.core.path_utils import (
+    get_collection_name_for_path,
+    resolve_file_filter_pattern,
+    validate_indexable_path,
+)
 from aci.infrastructure import GrepSearcher
 from aci.infrastructure.codebase_registry import (
     CodebaseRegistryStore,
@@ -51,7 +55,12 @@
     context_settings={"color": False},
 )
 
-def get_services():
+def _project_metadata_db_path(path: Path) -> Path:
+    """Return the metadata DB path scoped to a project root."""
+    return path.resolve() / ".aci" / "index.db"
+
+
+def get_services(metadata_db_path: Path | None = None):
     """
     Initialize services from .env with config-driven settings.
 
@@ -63,7 +72,7 @@ def get_services():
         Tuple of (config, embedding_client, vector_store, metadata_store,
                   file_scanner, chunker, reranker)
     """
-    container = create_services()
+    container = create_services(metadata_db_path=metadata_db_path)
     return (
         container.config,
         container.embedding_client,
@@ -100,7 +109,7 @@ def index(
             file_scanner,
             chunker,
             reranker,
-        ) = get_services()
+        ) = get_services(metadata_db_path=_project_metadata_db_path(path))
 
         # Use config workers if not overridden by CLI
         actual_workers = workers if workers is not None else cfg.indexing.max_workers
@@ -217,6 +226,9 @@ def search(
 ):
     """Search the indexed codebase."""
     try:
+        # Determine the search base path
+        search_base = path if path is not None else Path.cwd()
+
         (
             cfg,
             embedding_client,
@@ -225,17 +237,15 @@ def search(
             file_scanner,
             chunker,
             config_reranker,
-        ) = get_services()
-
-        # Determine the search base path
-        search_base = path if path is not None else Path.cwd()
+        ) = get_services(metadata_db_path=_project_metadata_db_path(search_base))
 
         # Use centralized repository resolution for path validation and collection name
         resolution = resolve_repository(search_base, metadata_store)
         if not resolution.valid:
             console.print(f"[bold red]Error:[/bold red] {resolution.error_message}")
             raise typer.Exit(1)
         collection_name = resolution.collection_name
+        normalized_file_filter = resolve_file_filter_pattern(file_filter, resolution.indexed_root)
 
         # Use config values if not overridden by CLI
         actual_limit = limit if limit is not None else cfg.search.default_limit
@@ -291,7 +301,7 @@ def search(
                 search_service.search(
                     query=query,
                     limit=actual_limit,
-                    file_filter=file_filter,  # User-provided filter only
+                    file_filter=normalized_file_filter,
                     use_rerank=use_rerank and reranker is not None,
                     search_mode=search_mode,  # Pass search mode
                     collection_name=collection_name,  # Pass explicitly, no state mutation
@@ -375,7 +385,7 @@ def update(
             file_scanner,
             chunker,
             reranker,
-        ) = get_services()
+        ) = get_services(metadata_db_path=_project_metadata_db_path(path))
 
         with Progress(
             SpinnerColumn(),

src/aci/core/path_utils.py

@@ -173,6 +173,41 @@ def parse_runtime_path_mappings(raw_value: str | None) -> list[RuntimePathMappin
     return mappings
 
 
+def resolve_file_filter_pattern(
+    file_filter: str | None, indexed_root: str | Path | None
+) -> str | None:
+    """Resolve relative file-filter prefixes against the indexed root path.
+
+    Keeps broad wildcard-only patterns unchanged (e.g. ``*.py`` or ``**/*.py``),
+    but expands relative directory-prefixed patterns (e.g. ``src/**/*.py``)
+    to absolute patterns rooted at ``indexed_root``.
+    """
+    if not file_filter:
+        return file_filter
+    if indexed_root is None:
+        return file_filter
+
+    raw_filter = file_filter.strip()
+    if not raw_filter:
+        return raw_filter
+
+    # Already absolute (POSIX, Windows drive, or UNC path).
+    if raw_filter.startswith("/") or _looks_like_windows_path(raw_filter) or raw_filter.startswith("\\\\"):
+        return raw_filter
+
+    normalized = raw_filter.replace("\\", "/")
+    has_directory_prefix = "/" in normalized
+    starts_with_wildcard = normalized.startswith(("*", "?", "["))
+    if not has_directory_prefix or starts_with_wildcard:
+        return raw_filter
+
+    relative_filter = normalized.lstrip("./")
+    if not relative_filter:
+        return raw_filter
+
+    return str(Path(indexed_root).resolve() / Path(relative_filter))
+
+
 def _apply_runtime_path_mapping(
     path_str: str,
     path_mappings: Sequence[RuntimePathMapping],

src/aci/http_server.py

@@ -11,7 +11,11 @@
 from fastapi import FastAPI, HTTPException, Query
 from pydantic import BaseModel
 
-from aci.core.path_utils import get_collection_name_for_path, is_system_directory
+from aci.core.path_utils import (
+    get_collection_name_for_path,
+    is_system_directory,
+    resolve_file_filter_pattern,
+)
 from aci.core.watch_config import WatchConfig
 from aci.infrastructure.codebase_registry import best_effort_update_registry
 from aci.infrastructure.file_watcher import FileWatcher
@@ -330,6 +334,7 @@ async def search(
                 raise HTTPException(status_code=400, detail=resolution.error_message)
 
             collection_name = resolution.collection_name
+            normalized_file_filter = resolve_file_filter_pattern(file_filter, resolution.indexed_root)
 
             apply_rerank = cfg.search.use_rerank if use_rerank is None else use_rerank
 
@@ -363,7 +368,7 @@ async def search(
             results = await search_service.search(
                 query=q,
                 limit=limit,
-                file_filter=file_filter,
+                file_filter=normalized_file_filter,
                 use_rerank=apply_rerank,
                 search_mode=search_mode,
                 collection_name=collection_name,

src/aci/mcp/handlers.py

@@ -11,6 +11,7 @@
 from aci.core.path_utils import (
     RuntimePathResolutionResult,
     get_collection_name_for_path,
+    resolve_file_filter_pattern,
     resolve_runtime_path,
     validate_indexable_path,
 )
@@ -243,14 +244,16 @@ async def _handle_search_code(arguments: dict, ctx: MCPContext) -> list[TextCont
                      f"Valid types: {', '.join(sorted(valid_artifact_types))}"
             )]
 
+    normalized_file_filter = resolve_file_filter_pattern(file_filter, indexed_root)
+
     # Request more results if filtering by subdirectory (to ensure enough after filtering)
     fetch_limit = limit * 3 if path_prefix_filter else limit
 
     # Pass collection_name explicitly to avoid shared state mutation
     results = await search_service.search(
         query=query,
         limit=fetch_limit,
-        file_filter=file_filter,
+        file_filter=normalized_file_filter,
         use_rerank=use_rerank,
         search_mode=search_mode,
         collection_name=collection_name,

tests/unit/test_chunker_core.py

@@ -17,7 +17,7 @@
     get_import_registry,
 )
 from aci.core.file_scanner import ScannedFile
-from aci.core.tokenizer import get_default_tokenizer
+from aci.core.tokenizer import CharacterTokenizer
 
 
 class TestCodeChunk:
@@ -143,7 +143,7 @@ class TestChunker:
 
     @pytest.fixture
     def tokenizer(self):
-        return get_default_tokenizer()
+        return CharacterTokenizer()
 
     @pytest.fixture
     def chunker(self, tokenizer):

tests/unit/test_chunker_splitter.py

@@ -7,15 +7,15 @@
 from aci.core.ast_parser import ASTNode, TreeSitterParser
 from aci.core.chunker import Chunker, SmartChunkSplitter
 from aci.core.file_scanner import ScannedFile
-from aci.core.tokenizer import get_default_tokenizer
+from aci.core.tokenizer import CharacterTokenizer
 
 
 class TestSmartChunkSplitter:
     """Tests for SmartChunkSplitter implementation."""
 
     @pytest.fixture
     def tokenizer(self):
-        return get_default_tokenizer()
+        return CharacterTokenizer()
 
     @pytest.fixture
     def splitter(self, tokenizer):
@@ -195,7 +195,7 @@ class TestChunkerWithSmartSplitter:
 
     @pytest.fixture
     def tokenizer(self):
-        return get_default_tokenizer()
+        return CharacterTokenizer()
 
     @pytest.fixture
     def parser(self):

tests/unit/test_cli_metadata_db_path.py

@@ -0,0 +1,51 @@
+from pathlib import Path
+
+import pytest
+import typer
+
+import aci.cli as cli
+
+
+def test_project_metadata_db_path_is_scoped_to_project_root(tmp_path: Path) -> None:
+    nested = tmp_path / "repo" / "subdir"
+    nested.mkdir(parents=True)
+
+    db_path = cli._project_metadata_db_path(nested)
+
+    assert db_path == nested.resolve() / ".aci" / "index.db"
+
+
+def test_index_uses_project_scoped_metadata_db_path(monkeypatch: pytest.MonkeyPatch, tmp_path: Path) -> None:
+    indexed_path = tmp_path / "repo"
+    indexed_path.mkdir()
+    captured: dict[str, Path | None] = {"metadata_db_path": None}
+
+    def fake_get_services(metadata_db_path: Path | None = None):
+        captured["metadata_db_path"] = metadata_db_path
+        raise RuntimeError("stop")
+
+    monkeypatch.setattr(cli, "get_services", fake_get_services)
+
+    with pytest.raises(typer.Exit):
+        cli.index(path=indexed_path, workers=None)
+
+    assert captured["metadata_db_path"] == indexed_path.resolve() / ".aci" / "index.db"
+
+
+def test_search_uses_explicit_path_for_project_scoped_metadata_db_path(
+    monkeypatch: pytest.MonkeyPatch, tmp_path: Path
+) -> None:
+    search_path = tmp_path / "repo"
+    search_path.mkdir()
+    captured: dict[str, Path | None] = {"metadata_db_path": None}
+
+    def fake_get_services(metadata_db_path: Path | None = None):
+        captured["metadata_db_path"] = metadata_db_path
+        raise RuntimeError("stop")
+
+    monkeypatch.setattr(cli, "get_services", fake_get_services)
+
+    with pytest.raises(typer.Exit):
+        cli.search(query="hello", path=search_path)
+
+    assert captured["metadata_db_path"] == search_path.resolve() / ".aci" / "index.db"

tests/unit/test_incremental_update_scopes_to_root.py

@@ -2,7 +2,9 @@
 import tempfile
 from pathlib import Path
 
+from aci.core.chunker import create_chunker
 from aci.core.file_scanner import FileScanner
+from aci.core.tokenizer import CharacterTokenizer
 from aci.infrastructure.fakes import InMemoryVectorStore, LocalEmbeddingClient
 from aci.infrastructure.metadata_store import IndexMetadataStore
 from aci.services.indexing_service import IndexingService
@@ -47,6 +49,7 @@ def test_update_incremental_does_not_delete_other_repo_metadata():
             vector_store=vector_store,
             metadata_store=metadata_store,
             file_scanner=file_scanner,
+            chunker=create_chunker(tokenizer=CharacterTokenizer()),
             max_workers=1,
         )
 

tests/unit/test_runtime_path_resolution.py

@@ -107,3 +107,28 @@ def test_mcp_index_codebase_uses_resolved_runtime_path(tmp_path: Path):
     payload = json.loads(result[0].text)
     assert payload["requested_path"] == r"D:\workspace"
     assert payload["indexed_path"] == str(mounted_repo.resolve())
+
+
+def test_resolve_file_filter_pattern_keeps_wildcard_only_pattern(tmp_path: Path):
+    from aci.core.path_utils import resolve_file_filter_pattern
+
+    resolved = resolve_file_filter_pattern("**/*.tsx", tmp_path)
+
+    assert resolved == "**/*.tsx"
+
+
+def test_resolve_file_filter_pattern_expands_relative_prefixed_pattern(tmp_path: Path):
+    from aci.core.path_utils import resolve_file_filter_pattern
+
+    resolved = resolve_file_filter_pattern("apps/web/**/*.tsx", tmp_path)
+
+    assert resolved == str(tmp_path.resolve() / "apps/web/**/*.tsx")
+
+
+def test_resolve_file_filter_pattern_keeps_absolute_pattern(tmp_path: Path):
+    from aci.core.path_utils import resolve_file_filter_pattern
+
+    absolute_pattern = str(tmp_path / "apps/**/*.tsx")
+    resolved = resolve_file_filter_pattern(absolute_pattern, "/another/root")
+
+    assert resolved == absolute_pattern