sourcecode 1.68.0__py3-none-any.whl → 1.69.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
sourcecode/__init__.py CHANGED
@@ -1,3 +1,3 @@
1
1
  """sourcecode — Deterministic codebase context maps for AI coding agents."""
2
2
 
3
- __version__ = "1.68.0"
3
+ __version__ = "1.69.0"
sourcecode/classifier.py CHANGED
@@ -1,5 +1,6 @@
1
1
  from __future__ import annotations
2
2
 
3
+ import re
3
4
  from collections.abc import Iterable, Sequence
4
5
  from dataclasses import replace
5
6
  from typing import Any, Literal
@@ -104,6 +105,21 @@ class TypeClassifier:
104
105
  stack_names = {stack.stack for stack in stacks}
105
106
  framework_names = {framework.name for stack in stacks for framework in stack.frameworks}
106
107
 
108
+ # BUG #4 (JobRunr field test): a framework present only in a small optional
109
+ # integration/adapter submodule must NOT label the whole repo as that
110
+ # framework's app type. JobRunr is a framework-agnostic background-job
111
+ # LIBRARY whose `core` module holds ~85% of the code; Quarkus/Micronaut/Spring
112
+ # appear only in tiny per-framework adapter modules — yet presence-based
113
+ # classification returned project_type="api"+Quarkus. We weight by code
114
+ # locality: if the DOMINANT source module (the one with the most source files)
115
+ # contains no evidence of the app-defining framework, the framework is an
116
+ # optional adapter and the repo is a library. A monolithic Spring app is
117
+ # unaffected — its dominant module *does* use the framework. Drop such
118
+ # localized frameworks from the set that drives the app-type decision below.
119
+ _app_frameworks = framework_names & (_WEB_FRAMEWORKS | _API_FRAMEWORKS)
120
+ _localized = self._localized_adapter_frameworks(file_tree, stacks, _app_frameworks)
121
+ framework_names = framework_names - _localized
122
+
107
123
  if len(stack_names) >= 2 and self._is_fullstack(stacks):
108
124
  return "fullstack"
109
125
 
@@ -129,9 +145,16 @@ class TypeClassifier:
129
145
  if framework_names & _API_FRAMEWORKS:
130
146
  return "api"
131
147
 
148
+ # All app-defining frameworks were localized to optional adapter submodules
149
+ # (multi-module library with per-framework integrations) — report library,
150
+ # never "unknown", when there is clearly source code present.
151
+ if _localized and not (framework_names & (_WEB_FRAMEWORKS | _API_FRAMEWORKS)):
152
+ return "library"
153
+
154
+ # Strong CLI signals: a CLI framework or an explicit cli entry point.
132
155
  if framework_names & _CLI_FRAMEWORKS or any(
133
156
  entry.kind == "cli" for entry in entry_points
134
- ) or any(path.startswith("bin/") for path in flat_paths):
157
+ ):
135
158
  return "cli"
136
159
 
137
160
  if stack_names:
@@ -141,8 +164,105 @@ class TypeClassifier:
141
164
  if single in {"cpp", "dotnet"} and any(entry.kind == "cli" for entry in entry_points):
142
165
  return "cli"
143
166
 
167
+ # BUG #4 (JobRunr field test): a multi-module JVM repo with no app-defining
168
+ # web/API framework is a library/toolkit, not an "unknown" — never let the
169
+ # first command of an audit emit a vacuous classification for a clearly
170
+ # structured codebase (e.g. JobRunr: core + per-framework adapter modules).
171
+ # This is checked BEFORE the weak `bin/`-directory CLI heuristic so a build
172
+ # output / wrapper `bin/` dir does not mislabel a library as a CLI.
173
+ if stack_names & {"java", "kotlin", "scala"} and self._is_multi_module(file_tree):
174
+ return "library"
175
+
176
+ # Weak CLI heuristic: a top-level bin/ directory (only when nothing stronger).
177
+ if any(path.startswith("bin/") for path in flat_paths):
178
+ return "cli"
179
+
144
180
  return "unknown" if stacks else None
145
181
 
182
+ def _is_multi_module(self, file_tree: dict[str, Any]) -> bool:
183
+ """True when the repo has >1 source module (distinct `*/src/...` roots)."""
184
+ _CODE_EXTS = (".java", ".kt", ".kts", ".scala", ".groovy")
185
+ modules = {
186
+ self._module_of(p)
187
+ for p in flatten_file_tree(file_tree)
188
+ if p.endswith(_CODE_EXTS)
189
+ }
190
+ modules.discard("")
191
+ return len(modules) >= 2
192
+
193
+ @staticmethod
194
+ def _module_of(path: str) -> str:
195
+ """Group a source path into its module root.
196
+
197
+ For Maven/Gradle layouts the module is everything before `/src/`
198
+ (e.g. `framework-support/jobrunr-quarkus/src/main/java/...` →
199
+ `framework-support/jobrunr-quarkus`). Otherwise the top-level directory.
200
+ """
201
+ norm = path.replace("\\", "/")
202
+ idx = norm.find("/src/")
203
+ if idx > 0:
204
+ return norm[:idx]
205
+ head, _, tail = norm.partition("/")
206
+ return head if tail else ""
207
+
208
+ _EVIDENCE_PATH_RE = re.compile(r"\(([^()]+)\)\s*$")
209
+
210
+ def _localized_adapter_frameworks(
211
+ self,
212
+ file_tree: dict[str, Any],
213
+ stacks: Sequence[StackDetection],
214
+ candidate_frameworks: set[str],
215
+ ) -> set[str]:
216
+ """Frameworks confined to a minority module while a framework-agnostic
217
+ module dominates the codebase (library + per-framework adapters).
218
+
219
+ Returns the subset of ``candidate_frameworks`` that should NOT drive the
220
+ project-type decision. A framework qualifies only when (a) the repo is
221
+ multi-module, (b) the framework's evidence files are all outside the
222
+ dominant source module, and (c) the framework has located evidence files
223
+ (a manifest-only/root detection applies repo-wide and never localizes).
224
+ """
225
+ if not candidate_frameworks:
226
+ return set()
227
+
228
+ _CODE_EXTS = (".java", ".kt", ".kts", ".scala", ".groovy")
229
+ module_file_counts: dict[str, int] = {}
230
+ for p in flatten_file_tree(file_tree):
231
+ if not p.endswith(_CODE_EXTS):
232
+ continue
233
+ mod = self._module_of(p)
234
+ module_file_counts[mod] = module_file_counts.get(mod, 0) + 1
235
+
236
+ # Need a genuine multi-module repo to reason about locality.
237
+ if len(module_file_counts) < 2:
238
+ return set()
239
+ dominant_module = max(module_file_counts, key=lambda m: module_file_counts[m])
240
+
241
+ # Collect evidence file paths per framework from detected_via.
242
+ evidence: dict[str, set[str]] = {}
243
+ for stack in stacks:
244
+ for fw in stack.frameworks:
245
+ if fw.name not in candidate_frameworks:
246
+ continue
247
+ paths = evidence.setdefault(fw.name, set())
248
+ for ev in fw.detected_via:
249
+ if ev.startswith("manifest:"):
250
+ continue
251
+ m = self._EVIDENCE_PATH_RE.search(ev)
252
+ if m:
253
+ paths.add(m.group(1).strip())
254
+
255
+ localized: set[str] = set()
256
+ for fw_name in candidate_frameworks:
257
+ files = evidence.get(fw_name) or set()
258
+ if not files:
259
+ # No locatable evidence (manifest-only) → applies repo-wide.
260
+ continue
261
+ modules = {self._module_of(f) for f in files}
262
+ if dominant_module not in modules:
263
+ localized.add(fw_name)
264
+ return localized
265
+
146
266
  def _is_fullstack(self, stacks: Sequence[StackDetection]) -> bool:
147
267
  has_web = False
148
268
  has_api = False
@@ -90,6 +90,19 @@ _GRADLE_JAVA_VERSION_RE = re.compile(
90
90
  _GRADLE_JAVA_ENUM_RE = re.compile(
91
91
  r"""(?:sourceCompatibility|targetCompatibility)\s*=\s*JavaVersion\.VERSION_(\d+)"""
92
92
  )
93
+ # BUG #4: a gradle line that genuinely declares a dependency or plugin. Matches a
94
+ # quoted Maven coordinate (group:artifact, requires the colon), a plugins-block
95
+ # `id '...'`, an `apply plugin:`, or a `classpath`/`platform(`/`enforcedPlatform(`
96
+ # declaration. Deliberately does NOT match `it.name.contains('quarkus')`-style
97
+ # subproject filters (a bare quoted name with no colon and no plugin/dep keyword).
98
+ _GRADLE_DEP_LINE_RE = re.compile(
99
+ r"""['"][\w.\-]+:[\w.\-]+""" # maven coordinate "group:artifact..."
100
+ r"""|\bclasspath\b"""
101
+ r"""|\b(?:enforced)?platform\s*\("""
102
+ r"""|\bapply\s+plugin\s*:"""
103
+ r"""|\bid\s*[('"]""",
104
+ re.IGNORECASE,
105
+ )
93
106
 
94
107
 
95
108
  class JavaDetector(AbstractDetector):
@@ -279,7 +292,29 @@ class JavaDetector(AbstractDetector):
279
292
  original = "\n".join(read_text_lines(path))
280
293
  content = original.lower()
281
294
  sb_version = self._extract_gradle_sb_version(original)
282
- return self._detect_jvm_frameworks(content, "build.gradle", sb_version=sb_version)
295
+ # BUG #4 (JobRunr field test): framework tokens must be matched only inside
296
+ # genuine dependency / plugin declarations, never anywhere in the file. A
297
+ # multi-module root build.gradle that EXCLUDES a subproject by name —
298
+ # `configure(subprojects.findAll { !it.name.contains('quarkus') })` — must
299
+ # not be read as "this project uses Quarkus". Restrict the scan to lines
300
+ # carrying a Maven coordinate / plugin id; the exclusion filter has neither.
301
+ dep_text = self._gradle_dependency_text(content)
302
+ return self._detect_jvm_frameworks(dep_text, "build.gradle", sb_version=sb_version)
303
+
304
+ @staticmethod
305
+ def _gradle_dependency_text(content: str) -> str:
306
+ """Keep only gradle lines that declare a dependency or plugin.
307
+
308
+ A real dependency carries a Maven coordinate (`group:artifact`...) and a
309
+ plugin carries an `id`/`plugin`/`classpath` token. Subproject-name filters
310
+ and arbitrary prose are dropped, so a framework substring there cannot
311
+ manufacture a phantom framework detection.
312
+ """
313
+ kept: list[str] = []
314
+ for line in content.splitlines():
315
+ if _GRADLE_DEP_LINE_RE.search(line):
316
+ kept.append(line)
317
+ return "\n".join(kept)
283
318
 
284
319
  def _extract_gradle_sb_version(self, content: str) -> str | None:
285
320
  m = _GRADLE_SB_PLUGIN_RE.search(content)
sourcecode/explain.py CHANGED
@@ -223,7 +223,89 @@ def _build_purpose(
223
223
  if role_anns:
224
224
  desc = f"{role_anns[0]} bean"
225
225
 
226
- return desc or "No stereotype detected — may be a plain class or utility."
226
+ if desc:
227
+ return desc
228
+
229
+ # BUG #4 (JobRunr field test): with no recognized framework annotation, the old
230
+ # fallback returned "No stereotype detected — may be a plain class or utility",
231
+ # which is actively misleading for a central domain class. Libraries and clean/
232
+ # hexagonal architectures model rich roles WITHOUT DI annotations. Infer a
233
+ # low-confidence structural role from signals the tool already computes:
234
+ # in-degree (coupling), lifecycle methods, and naming convention.
235
+ structural = _structural_purpose(class_fqn, raw_nodes, cir)
236
+ if structural:
237
+ return structural
238
+ return "No stereotype detected — may be a plain class or utility."
239
+
240
+
241
+ # Lifecycle method names that signal an orchestrator/component managing state.
242
+ _LIFECYCLE_METHODS = frozenset({
243
+ "start", "stop", "init", "initialize", "shutdown", "close",
244
+ "pause", "resume", "run", "destroy", "open", "restart",
245
+ })
246
+
247
+ # Class-name suffix → inferred role (no annotation required).
248
+ _NAME_ROLE_SUFFIXES: tuple[tuple[str, str], ...] = (
249
+ ("Server", "server/orchestrator"),
250
+ ("Manager", "manager/coordinator"),
251
+ ("Controller", "controller"),
252
+ ("Service", "service"),
253
+ ("Repository", "repository/data access"),
254
+ ("Factory", "factory"),
255
+ ("Builder", "builder"),
256
+ ("Handler", "handler"),
257
+ ("Listener", "listener"),
258
+ ("Provider", "provider"),
259
+ ("Registry", "registry"),
260
+ ("Scheduler", "scheduler"),
261
+ ("Dispatcher", "dispatcher"),
262
+ ("Processor", "processor"),
263
+ ("Filter", "filter"),
264
+ ("Interceptor", "interceptor"),
265
+ )
266
+
267
+
268
+ def _structural_purpose(
269
+ class_fqn: str,
270
+ raw_nodes: list[dict],
271
+ cir: "CanonicalRepositoryIR",
272
+ ) -> str:
273
+ """Infer a low-confidence stereotype from structural signals (no annotations)."""
274
+ try:
275
+ in_degree = len(_build_callers(class_fqn, cir))
276
+ except Exception:
277
+ in_degree = 0
278
+
279
+ try:
280
+ method_names = {m.split("(")[0].lower() for m in _build_public_methods(class_fqn, raw_nodes)}
281
+ except Exception:
282
+ method_names = set()
283
+ lifecycle = sorted(method_names & _LIFECYCLE_METHODS)
284
+
285
+ simple = _simple(class_fqn)
286
+ role = ""
287
+ for suffix, label in _NAME_ROLE_SUFFIXES:
288
+ if simple.endswith(suffix):
289
+ role = label
290
+ break
291
+ if not role and lifecycle:
292
+ role = "orchestrator/lifecycle component"
293
+
294
+ # Require at least one real signal — otherwise stay honestly silent.
295
+ if not role and in_degree < 3:
296
+ return ""
297
+
298
+ head = f"Likely {role} (no DI annotations found)" if role else \
299
+ "Likely a structurally significant class (no DI annotations found)"
300
+ signals: list[str] = []
301
+ if in_degree >= 3:
302
+ signals.append(f"high in-degree ({in_degree})")
303
+ elif in_degree:
304
+ signals.append(f"in-degree {in_degree}")
305
+ if lifecycle:
306
+ signals.append(f"lifecycle methods detected ({'/'.join(lifecycle)})")
307
+ suffix = f" — {', '.join(signals)}" if signals else ""
308
+ return f"{head}{suffix} — inferred from structure, not annotations (low confidence)."
227
309
 
228
310
 
229
311
  def _build_public_methods(class_fqn: str, raw_nodes: list[dict]) -> list[str]:
@@ -0,0 +1,26 @@
1
+ """GENERATED FILE — do not edit by hand.
2
+
3
+ Allowlist of sun.* / com.sun.* packages exported UNCONDITIONALLY by the
4
+ JDK (no `--add-exports` / `--add-opens` required on classpath or module
5
+ path). Consumed by migrate-check MIG-011 to suppress false positives.
6
+
7
+ Regenerate with: python scripts/generate_jdk_exports.py > src/sourcecode/jdk_exports.py
8
+ Generated from: java version "21.0.10" 2026-01-20 LTS
9
+ """
10
+ from __future__ import annotations
11
+
12
+
13
+ JDK_UNCONDITIONAL_EXPORTS: frozenset[str] = frozenset(
14
+ {
15
+ "com.sun.java.accessibility.util",
16
+ "com.sun.management",
17
+ "com.sun.net.httpserver",
18
+ "com.sun.net.httpserver.spi",
19
+ "com.sun.nio.sctp",
20
+ "com.sun.security.auth",
21
+ "com.sun.security.auth.callback",
22
+ "com.sun.security.auth.login",
23
+ "com.sun.security.auth.module",
24
+ "com.sun.security.jgss",
25
+ }
26
+ )
@@ -23,6 +23,7 @@ from pathlib import Path
23
23
  from typing import Optional, TYPE_CHECKING
24
24
 
25
25
  from sourcecode.path_filters import is_test_or_fixture_path
26
+ from sourcecode.jdk_exports import JDK_UNCONDITIONAL_EXPORTS
26
27
 
27
28
  if TYPE_CHECKING:
28
29
  from sourcecode.hibernate_strat import HibernateStratification
@@ -381,15 +382,22 @@ _JAVA_9_RULES: list[_Rule] = [
381
382
  severity="high",
382
383
  title="JDK internal API imports (sun.* / com.sun.net.*) — strong encapsulation since Java 9",
383
384
  explanation=(
384
- "Imports from sun.* and com.sun.net.* reference JDK-internal APIs that are "
385
- "not part of the public specification. Since Java 9 (JPMS), these packages are "
386
- "strongly encapsulated and require --add-exports / --add-opens JVM flags, "
387
- "which are cumbersome and may be removed in future Java releases."
385
+ "Imports from sun.* and com.sun.* (tools/jdi/source internals) reference "
386
+ "JDK-internal APIs that are not part of the public specification. Since Java 9 "
387
+ "(JPMS), these packages are strongly encapsulated and require --add-exports / "
388
+ "--add-opens JVM flags, which are cumbersome and may be removed in future Java "
389
+ "releases. Packages the JDK exports UNCONDITIONALLY (e.g. com.sun.net.httpserver "
390
+ "in jdk.httpserver, com.sun.management in jdk.management) are NOT flagged: they "
391
+ "need no JVM flags on any classpath or module path."
388
392
  ),
389
393
  fix_hint=(
390
394
  "Replace internal API usage with public equivalents. "
391
- "For com.sun.net.httpserver, migrate to java.net.http.HttpServer or a framework. "
392
- "Add '--add-exports java.base/sun.misc=ALL-UNNAMED' only as a last resort."
395
+ "For sun.misc.Unsafe migrate to java.lang.invoke.VarHandle; for com.sun.tools.* "
396
+ "use the public javax.tools / java.compiler API. "
397
+ "Add '--add-exports java.base/sun.misc=ALL-UNNAMED' only as a last resort. "
398
+ "Note: unconditionally-exported packages (com.sun.net.httpserver, "
399
+ "com.sun.management, com.sun.security.auth, ...) are auto-excluded — they are "
400
+ "public, stable, and require no migration."
393
401
  ),
394
402
  migration_target="java_9_plus",
395
403
  openrewrite_recipe=None,
@@ -585,6 +593,45 @@ def _is_no_migrate_javax(fqn: str) -> bool:
585
593
  return any(fqn.startswith(p) for p in _JAKARTA_NO_MIGRATE_PREFIXES)
586
594
 
587
595
 
596
+ # BUG #1 (JobRunr field test): MIG-011 flags `sun.*` / `com.sun.*` imports as
597
+ # strongly-encapsulated JDK internals on a pure PREFIX heuristic. That is wrong
598
+ # for packages the JDK exports UNCONDITIONALLY (no `to` clause) — e.g.
599
+ # `com.sun.net.httpserver` (module jdk.httpserver, public since Java 6, the basis
600
+ # of JEP 408) or `com.sun.management` (jdk.management, JMX/diagnostics). These
601
+ # need NO `--add-exports` / `--add-opens` on any classpath or module path, so they
602
+ # must not be flagged `high` / `manual_migration`. The allowlist is generated from
603
+ # the running JDK by scripts/generate_jdk_exports.py (see sourcecode/jdk_exports.py),
604
+ # never hand-maintained. Genuinely-internal packages (sun.misc.Unsafe, com.sun.tools.*,
605
+ # com.sun.jdi.*, com.sun.source.*) are NOT in the allowlist and keep `high` severity.
606
+
607
+
608
+ def _import_package(fqn: str) -> str:
609
+ """Extract the Java package of an import FQN.
610
+
611
+ Packages are lowercase by convention and types are Capitalized, so the
612
+ package is the maximal prefix of non-type segments. Handles wildcard
613
+ (`a.b.*`) and `static` imports. Conservative: an unrecognized shape yields
614
+ the leading lowercase run, never a broader prefix — so a sub-package like
615
+ `com.sun.management.internal` is never confused with `com.sun.management`.
616
+ """
617
+ fqn = fqn.strip().rstrip(";").strip()
618
+ if fqn.startswith("static "):
619
+ fqn = fqn[len("static "):].strip()
620
+ if fqn.endswith(".*"):
621
+ return fqn[:-2]
622
+ pkg_parts: list[str] = []
623
+ for seg in fqn.split("."):
624
+ if seg[:1].isupper(): # first type segment — package ends here
625
+ break
626
+ pkg_parts.append(seg)
627
+ return ".".join(pkg_parts)
628
+
629
+
630
+ def _is_jdk_unconditional_export(fqn: str) -> bool:
631
+ """True if an import targets a package the JDK exports unconditionally."""
632
+ return _import_package(fqn) in JDK_UNCONDITIONAL_EXPORTS
633
+
634
+
588
635
  # BUG #8: autogenerated source markers — path fragments and the JSR-250 marker.
589
636
  _GENERATED_PATH_FRAGMENTS: tuple[str, ...] = (
590
637
  "/generated-sources/", "/generated/", "/target/generated",
@@ -1698,6 +1745,16 @@ def _scan_file(
1698
1745
  # javax.annotation.processing.*, ...). These do NOT migrate to jakarta.
1699
1746
  if matches and rule.migration_target == "jakarta":
1700
1747
  matches = [m for m in matches if not _is_no_migrate_javax(m.group(1).strip())]
1748
+ # BUG #1: MIG-011 prefix heuristic must not flag sun.*/com.sun.* packages
1749
+ # the JDK exports unconditionally (no --add-exports/--add-opens needed).
1750
+ # Drop those imports; if none remain, the file produces no MIG-011 finding
1751
+ # (so it never inflates blocking_count / effort). Genuinely-internal
1752
+ # packages are absent from the allowlist and survive as `high`.
1753
+ if matches and rule.id == "MIG-011":
1754
+ matches = [
1755
+ m for m in matches
1756
+ if not _is_jdk_unconditional_export(m.group(1).strip())
1757
+ ]
1701
1758
  if matches:
1702
1759
  import_first_line = source[: matches[0].start()].count("\n") + 1
1703
1760
  matched_imports = [m.group(1) for m in matches]
@@ -3366,9 +3366,20 @@ def build_repo_ir(
3366
3366
  and not _INHERIT_PRESCAN_RE.search(source):
3367
3367
  pkg_m = _PKG_RE.search(source)
3368
3368
  _pkg = pkg_m.group(1) if pkg_m else ""
3369
- # Minimal class-name symbols for same-package map (no methods/fields)
3369
+ # Minimal class-name symbols for same-package map (no methods/fields).
3370
+ # BUG #2 (JobRunr field test): this fast-path regex previously ran over RAW
3371
+ # source, so prose inside Javadoc/comments and string literals (e.g.
3372
+ # "This class provides the entry point", "...the interface is Serializable
3373
+ # ...instead.") was tokenized into phantom type symbols like
3374
+ # `org.jobrunr.configuration.provides`. Those leaked into the symbol graph
3375
+ # and every consumer of it (modernize statically_unreferenced /
3376
+ # framework_dispatched, impact, export --c4). Strip comments AND string
3377
+ # literals before scanning so only real declarations are captured. The name
3378
+ # must also start uppercase ([A-Z]) — Java type convention, matching the
3379
+ # precision of the full _CLASS_DECL_RE used on annotated files.
3380
+ _decl_source = _STRING_LITERAL_RE.sub('', _strip_java_comments(source))
3370
3381
  _min_syms: list[SymbolRecord] = []
3371
- for _cm in re.finditer(r'(?:class|interface|enum)\s+(\w+)', source):
3382
+ for _cm in re.finditer(r'\b(?:class|interface|enum)\s+([A-Z]\w*)', _decl_source):
3372
3383
  _cls_name = _cm.group(1)
3373
3384
  _fqn = f"{_pkg}.{_cls_name}" if _pkg else _cls_name
3374
3385
  _min_syms.append(SymbolRecord(
@@ -4023,6 +4034,27 @@ def extract_java_endpoints(root: Path) -> "dict[str, Any]":
4023
4034
  _SERVLET_RE = _re.compile(r"\bextends\s+\w*HttpServlet\b")
4024
4035
  _nonspring: dict[str, int] = {"webscripts": 0, "jax_rs": 0, "servlets": 0}
4025
4036
 
4037
+ # BUG #3 (JobRunr field test): imperative router-DSL routes. Lightweight HTTP
4038
+ # frameworks that deliberately avoid Spring/JAX-RS (JobRunr's own dashboard
4039
+ # handler, Javalin, Spark Java, hand-rolled routers) register routes as method
4040
+ # calls `get("/path", handler)` / `post(...)` instead of annotations, so the
4041
+ # annotation surface above never sees them and `endpoints` returns 0 — a silent
4042
+ # total false negative that also disables the `validation` command downstream.
4043
+ # Detection is by SYNTACTIC SHAPE (no framework knowledge): an HTTP-verb method
4044
+ # name, a first argument that is a string literal looking like a path (starts
4045
+ # with "/", may contain :param / {param}), AND a second argument (the trailing
4046
+ # comma) — the comma is what distinguishes a route registration from a 1-arg
4047
+ # getter such as Map.get("/k"). Matches both bare `get(...)` (static-import /
4048
+ # Spark style) and `app.get(...)` (Javalin style); the lookbehind only rejects
4049
+ # an identifier char so `forget(` is not mistaken for `get(`. Reported at
4050
+ # confidence "medium" — an occasional flagged false positive beats a total
4051
+ # silent false negative.
4052
+ _DSL_ROUTE_RE = _re.compile(
4053
+ r'(?<![A-Za-z0-9_])(get|post|put|delete|patch|head|options)\s*\(\s*'
4054
+ r'"(/[^"\s]*)"\s*,'
4055
+ )
4056
+ _dsl_routes: list[dict] = []
4057
+
4026
4058
  for jf in java_files:
4027
4059
  try:
4028
4060
  source = jf.read_text(encoding="utf-8", errors="replace")
@@ -4051,6 +4083,27 @@ def extract_java_endpoints(root: Path) -> "dict[str, Any]":
4051
4083
  if m:
4052
4084
  extends_map[sym.symbol] = m.group(1)
4053
4085
 
4086
+ # BUG #3: scan for imperative router-DSL route registrations. Strip comments
4087
+ # first so commented-out / example routes don't leak in. The enclosing class
4088
+ # is the first class/interface symbol in the file (route DSLs live in one
4089
+ # handler class); fall back to the file stem when none was extracted.
4090
+ _dsl_src = _strip_java_comments(source)
4091
+ _dsl_matches = list(_DSL_ROUTE_RE.finditer(_dsl_src))
4092
+ if _dsl_matches:
4093
+ _cls_fqn = next(
4094
+ (s.symbol for s in symbols if s.type in ("class", "interface")), None
4095
+ )
4096
+ _cls_simple = (
4097
+ _cls_fqn.split(".")[-1] if _cls_fqn else Path(rel).stem
4098
+ )
4099
+ for _m in _dsl_matches:
4100
+ _dsl_routes.append({
4101
+ "method": _m.group(1).upper(),
4102
+ "path": _m.group(2),
4103
+ "controller": _cls_simple,
4104
+ "effective_class": _cls_fqn or _cls_simple,
4105
+ })
4106
+
4054
4107
  routes = _build_route_surface(
4055
4108
  all_symbols, route_diffs=None, extends_map=extends_map,
4056
4109
  custom_security=_custom_sec_tuple,
@@ -4226,6 +4279,32 @@ def extract_java_endpoints(root: Path) -> "dict[str, Any]":
4226
4279
  if e.get("security", {}).get("policy") == "none_detected"
4227
4280
  )
4228
4281
 
4282
+ # BUG #3: merge imperative router-DSL endpoints. Dedup against the annotation
4283
+ # surface (method+path) in case a handler is both annotated and DSL-registered.
4284
+ # These carry confidence "medium" (no unambiguous annotation) and explicit
4285
+ # provenance so a consumer can tell shape-detected routes from annotated ones.
4286
+ if _dsl_routes:
4287
+ _seen_mp = {(e.get("method"), e.get("path")) for e in endpoints}
4288
+ for _r in _dsl_routes:
4289
+ _key = (_r["method"], _r["path"])
4290
+ if _key in _seen_mp:
4291
+ continue
4292
+ _seen_mp.add(_key)
4293
+ endpoints.append({
4294
+ "method": _r["method"],
4295
+ "path": _r["path"],
4296
+ "controller": _r["controller"],
4297
+ "handler": _r["controller"],
4298
+ "return_type": "unknown",
4299
+ "security": {"policy": "none_detected"},
4300
+ "confidence": "medium",
4301
+ "source": "router_dsl",
4302
+ })
4303
+ no_security_signal = sum(
4304
+ 1 for e in endpoints
4305
+ if e.get("security", {}).get("policy") == "none_detected"
4306
+ )
4307
+
4229
4308
  # Append spec-recovered endpoints AFTER the security-model heuristics (which
4230
4309
  # are about annotation/filter/XML coverage of scanned source) so spec-sourced
4231
4310
  # entries don't skew those signals. They carry their own source provenance.
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: sourcecode
3
- Version: 1.68.0
3
+ Version: 1.69.0
4
4
  Summary: Persistent structural context and ultra-fast repeated analysis for AI coding agents
5
5
  License-File: LICENSE
6
6
  Keywords: agents,ai,codebase,context,developer-tools,llm
@@ -1,4 +1,4 @@
1
- sourcecode/__init__.py,sha256=gp3XLYiUZaatdehs90NAezSd-3vDxmo0g-OIVUqEBmg,103
1
+ sourcecode/__init__.py,sha256=mxqKQ416I6lEnKsI6lwjPPRQY5VpPWrIc3-FB4IRfK8,103
2
2
  sourcecode/adaptive_scanner.py,sha256=XffluXKzJUXrMtjEiAOnSNPZnztdIcts17T9ouHeID0,10521
3
3
  sourcecode/architecture_analyzer.py,sha256=liCwQmLgb5vplohy8arjYxs_HOIv5C9MjLh_gY6bc5Q,44115
4
4
  sourcecode/architecture_summary.py,sha256=z34_6v7cSwy98cof2UVciGho7SCrZ93tiqMmq5WNzRQ,20405
@@ -6,7 +6,7 @@ sourcecode/ast_extractor.py,sha256=sa6CmLpn-k5G3_Hzxn8hAlZ5-TS-EVzXDD0Gvxd2jzs,5
6
6
  sourcecode/cache.py,sha256=1V3vsaODAa2UBJAC0xpvxpmRdriCezQx5Q8JCcfgziE,31892
7
7
  sourcecode/canonical_ir.py,sha256=DEwucOPJguLsVtg5cV8mWXNi112l5jmBhv73KGGebVk,24849
8
8
  sourcecode/cir_graphs.py,sha256=9G0HHj1kw2325IDyzo2OpX73BNswEckecf4MZUXB4JM,12078
9
- sourcecode/classifier.py,sha256=hKzg-nQ47htqqIUzSGvYxv15cXrA3KgICTwJmdqal0o,8095
9
+ sourcecode/classifier.py,sha256=YTTCoRdcLEFRVcql9Ow1dE7eYQj0jq2rgx32bRDnb1k,13852
10
10
  sourcecode/cli.py,sha256=VqAiGRauTO-s8yBJh2JMldha3VKx0jTXC0tOvRpWQkI,283297
11
11
  sourcecode/code_notes_analyzer.py,sha256=EJemNCNc9Dn-1RZYu-aNbK0ELzmsyC4s6FdHi3XyNEI,9392
12
12
  sourcecode/confidence_analyzer.py,sha256=_jckZSxksV-OU38vbkxfVNBnWCtlCq8Vwfg23x1uspA,19054
@@ -20,7 +20,7 @@ sourcecode/doc_analyzer.py,sha256=05bjTUbDbmnbajD_cgRnACzS8T7xxBKVX4CjkJlhZg8,24
20
20
  sourcecode/entrypoint_classifier.py,sha256=jhTYlyqDJH2AtdEcLVaRU3lYRTJuF8DkxVzl4-W3zWE,5322
21
21
  sourcecode/env_analyzer.py,sha256=aNTyYgQk5noJDfJU6FmasmESOHfiomyJw5EvZqjy6qc,22213
22
22
  sourcecode/error_schema.py,sha256=uwosfNaSujtYm11_732Hu92z5ITV040fQDaIyefSvR4,1683
23
- sourcecode/explain.py,sha256=dVG35YBlpRmbtOXSmspEhoIwDMVApPmLISBy3iigUSc,16913
23
+ sourcecode/explain.py,sha256=GbcruAyzlmseV3o2rjeyxGQxToCfSYHJjKG3N05NVbQ,19897
24
24
  sourcecode/file_chunker.py,sha256=3vkM3mDQ5eE_yTPvUgjyjpGFBIjkW6_mrBmIbrylnA8,16444
25
25
  sourcecode/file_classifier.py,sha256=A0fEABqtfVu1MfoaxnPAvGpZgneGgVXlJDhT74NYXxE,15314
26
26
  sourcecode/flow_analyzer.py,sha256=dSiuY4w49k29jW_EPXUOND9B5uVbuCA7kjnuHi-pIWA,28781
@@ -30,10 +30,11 @@ sourcecode/git_analyzer.py,sha256=JStxTQXNjBWi_wLdwhsZs9mT-v50cSJIz4Agzn6Kh9I,13
30
30
  sourcecode/graph_analyzer.py,sha256=DHR8fY69oU_Pi4SYaWboX6EoEFrctQKB9dsjpqwGMzw,62403
31
31
  sourcecode/hibernate_strat.py,sha256=ZyI0sikxh7IemU8uqn1Z73z-tHeFeCJuYhJnyQL8pzU,57765
32
32
  sourcecode/integration_detector.py,sha256=ygZwXHJXJ4QM6oiqKb5S6WIfElByDYdz9moyuFlZpYI,12997
33
+ sourcecode/jdk_exports.py,sha256=fCrlwNAXUT9gge_joq6kMnY3zJxYB2pxqy-0w3o3MJI,874
33
34
  sourcecode/license.py,sha256=wckiLuiwaE35KMCStUf1gYzleJuFe6qsSyxUQJvit3s,23500
34
35
  sourcecode/mcp_nudge.py,sha256=5ELU_ixzh6uA83NXLOZT8h00OhL53okfQdji3jyKOjg,2917
35
36
  sourcecode/metrics_analyzer.py,sha256=m0ENgtqKeBL17kUIK3fmGkgo7UfXBNHxCMj0H_Y5K7c,22750
36
- sourcecode/migrate_check.py,sha256=Z0AZhWuZWfGuG3dsQugJn92FRpId5-SFAoyMc1RBZHk,91908
37
+ sourcecode/migrate_check.py,sha256=QNyFmrEqYUSqmi0iIQ1P6oI63g8prOluSvYA5fUJ7Ow,95040
37
38
  sourcecode/openapi_surface.py,sha256=BTt0K-woZbkbWTN77IkqeBm_Okag9owR0848fmot8sk,16207
38
39
  sourcecode/output_budget.py,sha256=Js9yUlfQtPhqBl9R6wn_9UHVjjJc3GtLcqyfjf5t50Q,9869
39
40
  sourcecode/path_filters.py,sha256=VnaD9jxZVNzluackNSTCdIddwzAqIseuSCs_A-gpCDM,6898
@@ -46,7 +47,7 @@ sourcecode/redactor.py,sha256=SB4hwIvg8h-hvcqKcDWaZvA-aSyn-at-BIRwa0tUv5E,3227
46
47
  sourcecode/relevance_scorer.py,sha256=0AgEt4KrV73nioMqBgjhGjtY7L2C7L7cSyKtj3IKcrw,9408
47
48
  sourcecode/rename_refactor.py,sha256=h6dNFlB9aZ_3q6heeHBkgXQeXaT03nvPSsYH6P8qxFg,12965
48
49
  sourcecode/repo_classifier.py,sha256=FG1vaWKdWXsWdl-S8hjVMiTqcwgaRXkDyvK4rPcOGtQ,22681
49
- sourcecode/repository_ir.py,sha256=Blo4rTCkY7EppwunxqlvI-1b9wU3EPLuHlZik3Ksz7c,224908
50
+ sourcecode/repository_ir.py,sha256=tP9L8v9K2xg-xb93N1aELJP-pHHxww8537GZ7Sf0_94,229324
50
51
  sourcecode/ris.py,sha256=RcqLVwC-doFcKKViYDkCjZLBqf_wzLES7-F6vHEeWzE,20419
51
52
  sourcecode/runtime_classifier.py,sha256=uTAD6BDCiBLUZEDRfqk718kM4RTT_vAbfkcOI2_Xx58,18432
52
53
  sourcecode/scanner.py,sha256=WdOQ78mMzjR1NjmKTlbxdgwinnCTfAhxCVLBEFQiFHU,8899
@@ -75,7 +76,7 @@ sourcecode/detectors/elixir.py,sha256=jCpvt5Yi6jvplc80ovRtWh17q-11ZGo9qX7o8b57TJ
75
76
  sourcecode/detectors/go.py,sha256=2r66uRQfeTWsqxr4HDhT6vExZErby0t46QXLHVBRv9w,2782
76
77
  sourcecode/detectors/heuristic.py,sha256=7cRxrip4yIaggYzZJB6ef8yHKh-gHgiH_pXMFcjlyFU,3723
77
78
  sourcecode/detectors/hybrid.py,sha256=IGFRUVsAZ1ooRlFdznCeJAV6vy1yVDx-VyghvLtddXc,9101
78
- sourcecode/detectors/java.py,sha256=HZcK94A3FzhTMXveicKSSDJBvRbgeqnflEP_mPgaOpo,30438
79
+ sourcecode/detectors/java.py,sha256=V9Eb2EMqEdonIhoegvFN2-z95R4Nu9q2y9G54baWShc,32298
79
80
  sourcecode/detectors/jvm_ext.py,sha256=EgHJ5W8EE-ZTN9V607mVzohyKgZE8Mc2jCi-DF8RAZU,2616
80
81
  sourcecode/detectors/nodejs.py,sha256=Hg3Gmr7yIMJFiLoDwOTk2wtu00wxIs6kZf-oQujTFUA,13187
81
82
  sourcecode/detectors/parsers.py,sha256=ugPg8yNUf0Ai1gA7Fnn6wAkYGFjTxRodSP3IeViYJJ4,2290
@@ -103,8 +104,8 @@ sourcecode/telemetry/consent.py,sha256=H5z2Wu63pZqbaKucRPoJQJ0zCo4cke9ZlBrJC-MaW
103
104
  sourcecode/telemetry/events.py,sha256=LtzYfaX9Ilckj5PTvAcTpDa9mLqDsYPDUiDkRa58piY,2580
104
105
  sourcecode/telemetry/filters.py,sha256=NHa5T-6DaZduQPFuC34jOqHWQgSizM-Ygq8aZ4j19ng,5834
105
106
  sourcecode/telemetry/transport.py,sha256=4gGHsq0WeY9VywEZXA3vUxykfiYnw9uuqfjAAec7F8o,1681
106
- sourcecode-1.68.0.dist-info/METADATA,sha256=_4nRHRSnFFBpmv9OJqSFf30pGcWo7XpOC0gyAJgoT4c,47341
107
- sourcecode-1.68.0.dist-info/WHEEL,sha256=QccIxa26bgl1E6uMy58deGWi-0aeIkkangHcxk2kWfw,87
108
- sourcecode-1.68.0.dist-info/entry_points.txt,sha256=ex3F9rmbXeyDIoFQHtkEqTsKSaJow8F0LrVu8XfIktQ,57
109
- sourcecode-1.68.0.dist-info/licenses/LICENSE,sha256=7DdHrU9Z_3e7dSvq4ISijZNjnuHo5NIHNiHDouMQ9JU,10491
110
- sourcecode-1.68.0.dist-info/RECORD,,
107
+ sourcecode-1.69.0.dist-info/METADATA,sha256=s7Ko18CIs4SrMOMsNrqCkc-nFZLcxTnC0M_r0zxXycU,47341
108
+ sourcecode-1.69.0.dist-info/WHEEL,sha256=QccIxa26bgl1E6uMy58deGWi-0aeIkkangHcxk2kWfw,87
109
+ sourcecode-1.69.0.dist-info/entry_points.txt,sha256=ex3F9rmbXeyDIoFQHtkEqTsKSaJow8F0LrVu8XfIktQ,57
110
+ sourcecode-1.69.0.dist-info/licenses/LICENSE,sha256=7DdHrU9Z_3e7dSvq4ISijZNjnuHo5NIHNiHDouMQ9JU,10491
111
+ sourcecode-1.69.0.dist-info/RECORD,,