cg-code-graph 0.10.1__tar.gz → 0.11.0__tar.gz

{cg_code_graph-0.10.1/cg_code_graph.egg-info → cg_code_graph-0.11.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cg-code-graph
-Version: 0.10.1
+Version: 0.11.0
 Summary: cg: a code graph of routes, calls, data access and platform boundaries for agents and humans (CLI + MCP server)
 Author: cyberchronos00
 License-Expression: MIT
@@ -17,7 +17,7 @@ Requires-Dist: tree-sitter-rust>=0.23
 Requires-Dist: tree-sitter-c>=0.23
 Requires-Dist: tree-sitter-cpp>=0.23
 Requires-Dist: tree-sitter-kotlin>=1.0
-Requires-Dist: tree-sitter-swift>=0.6
+Requires-Dist: tree-sitter-swift<0.7.4,>=0.6
 Provides-Extra: dev
 Requires-Dist: pytest>=8; extra == "dev"
 Dynamic: license-file
@@ -93,10 +93,11 @@ other tools; see [Prerequisites per language](#prerequisites-per-language).
 options and updates):
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
-#   or: uv tool install git+https://github.com/cyberchronos00/code-graph
-#   or: pipx install git+https://github.com/cyberchronos00/code-graph
-#   or from PyPI: pip install cg-code-graph   (uv tool install cg-code-graph / pipx install cg-code-graph)
+pipx install cg-code-graph    # or: uv tool install cg-code-graph   (PyPI; the commands are cg and cg-mcp)
+#   or the script (also installs exact-mode tools with --with c,rust,...):
+#     curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
+#   or the latest commit: uv tool install git+https://github.com/cyberchronos00/code-graph
+#                     or: pipx install git+https://github.com/cyberchronos00/code-graph
 cg doctor                     # what indexes exact / heuristic on this machine, and what to install for the rest
 git clone https://github.com/cyberchronos00/code-graph.git && cd code-graph    # the sample apps used below
 ```

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/README.md

@@ -69,10 +69,11 @@ other tools; see [Prerequisites per language](#prerequisites-per-language).
 options and updates):
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
-#   or: uv tool install git+https://github.com/cyberchronos00/code-graph
-#   or: pipx install git+https://github.com/cyberchronos00/code-graph
-#   or from PyPI: pip install cg-code-graph   (uv tool install cg-code-graph / pipx install cg-code-graph)
+pipx install cg-code-graph    # or: uv tool install cg-code-graph   (PyPI; the commands are cg and cg-mcp)
+#   or the script (also installs exact-mode tools with --with c,rust,...):
+#     curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
+#   or the latest commit: uv tool install git+https://github.com/cyberchronos00/code-graph
+#                     or: pipx install git+https://github.com/cyberchronos00/code-graph
 cg doctor                     # what indexes exact / heuristic on this machine, and what to install for the rest
 git clone https://github.com/cyberchronos00/code-graph.git && cd code-graph    # the sample apps used below
 ```

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0/cg_code_graph.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cg-code-graph
-Version: 0.10.1
+Version: 0.11.0
 Summary: cg: a code graph of routes, calls, data access and platform boundaries for agents and humans (CLI + MCP server)
 Author: cyberchronos00
 License-Expression: MIT
@@ -17,7 +17,7 @@ Requires-Dist: tree-sitter-rust>=0.23
 Requires-Dist: tree-sitter-c>=0.23
 Requires-Dist: tree-sitter-cpp>=0.23
 Requires-Dist: tree-sitter-kotlin>=1.0
-Requires-Dist: tree-sitter-swift>=0.6
+Requires-Dist: tree-sitter-swift<0.7.4,>=0.6
 Provides-Extra: dev
 Requires-Dist: pytest>=8; extra == "dev"
 Dynamic: license-file
@@ -93,10 +93,11 @@ other tools; see [Prerequisites per language](#prerequisites-per-language).
 options and updates):
 
 ```bash
-curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
-#   or: uv tool install git+https://github.com/cyberchronos00/code-graph
-#   or: pipx install git+https://github.com/cyberchronos00/code-graph
-#   or from PyPI: pip install cg-code-graph   (uv tool install cg-code-graph / pipx install cg-code-graph)
+pipx install cg-code-graph    # or: uv tool install cg-code-graph   (PyPI; the commands are cg and cg-mcp)
+#   or the script (also installs exact-mode tools with --with c,rust,...):
+#     curl -fsSL https://raw.githubusercontent.com/cyberchronos00/code-graph/main/install.sh | sh
+#   or the latest commit: uv tool install git+https://github.com/cyberchronos00/code-graph
+#                     or: pipx install git+https://github.com/cyberchronos00/code-graph
 cg doctor                     # what indexes exact / heuristic on this machine, and what to install for the rest
 git clone https://github.com/cyberchronos00/code-graph.git && cd code-graph    # the sample apps used below
 ```

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/cg_code_graph.egg-info/SOURCES.txt

@@ -34,8 +34,11 @@ codegraph/query.py
 codegraph/realtime.py
 codegraph/roundtrip.py
 codegraph/routes.py
+codegraph/rpc.py
+codegraph/sockets.py
 codegraph/starters.py
 codegraph/tests_index.py
+codegraph/udp_apps.py
 codegraph/xcode.py
 codegraph/core/__init__.py
 codegraph/core/cache.py
@@ -237,7 +240,9 @@ tests/test_python_web_routes.py
 tests/test_python_web_routes_more.py
 tests/test_roundtrip.py
 tests/test_routes_search_explain.py
+tests/test_rpc.py
 tests/test_scip_health.py
+tests/test_sockets.py
 tests/test_starters_budget.py
 tests/test_swift.py
 tests/test_swift_binding.py

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/cg_code_graph.egg-info/requires.txt

@@ -6,7 +6,7 @@ tree-sitter-rust>=0.23
 tree-sitter-c>=0.23
 tree-sitter-cpp>=0.23
 tree-sitter-kotlin>=1.0
-tree-sitter-swift>=0.6
+tree-sitter-swift<0.7.4,>=0.6
 
 [dev]
 pytest>=8

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/__init__.py

@@ -1,2 +1,2 @@
 """code-graph: deterministic code knowledge graph (beta)."""
-__version__ = "0.10.1"
+__version__ = "0.11.0"

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/cli.py

@@ -18,7 +18,7 @@
   python -m codegraph.cli bridges [PATTERN] --db ... [--protocol capacitor] [--unmatched]   (web / native bridge calls)
   python -m codegraph.cli external [PATTERN] --db ... [--protocol P] [--source S] [--tls-off]   (databases, caches, brokers, mail ... and who reaches them)
   python -m codegraph.cli tools [PATTERN] --db ... [--framework F] [--agent A] [--unmatched]   (LLM / MCP tools: handler, tables, agents, checks)
-  python -m codegraph.cli protocols [PATTERN] --db ... [--protocol P] [--side send|receive] [--unmatched]   (every protocol endpoint: senders, receivers, checks)
+  python -m codegraph.cli protocols [PATTERN] --db ... [--protocol P] [--side send|receive] [--unmatched] [--listeners]   (every protocol endpoint: senders, receivers, checks; listening sockets)
   python -m codegraph.cli platforms [summary|divergence] --db ... [--target ios]   (platform-specific code, gaps between variants)
   reaches / impact / downstream / path / routes / search take --platform TARGET: only code built for that target
 spec forms: table.column | connection:<name> (glob *) | env:<KEY*> | config:<a.b> | Class.method or Class::method (either
@@ -154,6 +154,8 @@ def main(argv=None):
     p.add_argument("--side", choices=["send", "receive"]); p.add_argument("--max-items", type=int, default=60)
     p.add_argument("--unmatched", action="store_true", help="only endpoints with a check (no_receiver, no_sender, ambiguous, "
                                                             "schema_mismatch, unguarded) or an external peer")
+    p.add_argument("--listeners", action="store_true", help="every listening TCP / UDP socket with its bind address, exposure "
+                                                            "(all interfaces / loopback / specific) and handler")
     p = sub.add_parser("external", help="external systems (databases, caches, brokers, mail, directories, file transfer, object "
                                          "stores, third-party HTTP hosts): users, entry points, address and credential sources")
     p.add_argument("pattern", nargs="?", help="external id, substring or glob (`external:postgres:*`, `redis`)")
@@ -429,7 +431,8 @@ def main(argv=None):
         return
     if a.cmd == "protocols":
         from .protocols.view import protocols, render_protocols
-        res = protocols(st, a.pattern, protocol=a.protocol, side=a.side, unmatched=a.unmatched, max_items=max(a.max_items, 200) if a.json else a.max_items)
+        res = protocols(st, a.pattern, protocol=a.protocol, side=a.side, unmatched=a.unmatched, max_items=max(a.max_items, 200) if a.json else a.max_items,
+                        listeners=a.listeners)
         print(json.dumps(res, indent=1, default=str) if a.json else render_protocols(res, max_items=a.max_items))
         return
     if a.cmd == "tests":

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/coverage.py

@@ -6,6 +6,7 @@ its normal search and file reading."""
 from __future__ import annotations
 
 import os
+import re
 from collections import Counter
 
 from . import presets
@@ -188,12 +189,27 @@ def scan(root: str | Path) -> Counter:
     return scan_tree(root).counts
 
 
+_TOOL_FILE = re.compile(r"(?:^|/)(?:[^/]*\.config\.[cm]?[jt]sx?|\.[^/]*rc\.[cm]?js|[Gg]runtfile\.js|[Gg]ulpfile\.[cm]?js)$")
+_TEST_NAME = re.compile(r"(?:\.(?:test|spec)\.[cm]?[jt]sx?$|(?:^|/)__tests__/|\.d\.[cm]?ts$)")    # + declarations
+
+
 def file_completeness(discovered: list[str], report: dict) -> dict:
     """Bucket every discovered file of one language from the plugin's per-file report
     ({seen, parse_failed, skipped_oversize, unmapped, excluded}). A discovered file the plugin never looked at
     (its own skip list) is `excluded`."""
     sets = {b: set(report.get(b) or ()) for b in BUCKETS}
     seen = set(report.get("seen") or ())
+    roots = report.get("roots")
+    if roots is not None:
+        # a plugin that reads only some directories (TypeScript: source dirs, bin / script files, test trees): a file
+        # outside all of them was never read, so it is unmapped (not indexed), not a deliberate exclusion. Tool
+        # configuration (`vite.config.ts`, `.eslintrc.js`) and test files by name stay excluded (#106)
+        dirs = tuple(r for r in roots if r.endswith("/"))
+        files = {r for r in roots if not r.endswith("/")}
+        for f in discovered:
+            if f in seen or f in files or f.startswith(dirs) or _TOOL_FILE.search(f) or _TEST_NAME.search(f):
+                continue
+            sets["unmapped"].add(f)
     out = {b: [] for b in BUCKETS}
     indexed = 0
     for f in sorted(discovered):
@@ -296,6 +312,9 @@ def compute(root: str | Path, plugins: dict, scip_imported: bool = False, report
             e["files_complete"] = missing_files(e) == 0
             if not e["files_complete"] and lang == "python" and e.get("unmapped"):
                 e["hint"] = _python_unmapped_hint(st)
+            elif not e["files_complete"] and lang == "typescript" and e.get("unmapped"):
+                e["hint"] = ("files outside the source dirs and test trees were not read: list the directories that hold "
+                             "source in .cg.yaml `include` (or a tsconfig.json `include`) and re-index")
         se = (rep or {}).get("syntax_errors")
         if isinstance(se, list) and se and status not in ("skipped", "not_indexed"):
             # files parsed with syntax errors (#73): error spans and the declarations lost there

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/indexer.py

@@ -234,6 +234,18 @@ def index_project(root: str | Path, db_path: str | Path, name: str | None = None
     from .process_runs import apply as apply_process_runs
     if (pr := apply_process_runs(project, builder)):
         stats["process_runs"] = pr
+    # raw TCP / UDP sockets paired by port (#39): endpoint:tcp:<port> / endpoint:udp:<port>
+    from .sockets import apply as apply_sockets
+    t_so = time.time()
+    if (so := apply_sockets(project, builder)):
+        so["seconds"] = round(time.time() - t_so, 2)
+        stats["sockets"] = so
+    # RPC contracts (#33): gRPC services of the .proto files, their implementations and stub calls
+    from .rpc import apply as apply_rpc
+    t_rpc = time.time()
+    if (rp := apply_rpc(project, builder)):
+        rp["seconds"] = round(time.time() - t_rpc, 2)
+        stats["rpc"] = rp
     # test code (tests/, *.spec.ts ...) never feeds the application graph: its edges become TEST_* kinds
     from .tests_index import isolate_tests
     stats["tests"] = isolate_tests(builder)

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/mcp_server.py

@@ -698,20 +698,21 @@ def bridges(pattern: str | None = None, protocol: str | None = None, unmatched:
 
 @tool
 def protocol_links(pattern: str | None = None, protocol: str | None = None, side: str | None = None,
-                   unmatched: bool = False) -> str:
+                   unmatched: bool = False, listeners: bool = False) -> str:
     """Every protocol endpoint in one view (#31 model): HTTP client endpoints and routes (http / ws / graphql), Pusher
     broadcast channels and subscriptions, NestJS microservice messages (nest-rpc / nest-event / nest-ws / grpc), job
     queues (bull, laravel-queue, celery), application events (laravel-event, nest-event-emitter, django-signal), web /
     native bridges and desktop IPC (capacitor, react-native, flutter, electron-ipc, tauri ...) and the generic
     endpoint:<protocol>:<name> nodes (code -SENDS_TO-> endpoint -RECEIVED_BY-> handler, MATCHES_ENDPOINT for
-    wildcard / template matches; MQTT, NATS, AMQP, Kafka, Redis pub/sub, Socket.IO). Without arguments: one line per
+    wildcard / template matches; MQTT, NATS, AMQP, Kafka, Redis pub/sub, Socket.IO, raw TCP / UDP sockets by port).
+    listeners: every listening TCP / UDP socket with bind address, exposure and handler. Without arguments: one line per
     protocol (endpoints, send / receive sides, linked, check counts). With a pattern (name, id, substring or glob) /
     protocol / side (send | receive) / unmatched: one block per endpoint with senders (and, for a few endpoints, the
     entry points reaching them), receivers, guards, matches and checks: no_receiver, no_sender, test_sender_only,
     ambiguous, schema_mismatch, unguarded, external (.cg.yaml protocols.external, third-party origins, bridge
     packages). impact / reaches / path / downstream follow SENDS_TO / RECEIVED_BY / MATCHES_ENDPOINT as usual."""
     from .protocols.view import protocols as _pr, render_protocols
-    return render_protocols(_pr(_st(), pattern, protocol=protocol, side=side, unmatched=unmatched))
+    return render_protocols(_pr(_st(), pattern, protocol=protocol, side=side, unmatched=unmatched, listeners=listeners))
 
 
 @tool

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/platforms.py

@@ -944,11 +944,12 @@ def _alt_groups(builder, nvals: dict) -> dict[str, list[str]]:
         if g in alts and any(builder.nodes[i].file == file or builder.nodes[i].kind == "macro" for i in alts[g]):
             alts[g] += alts.pop(k)
     # same-named functions of separate programs (libuv docs/code/*/main.c `alloc_buffer`, each file with its own
-    # main()) are not one symbol: members in two or more files that define main() leave the group (#63)
+    # main()) are not one symbol: members in two or more files that define main() leave the group (#63); the same
+    # holds for the classes and structs of those programs (grpc's examples/cpp: a `GreeterServiceImpl` per server, #131)
     mains = {n.file for n in builder.nodes.values() if n.lang in ("c", "cpp") and n.kind == "function" and n.name == "main"}
     if mains:
         for k, ids in list(alts.items()):
-            if not k.startswith("c-fn:"):
+            if not k.startswith("c-fn:") and not all(builder.nodes[i].lang in ("c", "cpp") for i in ids):
                 continue
             prog = [i for i in ids if builder.nodes[i].file in mains]
             if len({builder.nodes[i].file for i in prog}) >= 2:

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/plugins/cfamily/plugin.py

@@ -655,6 +655,20 @@ class CFamilyPlugin(LanguagePlugin):
             if it.kind in TYPES:
                 types[it.name].append(it)
 
+        cls_files = defaultdict(set)
+        for it in self.items:
+            if it.kind in TYPES:
+                cls_files[it.qual].add(it.file)
+
+        def _visible(m, rel) -> bool:
+            """A method's class defined only in other source files (not headers) that `rel` does not include is
+            out of reach of a member call there (`writer->Write(..)` on a library type is not `HelloReactor::Write`
+            of another program's .cc)."""
+            files = cls_files.get(m.qual.rsplit("::", 1)[0]) if "::" in m.qual else None
+            if not files:
+                return True
+            return any(f == rel or os.path.splitext(f)[1] in HDR_EXT for f in files) or bool(files & self._included(rel))
+
         def pick(cands, rel):
             if not cands:
                 return []
@@ -685,7 +699,8 @@ class CFamilyPlugin(LanguagePlugin):
                     if name in macros and macros[name].attrs.get("function_like"):
                         targets = [macros[name]]
                     else:
-                        targets = pick(funcs.get(name), rel) or (pick(methods.get(name), rel) if owner.kind == "method" else [])
+                        targets = pick(funcs.get(name), rel) or (
+                            [m for m in pick(methods.get(name), rel) if _visible(m, rel)] if owner.kind == "method" else [])
                 elif form == "qualified":
                     q = quals.get(txt) or next((v for k, v in quals.items() if k.endswith("::" + txt)), None)
                     targets = [q] if q else []
@@ -696,6 +711,12 @@ class CFamilyPlugin(LanguagePlugin):
                         roots = [m for m in c if self._is_virtual(m) and not self._is_override(m)]
                         c = roots if len(roots) == 1 else c
                     targets = c if len(c) == 1 else []
+                    # the only method of that name, but its class is defined only in another source file: not the
+                    # receiver here. (Dropping such candidates before the uniqueness test would turn `s.data()` on a
+                    # std::string into leveldb's `Slice::data`, so it only removes.)
+                    if targets and not _visible(targets[0], rel):
+                        stats["heuristic_member_call_class_not_visible"] += 1
+                        targets = []
                 if not targets:
                     stats["heuristic_unresolved_calls"] += 1
                     continue
@@ -720,6 +741,10 @@ class CFamilyPlugin(LanguagePlugin):
                 if owner is None:
                     continue
                 for x in pick(types.get(name), rel):
+                    # a type is not linked across files: one defined only in another source file is not this one
+                    if x.file != rel and os.path.splitext(x.file)[1] not in HDR_EXT and x.file not in self._included(rel):
+                        stats["heuristic_type_ref_not_visible"] += 1
+                        continue
                     self._ref_edge(self.nid(owner), self.nid(x), rel, line, HEURISTIC, "", stats)
 
     def _is_virtual(self, it) -> bool:

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/plugins/cfamily/syntax.py

@@ -153,7 +153,17 @@ class Extractor:
         self.f = CFile(path, lang)
 
     def run(self) -> CFile:
-        tree = parser(self.f.lang).parse(mask_annotations(self.src, self.blank) if self.blank is not None else self.src)
+        src = mask_lock_annotations(mask_annotations(self.src, self.blank) if self.blank is not None else self.src)
+        tree = parser(self.f.lang).parse(src)
+        if tree.root_node.has_error:
+            # `ABSL_FLAG(uint16_t, port, 50051, "..");` at file level parses as a broken function definition that
+            # runs to the next `}` and swallows the class after it (#131): blank such macro statements and keep
+            # the parse that loses fewer lines
+            masked = mask_macro_statements(src)
+            if masked != src:
+                t2 = parser(self.f.lang).parse(masked)
+                if _error_lines(t2.root_node) < _error_lines(tree.root_node):
+                    tree = t2
         self.f.lines = self.src.decode("utf-8", "replace").split("\n")
         if tree.root_node.has_error:
             from ...core.syntax_errors import tree_spans
@@ -200,7 +210,8 @@ class Extractor:
             sub = Extractor(self.path, chunk.encode("utf-8"), self.f.lang, self.module, self.blank)
             sub.f = self.f
             n0 = len(self.f.items)
-            t = parser(self.f.lang).parse(mask_annotations(sub.src, self.blank) if self.blank is not None else sub.src)
+            t = parser(self.f.lang).parse(mask_lock_annotations(
+                mask_annotations(sub.src, self.blank) if self.blank is not None else sub.src))
             if not t.root_node.has_error:
                 sub.container(t.root_node, {"ns": [], "cls": None, "access": None, "anon": False, "extern_c": False})
             if len(self.f.items) == n0:
@@ -879,3 +890,94 @@ def mask_annotations(src: bytes, rx: re.Pattern) -> bytes:
         cont = directive and line.rstrip().endswith(b"\\")
         out.append(line if directive else rx.sub(lambda m: b" " * (m.end() - m.start()), line))
     return b"\n".join(out)
+
+
+MACRO_STMT = re.compile(rb"^[A-Z][A-Z0-9_]*[ \t]*\(", re.M)
+
+
+def mask_macro_statements(src: bytes) -> bytes:
+    """File-level `NAME(...);` statements of an upper-case macro starting in column 0 blanked (same length, newlines
+    kept). Strings and comments in the arguments are skipped while matching the parentheses."""
+    out = bytearray(src)
+    for m in MACRO_STMT.finditer(src):
+        i, depth, n = m.end() - 1, 0, len(src)
+        while i < n:
+            ch = src[i:i + 1]
+            if ch in (b'"', b"'"):
+                j = i + 1
+                while j < n and src[j:j + 1] not in (ch, b"\n"):
+                    j += 2 if src[j:j + 1] == b"\\" else 1
+                i = j + 1
+                continue
+            if src[i:i + 2] == b"/*":
+                j = src.find(b"*/", i + 2)
+                i = n if j < 0 else j + 2
+                continue
+            if src[i:i + 2] == b"//":
+                j = src.find(b"\n", i)
+                i = n if j < 0 else j
+                continue
+            if ch == b"(":
+                depth += 1
+            elif ch == b")":
+                depth -= 1
+                if depth == 0:
+                    break
+            elif ch in (b"{", b"}", b";"):
+                i = n
+                break
+            i += 1
+        if i >= n:
+            continue
+        rest = re.match(rb"[ \t]*;", src[i + 1:i + 40])
+        if not rest:
+            continue
+        end = i + 1 + rest.end()
+        for k in range(m.start(), end):
+            if out[k] != 10:
+                out[k] = 32
+    return bytes(out)
+
+
+def _error_lines(root) -> int:
+    from ...core.syntax_errors import tree_spans
+    return sum(b - a + 1 for a, b in tree_spans(root))
+
+
+# Clang thread-safety annotations after a declarator: `void Write() ABSL_EXCLUSIVE_LOCKS_REQUIRED(&mu_) {`,
+# `bool done_ GUARDED_BY(mu_) = false;`, `GTEST_LOCK_EXCLUDED_(mutex_)`. tree-sitter reads them as the declarator
+# (a method `mu_`), so they are blanked (same length) like the name-only annotation macros.
+LOCK_ANNOT = re.compile(
+    rb"\b(?:(?:[A-Z][A-Z0-9]*_)*(?:EXCLUSIVE_LOCKS_REQUIRED|SHARED_LOCKS_REQUIRED|LOCKS_EXCLUDED|LOCK_EXCLUDED|"
+    rb"LOCK_RETURNED|PT_GUARDED_BY|GUARDED_BY|ACQUIRED_BEFORE|ACQUIRED_AFTER|EXCLUSIVE_LOCK_FUNCTION|SHARED_LOCK_FUNCTION|"
+    rb"UNLOCK_FUNCTION|EXCLUSIVE_TRYLOCK_FUNCTION|SHARED_TRYLOCK_FUNCTION|ASSERT_EXCLUSIVE_LOCK|ASSERT_SHARED_LOCK|"
+    rb"CAPABILITY)|"
+    rb"[A-Z][A-Z0-9]*_(?:REQUIRES_SHARED|REQUIRES|EXCLUDES|ACQUIRE_SHARED|ACQUIRE|RELEASE_SHARED|RELEASE|TRY_ACQUIRE|"
+    rb"RETURN_CAPABILITY|ASSERT_CAPABILITY))_?[ \t]*\((?:[^()\n]|\([^()\n]*\))*\)")
+
+
+LOCK_NAME = re.compile(rb"\b(?:[A-Z][A-Z0-9]*_)*(?:SCOPED_LOCKABLE|LOCKABLE|SCOPED_CAPABILITY|NO_THREAD_SAFETY_ANALYSIS)_?\b(?![ \t]*\()")
+
+
+def mask_lock_annotations(src: bytes) -> bytes:
+    if b"LOCK" not in src and b"GUARDED_BY" not in src and b"_REQUIRES" not in src and b"_EXCLUDES" not in src \
+            and b"_ACQUIRE" not in src and b"_RELEASE" not in src and b"CAPABILITY" not in src:
+        return src
+    out = bytearray(src)
+    for m in LOCK_ANNOT.finditer(src):
+        ls = src.rfind(b"\n", 0, m.start()) + 1
+        before = src[ls:m.start()].rstrip()
+        if not before.strip() and ls > 0:
+            # on a line of its own after the declarator: `void Mock::Allow(uintptr_t obj)\n    GTEST_LOCK_EXCLUDED_(mu) {`
+            pl = src[src.rfind(b"\n", 0, ls - 1) + 1:ls - 1].rstrip()
+            before = pl if re.search(rb"(?:\)|\bconst|\boverride|\bnoexcept)$", pl) and not pl.lstrip().startswith(b"#") else b""
+        if before.lstrip().startswith(b"#") or not before or not re.search(rb"[\w)&*\]]$", before):
+            continue        # a directive, or not after a declarator (a statement / call of its own)
+        if re.search(rb"\b(?:return|if|while|case)\b[^;{}]*$", before):
+            continue
+        out[m.start():m.end()] = b" " * (m.end() - m.start())
+    for m in LOCK_NAME.finditer(src):       # `class SCOPED_LOCKABLE MutexLock {`, `void f() NO_THREAD_SAFETY_ANALYSIS {`
+        ls = src.rfind(b"\n", 0, m.start()) + 1
+        if not src[ls:m.start()].lstrip().startswith(b"#"):
+            out[m.start():m.end()] = b" " * (m.end() - m.start())
+    return bytes(out)

{cg_code_graph-0.10.1 → cg_code_graph-0.11.0}/codegraph/plugins/kotlin/exact.py

@@ -40,6 +40,7 @@ CALL_KINDS = ("CALLS", "INSTANTIATES")
 # `valueOf` / `entries` (`<init>` and `<anonymous object at ...>` start with "<")
 SYNTHETIC = re.compile(r"copy|component\d+|values|valueOf|entries")
 FACADE = re.compile(r"(?<=/)\w+Kt#(?=[^#]+\.$)")   # the JVM facade class of a Kotlin file's top-level functions
+ACCESSOR = re.compile(r"[gs]et[A-Z]\w*")                # a property accessor's JVM name
 
 
 def _has_kotlin_docs(path: str) -> bool:
@@ -301,10 +302,19 @@ class ExactLayer:
                         if d.name == name and d.line <= o.line + 1 <= d.end and (suf == "#") == (d.kind == "class"):
                             nid = d.id
                             break
+                if nid is None and suf == "(" and ACCESSOR.fullmatch(name):
+                    # an explicit `get()` / `set(v)` accessor: scip-java defines `getX().` at the keyword (0.13)
+                    pn = name[3].lower() + name[4:]
+                    for d in by_file.get(rel, ()):
+                        if d.name == pn and d.id in self.p.props and d.line <= o.line + 1 <= d.end:
+                            nid = d.id
+                            break
                 if nid is None:
                     if (suf in ("(", "#") and not name.startswith("<") and not SYNTHETIC.fullmatch(name)
                             and (rel, o.line, o.col) not in term_pos):
                         unmatched += 1
+                        if len(st.setdefault("scip_defs_unmatched_samples", [])) < 10:
+                            st["scip_defs_unmatched_samples"].append(f"{rel}:{o.line + 1}:{o.col + 1} {name}{suf}")
                     continue
                 sym[s] = nid
                 matched += 1