sourcecode 1.33.23__tar.gz → 1.33.25__tar.gz

{sourcecode-1.33.23 → sourcecode-1.33.25}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: sourcecode
-Version: 1.33.23
+Version: 1.33.25
 Summary: Persistent structural context and ultra-fast repeated analysis for AI coding agents
 License-File: LICENSE
 Keywords: agents,ai,codebase,context,developer-tools,llm
@@ -39,7 +39,7 @@ Description-Content-Type: text/markdown
 
 **Persistent structural context and ultra-fast repeated analysis for AI coding agents.**
 
-![Version](https://img.shields.io/badge/version-1.33.23-blue)
+![Version](https://img.shields.io/badge/version-1.33.25-blue)
 ![Python](https://img.shields.io/badge/python-3.10%2B-green)
 
 ---

{sourcecode-1.33.23 → sourcecode-1.33.25}/README.md

@@ -2,7 +2,7 @@
 
 **Persistent structural context and ultra-fast repeated analysis for AI coding agents.**
 
-![Version](https://img.shields.io/badge/version-1.33.23-blue)
+![Version](https://img.shields.io/badge/version-1.33.25-blue)
 ![Python](https://img.shields.io/badge/python-3.10%2B-green)
 
 ---

{sourcecode-1.33.23 → sourcecode-1.33.25}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "sourcecode"
-version = "1.33.23"
+version = "1.33.25"
 description = "Persistent structural context and ultra-fast repeated analysis for AI coding agents"
 readme = "README.md"
 requires-python = ">=3.9"

{sourcecode-1.33.23 → sourcecode-1.33.25}/src/sourcecode/__init__.py

@@ -1,3 +1,3 @@
 """sourcecode — Deterministic codebase context maps for AI coding agents."""
 
-__version__ = "1.33.23"
+__version__ = "1.33.25"

{sourcecode-1.33.23 → sourcecode-1.33.25}/src/sourcecode/entrypoint_classifier.py

@@ -25,6 +25,31 @@ _DEVELOPMENT_DIRS = frozenset({
 _DEV_MARKERS = ("rspress", "vite", "storybook", "playground", "dev-server")
 _PRODUCTION_SCRIPT_REASONS = {"script:start", "script:serve", "script:server"}
 
+# JVM source trees: directories below these markers are package namespaces,
+# not functional directory roles — "com/example/spaghetti" must not match
+# _AUXILIARY_DIRS just because "example" is a package name component.
+_JVM_SRC_ROOTS = (
+    "src/main/java/", "src/test/java/",
+    "src/main/kotlin/", "src/test/kotlin/",
+    "src/main/scala/", "src/test/scala/",
+    "src/integration-test/java/", "src/integrationTest/java/",
+)
+
+
+def _dir_check_parts(path: str) -> frozenset[str]:
+    """Return path segments used for auxiliary/development dir-name matching.
+
+    For JVM source files the package path (below the source root) is a
+    namespace, not a functional directory hierarchy.  Truncate there so
+    package components like 'example' in com.example.* don't false-match
+    _AUXILIARY_DIRS entries.
+    """
+    for root in _JVM_SRC_ROOTS:
+        idx = path.find(root)
+        if idx >= 0:
+            return frozenset(path[: idx + len(root)].split("/"))
+    return frozenset(path.split("/"))
+
 
 def classify_entry_point(ep: EntryPoint) -> Classification:
     """Return the operational class for an entry point.
@@ -33,7 +58,7 @@ def classify_entry_point(ep: EntryPoint) -> Classification:
     and auxiliary path evidence wins over detector-provided production labels.
     """
     path = ep.path.replace("\\", "/").lower()
-    parts = set(path.split("/"))
+    parts = _dir_check_parts(path)
     reason = (ep.reason or "").lower()
     evidence = (ep.evidence or "").lower()
     marker_text = f"{path} {reason} {evidence}"

{sourcecode-1.33.23 → sourcecode-1.33.25}/src/sourcecode/prepare_context.py

@@ -864,6 +864,10 @@ _SYMPTOM_STOP_WORDS: frozenset[str] = frozenset({
     "error", "issue", "problem", "bug",
     "from", "into", "via", "due", "also", "after", "before",
     "slow", "fast", "new", "old",
+    # Diagnostic abbreviations — too ubiquitous in comments/javadoc to be useful
+    # as path/content signals (e.g. "npe" matches "NPE" in thousands of comments).
+    "npe", "oom", "oob", "iae", "ise", "npe", "cce", "aioobe",
+    "exception", "throwable", "stacktrace",
 })
 
 # Repo-scale threshold: above this file count, use stricter injection logic.
@@ -2363,6 +2367,26 @@ class TaskContextBuilder:
                 if _is_large_repo and len(relevant_files) > 40:
                     relevant_files = relevant_files[:40]
 
+                # Score normalization: prevent all-1.0 saturation when many files
+                # accumulate enough boost to hit the cap.  Normalize display scores
+                # relative to max raw signal so the ranked gradient is preserved.
+                _norm_max = max(
+                    (_raw_signals.get(rf.path, rf.score) for rf in relevant_files),
+                    default=1.0,
+                ) or 1.0
+                if _norm_max > 1.0:
+                    relevant_files = [
+                        RelevantFile(
+                            path=rf.path,
+                            role=rf.role,
+                            score=round(min(_raw_signals.get(rf.path, rf.score) / _norm_max, 1.0), 3),
+                            reason=rf.reason,
+                            why=rf.why,
+                            tier=rf.tier,
+                        )
+                        for rf in relevant_files
+                    ]
+
                 # Synonym note (only when synonyms actually fired)
                 if _frontend_kws and _sx_synonyms:
                     symptom_note = (
@@ -2491,12 +2515,27 @@ class TaskContextBuilder:
                     key=lambda x: -(x["public_method_count"] * (1.5 if x["has_framework_annotations"] else 1.0))
                 )
                 _top = _java_candidates if all_gaps else _java_candidates[:20]
+                _max_rank = max(
+                    (c["public_method_count"] * (1.5 if c["has_framework_annotations"] else 1.0)
+                     for c in _java_candidates),
+                    default=1.0,
+                ) or 1.0
                 test_gaps = [
                     {
                         "path": c["path"],
                         "public_method_count": c["public_method_count"],
                         "has_framework_annotations": c["has_framework_annotations"],
                         "rank_score": round(c["public_method_count"] * (1.5 if c["has_framework_annotations"] else 1.0), 1),
+                        "score": round(min(
+                            c["public_method_count"] * (1.5 if c["has_framework_annotations"] else 1.0) / _max_rank,
+                            1.0,
+                        ), 3),
+                        "reason": (
+                            (f"{c['public_method_count']} public method{'s' if c['public_method_count'] != 1 else ''}; "
+                             if c["public_method_count"] else "")
+                            + ("has Spring/Jakarta framework annotations" if c["has_framework_annotations"]
+                               else "no test coverage detected")
+                        ).strip("; ") or "untested source file",
                     }
                     for c in _top
                 ]

{sourcecode-1.33.23 → sourcecode-1.33.25}/src/sourcecode/relevance_scorer.py

@@ -88,6 +88,31 @@ _AUXILIARY_DIR_PATTERNS: list[re.Pattern[str]] = [
     re.compile(r"(?:^|/)stories(?:/|$)"),
 ]
 
+# JVM source roots: path components below these are package namespaces, not
+# functional directories. "com/example/foo" must not match the "example"
+# auxiliary-dir pattern just because "example" is a package name component.
+_JVM_SRC_ROOTS: tuple[str, ...] = (
+    "src/main/java/", "src/test/java/",
+    "src/main/kotlin/", "src/test/kotlin/",
+    "src/main/scala/", "src/test/scala/",
+    "src/integration-test/java/", "src/integrationTest/java/",
+)
+
+
+def _jvm_strip_path(path: str) -> str:
+    """Return the path segment used for auxiliary-dir pattern matching.
+
+    For JVM source trees the package path below the source root is a namespace,
+    not a functional directory hierarchy.  Truncate at the source root so that
+    package components like 'example' in com.example.* don't false-match
+    _AUXILIARY_DIR_PATTERNS entries.
+    """
+    for root in _JVM_SRC_ROOTS:
+        idx = path.find(root)
+        if idx >= 0:
+            return path[: idx + len(root)]
+    return path
+
 # Test file patterns — scored low, excluded from default contract output
 _TEST_FILE_PATTERNS: tuple[str, ...] = (
     "_test.", ".test.", ".spec.", "test_", "conftest", "_spec.",
@@ -221,7 +246,7 @@ class RelevanceScorer:
         return self._is_auxiliary(path.replace("\\", "/"))
 
     def _is_auxiliary(self, norm: str) -> bool:
-        return any(p.search(norm) for p in _AUXILIARY_DIR_PATTERNS)
+        return any(p.search(_jvm_strip_path(norm)) for p in _AUXILIARY_DIR_PATTERNS)
 
     def package_role(self, path: str) -> str:
         """Return the monorepo package role for this path, or empty string."""

{sourcecode-1.33.23 → sourcecode-1.33.25}/src/sourcecode/ris.py

@@ -422,7 +422,9 @@ def get_cold_start_context(repo_root: Path) -> dict:
         if _validation_status == "incomplete_snapshot" and not stale:
             _status_base = "cold_start_incomplete"
 
+        _compact = ris.compact_summary or {}
         result: dict = {
+            "schema_version": "1.1",
             "status": _status_base,
             "repo_id": ris.repo_id,
             "git_head": ris.git_head,
@@ -433,7 +435,10 @@ def get_cold_start_context(repo_root: Path) -> dict:
             "cache_source": "RIS",
             "data_scope": "RIS_BOOTSTRAP",
             "api_surface_complete": _api_complete,
-            "summary": ris.compact_summary,
+            # Backward-compat alias: old consumers read cold-start.project_type directly.
+            # New consumers should use cold-start.summary.project_type.
+            "project_type": _compact.get("project_type"),
+            "summary": _compact,
             "entrypoints": ris.structural_map.get("entrypoints", []),
             "endpoints": endpoints,
             "hotspots": ris.git_context_snapshot.get("hotspots", []),