Merge pull request #815 from DeusData/distill/699-receiver-recursion

fix(extract): receiver-aware self-recursion detection

Author: DeusData

internal/cbm/cbm.c

@@ -439,6 +439,77 @@ static bool is_alloc_name(const char *n) {
     return name_in_set(n, set);
 }
 
+// Extract the receiver identifier from a def's receiver text — Go's
+// "(s *Store)" / "(s Store)" → "s". Stores the identifier start in *out and
+// returns its length; returns 0 for unnamed receivers ("(*Store)", "(Store)"),
+// where no second token follows the identifier (a lone token is the TYPE, not
+// a name — such methods have no receiver variable to call through anyway).
+static size_t receiver_ident(const char *recv_text, const char **out) {
+    const char *p = recv_text;
+    if (*p == '(') {
+        p++;
+    }
+    while (*p == ' ' || *p == '\t') {
+        p++;
+    }
+    const char *start = p;
+    while ((*p >= 'a' && *p <= 'z') || (*p >= 'A' && *p <= 'Z') || (*p >= '0' && *p <= '9') ||
+           *p == '_') {
+        p++;
+    }
+    size_t len = (size_t)(p - start);
+    if (len == 0) {
+        return 0; // "(*Store)": leading '*', no identifier
+    }
+    while (*p == ' ' || *p == '\t') {
+        p++;
+    }
+    if (*p == ')' || *p == '\0') {
+        return 0; // "(Store)": single token is the type, receiver unnamed
+    }
+    *out = start;
+    return len;
+}
+
+// Whether a callee expression targets the same instance/class as the enclosing
+// def, i.e. counts as genuine self-recursion rather than a same-named call on a
+// different receiver. callee_name may be bare ("recur") or qualified
+// ("self.recur", "this.recur", "super().save", "axios.get", "self.obj.recur").
+//
+// Bare names have no receiver → assume self-call (free function calling itself
+// by bare name; preserves prior behavior). Qualified names: the receiver chain
+// is everything before the LAST '.', and the WHOLE chain must name the same
+// object — self/this/cls/@self, or the enclosing def's own receiver identifier
+// (Go: `s` in `func (s *Store) save()`, from CBMDefinition.receiver). Matching
+// the whole chain (not its first segment) keeps self.obj.recur() out: it
+// targets self's FIELD obj, a different object. super() is the parent class and
+// any other receiver (axios, console, ...) a different target. See #599.
+static bool is_self_receiver(const char *callee_name, const char *def_receiver) {
+    if (!callee_name || !callee_name[0]) {
+        return false;
+    }
+    const char *dot = strrchr(callee_name, '.');
+    if (!dot) {
+        return true; // bare name → self-recursion candidate
+    }
+    size_t rlen = (size_t)(dot - callee_name);
+    static const char *const self_receivers[] = {"self", "this", "cls", "@self", NULL};
+    for (int i = 0; self_receivers[i]; i++) {
+        size_t sl = strlen(self_receivers[i]);
+        if (rlen == sl && strncmp(callee_name, self_receivers[i], sl) == 0) {
+            return true;
+        }
+    }
+    if (def_receiver) {
+        const char *rid = NULL;
+        size_t ril = receiver_ident(def_receiver, &rid);
+        if (ril > 0 && ril == rlen && strncmp(callee_name, rid, ril) == 0) {
+            return true; // call through the enclosing method's own receiver
+        }
+    }
+    return false; // super() / axios / console / self.obj / any other receiver
+}
+
 // Count parameters from a signature string like "(int a, Foo* b, cb (*)(int,int))".
 // Fallback for languages where param_names isn't populated (e.g. C keeps only the
 // signature text). Counts commas at the top paren level; treats "()"/"(void)" as 0.
@@ -772,12 +843,16 @@ CBMFileResult *cbm_extract_file(const char *source, int source_len, CBMLanguage
             continue;
         }
         CBMDefinition *d = &result->defs.items[best];
-        // callee_name may be bare ("recur") or qualified ("pkg.recur", "self.recur")
+        // callee_name may be bare ("recur") or qualified ("self.recur",
+        // "super().save", "axios.get"). A short-name match alone is not
+        // self-recursion: the callee must also target the same object
+        // (is_self_receiver), or super().save() inside save and axios.get
+        // inside get are false positives (#599).
         const char *dot = strrchr(c->callee_name, '.');
         const char *callee_short = dot ? dot + 1 : c->callee_name;
         bool in_loop = c->loop_depth > 0;
 
-        if (strcmp(callee_short, d->name) == 0) {
+        if (strcmp(callee_short, d->name) == 0 && is_self_receiver(c->callee_name, d->receiver)) {
             // Direct self-recursion. The call graph omits self-edges (pass_calls
             // skips source==target), so detect it here; seeds "recursive".
             d->is_recursive = true;

tests/test_extraction.c

@@ -2987,6 +2987,132 @@ TEST(complexity_guarded_recursion) {
     PASS();
 }
 
+/* #599: super().save() inside a method named save is a parent-class call —
+ * the receiver is super(), never self — so it must NOT flag self-recursion.
+ * super()-ONLY fixture: no self.save() alongside, so the assertion cannot pass
+ * vacuously off a genuine self-call. */
+TEST(complexity_super_only_not_recursive) {
+    CBMFileResult *r = extract("class B(A):\n"
+                               "    def save(self):\n"
+                               "        super().save()\n",
+                               CBM_LANG_PYTHON, "t", "super_only.py");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *d = find_def(r, "save");
+    ASSERT_NOT_NULL(d);
+    ASSERT_FALSE(d->is_recursive); /* parent-class call, not self-recursion */
+    ASSERT_FALSE(d->unguarded_recursion);
+    cbm_free_result(r);
+    PASS();
+}
+
+/* #599: a same-named call on an unrelated receiver (axios.get inside a
+ * function also named get) is delegation, not self-recursion. */
+TEST(complexity_same_name_other_receiver_not_recursive) {
+    CBMFileResult *r = extract("function get(url) {\n"
+                               "    return axios.get(url);\n"
+                               "}\n",
+                               CBM_LANG_JAVASCRIPT, "t", "axios_get.js");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *d = find_def(r, "get");
+    ASSERT_NOT_NULL(d);
+    ASSERT_FALSE(d->is_recursive); /* axios.get targets axios, not this fn */
+    ASSERT_FALSE(d->unguarded_recursion);
+    cbm_free_result(r);
+    PASS();
+}
+
+/* Guard: genuine self-recursion through self/this receivers still trips the
+ * detector after the receiver-aware narrowing (#599). */
+TEST(complexity_self_receiver_still_recursive) {
+    /* Python: self.recur() — same object. */
+    CBMFileResult *r = extract("class C:\n"
+                               "    def recur(self, n):\n"
+                               "        if n > 0:\n"
+                               "            self.recur(n - 1)\n",
+                               CBM_LANG_PYTHON, "t", "self_recur.py");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *d = find_def(r, "recur");
+    ASSERT_NOT_NULL(d);
+    ASSERT_TRUE(d->is_recursive);
+    ASSERT_FALSE(d->unguarded_recursion); /* guarded by `if n > 0` */
+    cbm_free_result(r);
+
+    /* JS: this.step() — same object. */
+    r = extract("class C {\n"
+                "    step(n) {\n"
+                "        if (n > 0) { this.step(n - 1); }\n"
+                "    }\n"
+                "}\n",
+                CBM_LANG_JAVASCRIPT, "t", "this_step.js");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    d = find_def(r, "step");
+    ASSERT_NOT_NULL(d);
+    ASSERT_TRUE(d->is_recursive);
+    ASSERT_FALSE(d->unguarded_recursion); /* guarded by `if (n > 0)` */
+    cbm_free_result(r);
+    PASS();
+}
+
+/* #599: chained receiver — self.obj.recur() inside recur targets self's FIELD
+ * obj, a different object. The whole receiver chain ("self.obj") must be
+ * compared, not just its first segment ("self"). */
+TEST(complexity_chained_receiver_not_self) {
+    CBMFileResult *r = extract("class C:\n"
+                               "    def recur(self, n):\n"
+                               "        self.obj.recur(n)\n",
+                               CBM_LANG_PYTHON, "t", "chained_recur.py");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *d = find_def(r, "recur");
+    ASSERT_NOT_NULL(d);
+    ASSERT_FALSE(d->is_recursive); /* self.obj is not self */
+    ASSERT_FALSE(d->unguarded_recursion);
+    cbm_free_result(r);
+    PASS();
+}
+
+/* #599: Go method receiver — the enclosing def's own receiver identifier
+ * (`s` in `func (s *Store) save()`) is whitelisted dynamically from
+ * CBMDefinition.receiver, so s.save() still counts as self-recursion while
+ * s.backup.save() (a field's same-named method) does not. */
+TEST(complexity_go_method_receiver_self_recursion) {
+    CBMFileResult *r = extract("package p\n"
+                               "type Store struct{}\n"
+                               "func (s *Store) save(n int) {\n"
+                               "    if n > 0 {\n"
+                               "        s.save(n - 1)\n"
+                               "    }\n"
+                               "}\n",
+                               CBM_LANG_GO, "t", "store.go");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    const CBMDefinition *d = find_def(r, "save");
+    ASSERT_NOT_NULL(d);
+    ASSERT_TRUE(d->is_recursive); /* s.save() == receiver s → self */
+    ASSERT_FALSE(d->unguarded_recursion); /* guarded by `if n > 0` */
+    cbm_free_result(r);
+
+    /* Same-named method on a field of the receiver: NOT self-recursion. */
+    r = extract("package p\n"
+                "type Store struct{ backup *Store }\n"
+                "func (s *Store) save(n int) {\n"
+                "    s.backup.save(n)\n"
+                "}\n",
+                CBM_LANG_GO, "t", "store_backup.go");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+    d = find_def(r, "save");
+    ASSERT_NOT_NULL(d);
+    ASSERT_FALSE(d->is_recursive); /* s.backup is not s */
+    ASSERT_FALSE(d->unguarded_recursion);
+    cbm_free_result(r);
+    PASS();
+}
+
 /* Deep chained member access + parameter count structure smells. */
 TEST(complexity_access_depth_and_params) {
     CBMFileResult *r = extract("package p\n"
@@ -3380,6 +3506,11 @@ SUITE(extraction) {
     RUN_TEST(complexity_linear_scan_in_loop);
     RUN_TEST(complexity_recursion_in_loop_unguarded);
     RUN_TEST(complexity_guarded_recursion);
+    RUN_TEST(complexity_super_only_not_recursive);
+    RUN_TEST(complexity_same_name_other_receiver_not_recursive);
+    RUN_TEST(complexity_self_receiver_still_recursive);
+    RUN_TEST(complexity_chained_receiver_not_self);
+    RUN_TEST(complexity_go_method_receiver_self_recursion);
     RUN_TEST(complexity_access_depth_and_params);
     RUN_TEST(walk_defs_no_truncation_over_4096_issue668);