sourcecode 1.30.17__tar.gz → 1.30.19__tar.gz

sourcecode-1.30.19/.continue-here.md

@@ -0,0 +1,150 @@
+# Continue Here — atlas-cli sesión 15
+
+**Paused:** 2026-05-18 (sesión 15)
+**Repo:** `/Users/user/Downloads/atlas-cli`
+**Branch:** master
+**Working tree:** 3 ficheros modificados, SIN commit
+
+---
+
+## Trabajo de esta sesión
+
+### Ya commiteado (66db10b)
+
+`fix(cli): three flag-routing bugs in compact/repo-ir paths`
+
+| Bug | Fichero | Fix |
+|---|---|---|
+| `--changed-only` = `--compact` (byte idéntico) | `cli.py` | Filtrar `sm.file_paths`, `entry_points`, `code_notes` antes de `compact_view`; `_allowed_changed_files` se computaba pero nunca se aplicaba porque `compact` forzaba `mode=raw` |
+| `--compact --format yaml` ignorado | `cli.py` | Rama compact hardcodeaba `json.dumps`; añadir bloque `if format == "yaml"` igual al de contract_view |
+| `sourcecode repo-ir .` → "No such command '.'" | `cli.py` | `"repo-ir"` ausente de `_SUBCOMMANDS`; preprocessor consumía token como PATH antes de que Click pudiera dispatchar |
+
+---
+
+### Pendiente de commit (3 ficheros modificados)
+
+**Para commitear (primera acción al retomar):**
+
+```bash
+cd /Users/user/Downloads/atlas-cli
+git add src/sourcecode/contract_pipeline.py \
+        src/sourcecode/repository_ir.py \
+        src/sourcecode/prepare_context.py
+
+git commit -m "fix(encoding): utf-8 en subprocess + symptom boost por commit messages
+
+- repository_ir.py _get_git_old_content: añadir encoding='utf-8', errors='replace'
+  al subprocess.run de git show — causa raíz del UnicodeDecodeError cp1252 en Windows
+- contract_pipeline.py grep -rl: mismo fix (fallback Python en Windows, defensivo)
+- prepare_context.py: symptom boost refactor en 6 passes
+  - Nuevo: _recent_commits_for_symptom extraído en bloque 5b (sin coste extra)
+  - Nuevo: commit message index (file → n commits con keyword en mensaje)
+  - Nuevo: inyección desde commit index para ficheros no en candidate pool
+  - Fix: content scan +0.05/hit, cap +0.50 (era +0.02, cap +0.30)
+  - Fix: code notes matching → back-propagate score al fichero (+0.20/note)
+  - Fix: path-kw boost aplicado a candidatos pre-existentes (no solo inyectados)
+  - Multi-signal boost en un solo file-read pass; reason string explica cada boost"
+```
+
+---
+
+## Qué hace cada cambio pendiente
+
+### `repository_ir.py` + `contract_pipeline.py` — encoding fix
+
+```python
+# Antes (crash en Windows cp1252):
+result = subprocess.run([...], text=True, timeout=5)
+
+# Después:
+result = subprocess.run([...], text=True, encoding="utf-8", errors="replace", timeout=5)
+```
+
+Causa raíz del bug reportado: `sourcecode repo-ir . --since HEAD~3` en Windows 11 Python 3.10
+crasheaba con `UnicodeDecodeError: 'charmap' codec can't decode byte 0x81` porque
+`subprocess` hereda `cp1252` como encoding por defecto en Windows.
+
+### `prepare_context.py` — symptom boost overhaul
+
+**Problema original:** `--symptom "sesiones"` = output idéntico al baseline porque:
+- Ningún path Java contiene "sesiones" (clases no se nombran en español)
+- Content scan era demasiado débil (+0.02/hit → max +0.20 para 10 hits)
+- Git commit messages no se escaneaban
+
+**Nuevo pipeline (6 passes):**
+
+| Pass | Señal | Boost | Cap |
+|---|---|---|---|
+| 1 | Code notes text match → `related_notes` + `_note_matched_paths` | — | — |
+| 2 | **Commit index**: `file → n_commits` con keyword en mensaje | — | — |
+| 3 | **Inyección desde commit index** (ficheros no en pool) | +0.15/commit desde 0.5 | 0.85 |
+| 4 | Path injection (sin cambios lógicos) | +0.20/kw desde 0.5 | 1.0 |
+| 5 | **Multi-signal boost** en 1 file-read: commit (+0.25/commit, cap+0.40), note (+0.20/note, cap+0.30), path-kw (+0.20/kw), content (+0.05/hit, cap+0.50) | acumulativo | 1.0 |
+| 6 | Cross-layer synonym frontend→backend (sin cambios) | sin cambios | — |
+
+**Before/after verificado** en BroadleafCommerce:
+- `--symptom "pagination"`: top 10 pasa de servicios genéricos (0.67) a `PageDao.java` (0.90, role=`symptom_match`, reason="commit-msg symptom ×1 (+0.25)")
+- `--symptom "zookeeper"`: `pom.xml` (1.00, 2 commit hits), `ZookeeperUtil.java` (1.00, path+content)
+
+---
+
+## Estado de tests
+
+```
+920 passed, 3 skipped
+```
+
+Pre-existing failures (no tocar):
+- `tests/test_block2_coverage.py` — bug `DocRecord` en `doc_analyzer.py`
+- `tests/test_dependency_analyzer_node_python.py::test_python_requirements_without_lockfile_keeps_declared_versions`
+
+---
+
+## Pendientes (próxima sesión)
+
+### Heredados de sesión 14 (aún vigentes)
+
+1. **Spring bean wiring graph** — `_build_relations` emite `injects` edges para `@Autowired`. Falta: trazar `@Configuration → @Bean → @Autowired` completo, exponer como `wiring_graph`.
+
+2. **Smoke test en repo spring-boot real:**
+   ```bash
+   cd ~/Documents/workspace/spring-boot-realworld-example-app
+   sourcecode repo-ir . --since HEAD~1 2>/dev/null | python3 -c "
+   import json, sys; d=json.load(sys.stdin)
+   print('nodes:', len(d['graph']['nodes']))
+   print('route_surface:', d.get('route_surface', []))
+   print('reverse_graph keys:', len(d.get('reverse_graph', {})))
+   "
+   ```
+
+3. **Transaction boundary tracking** — `@Transactional` detectado. Falta: emitir `transactional_boundary` edges.
+
+4. **`_diff_routes` para `@RequestMapping` en clase** — class-level mapping change debería propagarse a todos los endpoints del controller.
+
+### Nuevos de sesión 15
+
+5. **Tests para el symptom fix** — validar que `--symptom "sesiones"` con commits mock que contengan "sesiones" en mensaje produce ranking diferente al baseline. Cobertura en `tests/test_prepare_context_symptom.py` o similar.
+
+6. **`_FRONTEND_SYMPTOM_MAP`** — añadir "sesiones"/"session" como frontend keyword con backend terms `["httpsession", "sessionmanager", "sessionservice", "sessionrepository"]`.
+
+---
+
+## Para retomar
+
+```bash
+cd /Users/user/Downloads/atlas-cli
+git log --oneline -3   # debe mostrar 66db10b como más reciente
+git status             # 3 ficheros M: contract_pipeline, prepare_context, repository_ir
+
+# Primera acción: commitear (comando arriba)
+
+python3 -m pytest tests/ \
+  --ignore=tests/test_block2_coverage.py \
+  --ignore=tests/test_dependency_analyzer_node_python.py \
+  -q
+# Expected: 920 passed, 3 skipped
+```
+
+---
+
+*Pausado 2026-05-18 sesión 15 — gsd:pause-work*

{sourcecode-1.30.17 → sourcecode-1.30.19}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 1.30.17
+Version: 1.30.19
 Summary: Deterministic codebase context for AI coding agents
 License:                                  Apache License
                                    Version 2.0, January 2004
@@ -221,7 +221,7 @@ Description-Content-Type: text/markdown
 
 **Deterministic, behavior-aware codebase context for AI agents and PR review.**
 
-![Version](https://img.shields.io/badge/version-1.30.17-blue)
+![Version](https://img.shields.io/badge/version-1.30.19-blue)
 ![Python](https://img.shields.io/badge/python-3.10%2B-green)
 
 ---
@@ -257,7 +257,7 @@ pipx install sourcecode
 
 ```bash
 sourcecode version
-# sourcecode 1.30.17
+# sourcecode 1.30.19
 ```
 
 ---

{sourcecode-1.30.17 → sourcecode-1.30.19}/README.md

@@ -2,7 +2,7 @@
 
 **Deterministic, behavior-aware codebase context for AI agents and PR review.**
 
-![Version](https://img.shields.io/badge/version-1.30.17-blue)
+![Version](https://img.shields.io/badge/version-1.30.19-blue)
 ![Python](https://img.shields.io/badge/python-3.10%2B-green)
 
 ---
@@ -38,7 +38,7 @@ pipx install sourcecode
 
 ```bash
 sourcecode version
-# sourcecode 1.30.17
+# sourcecode 1.30.19
 ```
 
 ---

{sourcecode-1.30.17 → sourcecode-1.30.19}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sourcecode"
-version = "1.30.17"
+version = "1.30.19"
 description = "Deterministic codebase context for AI coding agents"
 readme = "README.md"
 requires-python = ">=3.9"

{sourcecode-1.30.17 → sourcecode-1.30.19}/src/sourcecode/__init__.py

@@ -1,3 +1,3 @@
 """sourcecode — Deterministic codebase context maps for AI coding agents."""
 
-__version__ = "1.30.17"
+__version__ = "1.30.19"

{sourcecode-1.30.17 → sourcecode-1.30.19}/src/sourcecode/cli.py

@@ -1648,6 +1648,12 @@ def prepare_context_cmd(
         "--format",
         help="Output format: json (default) | github-comment (Markdown PR comment for review-pr task)",
     ),
+    debug_perf: bool = typer.Option(
+        False,
+        "--debug-perf",
+        help="Emit per-phase timing to stderr (git scan ms, symptom scoring ms, total ms)",
+        hidden=True,
+    ),
 ) -> None:
     """Task-specific context for AI coding agents.
 
@@ -1714,9 +1720,21 @@ def prepare_context_cmd(
         raise typer.Exit()
 
     from dataclasses import asdict
+    import time as _time
 
     builder = TaskContextBuilder(target)
+    _t0 = _time.perf_counter()
     output = builder.build(task, since=since, symptom=symptom)
+    _t_total = (_time.perf_counter() - _t0) * 1000
+
+    if debug_perf:
+        _perf = getattr(output, "_perf_ms", {}) or {}
+        typer.echo(
+            f"[debug-perf] total={_t_total:.0f}ms"
+            + (f"  git_scan={_perf.get('git_scan_ms', '?')}ms" if "git_scan_ms" in _perf else "")
+            + (f"  symptom={_perf.get('symptom_ms', '?')}ms" if "symptom_ms" in _perf else ""),
+            err=True,
+        )
 
     # Task-specific content-filter: each task emphasizes different output fields.
     # Fields marked False are suppressed from this task's output to reduce noise.
@@ -1918,6 +1936,8 @@ def prepare_context_cmd(
         out["related_notes"] = output.related_notes
     if output.symptom_note:
         out["symptom_note"] = output.symptom_note
+    if output.symptom_explain:
+        out["symptom_explain"] = output.symptom_explain
     if llm_prompt:
         out["llm_prompt"] = builder.render_prompt(output)
 
@@ -2012,12 +2032,33 @@ def repo_ir_cmd(
         "--include-tests",
         help="Include test files in analysis (excluded by default)",
     ),
+    max_nodes: Optional[int] = typer.Option(
+        None,
+        "--max-nodes",
+        help="Limit graph.nodes to top N by impact score (reduces output size)",
+    ),
+    max_edges: Optional[int] = typer.Option(
+        None,
+        "--max-edges",
+        help="Limit graph.edges to N (priority: edges between kept nodes)",
+    ),
+    summary_only: bool = typer.Option(
+        False,
+        "--summary-only",
+        help="Omit full graph.nodes/edges; keep analysis summary, impact, and change_set (<300KB typical)",
+    ),
 ) -> None:
     """Deterministic symbol-level IR for Java repositories.
 
     \b
     Extracts symbols, relations, Spring roles, and (with --since) symbol-level diffs.
-    Output is JSON: symbols[], relations[], changed_symbols[], spring_summary, graph_metadata.
+    Output is JSON: graph{nodes,edges}, analysis, impact, subsystems, change_set.
+
+    \b
+    Size control:
+      --summary-only          Omit full graph; keep analysis + impact (smallest output)
+      --max-nodes N           Keep top N nodes by score
+      --max-edges N           Keep top N edges (priority: both endpoints kept)
 
     \b
     Examples:
@@ -2025,11 +2066,13 @@ def repo_ir_cmd(
       sourcecode repo-ir /path/to/repo --since HEAD~1
       sourcecode repo-ir --files src/main/java/UserService.java
       sourcecode repo-ir --since main --output ir.json
+      sourcecode repo-ir --since HEAD~3 --summary-only --output ir-small.json
+      sourcecode repo-ir --max-nodes 200 --max-edges 500
     """
     import json as _json
     import sys as _sys
 
-    from sourcecode.repository_ir import build_repo_ir, find_java_files
+    from sourcecode.repository_ir import apply_ir_size_limits, build_repo_ir, find_java_files
 
     root = path.resolve()
     if not root.is_dir():
@@ -2063,11 +2106,24 @@ def repo_ir_cmd(
         return
 
     ir = build_repo_ir(file_list, root, since=since)
+    ir = apply_ir_size_limits(
+        ir,
+        max_nodes=max_nodes,
+        max_edges=max_edges,
+        summary_only=summary_only,
+    )
     output = _json.dumps(ir, indent=2, ensure_ascii=False)
 
     if output_path:
         output_path.write_text(output, encoding="utf-8")
-        typer.echo(f"IR written to {output_path}", err=True)
+        n_nodes = len((ir.get("graph") or {}).get("nodes") or [])
+        n_edges = len((ir.get("graph") or {}).get("edges") or [])
+        size_kb = len(output.encode("utf-8")) // 1024
+        typer.echo(
+            f"IR written to {output_path} "
+            f"({size_kb}KB, {n_nodes} nodes, {n_edges} edges)",
+            err=True,
+        )
     else:
         _sys.stdout.write(output)
         _sys.stdout.write("\n")

{sourcecode-1.30.17 → sourcecode-1.30.19}/src/sourcecode/contract_pipeline.py

@@ -623,6 +623,8 @@ def _find_symbol_files(
             cwd=str(root),
             capture_output=True,
             text=True,
+            encoding="utf-8",
+            errors="replace",
             timeout=20,
         )
         for line in result.stdout.splitlines():

{sourcecode-1.30.17 → sourcecode-1.30.19}/src/sourcecode/git_analyzer.py

@@ -52,8 +52,15 @@ _HOTSPOT_AUX_DIRS: frozenset[str] = frozenset({
     "sandbox", "sandboxes",
     "ci", "translations", "locales", "locale", "i18n", "l10n",
     ".planning",
+    # vendor / build dirs: high path count, zero operational signal
+    "vendor", "node_modules", "dist", "target", "build",
+    ".gradle", ".mvn", "generated-sources", "generated-resources",
 })
 
+# When git diff returns more changed files than this, degrade gracefully:
+# truncate hotspot analysis depth and emit a structured warning.
+_CHANGED_FILES_DEGRADATION_THRESHOLD = 200
+
 
 def _run_git(args: list[str], cwd: Path, timeout: int = 15) -> tuple[str, int]:
     result = subprocess.run(