Merge pull request #2695 from DeusData/fix/py-class-body-sibling-base

fix(python): bind a nested class base to an earlier sibling of its class body

src/pipeline/pass_lsp_cross.c

@@ -305,10 +305,16 @@ static const char *pxc_python_dotted_base_qn(const pxc_base_scope_t *bs, const c
  * Returns NULL when the base is not a confidently-known project type;
  * stdlib and third-party bases land here and keep their raw spelling. */
 static const char *pxc_resolve_base_qn(const pxc_base_scope_t *bs, const char *raw,
-                                       const char *module_qn) {
+                                       const char *module_qn, const char *class_qn) {
     if (!bs || !bs->reg || !raw || !raw[0]) {
         return NULL;
     }
+    /* A base of a class nested in a class body: an earlier sibling first. */
+    const char *sibling =
+        cbm_registry_resolve_python_class_body_base(bs->reg, bs->gbuf, class_qn, raw, bs->lang);
+    if (sibling) {
+        return sibling;
+    }
     const char *dotted = pxc_python_dotted_base_qn(bs, raw);
     if (dotted) {
         return dotted;
@@ -338,7 +344,8 @@ static const char *pxc_resolve_base_qn(const pxc_base_scope_t *bs, const char *r
  * source spelling. Unresolved entries pass through verbatim so a base the
  * registry does not know keeps working exactly as before. */
 static const char *pxc_join_base_qns(CBMArena *arena, const char *const *bases,
-                                     const pxc_base_scope_t *bs, const char *module_qn) {
+                                     const pxc_base_scope_t *bs, const char *module_qn,
+                                     const char *class_qn) {
     if (!bases || !bases[0]) {
         return NULL;
     }
@@ -352,7 +359,7 @@ static const char *pxc_join_base_qns(CBMArena *arena, const char *const *bases,
         return pxc_join_pipe(arena, bases);
     }
     for (int i = 0; i < count; i++) {
-        const char *qn = pxc_resolve_base_qn(bs, bases[i], module_qn);
+        const char *qn = pxc_resolve_base_qn(bs, bases[i], module_qn, class_qn);
         resolved[i] = qn ? qn : bases[i];
     }
     resolved[count] = NULL;
@@ -520,9 +527,10 @@ static int pxc_build_lsp_def(CBMArena *arena, const CBMDefinition *src, const ch
     /* Languages whose cross registrars read embedded_types as QNs get their
      * bases resolved against the project registry; everyone else keeps the
      * raw source spelling their own registrar already knows how to handle. */
-    dst->embedded_types = (bs && bs->reg && pxc_lang_resolves_base_qns(lang))
-                              ? pxc_join_base_qns(arena, src->base_classes, bs, module_qn)
-                              : pxc_join_pipe(arena, src->base_classes);
+    dst->embedded_types =
+        (bs && bs->reg && pxc_lang_resolves_base_qns(lang))
+            ? pxc_join_base_qns(arena, src->base_classes, bs, module_qn, src->qualified_name)
+            : pxc_join_pipe(arena, src->base_classes);
     dst->signature_param_types = NULL;
     dst->signature_param_count = 0;
     if (src->signature_param_types && src->signature_param_count > 0) {
@@ -727,7 +735,7 @@ static void pxc_fold_py_field_types(CBMArena *arena, const CBMFileResult *result
         fields[f].type_qn = NULL;
         CBMLSPDef *owner = ft->class_qn ? (CBMLSPDef *)cbm_ht_get(by_qn, ft->class_qn) : NULL;
         const char *name = owner ? pxc_py_annotation_type_name(arena, ft->type_text) : NULL;
-        const char *qn = name ? pxc_resolve_base_qn(bs, name, module_qn) : NULL;
+        const char *qn = name ? pxc_resolve_base_qn(bs, name, module_qn, NULL) : NULL;
         if (!qn || !ft->field_name || !ft->field_name[0]) {
             continue;
         }

src/pipeline/pass_parallel.c

@@ -3910,13 +3910,17 @@ static void resolve_def_inherits(resolve_ctx_t *rc, resolve_worker_state_t *ws,
         if (cbm_pipeline_php_vendor_bound(php_vendor, def->base_classes[b])) {
             continue;
         }
-        const char *bqn =
-            resolve_as_class(rc->registry, def->base_classes[b], mq, ik, iv, ic, fs->lang);
-        /* Same external-base gate as the sequential semantic pass. */
-        if (!bqn ||
-            cbm_python_external_base_contradicts(fs->lang, fs->imports, def->base_classes[b], bqn,
-                                                 rc->main_gbuf, rc->project_name, fs->rel)) {
-            continue;
+        /* Same order as the sequential semantic pass: an earlier sibling of
+         * the class body first (no external-base gate), then the registry. */
+        const char *bqn = cbm_registry_resolve_python_class_body_base(
+            rc->registry, rc->main_gbuf, def->qualified_name, def->base_classes[b], fs->lang);
+        if (!bqn) {
+            bqn = resolve_as_class(rc->registry, def->base_classes[b], mq, ik, iv, ic, fs->lang);
+            if (!bqn || cbm_python_external_base_contradicts(
+                            fs->lang, fs->imports, def->base_classes[b], bqn, rc->main_gbuf,
+                            rc->project_name, fs->rel)) {
+                continue;
+            }
         }
         const cbm_gbuf_node_t *bn = cbm_gbuf_find_by_qn(rc->main_gbuf, bqn);
         if (bn && node->id != bn->id) {

src/pipeline/pass_semantic.c

@@ -643,12 +643,18 @@ static void sem_process_def_edges(cbm_pipeline_ctx_t *ctx, const CBMDefinition *
             if (cbm_pipeline_php_vendor_bound(php_vendor, def->base_classes[b])) {
                 continue;
             }
-            const char *base_qn = resolve_as_class(ctx->registry, def->base_classes[b], module_qn,
-                                                   imp_keys, imp_vals, imp_count, lang);
-            if (!base_qn || cbm_python_external_base_contradicts(
-                                fi->language, &result->imports, def->base_classes[b], base_qn,
-                                ctx->gbuf, ctx->project_name, fi->rel_path)) {
-                continue; /* unresolved, or an external base (`unittest.TestCase`) */
+            /* An earlier sibling of the class body shadows every module or
+             * imported name, so it needs no external-base gate. */
+            const char *base_qn = cbm_registry_resolve_python_class_body_base(
+                ctx->registry, ctx->gbuf, def->qualified_name, def->base_classes[b], lang);
+            if (!base_qn) {
+                base_qn = resolve_as_class(ctx->registry, def->base_classes[b], module_qn, imp_keys,
+                                           imp_vals, imp_count, lang);
+                if (!base_qn || cbm_python_external_base_contradicts(
+                                    fi->language, &result->imports, def->base_classes[b], base_qn,
+                                    ctx->gbuf, ctx->project_name, fi->rel_path)) {
+                    continue; /* unresolved, or an external base (`unittest.TestCase`) */
+                }
             }
             const cbm_gbuf_node_t *base_node = cbm_gbuf_find_by_qn(ctx->gbuf, base_qn);
             if (base_node && node->id != base_node->id) {

src/pipeline/pipeline_internal.h

@@ -270,6 +270,14 @@ cbm_resolution_t cbm_registry_resolve_call_scoped(const cbm_registry_t *r, const
                                                   const char **import_map_vals,
                                                   int import_map_count, CBMLanguage lang,
                                                   const char *scope_qn);
+/* Python: the bare base `base` of the class `class_qn` nested in a class body
+ * binds the sibling class of that body (`<enclosing class>.<base>`) when the
+ * sibling is defined earlier in the same file (Python evaluates the bases in
+ * the class body). Exact lookups only; the registry-owned sibling QN, or NULL
+ * to resolve the base as before. Every base-class resolver asks this first. */
+const char *cbm_registry_resolve_python_class_body_base(const cbm_registry_t *r,
+                                                        const cbm_gbuf_t *gb, const char *class_qn,
+                                                        const char *base, CBMLanguage lang);
 #ifdef CBM_ENABLE_TEST_SEAMS
 void cbm_pxc_test_import_probe_fail_after(int successful_allocations);
 #endif

src/pipeline/registry.c

@@ -1383,6 +1383,47 @@ const char *cbm_registry_value_target(const cbm_registry_t *r, const char *qn) {
     return owned ? owned : qn;
 }
 
+/* Python evaluates a class statement's bases in the namespace of the body
+ * that encloses it. For a class nested in a class, a bare base naming a class
+ * defined EARLIER in that same class body is that sibling
+ * (`class IntegrityError(DatabaseError)` in sqlalchemy's asyncpg adapter),
+ * whatever same-named classes the project has elsewhere. A sibling defined
+ * later is not bound yet when the base is evaluated. Exact lookups only. */
+const char *cbm_registry_resolve_python_class_body_base(const cbm_registry_t *r,
+                                                        const cbm_gbuf_t *gb, const char *class_qn,
+                                                        const char *base, CBMLanguage lang) {
+    if (lang != CBM_LANG_PYTHON || !r || !gb || !class_qn || !base || !base[0] ||
+        strchr(base, '.')) {
+        return NULL;
+    }
+    const char *dot = strrchr(class_qn, '.');
+    if (!dot || dot == class_qn) {
+        return NULL;
+    }
+    char sibling[CBM_SZ_1K];
+    int enclosing_len = (int)(dot - class_qn);
+    int n = snprintf(sibling, sizeof(sibling), "%.*s", enclosing_len, class_qn);
+    if (n < 0 || (size_t)n >= sizeof(sibling) ||
+        !cbm_label_is_type_like(cbm_registry_label_of(r, sibling))) {
+        return NULL; /* not nested in a class body (module level, or a def) */
+    }
+    n = snprintf(sibling, sizeof(sibling), "%.*s.%s", enclosing_len, class_qn, base);
+    if (n < 0 || (size_t)n >= sizeof(sibling) || strcmp(sibling, class_qn) == 0) {
+        return NULL;
+    }
+    const char *owned = cbm_ht_get_key(r->exact, sibling);
+    if (!owned || !cbm_label_is_type_like(cbm_registry_label_of(r, owned))) {
+        return NULL;
+    }
+    const cbm_gbuf_node_t *cls = cbm_gbuf_find_by_qn(gb, class_qn);
+    const cbm_gbuf_node_t *sib = cbm_gbuf_find_by_qn(gb, owned);
+    if (!cls || !sib || !cls->file_path || !sib->file_path ||
+        strcmp(cls->file_path, sib->file_path) != 0 || sib->start_line >= cls->start_line) {
+        return NULL;
+    }
+    return owned;
+}
+
 int cbm_registry_find_by_name(const cbm_registry_t *r, const char *name, const char ***out,
                               int *count) {
     if (!r || !out || !count) {

tests/test_pipeline.c

@@ -15416,6 +15416,164 @@ TEST(pipeline_python_init_function_shares_submodule_qn_parallel_issue1548) {
     return pyn_run_init_collision(52);
 }
 
+/* ── F15: a nested class's bare base resolves in its class body first ── */
+
+/* The sqlalchemy asyncpg shape: an adapter class re-declares the DBAPI
+ * exception hierarchy as nested classes, each based on an EARLIER sibling
+ * spelled bare (`class IntegrityError(DatabaseError)`). Python evaluates a
+ * class statement's bases in the enclosing class body, so the sibling wins
+ * over the same-named module classes of exc.py, which the adapter imports
+ * as a module (so a name guess prefers them, as on sqlalchemy once its
+ * relative `from ... import exc` resolves). A sibling defined LATER in the
+ * body is not bound yet: `Early(Later)` names the module-level Later. */
+static const char pyb_exc_py[] = "class DBAPIError(Exception):\n"
+                                 "    pass\n"
+                                 "\n"
+                                 "class DatabaseError(DBAPIError):\n"
+                                 "    pass\n"
+                                 "\n"
+                                 "class IntegrityError(DatabaseError):\n"
+                                 "    pass\n";
+
+static const char pyb_adapter_py[] = "import exc\n"
+                                     "from exc import DBAPIError\n"
+                                     "\n"
+                                     "class Later(Exception):\n"
+                                     "    pass\n"
+                                     "\n"
+                                     "class AsyncAdapt_dbapi:\n"
+                                     "    class Error(DBAPIError):\n"
+                                     "        pass\n"
+                                     "\n"
+                                     "    class DatabaseError(Error):\n"
+                                     "        pass\n"
+                                     "\n"
+                                     "    class IntegrityError(DatabaseError):\n"
+                                     "        pass\n"
+                                     "\n"
+                                     "    class NotNullViolationError(IntegrityError):\n"
+                                     "        pass\n"
+                                     "\n"
+                                     "    class Early(Later):\n"
+                                     "        pass\n"
+                                     "\n"
+                                     "    class Later(Exception):\n"
+                                     "        pass\n";
+
+/* A second adapter with the same nested names, as sqlalchemy has one per
+ * driver: a name-only guess has several same-named candidates to pick. */
+static const char pyb_adapter2_py[] = "class AsyncAdapt_other_dbapi:\n"
+                                      "    class Error(Exception):\n"
+                                      "        pass\n"
+                                      "\n"
+                                      "    class DatabaseError(Error):\n"
+                                      "        pass\n"
+                                      "\n"
+                                      "    class IntegrityError(DatabaseError):\n"
+                                      "        pass\n";
+
+static int pyb_run_class_body_base(int fillers) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "/tmp/cbm_py_cbbase_XXXXXX");
+    if (!cbm_mkdtemp(tmp)) {
+        FAIL("tmpdir");
+    }
+    write_temp_file(tmp, "exc.py", pyb_exc_py);
+    write_temp_file(tmp, "adapter.py", pyb_adapter_py);
+    write_temp_file(tmp, "adapter2.py", pyb_adapter2_py);
+    for (int i = 0; i < fillers; i++) {
+        char name[64];
+        char body[128];
+        snprintf(name, sizeof(name), "filler%d.py", i);
+        snprintf(body, sizeof(body), "def filler%d():\n    return %d\n", i, i);
+        write_temp_file(tmp, name, body);
+    }
+    char *old_workers = getenv("CBM_WORKERS");
+    char *saved = old_workers ? strdup(old_workers) : NULL;
+    if (fillers > 0) {
+        cbm_setenv("CBM_WORKERS", "4", 1);
+    }
+    char db_path[512];
+    snprintf(db_path, sizeof(db_path), "%s/cbbase.db", tmp);
+    int rc[2] = {-1, -1};
+    bool db_sib[2] = {false, false};
+    bool db_mod[2] = {true, true};
+    bool ie_sib[2] = {false, false};
+    bool ie_mod[2] = {true, true};
+    bool nn_sib[2] = {false, false};
+    bool err_imp[2] = {false, false};
+    bool early_mod[2] = {false, false};
+    bool early_later_sib[2] = {true, true};
+    bool other_sib[2] = {false, false};
+    for (int run = 0; run < 2; run++) {
+        if (run == 1) {
+            char path[512];
+            snprintf(path, sizeof(path), "%s/adapter.py", tmp);
+            th_append_file(path, "\n# edited: the second run re-indexes incrementally\n");
+        }
+        cbm_pipeline_t *p = cbm_pipeline_new(tmp, db_path, CBM_MODE_FULL);
+        rc[run] = p ? cbm_pipeline_run(p) : -1;
+        cbm_store_t *s = rc[run] == 0 ? cbm_store_open_path(db_path) : NULL;
+        if (s) {
+            const char *pr = cbm_pipeline_project_name(p);
+            const char *cls = "adapter.AsyncAdapt_dbapi";
+            char a[256];
+            char b[256];
+            snprintf(a, sizeof(a), "%s.DatabaseError", cls);
+            snprintf(b, sizeof(b), "%s.Error", cls);
+            db_sib[run] = pyn_edge(s, pr, a, b, "INHERITS");
+            db_mod[run] = pyn_edge(s, pr, a, "adapter2.AsyncAdapt_other_dbapi.Error", "INHERITS");
+            snprintf(a, sizeof(a), "%s.IntegrityError", cls);
+            snprintf(b, sizeof(b), "%s.DatabaseError", cls);
+            ie_sib[run] = pyn_edge(s, pr, a, b, "INHERITS");
+            ie_mod[run] = pyn_edge(s, pr, a, "exc.DatabaseError", "INHERITS");
+            snprintf(a, sizeof(a), "%s.NotNullViolationError", cls);
+            snprintf(b, sizeof(b), "%s.IntegrityError", cls);
+            nn_sib[run] = pyn_edge(s, pr, a, b, "INHERITS");
+            snprintf(a, sizeof(a), "%s.Error", cls);
+            err_imp[run] = pyn_edge(s, pr, a, "exc.DBAPIError", "INHERITS");
+            snprintf(a, sizeof(a), "%s.Early", cls);
+            snprintf(b, sizeof(b), "%s.Later", cls);
+            early_mod[run] = pyn_edge(s, pr, a, "adapter.Later", "INHERITS");
+            early_later_sib[run] = pyn_edge(s, pr, a, b, "INHERITS");
+            other_sib[run] = pyn_edge(s, pr, "adapter2.AsyncAdapt_other_dbapi.IntegrityError",
+                                      "adapter2.AsyncAdapt_other_dbapi.DatabaseError", "INHERITS");
+            cbm_store_close(s);
+        }
+        if (p) {
+            cbm_pipeline_free(p);
+        }
+    }
+    if (saved) {
+        cbm_setenv("CBM_WORKERS", saved, 1);
+        free(saved);
+    } else {
+        cbm_unsetenv("CBM_WORKERS");
+    }
+    th_rmtree(tmp);
+    for (int run = 0; run < 2; run++) {
+        ASSERT_EQ(rc[run], 0);
+        ASSERT_TRUE(db_sib[run]);           /* DatabaseError(Error): the earlier sibling */
+        ASSERT_FALSE(db_mod[run]);          /* ... not the other adapter's Error */
+        ASSERT_TRUE(ie_sib[run]);           /* IntegrityError(DatabaseError): the sibling */
+        ASSERT_FALSE(ie_mod[run]);          /* ... not exc.DatabaseError */
+        ASSERT_TRUE(nn_sib[run]);           /* a chain of siblings */
+        ASSERT_TRUE(err_imp[run]);          /* control: a non-sibling base keeps the import */
+        ASSERT_TRUE(early_mod[run]);        /* a LATER sibling is not bound yet: module Later */
+        ASSERT_FALSE(early_later_sib[run]); /* ... never the later sibling */
+        ASSERT_TRUE(other_sib[run]);        /* the second adapter binds its own sibling */
+    }
+    PASS();
+}
+
+TEST(pipeline_python_class_body_sibling_base) {
+    return pyb_run_class_body_base(0);
+}
+
+TEST(pipeline_python_class_body_sibling_base_parallel) {
+    return pyb_run_class_body_base(52);
+}
+
 /* #1548 keeps TESTS for calls made through a test's own nested helper: the
  * call now starts at the extracted helper def (not at the test FILE node, as
  * before #1548), so TESTS is attributed to the enclosing test function. A
@@ -25114,6 +25272,8 @@ SUITE(pipeline) {
     RUN_TEST(pipeline_python_super_follows_c3_mro_parallel);
     RUN_TEST(pipeline_python_init_function_shares_submodule_qn_issue1548);
     RUN_TEST(pipeline_python_init_function_shares_submodule_qn_parallel_issue1548);
+    RUN_TEST(pipeline_python_class_body_sibling_base);
+    RUN_TEST(pipeline_python_class_body_sibling_base_parallel);
     RUN_TEST(pipeline_python_nested_helper_keeps_tests_issue1548);
     RUN_TEST(pipeline_python_nested_helper_keeps_tests_parallel_issue1548);
     RUN_TEST(pipeline_python_decorated_nested_class_extracted);