codemap-core 0.3.3__tar.gz → 0.3.5__tar.gz

{codemap_core-0.3.3 → codemap_core-0.3.5}/CHANGELOG.md

@@ -8,6 +8,77 @@ During `0.x`, MINOR may introduce breaking changes — they will be marked `BREA
 
 ## [Unreleased]
 
+## [0.3.5] — 2026-06-26
+
+Lockstep version-only bump across all **20 packages**; the only source
+change is in `codemap-aimemory` — first half of the AI-Enterprise-
+Delivery-System roadmap **P3-1**: a CLI that downstream tools (specode)
+can call to recall relevant knowledge before drafting requirements.
+
+### `codemap-aimemory` — `codemap recall <query>` CLI (P3-1, half 1)
+
+* New module `codemap_aimemory.recall` with a pure scorer
+  `recall(query, project_root, top_k, types)` that scans
+  `<project_root>/.ai-memory/knowledge/{rules,business,modules,cases,
+  pitfalls}/*.yml` and ranks by token overlap with each yml's
+  searchable fields (title / tags / statement / related_code / fix /
+  symptom / …). Title hits weighted higher than body hits.
+* New `codemap_aimemory.recall_cli` registers the **`codemap recall`**
+  subcommand via the `codemap.cli_commands` entry-point (the same
+  group that hosts `enrich` and `llm`). Flags:
+  `--project/-p <abs>` (default cwd), `--top-k/-k N` (default 5),
+  `--types/-t rules,business,…` (category filter),
+  `--output/-o yaml|json` (default yaml).
+* Output carries `query`, parsed `tokens`, `matched_entities` (id-based
+  hits against `_global/entities.yml`), and `knowledge` rows
+  (`knowledge_id`, `type`, `category`, `title`, `summary`, `score`,
+  `file`).
+* Tokenizer is dependency-free: English/digit words ≥ 2 chars +
+  Chinese char-bigrams. Good enough for the few-word requirements
+  query; embeddings land in P1-3.
+* 18 new unit tests (`test_recall.py`) cover tokenizer edge cases
+  (English / Chinese / mixed / empty), ranking, top-k, type filter,
+  title-weight, related_code/symptom/fix matching, malformed yml,
+  matched_entities, and full output shape.
+
+### Roadmap pairing
+
+The companion change in `pluginhub` extends `specode`'s requirements
+phase to invoke `codemap recall` before writing `requirements.md` and
+inject the top hits into the "已知约束 / 历史坑" section — closing the
+P3-1 loop end-to-end.
+
+## [0.3.4] — 2026-06-25
+
+Lockstep version-only bump across all **20 packages**; the only source
+change is in `codemap-aimemory` — closes the AI-Enterprise-Delivery-
+System roadmap **P1-2** (global entity index — business ↔ code).
+
+### `codemap-aimemory` — global entity cross-walk (P1-2)
+
+* New module `codemap_aimemory.global_entities` with a pure aggregator
+  `build_global_entities(out_dir, code_entity_ids)` that walks the
+  emitter's own entity vocabulary (`fn-*`, `cls-*`, `tbl-*`, `mod-*`)
+  alongside any `<out_dir>/knowledge/{rules,business,modules,cases,
+  pitfalls}/*.yml` produced by `spec-distill v2`. Knowledge yml is
+  scanned for three reference channels: `related_code[].entity`,
+  `related_knowledge[]`, and `affects[]` (pitfalls). Malformed yml
+  is silently skipped — a hand-edited knowledge file never crashes
+  the build.
+* `AiMemoryEmitter` now writes **`.ai-memory/_global/entities.yml`**:
+  one row per entity with `id`, inferred `type` (from id prefix),
+  `source` (`code` / `knowledge` / `both`), and `knowledge_refs`
+  (every knowledge_id that mentions this entity). Top-level `counts`
+  partition: `total` / `code_only` / `knowledge_only` / `both`.
+* Type inference covers all current prefixes: `fn-` → function,
+  `cls-` → class, `tbl-` → table, `mod-` → module, `rule-` → rule,
+  `biz-` → business_process, `case-` → case, `pit-` → pitfall.
+* Closes the L1 ↔ L2/L3 lookup gap: an Agent answering "where is
+  `sf_coupon` referenced?" now hits one yml instead of walking the
+  whole knowledge tree.
+* 13 new unit tests (`test_global_entities.py`) + 1 emitter integration
+  test cover empty / partial / cross-linked / malformed scenarios.
+
 ## [0.3.3] — 2026-06-25
 
 Lockstep version-only bump across all **20 packages**; the only source

{codemap_core-0.3.3 → codemap_core-0.3.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codemap-core
-Version: 0.3.3
+Version: 0.3.5
 Summary: Language-neutral code index for AI agents
 Project-URL: Homepage, https://github.com/qxbyte/codemap
 Project-URL: Repository, https://github.com/qxbyte/codemap

{codemap_core-0.3.3 → codemap_core-0.3.5}/pyproject.toml

@@ -12,7 +12,7 @@ build-backend = "hatchling.build"
 # (`codemap-java`, `codemap-go`, …) — this is the engine; the others
 # are language adapters.
 name = "codemap-core"
-version = "0.3.3"
+version = "0.3.5"
 description = "Language-neutral code index for AI agents"
 readme = "README.md"
 license = { file = "LICENSE" }