java-codebase-rag 0.6.3__py3-none-any.whl → 0.6.5__py3-none-any.whl

graph_enrich.py

@@ -23,7 +23,7 @@ import sys
 from dataclasses import dataclass, field, replace
 from functools import lru_cache
 from pathlib import Path
-from typing import Any
+from typing import Any, TypeVar
 from ast_java import (
     AnnotationRef,
     JavaFileAst,
@@ -820,7 +820,15 @@ def _route_path_atom(raw_value: str, value_kind: str | None) -> tuple[str, str,
     return "", "constant_ref", 0.7, False
 
 
-def _route_hint_lookup(ann: AnnotationRef, hints: dict[str, RouteHint]) -> RouteHint | None:
+_HINT = TypeVar("_HINT")
+
+
+def _hint_lookup(ann: AnnotationRef, hints: dict[str, _HINT]) -> _HINT | None:
+    """Resolve a brownfield hint by qualified name, then simple name, then suffix.
+
+    Shared by route / http-client / async-producer hint resolution; the three
+    former copies differed only in the hint value type.
+    """
     q = ann.qualified.strip()
     if q in hints:
         return hints[q]
@@ -1118,7 +1126,7 @@ def resolve_routes_for_method(
 
     # ----- Step 2: Layer B — annotation route hints -----
     for _is_m, ann in combined_anns:
-        hint = _route_hint_lookup(ann, overrides.annotation_to_route_hint)
+        hint = _hint_lookup(ann, overrides.annotation_to_route_hint)
         if hint is None:
             continue
         working.append(
@@ -1172,36 +1180,6 @@ def resolve_routes_for_method(
     return working
 
 
-def _client_hint_lookup(
-    ann: AnnotationRef,
-    hints: dict[str, HttpClientHint],
-) -> HttpClientHint | None:
-    q = ann.qualified.strip()
-    if q in hints:
-        return hints[q]
-    if ann.name in hints:
-        return hints[ann.name]
-    for k, h in sorted(hints.items(), key=lambda kv: kv[0]):
-        if k.endswith("." + ann.name):
-            return h
-    return None
-
-
-def _async_hint_lookup(
-    ann: AnnotationRef,
-    hints: dict[str, AsyncProducerHint],
-) -> AsyncProducerHint | None:
-    q = ann.qualified.strip()
-    if q in hints:
-        return hints[q]
-    if ann.name in hints:
-        return hints[ann.name]
-    for k, h in sorted(hints.items(), key=lambda kv: kv[0]):
-        if k.endswith("." + ann.name):
-            return h
-    return None
-
-
 def _call_from_http_hint(
     *,
     hint: HttpClientHint,
@@ -1296,7 +1274,7 @@ def resolve_http_client_for_method(
     anchor = builtin_http[0] if builtin_http else (layer_c_src[0] if layer_c_src else None)
 
     for _is_m, ann in combined_anns:
-        hint = _client_hint_lookup(ann, overrides.annotation_to_http_client_hint)
+        hint = _hint_lookup(ann, overrides.annotation_to_http_client_hint)
         if hint is None:
             continue
         brownfield_calls.append(
@@ -1388,7 +1366,7 @@ def resolve_async_producer_for_method(
     anchor = builtin_async[0] if builtin_async else (layer_c_src[0] if layer_c_src else None)
 
     for _is_m, ann in combined_anns:
-        hint = _async_hint_lookup(ann, overrides.annotation_to_async_producer_hint)
+        hint = _hint_lookup(ann, overrides.annotation_to_async_producer_hint)
         if hint is None:
             continue
         brownfield_calls.append(

java_codebase_rag/cli.py

@@ -251,7 +251,7 @@ def _startup_hints(cfg: ResolvedOperatorConfig) -> None:
 
 def _add_index_embedding_flags(p: argparse.ArgumentParser) -> None:
     p.add_argument("--source-root", type=str, default=None, help="Java repository root (default: cwd)")
-    p.add_argument("--index-dir", type=str, default=None, help="Index directory (Lance + Kuzu + cocoindex state)")
+    p.add_argument("--index-dir", type=str, default=None, help="Index directory (Lance + LadybugDB + cocoindex state)")
     p.add_argument("--embedding-model", type=str, default=None, help="Override SBERT_MODEL / YAML embedding.model")
     p.add_argument("--embedding-device", type=str, default=None, help="Override SBERT_DEVICE / YAML embedding.device")
 
@@ -536,7 +536,7 @@ def _cmd_reprocess(args: argparse.Namespace) -> int:
             _emit_reprocess_outcome(payload, selective_tty_mode="graph" if ok else None)
             return _reprocess_exit_code(payload)
 
-        import server  # lazy: pulls sentence_transformers/torch/lancedb/kuzu
+        import server  # lazy: pulls sentence_transformers/torch/lancedb/ladybug
 
         result = asyncio.run(
             server.run_refresh_pipeline(
@@ -715,7 +715,7 @@ def _cmd_unresolved_calls_list(args: argparse.Namespace) -> int:
     from ladybug_queries import LadybugGraph  # lazy
 
     if not LadybugGraph.exists():
-        _emit({"success": False, "message": "Kuzu graph not found"})
+        _emit({"success": False, "message": "LadybugDB graph not found"})
         return 1
     graph = LadybugGraph.get()
     rows = graph.list_unresolved_call_sites(
@@ -736,7 +736,7 @@ def _cmd_unresolved_calls_stats(args: argparse.Namespace) -> int:
     from ladybug_queries import LadybugGraph  # lazy
 
     if not LadybugGraph.exists():
-        _emit({"success": False, "message": "Kuzu graph not found"})
+        _emit({"success": False, "message": "LadybugDB graph not found"})
         return 1
     graph = LadybugGraph.get()
     buckets = graph.stats_unresolved_call_sites(by=args.by)
@@ -761,7 +761,7 @@ def _cmd_analyze_pr(args: argparse.Namespace) -> int:
     from ladybug_queries import LadybugGraph  # lazy
 
     if not LadybugGraph.exists():
-        _emit({"success": False, "message": "Kuzu graph not found"})
+        _emit({"success": False, "message": "LadybugDB graph not found"})
         return 1
     graph = LadybugGraph.get()
     report = pr_analysis.analyze_pr_pipeline(graph, diff_text)
@@ -801,7 +801,7 @@ def build_parser() -> argparse.ArgumentParser:
         help="Create a fresh index from a Java repository.",
         description=(
             "First-time index creation. Refuses if the resolved index directory "
-            "already contains a Kuzu graph or Lance tables. Exit 2 on refusal."
+            "already contains a LadybugDB graph or Lance tables. Exit 2 on refusal."
         ),
     )
     _add_index_embedding_flags(init)
@@ -870,7 +870,7 @@ def build_parser() -> argparse.ArgumentParser:
     increment = subparsers.add_parser(
         "increment",
         help="Pick up changes since the last index update.",
-        description="Runs cocoindex catch-up and incremental Kuzu graph update. Use --vectors-only to skip graph update.",
+        description="Runs cocoindex catch-up and incremental LadybugDB graph update. Use --vectors-only to skip graph update.",
     )
     _add_index_embedding_flags(increment)
     _add_verbosity_flags(increment)
@@ -883,9 +883,9 @@ def build_parser() -> argparse.ArgumentParser:
 
     reprocess = subparsers.add_parser(
         "reprocess",
-        help="Rebuild vectors and/or Kuzu (default: both full phases).",
+        help="Rebuild vectors and/or LadybugDB (default: both full phases).",
         description=(
-            "Default: full Lance reprocess (cocoindex --full-reprocess) then full Kuzu graph rebuild. "
+            "Default: full Lance reprocess (cocoindex --full-reprocess) then full LadybugDB graph rebuild. "
             "Use --vectors-only or --graph-only to run a single phase (mutually exclusive)."
         ),
     )
@@ -907,7 +907,7 @@ def build_parser() -> argparse.ArgumentParser:
     erase = subparsers.add_parser(
         "erase",
         help="Delete the index from disk.",
-        description="Runs cocoindex drop, removes Kuzu, and drops Lance tables. Requires --yes or TTY confirmation.",
+        description="Runs cocoindex drop, removes LadybugDB, and drops Lance tables. Requires --yes or TTY confirmation.",
     )
     _add_index_embedding_flags(erase)
     erase.add_argument("--yes", action="store_true", help="Confirm destructive deletion (required in CI)")

java_codebase_rag/install_data/agents/explorer-rag-enhanced.md

@@ -151,15 +151,15 @@ Simple types in parentheses; generics erased. No spaces after commas. No-arg: `(
 
 ### Shared NodeFilter
 
-For `find`, `filter` is required — `{}` means no predicates. **Strict frame:** unknown keys or inapplicable populated fields → `success=false`.
+For `find`, `filter` is required — `{}` means no predicates. **Strict frame:** unknown keys or inapplicable populated fields → `success=false`; invalid enum values (e.g. wrong case) are rejected earlier at the schema layer with the valid set listed.
 
 | Keys | Applies to |
 | ---- | ---------- |
 | `microservice`, `module` | All kinds |
 | `role`, `exclude_roles`, `annotation`, `capability`, `fqn_prefix`, `symbol_kind`, `symbol_kinds` | **symbol** |
 | `http_method`, `path_prefix`, `framework` | **route** |
-| `client_kind`, `target_service`, `target_path_prefix`, `http_method` | **client** |
-| `producer_kind`, `topic_prefix` | **producer** |
+| `source_layer`, `client_kind`, `target_service`, `target_path_prefix`, `http_method` | **client** |
+| `source_layer`, `producer_kind`, `topic_prefix` | **producer** |
 
 No wildcards in prefix fields — use `search(query=…)` for fuzzy text.
 

java_codebase_rag/install_data/skills/explore-codebase/SKILL.md

@@ -125,15 +125,15 @@ Use these strings **verbatim** in `neighbors(..., edge_types=[...])`.
 
 ### NodeFilter (`find`, `search.filter`, `neighbors.filter`)
 
-For `find`, `filter` is required — `{}` means no predicates. **Strict frame:** unknown keys or inapplicable populated fields → `success=false`.
+For `find`, `filter` is required — `{}` means no predicates. **Strict frame:** unknown keys or inapplicable populated fields → `success=false`; invalid enum values (e.g. wrong case) are rejected earlier at the schema layer with the valid set listed.
 
 | Applicable to | Keys |
 | ------------- | ---- |
 | All kinds | `microservice`, `module` |
 | **symbol** only | `role`, `exclude_roles`, `annotation`, `capability`, `fqn_prefix`, `symbol_kind`, `symbol_kinds` |
 | **route** only | `http_method`, `path_prefix`, `framework` |
-| **client** only | `client_kind`, `target_service`, `target_path_prefix`, `http_method` |
-| **producer** only | `producer_kind`, `topic_prefix` |
+| **client** only | `source_layer`, `client_kind`, `target_service`, `target_path_prefix`, `http_method` |
+| **producer** only | `source_layer`, `producer_kind`, `topic_prefix` |
 
 No wildcards in prefix fields — use `search(query=…)` for ranked text.
 
@@ -166,8 +166,8 @@ Exclude `DTO`, `OTHER`, `MAPPER` with `exclude_roles` when tracing business logi
 
 **Symbol kinds:** `class`, `interface`, `enum`, `record`, `annotation`, `method`, `constructor`.
 
-**Route frameworks:** `spring_mvc`, `webflux`, `kafka`, `rabbitmq`, `jms`, `stream`, `codebase_async_route`, …
-**Client kinds:** `feign_method`, `rest_template`, `web_client`. **Producer kinds:** `kafka_send`, `stream_bridge_send`.
+**Route frameworks:** `spring_mvc`, `webflux`. (Route *kinds* are `http_endpoint`, `http_consumer`, `kafka_topic`, `rabbit_queue`, `jms_destination`, `stream_binding`.)
+**Client kinds:** `feign_method`, `rest_template`, `web_client`. **Producer kinds:** `kafka_send`, `stream_bridge_send`. **Source layers (client/producer):** `builtin`, `layer_a_meta`, `layer_b_ann`, `layer_b_fqn`, `layer_c_source`.
 **Match types:** `cross_service`, `intra_service`, `ambiguous`, `phantom`, `unresolved`.
 
 ---

{java_codebase_rag-0.6.3.dist-info → java_codebase_rag-0.6.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: java-codebase-rag
-Version: 0.6.3
+Version: 0.6.5
 Summary: MCP server for semantic + structural search over Java codebases
 Author: HumanBean17
 License-Expression: MIT
@@ -44,7 +44,7 @@ Dynamic: license-file
 
 A graph-native code intelligence layer for Java microservice estates, exposed to LLM agents via the **Model Context Protocol (MCP)**.
 
-The system extracts a deterministic property graph from Java source (tree-sitter), stores it in **Kuzu** (graph) alongside a **LanceDB** vector index (chunks), and exposes a deliberately small MCP surface — **five tools**: `search`, `find`, `describe`, `neighbors`, `resolve` — that collapse onto three primitive agent operations: **locate**, **inspect**, **walk**.
+The system extracts a deterministic property graph from Java source (tree-sitter), stores it in **LadybugDB** (graph) alongside a **LanceDB** vector index (chunks), and exposes a deliberately small MCP surface — **five tools**: `search`, `find`, `describe`, `neighbors`, `resolve` — that collapse onto three primitive agent operations: **locate**, **inspect**, **walk**.
 
 > **What this MCP is:** a **GPS for code navigation**, not a reasoning engine.
 > Agents use a simple loop:
@@ -63,9 +63,9 @@ For the design rationale, the GPS metaphor, and the full ontology, see [`docs/pa
 
 Generic code-search tools (grep, ctags, vector-only RAG) hit a ceiling on real Java microservice estates: they find files but lose the structure that makes a Spring/JAX-RS system navigable. This project is built around five choices that target that gap.
 
-- **Hybrid RAG + GraphRAG, not either-or.** Semantic recall (LanceDB chunk vectors) and structural navigation (Kuzu property graph) are composed in one surface. `search` finds candidate nodes by meaning; `neighbors` walks the exact edge you care about (`CALLS`, `IMPLEMENTS`, `INJECTS`, `DECLARES_ROUTE`, …). The agent picks the right primitive per step instead of being forced into pure-vector or pure-symbol search.
+- **Hybrid RAG + GraphRAG, not either-or.** Semantic recall (LanceDB chunk vectors) and structural navigation (LadybugDB property graph) are composed in one surface. `search` finds candidate nodes by meaning; `neighbors` walks the exact edge you care about (`CALLS`, `IMPLEMENTS`, `INJECTS`, `EXPOSES`, …). The agent picks the right primitive per step instead of being forced into pure-vector or pure-symbol search.
 
-- **A Java-tuned role model.** Symbols are labelled with stereotypes inferred from Spring and JAX-RS conventions — `CONTROLLER`, `SERVICE`, `REPOSITORY`, `CLIENT`, `PRODUCER`, `MAPPER`, `DTO`. Agents can ask "list controllers" or "who injects this repository" directly, instead of grep-ing for `@RestController` and hoping for the best. Roles drive both filtering (`find` with a `NodeFilter`) and ranking.
+- **A Java-tuned role model.** Symbols are labelled with stereotypes inferred from Spring and JAX-RS conventions — `CONTROLLER`, `SERVICE`, `REPOSITORY`, `COMPONENT`, `CONFIG`, `ENTITY`, `CLIENT`, `MAPPER`, `DTO`. Agents can ask "list controllers" or "who injects this repository" directly, instead of grep-ing for `@RestController` and hoping for the best. Roles drive both filtering (`find` with a `NodeFilter`) and ranking.
 
 - **Ranking specialized for Java codebases.** The composite ranker is aware of role, microservice, and FQN structure — not a generic BM25. A search for `"chat ingress"` surfaces controllers before utility classes; a search scoped to one microservice doesn't drown in matches from the other 19. Defaults are tuned on the bank-chat fixture and exposed in `docs/CONFIGURATION.md` for per-repo overrides.
 
@@ -113,7 +113,7 @@ All indexing lifecycle commands (`init`, `increment`, `reprocess`, `install`, `u
 
 If you prefer manual configuration, see [`docs/JAVA-CODEBASE-RAG-CLI.md`](./docs/JAVA-CODEBASE-RAG-CLI.md) for the full CLI reference.
 
-> **Stability disclaimer.** This package does **not** promise backward compatibility. MCP tool contracts, env vars, Lance/Kuzu schemas, config files, and Python APIs may change without a deprecation period. Track `main` and rebuild indexes when ontology or embedding settings change.
+> **Stability disclaimer.** This package does **not** promise backward compatibility. MCP tool contracts, env vars, Lance/LadybugDB schemas, config files, and Python APIs may change without a deprecation period. Track `main` and rebuild indexes when ontology or embedding settings change.
 
 ---
 
@@ -126,7 +126,7 @@ This repo ships a small multi-module Spring fixture under [`tests/bank-chat-syst
 git clone https://github.com/HumanBean17/java-codebase-rag
 cd java-codebase-rag
 
-# 2. Build the index (Lance vectors + Kuzu graph). First run downloads the
+# 2. Build the index (Lance vectors + LadybugDB graph). First run downloads the
 #    embedding model (~90 MB) and takes ~30-60s on the fixture.
 java-codebase-rag init --source-root tests/bank-chat-system --index-dir /tmp/bank-chat-index
 
@@ -141,7 +141,7 @@ Smoke-test the index with two checks (`search_lancedb` ships with the package):
 JAVA_CODEBASE_RAG_INDEX_DIR=/tmp/bank-chat-index \
   python -m search_lancedb "chat ingress controller" --table java --limit 3
 
-# Vector + graph expansion — proves Kuzu is wired in
+# Vector + graph expansion — proves LadybugDB is wired in
 JAVA_CODEBASE_RAG_INDEX_DIR=/tmp/bank-chat-index \
   python -m search_lancedb "chat ingress controller" --table java --limit 3 \
     --graph-expand --expand-depth 2
@@ -241,8 +241,8 @@ Run `java-codebase-rag --help` to list grouped subcommands. Operator playbook wi
 | Setup | `install` | Interactive setup wizard: config, MCP registration, skill/agent deployment, indexing. |
 | Setup | `update` | Refresh shipped artifacts (skill, agent, MCP entry) + incremental Lance/graph catch-up after pip upgrade. |
 | Lifecycle | `init` | First-time index. Refuses if artifacts already exist. |
-| Lifecycle | `increment` | CocoIndex catch-up + incremental Kuzu update. `--vectors-only` for Lance only. |
-| Lifecycle | `reprocess` | Full Lance + Kuzu rebuild. `--vectors-only` / `--graph-only` for a single phase. |
+| Lifecycle | `increment` | CocoIndex catch-up + incremental LadybugDB update. `--vectors-only` for Lance only. |
+| Lifecycle | `reprocess` | Full Lance + LadybugDB rebuild. `--vectors-only` / `--graph-only` for a single phase. |
 | Lifecycle | `erase` | Delete index artifacts. Requires `--yes` or TTY confirm. |
 | Introspection | `meta`, `tables`, `diagnose-ignore`, `unresolved-calls` | Health, table listing, ignore-layer diagnostics, receiver-failure call sites. |
 | Analysis | `analyze-pr` | Blast-radius / risk from a unified diff. |
@@ -277,7 +277,7 @@ python3 -m venv .venv
 
 The `cocoindex` package powers lifecycle commands that run the indexer (`init`, `increment`, `reprocess`, `erase`). Search and MCP navigation do not invoke it directly.
 
-The default embedding model is `sentence-transformers/all-MiniLM-L6-v2` (downloaded on first `init`). Override via the `EMBEDDING_MODEL` env var — see [`docs/CONFIGURATION.md` §1](./docs/CONFIGURATION.md#1-environment-variables).
+The default embedding model is `sentence-transformers/all-MiniLM-L6-v2` (downloaded on first `init`). Override via the `SBERT_MODEL` env var — see [`docs/CONFIGURATION.md` §1](./docs/CONFIGURATION.md#1-environment-variables).
 
 ---
 

{java_codebase_rag-0.6.3.dist-info → java_codebase_rag-0.6.5.dist-info}/RECORD

@@ -1,22 +1,22 @@
-ast_java.py,sha256=NQgZzstbsMq-PdowoD6r_ixJKxEEFzTP9xUzqDpiXeU,99661
+ast_java.py,sha256=Paee3ZV9G5iy8LqfkVq0Ah_1fV0k632oS5JPkiWzwB8,99206
 brownfield_events.py,sha256=yxXkKDgMb3VPtaiakGzncHM_EGnda8xIue6w90yYp8s,2055
-build_ast_graph.py,sha256=LNU1rIHwgEpbrWnWoywg4JdqbwwT4UKcx1iDay_GemM,156776
+build_ast_graph.py,sha256=jubb9Ex_6B3ktsInqNkk-EksBtQjofoFhE92lmHr308,169659
 chunk_heuristics.py,sha256=aQk2NOKxzUdqoUAJUO3G3LE0MN_bYZWNLQ0tkmj5uts,1813
-graph_enrich.py,sha256=POT4LwSkTsrjUmP67bsm2UezUam70cunuPDYDh-v1Bs,63332
+graph_enrich.py,sha256=W_OQK7YRU1K8wi-vfrxZWUd-EErTqWx840UBrhVpNsM,62788
 index_common.py,sha256=HT6FKHFJ084eFvd3fR1j8z8gf4eWoPHVW8GXLpw464I,285
-java_index_flow_lancedb.py,sha256=7CS-sdbzhCkUsHfpCZ85uM7vy4aoZI8n-RowLocfiKc,20267
+java_index_flow_lancedb.py,sha256=JHdRWDZoZ3wnxi8NNG2ck8zEmSALHB2UVB1N8TSvlmg,24109
 java_index_v1_common.py,sha256=nF1KrSqboF_RRvWerG9knRRFmWwsrG_CvhgnsoZ8KqA,1154
 java_ontology.py,sha256=71bCLDNvMy0SpZPzSR5apJ0qJXNd6y5ggkLdBEw_PFo,16682
-ladybug_queries.py,sha256=912j9VAYDjcU4ReVorWQ6R4DZl0tteKic-Pqu0jyBS0,90837
+ladybug_queries.py,sha256=7vSP7WsUKQYXHFenBMlCdpUFed39h3oul-WRkf55YVc,90330
 mcp_hints.py,sha256=3swh05LSiWur3tm3-yssndBsLxIxFhy501kBtJI8jJ0,42509
-mcp_v2.py,sha256=o94GJI7j6dLJDIA3R_1ZiQhjzQfMAEW3etdeZYnHOUc,80637
-path_filtering.py,sha256=-oX16SYLWYwX9pcV1fu3vbVTIhY1GzFflT7J1E2tqPY,17122
-pr_analysis.py,sha256=3-5L8_G5XupdJsl9RN73Lq-ejPoK11B3m_VzAx2fGG8,18413
-search_lancedb.py,sha256=scG6HBUrsgIeSWFrGcLcGdhWv1qODOx4JOBMAlLDY_E,36793
+mcp_v2.py,sha256=S8PiVGDlyI7cvTmeMdQobhKUxWqKoY2YxXXr4pu9lQI,80348
+path_filtering.py,sha256=R--XzI51LXBu5IBKMCnJWbkNr6I5d-SDmltyQQnWco0,17674
+pr_analysis.py,sha256=zrmZZD5yotJtM02Kif6_jgI_oeformOao793akp0N6Y,18394
+search_lancedb.py,sha256=ga_qySeCzA6hCibxUPXpCbTxW9mCjTdwirMhMhfNCz4,36801
 server.py,sha256=DcJwocSknBy0vwsUvYiC5hr-hrD_rRT-u9_DmuRTC9k,35035
 java_codebase_rag/__init__.py,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 java_codebase_rag/_fdlimit.py,sha256=WroFdfSNbcriKok6q8znTf74dqlznxea_1Fd5bHl_3o,1930
-java_codebase_rag/cli.py,sha256=uTCpgsE3gbDPSqPDNJeJ7hZu_MhyKYm4tqHIMkM67LM,38650
+java_codebase_rag/cli.py,sha256=TIrKNQMxu3_dSF2LXKsRKWY5RqRnSLQMrbKAoN4uQBQ,38698
 java_codebase_rag/cli_format.py,sha256=CT7-xdwZ0bMCdP68_UOwkvm-mnLluU3LutlM-mDNk60,1839
 java_codebase_rag/cli_progress.py,sha256=q6Wh97yzLGs1B8UFk_WAKivfQu7Y5RnUUE-T2YHWkIs,3237
 java_codebase_rag/config.py,sha256=bfwYI4R8PU9YV_M4r8-03iaUZ_0TW-qN_NuhIsDXy2M,18769
@@ -24,11 +24,11 @@ java_codebase_rag/installer.py,sha256=sXsHPo24aoDFoTr0D_vYLg0MFdGAV2wdL05FqRaul6
 java_codebase_rag/lance_optimize.py,sha256=25Rwj7HNO8F-35MxhFK6naqgbjd3H-T0zKb3pXB4H0s,9268
 java_codebase_rag/pipeline.py,sha256=ydNktEGL1YniAjJsr37yKBo_bGV4cN_LTGVTmmrsrZw,14688
 java_codebase_rag/progress.py,sha256=2IxdMALDM0wAQCyJrrfZ975zM_85C-4BfHxf4AtYifE,23212
-java_codebase_rag/install_data/agents/explorer-rag-enhanced.md,sha256=APl9d-No12qZNZLjU7mwNRwxHIgnT3ZtQZiD4clWlyU,14413
-java_codebase_rag/install_data/skills/explore-codebase/SKILL.md,sha256=pIM-Xdwq_fXkhhBJCdb-fA2nes5c_mMPcdUXb7Adyxo,12040
-java_codebase_rag-0.6.3.dist-info/licenses/LICENSE,sha256=gxvtiHtuviR_q8ZAjWw-QTcF3DyPzg6ZY-lQrr8OPpw,1068
-java_codebase_rag-0.6.3.dist-info/METADATA,sha256=zDwRolQN2GTSexo7J2ngIqB-RnqmsnTZLRQ-lWuqBL4,17197
-java_codebase_rag-0.6.3.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
-java_codebase_rag-0.6.3.dist-info/entry_points.txt,sha256=wsPZwot0Ui4JI3TIgW8LcbN8bNtKFbwQAlHAAJXfYgQ,117
-java_codebase_rag-0.6.3.dist-info/top_level.txt,sha256=syQgi8XPBwY2ws_NZ1uRCxTf_s41NpshwEHNdcdnk3A,245
-java_codebase_rag-0.6.3.dist-info/RECORD,,
+java_codebase_rag/install_data/agents/explorer-rag-enhanced.md,sha256=BkdQpBEWqSdvGHgbqMdRb5CWfEiFRJK4Dgqbyal3l6s,14551
+java_codebase_rag/install_data/skills/explore-codebase/SKILL.md,sha256=YkRnrM7Wh5E8raFjAW3RrN2V9-ov8upaGC3UdpSx6U8,12346
+java_codebase_rag-0.6.5.dist-info/licenses/LICENSE,sha256=gxvtiHtuviR_q8ZAjWw-QTcF3DyPzg6ZY-lQrr8OPpw,1068
+java_codebase_rag-0.6.5.dist-info/METADATA,sha256=jbKFqnBab9XoMAtIhfX-A9n0ztxhOEPrNcuH5G9qrEM,17242
+java_codebase_rag-0.6.5.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+java_codebase_rag-0.6.5.dist-info/entry_points.txt,sha256=wsPZwot0Ui4JI3TIgW8LcbN8bNtKFbwQAlHAAJXfYgQ,117
+java_codebase_rag-0.6.5.dist-info/top_level.txt,sha256=syQgi8XPBwY2ws_NZ1uRCxTf_s41NpshwEHNdcdnk3A,245
+java_codebase_rag-0.6.5.dist-info/RECORD,,

java_index_flow_lancedb.py

@@ -16,6 +16,7 @@ Usage:
 """
 from __future__ import annotations
 
+import asyncio
 import inspect
 import os
 import sys
@@ -50,7 +51,7 @@ from java_index_v1_common import (
 )
 from path_filtering import LayeredIgnore
 from ast_java import ONTOLOGY_VERSION, parse_java
-from graph_enrich import enrich_chunk
+from graph_enrich import collect_annotation_meta_chain, enrich_chunk, load_brownfield_overrides
 
 # Older cocoindex (e.g. 1.0.0a43) uses ``tracked=False``; newer releases renamed
 # the flag to ``detect_change`` (default False) and reject ``tracked``.
@@ -59,16 +60,21 @@ if "detect_change" in _ck_params:
     PROJECT_ROOT = coco.ContextKey[Path]("java_lance_project_root")
     LANCE_DB = coco.ContextKey("java_lance_async_conn")
     EMBEDDER = coco.ContextKey[SentenceTransformerEmbedder]("java_lance_embedder")
+    IGNORE = coco.ContextKey[LayeredIgnore]("java_lance_layered_ignore")
 elif "tracked" in _ck_params:
     PROJECT_ROOT = coco.ContextKey[Path]("java_lance_project_root", tracked=False)
     LANCE_DB = coco.ContextKey("java_lance_async_conn", tracked=False)
     EMBEDDER = coco.ContextKey[SentenceTransformerEmbedder](
         "java_lance_embedder", tracked=False
     )
+    IGNORE = coco.ContextKey[LayeredIgnore](
+        "java_lance_layered_ignore", tracked=False
+    )
 else:
     PROJECT_ROOT = coco.ContextKey[Path]("java_lance_project_root")
     LANCE_DB = coco.ContextKey("java_lance_async_conn")
     EMBEDDER = coco.ContextKey[SentenceTransformerEmbedder]("java_lance_embedder")
+    IGNORE = coco.ContextKey[LayeredIgnore]("java_lance_layered_ignore")
 
 splitter = RecursiveSplitter()
 
@@ -198,12 +204,12 @@ def _approximate_vectors_total(project_root: Path) -> int:
                 continue
             # Java: **/*.java
             if fn.endswith(".java"):
-                if not ignore.is_ignored(full)[0]:
+                if not ignore.is_ignored(full):
                     total += 1
                 continue
             # SQL: **/src/main/resources/db/migration/*.sql
             if fn.endswith(".sql") and "/db/migration/" in rel:
-                if not ignore.is_ignored(full)[0]:
+                if not ignore.is_ignored(full):
                     total += 1
                 continue
             # YAML: **/src/main/resources/application*.yml / .yaml
@@ -214,7 +220,7 @@ def _approximate_vectors_total(project_root: Path) -> int:
             # total below the actual done count. The ``rel``-based
             # ``"/src/main/resources/"`` gate stays (full path component).
             if fn.endswith((".yml", ".yaml")) and fn.startswith("application") and "/src/main/resources/" in rel:
-                if not ignore.is_ignored(full)[0]:
+                if not ignore.is_ignored(full):
                     total += 1
     return total
 
@@ -291,6 +297,7 @@ async def coco_lifespan(builder: coco.EnvironmentBuilder) -> AsyncIterator[None]
         trust_remote_code=True,
     )
     builder.provide(EMBEDDER, embedder)
+    builder.provide(IGNORE, LayeredIgnore(root))
 
     uri = str(index_dir)
 
@@ -306,6 +313,40 @@ async def coco_lifespan(builder: coco.EnvironmentBuilder) -> AsyncIterator[None]
     yield
 
 
+def _parse_and_enrich_java(
+    content_bytes: bytes,
+    chunks: list[Any],
+    rel: str,
+    project_root: Path,
+) -> list[Any]:
+    """Parse one Java file and enrich every chunk, off the event loop.
+
+    Returns a list of :class:`graph_enrich.ChunkEnrichment` aligned 1:1 with
+    ``chunks``. Intended to run via ``asyncio.to_thread`` from
+    ``process_java_file`` (vectors perf lever #2): while the worker thread
+    parses + enriches, the event loop is free to drive other files and keep the
+    embedder's batching queue fed.
+
+    Thread-safety: ``parse_java`` uses a per-thread tree-sitter ``Parser``
+    (see ``ast_java._parser``), so it is safe to call concurrently from these
+    worker threads — including the transitive ``parse_java`` that ``enrich_chunk``
+    triggers via ``collect_annotation_meta_chain`` → ``_collect_annotation_decl_index``.
+    ``enrich_chunk`` is otherwise pure-Python over the now-immutable AST; its
+    ``lru_cache`` reads are thread-safe under the GIL.
+    """
+    ast = parse_java(content_bytes)
+    return [
+        enrich_chunk(
+            ast,
+            chunk_start_byte=ch.start.byte_offset,
+            chunk_end_byte=ch.end.byte_offset,
+            file_path=rel,
+            project_root=project_root,
+        )
+        for ch in chunks
+    ]
+
+
 @coco.fn(memo=True)
 async def process_java_file(
     file: localfs.File,
@@ -313,7 +354,8 @@ async def process_java_file(
 ) -> None:
     embedder = coco.use_context(EMBEDDER)
     project_root = coco.use_context(PROJECT_ROOT)
-    if LayeredIgnore(project_root).is_ignored((project_root / file.file_path.path).resolve())[0]:
+    ignore = coco.use_context(IGNORE)
+    if ignore.is_ignored((project_root / file.file_path.path).resolve()):
         return
     try:
         content = await file.read_text()
@@ -326,6 +368,9 @@ async def process_java_file(
 
     language = detect_code_language(filename=file.file_path.path.name) or "text"
     cs, mn, ov = JAVA_CHUNK
+    # ``splitter.split`` stays inline: the module-level ``RecursiveSplitter``
+    # shares one Rust object, so keeping split on the event loop preserves its
+    # existing single-threaded access (no new cross-file concurrency hazard).
     chunks = splitter.split(
         content,
         cs,
@@ -335,18 +380,21 @@ async def process_java_file(
     )
     rel = file.file_path.path.as_posix()
     content_bytes = content.encode("utf-8", errors="replace")
-    ast = parse_java(content_bytes)
 
-    for ch in chunks:
+    # (vectors perf lever #2) parse + enrich off the event loop so the loop can
+    # keep the embedder's batching queue fed while this file is being parsed.
+    # parse_java is thread-safe (per-thread tree-sitter Parser in ast_java).
+    enrichments = await asyncio.to_thread(
+        _parse_and_enrich_java, content_bytes, chunks, rel, project_root
+    )
+    # (vectors perf lever #1) embed all chunks concurrently so the batched
+    # embedder groups them into one ``model.encode(...)`` (max_batch_size=64)
+    # instead of N serial batch-of-1 calls. Dominant win for ``increment``
+    # (few changed files → little cross-file concurrency → otherwise no batching).
+    embeddings = await asyncio.gather(*(embedder.embed(ch.text) for ch in chunks))
+
+    for ch, enrich, emb in zip(chunks, enrichments, embeddings):
         rs, re = chunk_key_range(ch)
-        enrich = enrich_chunk(
-            ast,
-            chunk_start_byte=ch.start.byte_offset,
-            chunk_end_byte=ch.end.byte_offset,
-            file_path=rel,
-            project_root=project_root,
-        )
-        emb = await embedder.embed(ch.text)
         table.declare_row(
             row=JavaLanceChunk(
                 id=str(uuid.uuid4()),
@@ -379,7 +427,8 @@ async def process_sql_file(
 ) -> None:
     embedder = coco.use_context(EMBEDDER)
     project_root = coco.use_context(PROJECT_ROOT)
-    if LayeredIgnore(project_root).is_ignored((project_root / file.file_path.path).resolve())[0]:
+    ignore = coco.use_context(IGNORE)
+    if ignore.is_ignored((project_root / file.file_path.path).resolve()):
         return
     try:
         content = await file.read_text()
@@ -401,9 +450,11 @@ async def process_sql_file(
     )
     rel = file.file_path.path.as_posix()
 
-    for ch in chunks:
+    # (vectors perf lever #1) embed chunks concurrently → batched encode.
+    embeddings = await asyncio.gather(*(embedder.embed(ch.text) for ch in chunks))
+
+    for ch, emb in zip(chunks, embeddings):
         rs, re = chunk_key_range(ch)
-        emb = await embedder.embed(ch.text)
         table.declare_row(
             row=SqlLanceChunk(
                 id=str(uuid.uuid4()),
@@ -425,7 +476,8 @@ async def process_yaml_file(
 ) -> None:
     embedder = coco.use_context(EMBEDDER)
     project_root = coco.use_context(PROJECT_ROOT)
-    if LayeredIgnore(project_root).is_ignored((project_root / file.file_path.path).resolve())[0]:
+    ignore = coco.use_context(IGNORE)
+    if ignore.is_ignored((project_root / file.file_path.path).resolve()):
         return
     try:
         content = await file.read_text()
@@ -448,9 +500,11 @@ async def process_yaml_file(
     )
     rel = file.file_path.path.as_posix()
 
-    for ch in chunks:
+    # (vectors perf lever #1) embed chunks concurrently → batched encode.
+    embeddings = await asyncio.gather(*(embedder.embed(ch.text) for ch in chunks))
+
+    for ch, emb in zip(chunks, embeddings):
         rs, re = chunk_key_range(ch)
-        emb = await embedder.embed(ch.text)
         table.declare_row(
             row=YamlLanceChunk(
                 id=str(uuid.uuid4()),
@@ -501,6 +555,26 @@ async def app_main() -> None:
     )
 
     project_root = coco.use_context(PROJECT_ROOT)
+    # Warm per-project enrichment caches ONCE on the event-loop thread, BEFORE
+    # coco.mount_each fans files into worker threads. collect_annotation_meta_chain
+    # and load_brownfield_overrides are lru_cached per (resolved) project root;
+    # without warming, the first wave of concurrent process_java_file worker
+    # threads each cold-miss and redundantly walk+parse the ENTIRE project (a
+    # thundering herd that would offset the embedding-batching win on large
+    # repos — perf lever #2 made enrich concurrent). With warming, every worker
+    # hits a populated cache (lru_cache reads are thread-safe). Key derivation
+    # mirrors enrich_chunk exactly so the warmed entries are the ones workers hit.
+    try:
+        load_brownfield_overrides(project_root)
+        try:
+            prs = str(Path(project_root).resolve())
+        except OSError:
+            prs = str(project_root)
+        collect_annotation_meta_chain(prs)
+    except Exception:
+        # Warm-up must never break indexing — a failure just means workers
+        # cold-miss lazily (the pre-warming behavior). Swallow and continue.
+        pass
     _ignore = LayeredIgnore(project_root)
     _walk_excludes = _ignore.cocoindex_excluded_patterns()
     # Emit ONE approximate total so the parent's renderer can show a determinate