entrygraph 0.1.42__tar.gz → 0.1.44__tar.gz

{entrygraph-0.1.42 → entrygraph-0.1.44}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: entrygraph
-Version: 0.1.42
+Version: 0.1.44
 Summary: Language-agnostic code graph: query symbols, entrypoints, and source-to-sink call paths from a SQLite index
 Project-URL: Repository, https://github.com/brettbergin/entrygraph
 Author-email: Brett Bergin <brettberginbc@yahoo.com>

{entrygraph-0.1.42 → entrygraph-0.1.44}/src/entrygraph/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.1.42'
-__version_tuple__ = version_tuple = (0, 1, 42)
+__version__ = version = '0.1.44'
+__version_tuple__ = version_tuple = (0, 1, 44)
 
 __commit_id__ = commit_id = None

{entrygraph-0.1.42 → entrygraph-0.1.44}/src/entrygraph/detect/entrypoints/golang.py

@@ -22,6 +22,12 @@ _GIN_METHODS = (
 _NETHTTP_REGISTER = frozenset({"HandleFunc", "Handle"})
 
 
+def _is_go_test(path: str) -> bool:
+    """Go test files (`*_test.go`) register routes only to exercise handlers; those
+    are test-harness calls, not the production route surface (#33)."""
+    return path.endswith("_test.go")
+
+
 def _go_main(x: FileExtraction) -> list[EntrypointHint]:
     hints = []
     for symbol in x.symbols:
@@ -43,6 +49,8 @@ def _go_main(x: FileExtraction) -> list[EntrypointHint]:
 
 
 def _nethttp_routes(x: FileExtraction) -> list[EntrypointHint]:
+    if _is_go_test(x.path):
+        return []
     hints = []
     for ref in x.references:
         if (
@@ -67,7 +75,43 @@ def _nethttp_routes(x: FileExtraction) -> list[EntrypointHint]:
     return hints
 
 
+def _group_prefixes(x: FileExtraction) -> dict[str, str]:
+    """Map each gin/fiber router-group variable to its composed path prefix.
+
+    `api := app.Group("/api"); v1 := api.Group("/v1")` -> {api: /api, v1: /api/v1}.
+    A nested group inherits its parent group's prefix; Go requires a variable to be
+    declared before use, so ascending line order resolves the chain in one pass. A
+    group with no assign target (`h(app.Group("/x"))`, passed straight into another
+    function) is out of static reach and contributes no prefix.
+    """
+    prefixes: dict[str, str] = {}
+    groups = sorted(
+        (
+            ref
+            for ref in x.references
+            if ref.kind == "call"
+            and ref.callee_name == "Group"
+            and ref.assign_target
+            and ref.receiver_text is not None
+            and ref.arg_preview
+        ),
+        key=lambda r: r.span.start_line,
+    )
+    for ref in groups:
+        if ref.arg_preview is None:  # already filtered above; narrows for the type-checker
+            continue
+        local = first_string_arg("(" + ref.arg_preview.lstrip("("))
+        if local is None:
+            continue  # closure-only group (chi-style middleware group) — no prefix
+        parent = prefixes.get(ref.receiver_text or "", "")
+        prefixes[ref.assign_target or ""] = _compose_prefixes([parent, local])
+    return prefixes
+
+
 def _gin_routes(x: FileExtraction) -> list[EntrypointHint]:
+    if _is_go_test(x.path):
+        return []
+    prefixes = _group_prefixes(x)
     hints = []
     for ref in x.references:
         if (
@@ -77,20 +121,30 @@ def _gin_routes(x: FileExtraction) -> list[EntrypointHint]:
             and ref.arg_preview
         ):
             route = first_string_arg("(" + ref.arg_preview.lstrip("("))
-            if route is not None and route.startswith("/"):
-                verb = ref.callee_name.upper()
-                method = verb if verb in _HTTP_VERBS else "*"
-                hints.append(
-                    EntrypointHint(
-                        rule_id="go.gin.route",
-                        kind=EntrypointKind.HTTP_ROUTE,
-                        handler_qualified_name=ref.caller_qualified_name,
-                        route=route,
-                        http_methods=[method],
-                        framework="gin",
-                        metadata={"registration": ref.arg_preview},
-                    )
+            if route is None:
+                continue
+            verb = ref.callee_name.upper()
+            is_verb = verb in _HTTP_VERBS
+            group_prefix = prefixes.get(ref.receiver_text)
+            # A grouped route often drops the leading slash (`v1.GET("users")`) since
+            # the group carries the prefix; accept that only for a concrete verb on a
+            # known group var. Otherwise keep the conservative leading-slash check
+            # (guards against Handle/Any's method-first arg and stray string args).
+            if not route.startswith("/") and not (is_verb and group_prefix is not None):
+                continue
+            method = verb if is_verb else "*"
+            full = _compose_prefixes([group_prefix or "", route])
+            hints.append(
+                EntrypointHint(
+                    rule_id="go.gin.route",
+                    kind=EntrypointKind.HTTP_ROUTE,
+                    handler_qualified_name=ref.caller_qualified_name,
+                    route=full,
+                    http_methods=[method],
+                    framework="gin",
+                    metadata={"registration": ref.arg_preview},
                 )
+            )
     return hints
 
 
@@ -128,6 +182,8 @@ def _chi_routes(x: FileExtraction) -> list[EntrypointHint]:
     falls inside a Route call's span gets that prefix (nested Routes stack). Mount
     (`r.Mount("/api", subRouter())`) points at a router built in another function,
     which is out of static reach, so its prefix isn't composed."""
+    if _is_go_test(x.path):
+        return []
     scopes: list[tuple[int, int, str]] = []
     for ref in x.references:
         if (
@@ -185,6 +241,8 @@ def _gorilla_routes(x: FileExtraction) -> list[EntrypointHint]:
     per line. Previously only bare HandleFunc was matched — the `.Path().Methods()`
     form and any `.Methods()` verb were dropped (loki: ~0% of gorilla routes).
     """
+    if _is_go_test(x.path):
+        return []
     routes: dict[int, str] = {}
     methods: dict[int, list[str]] = {}
     handlers: dict[int, str | None] = {}
@@ -253,7 +311,7 @@ def _grpc_services(x: FileExtraction) -> list[EntrypointHint]:
     `route` — the Entrypoint row has no name column, and it keeps distinct services
     from collapsing in dedup (they all share `handler_qualified_name=None`).
     """
-    if x.path.endswith("_test.go"):
+    if _is_go_test(x.path):
         return []  # test harnesses spin up real services; not production surface (#33)
     hints = []
     seen: set[str] = set()

{entrygraph-0.1.42 → entrygraph-0.1.44}/src/entrygraph/detect/entrypoints/ruby.py

@@ -21,6 +21,15 @@ _SINATRA_VERBS = frozenset({"get", "post", "put", "delete", "patch"})
 _RAILS_VERBS = frozenset({"get", "post", "put", "patch", "delete", "resources", "resource", "root"})
 
 
+def _is_ruby_test(path: str) -> bool:
+    """RSpec / Test::Unit files. Rack::Test reuses the bare `get '/x'` verb DSL
+    inside specs, indistinguishable from a Sinatra/Grape route declaration, so those
+    files must not contribute to the route surface (#33)."""
+    return (
+        path.endswith(("_spec.rb", "_test.rb")) or "/spec/" in f"/{path}" or "/test/" in f"/{path}"
+    )
+
+
 def _is_rails_routes_file(path: str) -> bool:
     """The main `config/routes.rb` plus split route files loaded via `draw(:x)`
     (`config/routes/api.rb`, `config/routes/admin.rb`, ...). Missing the split
@@ -29,6 +38,8 @@ def _is_rails_routes_file(path: str) -> bool:
 
 
 def _sinatra_routes(x: FileExtraction) -> list[EntrypointHint]:
+    if _is_ruby_test(x.path):
+        return []
     hints = []
     for ref in x.references:
         if (
@@ -106,6 +117,8 @@ def _rake_tasks(x: FileExtraction) -> list[EntrypointHint]:
 
 def _grape_routes(x: FileExtraction) -> list[EntrypointHint]:
     """Grape API classes: class-body `get '/x'` / `post '/y'` declarations."""
+    if _is_ruby_test(x.path):
+        return []
     hints = []
     for ref in x.references:
         if (

{entrygraph-0.1.42 → entrygraph-0.1.44}/src/entrygraph/extract/golang.py

@@ -263,6 +263,7 @@ class GoExtractor:
                     caller_qualified_name=caller,
                     arg_count=len(args.named_children) if args is not None else 0,
                     arg_preview=truncate(node_text(args)) if args is not None else None,
+                    assign_target=self._assign_target(node),
                 )
             )
             self._emit_callbacks(args, caller, out)
@@ -280,6 +281,25 @@ class GoExtractor:
                 )
             )
 
+    def _assign_target(self, call: Node) -> str | None:
+        """LHS variable when `call` is the sole RHS of a single-var `:=`/`=`.
+
+        `api := app.Group("/api")` -> "api". Multi-assignment (`a, b := f()`) and
+        calls nested in a larger expression (`h(app.Group("/x"))`) return None —
+        the group there isn't bound to a reachable variable.
+        """
+        right = call.parent
+        if right is None or right.type != "expression_list":
+            return None
+        stmt = right.parent
+        if stmt is None or stmt.type not in ("short_var_declaration", "assignment_statement"):
+            return None
+        left = stmt.child_by_field_name("left")
+        if left is None or len(left.named_children) != 1 or len(right.named_children) != 1:
+            return None
+        ident = left.named_children[0]
+        return node_text(ident) if ident.type == "identifier" else None
+
     def _emit_callbacks(self, args: Node | None, caller: str | None, out: FileExtraction) -> None:
         """Bare-identifier arguments passed to a call — a function value that may be
         invoked later, e.g. ``http.HandleFunc("/", handler)``. Resolution keeps only

{entrygraph-0.1.42 → entrygraph-0.1.44}/src/entrygraph/extract/ir.py

@@ -58,6 +58,10 @@ class RawReference:
     caller_qualified_name: str | None  # FQN of enclosing def; None = module level
     arg_count: int = 0
     arg_preview: str | None = None  # truncated literal args, for sink triage
+    # LHS variable when this call is the sole RHS of a single-var `:=`/`=`
+    # (`api := app.Group("/api")`). Lets router-group rules link routes registered
+    # on the group var back to its path prefix. Populated by the Go extractor.
+    assign_target: str | None = None
 
 
 @dataclass(slots=True)

{entrygraph-0.1.42 → entrygraph-0.1.44}/tests/test_entrypoints_go.py

@@ -19,7 +19,7 @@ def _go_ext(references, path="main.go"):
     )
 
 
-def _call(callee, receiver, arg, start=1, end=None):
+def _call(callee, receiver, arg, start=1, end=None, assign=None):
     return RawReference(
         kind="call",
         callee_text=f"{receiver}.{callee}" if receiver else callee,
@@ -28,6 +28,7 @@ def _call(callee, receiver, arg, start=1, end=None):
         span=Span(start, 0, end if end is not None else start, 40),
         caller_qualified_name="app.routes",
         arg_preview=arg,
+        assign_target=assign,
     )
 
 
@@ -118,3 +119,109 @@ def test_grpc_test_harness_registrations_excluded():
     # A real service registered inside a *_test.go harness is not production surface.
     refs = [_call("RegisterFrontendServer", "frontendv1pb", "(grpcServer, v1)")]
     assert _grpc_rule().match(_go_ext(refs, path="pkg/frontend/frontend_test.go")) == []
+
+
+def _gin_rule():
+    return {r.id: r for r in rules_for("go", {"gin"})}["go.gin.route"]
+
+
+def _fiber_rule():
+    return {r.id: r for r in rules_for("go", {"fiber"})}["go.fiber.route"]
+
+
+def test_gin_group_prefix_composed():
+    # api := router.Group("/api"); api.GET("/users", h) -> /api/users (#36). The
+    # group var is bound via the call's assign_target.
+    refs = [
+        _call("Group", "router", '("/api")', start=1, assign="api"),
+        _call("GET", "api", '("/users", listUsers)', start=2),
+        _call("GET", "router", '("/health", health)', start=3),  # ungrouped
+    ]
+    got = {(h.http_methods[0], h.route) for h in _gin_rule().match(_go_ext(refs))}
+    assert got == {("GET", "/api/users"), ("GET", "/health")}
+
+
+def test_gin_nested_groups_stack():
+    # api := app.Group("/api"); v1 := api.Group("/v1"); v1.GET("/users", h)
+    refs = [
+        _call("Group", "app", '("/api")', start=1, assign="api"),
+        _call("Group", "api", '("/v1")', start=2, assign="v1"),
+        _call("GET", "v1", '("/users", h)', start=3),
+    ]
+    got = {(h.http_methods[0], h.route) for h in _gin_rule().match(_go_ext(refs))}
+    assert got == {("GET", "/api/v1/users")}
+
+
+def test_gin_grouped_route_without_leading_slash():
+    # Grouped routes commonly drop the leading slash (gin versioning example);
+    # accept it for a concrete verb on a known group var and still prefix it.
+    refs = [
+        _call("Group", "router", '("/v1")', start=1, assign="apiV1"),
+        _call("GET", "apiV1", '("users", h)', start=2),
+    ]
+    got = {(h.http_methods[0], h.route) for h in _gin_rule().match(_go_ext(refs))}
+    assert got == {("GET", "/v1/users")}
+
+
+def test_gin_ungrouped_route_still_requires_leading_slash():
+    # A slash-less path on a non-group receiver stays filtered (guards against
+    # Handle/Any's method-first arg and stray string args).
+    refs = [_call("GET", "router", '("users", h)', start=1)]
+    assert _gin_rule().match(_go_ext(refs)) == []
+
+
+def test_gin_group_passed_to_function_has_no_prefix():
+    # h(app.Group("/api")) has no assign target — out of static reach, so a later
+    # same-named receiver must not pick up a phantom prefix.
+    refs = [
+        _call("Group", "app", '("/api")', start=1, assign=None),
+        _call("GET", "app", '("/health", h)', start=2),
+    ]
+    got = {(h.http_methods[0], h.route) for h in _gin_rule().match(_go_ext(refs))}
+    assert got == {("GET", "/health")}
+
+
+def test_fiber_group_prefix_end_to_end():
+    # Exercises the real Go extractor (assign_target capture) + fiber rule together.
+    from entrygraph.extract.base import FileContext
+    from entrygraph.extract.golang import GoExtractor
+    from entrygraph.parsing.parsers import parse
+
+    src = (
+        b"package router\n"
+        b"func SetupRoutes(app *fiber.App) {\n"
+        b'	api := app.Group("/api")\n'
+        b'	api.Get("/", handler.Hello)\n'
+        b'	auth := api.Group("/auth")\n'
+        b'	auth.Post("/login", handler.Login)\n'
+        b"}\n"
+    )
+    ctx = FileContext(
+        path="router/router.go",
+        language="go",
+        module_path="router",
+        source=src,
+        is_package=False,
+    )
+    x = GoExtractor().extract(parse("go", src), ctx)
+    got = {(h.http_methods[0], h.route) for h in _fiber_rule().match(x)}
+    assert got == {("GET", "/api"), ("POST", "/api/auth/login")}
+
+
+def test_go_route_rules_skip_test_files():
+    # *_test.go registers routes only to exercise handlers (chi corpus: 152/204
+    # routes came from _test.go). All Go HTTP-route rules must skip them (#33).
+    refs = [_call("Get", "r", '("/users", h)')]
+    assert _chi_rule().match(_go_ext(refs, path="middleware/compress_test.go")) == []
+    assert _gin_rule().match(_go_ext(refs, path="api/routes_test.go")) == []
+    assert _fiber_rule().match(_go_ext(refs, path="api/routes_test.go")) == []
+    # sanity: the same refs in a non-test file still produce a route
+    assert _chi_rule().match(_go_ext(refs, path="middleware/compress.go"))
+
+
+def test_go_nethttp_and_grpc_skip_test_files():
+    nethttp = {r.id: r for r in rules_for("go", {"net/http"})}["go.nethttp.route"]
+    refs = [_call("HandleFunc", "http", '("/x", h)')]
+    assert nethttp.match(_go_ext(refs, path="server_test.go")) == []
+    grpc = [_call("RegisterPusherServer", "pb", "(s, impl)")]
+    assert _grpc_rule().match(_go_ext(grpc, path="server_test.go")) == []

entrygraph-0.1.44/tests/test_entrypoints_ruby.py

@@ -0,0 +1,63 @@
+"""Ruby entrypoint-rule tests (kept per-language so PRs touching different
+languages don't serially conflict on one shared test module)."""
+
+from __future__ import annotations
+
+from entrygraph.detect.entrypoints import rules_for
+from entrygraph.extract.ir import FileExtraction, RawReference, Span
+
+
+def _ruby_ext(references, path):
+    return FileExtraction(
+        path=path,
+        language="ruby",
+        module_path=path.replace("/", ".").removesuffix(".rb"),
+        parse_ok=True,
+        error_count=0,
+        symbols=[],
+        references=list(references),
+    )
+
+
+def _verb(verb, route):
+    return RawReference(
+        kind="call",
+        callee_text=verb,
+        callee_name=verb,
+        receiver_text=None,
+        span=Span(1, 0, 1, 40),
+        caller_qualified_name=None,
+        arg_preview=f"('{route}')",
+    )
+
+
+def _sinatra_rule():
+    return {r.id: r for r in rules_for("ruby", {"sinatra"})}["ruby.sinatra.route"]
+
+
+def _grape_rule():
+    return {r.id: r for r in rules_for("ruby", {"grape"})}["ruby.grape.route"]
+
+
+def test_sinatra_route_in_app_file_detected():
+    hints = _sinatra_rule().match(_ruby_ext([_verb("get", "/health")], "app.rb"))
+    assert [h.route for h in hints] == ["/health"]
+
+
+def test_sinatra_ignores_rack_test_calls_in_spec_and_test_files():
+    # Rack::Test uses the same bare `get '/x'` DSL inside specs; those files are not
+    # the route surface (sinatra corpus: 256/274 routes came from test/spec) (#33).
+    for path in (
+        "test/routing_test.rb",
+        "spec/app_spec.rb",
+        "rack-protection/spec/lib/rack/protection/ip_spoofing_spec.rb",
+        "some/nested/test/helper_test.rb",
+    ):
+        assert _sinatra_rule().match(_ruby_ext([_verb("get", "/")], path)) == []
+
+
+def test_grape_ignores_test_files():
+    assert _grape_rule().match(_ruby_ext([_verb("post", "/x")], "spec/api_spec.rb")) == []
+    assert [h.route for h in _grape_rule().match(_ruby_ext([_verb("post", "/x")], "api.rb"))] == [
+        "/x"
+    ]