Merge Java hybrid LSP

Integrates the type-aware Java semantic resolver from worktree-adding-java-lsp-support (based on 673ac4e): java_lsp.c + generated java.lang/util stdlib registry, wired into the per-file LSP pass, with 94 java_lsp tests + ~170 java_lsp_coverage tests. Wiring conflicts (the shared files diverged 146 commits on main) resolved by keeping main's existing LSP registrations (go/c/php/py/ts/cs) and adding the Java entries in lsp_all.c, cbm.c (cbm_run_java_lsp dispatch), test_main.c and Makefile.cbm. Adapted to the current unity-build: renamed java_lsp.c's static helpers node_text/return_type_of to java_node_text/java_return_type_of so they no longer collide with ts_lsp.c inside the shared lsp_all.c translation unit. 4082 tests pass (incl. java_lsp + java_lsp_coverage); ASan/UBSan clean.
2026-10-27 06:53:11 +08:00 · 2026-06-02 19:06:02 +02:00
parent 12683b84a9 a600e80cb1
commit d849aeaee7
9 changed files with 9967 additions and 1 deletions
@@ -333,6 +333,7 @@ TEST_PY_LSP_STRESS_SRCS = tests/test_py_lsp_stress.c
 TEST_PY_LSP_SCALE_SRCS = tests/test_py_lsp_scale.c

 TEST_TS_LSP_SRCS = tests/test_ts_lsp.c
+TEST_JAVA_LSP_SRCS = tests/test_java_lsp.c tests/test_java_lsp_coverage.c

 TEST_INTEGRATION_SRCS = tests/test_integration.c tests/test_incremental.c

@@ -353,7 +354,7 @@ TEST_SIMHASH_SRCS = tests/test_simhash.c

 TEST_STACK_OVERFLOW_SRCS = tests/test_stack_overflow.c

-ALL_TEST_SRCS = $(TEST_FOUNDATION_SRCS) $(TEST_EXTRACTION_SRCS) $(TEST_STORE_SRCS) $(TEST_CYPHER_SRCS) $(TEST_MCP_SRCS) $(TEST_DISCOVER_SRCS) $(TEST_GRAPH_BUFFER_SRCS) $(TEST_PIPELINE_SRCS) $(TEST_WATCHER_SRCS) $(TEST_LZ4_SRCS) $(TEST_ZSTD_SRCS) $(TEST_ARTIFACT_SRCS) $(TEST_SQLITE_WRITER_SRCS) $(TEST_GO_LSP_SRCS) $(TEST_C_LSP_SRCS) $(TEST_PHP_LSP_SRCS) $(TEST_CS_LSP_SRCS) $(TEST_CS_LSP_BENCH_SRCS) $(TEST_SCOPE_SRCS) $(TEST_TYPE_REP_SRCS) $(TEST_PY_LSP_SRCS) $(TEST_PY_LSP_BENCH_SRCS) $(TEST_PY_LSP_STRESS_SRCS) $(TEST_PY_LSP_SCALE_SRCS) $(TEST_TS_LSP_SRCS) $(TEST_TRACES_SRCS) $(TEST_CLI_SRCS) $(TEST_MEM_SRCS) $(TEST_UI_SRCS) $(TEST_SECURITY_SRCS) $(TEST_YAML_SRCS) $(TEST_SIMHASH_SRCS) $(TEST_STACK_OVERFLOW_SRCS) $(TEST_INTEGRATION_SRCS)
+ALL_TEST_SRCS = $(TEST_FOUNDATION_SRCS) $(TEST_EXTRACTION_SRCS) $(TEST_STORE_SRCS) $(TEST_CYPHER_SRCS) $(TEST_MCP_SRCS) $(TEST_DISCOVER_SRCS) $(TEST_GRAPH_BUFFER_SRCS) $(TEST_PIPELINE_SRCS) $(TEST_WATCHER_SRCS) $(TEST_LZ4_SRCS) $(TEST_ZSTD_SRCS) $(TEST_ARTIFACT_SRCS) $(TEST_SQLITE_WRITER_SRCS) $(TEST_GO_LSP_SRCS) $(TEST_C_LSP_SRCS) $(TEST_PHP_LSP_SRCS) $(TEST_CS_LSP_SRCS) $(TEST_CS_LSP_BENCH_SRCS) $(TEST_SCOPE_SRCS) $(TEST_TYPE_REP_SRCS) $(TEST_PY_LSP_SRCS) $(TEST_PY_LSP_BENCH_SRCS) $(TEST_PY_LSP_STRESS_SRCS) $(TEST_PY_LSP_SCALE_SRCS) $(TEST_TS_LSP_SRCS) $(TEST_JAVA_LSP_SRCS) $(TEST_TRACES_SRCS) $(TEST_CLI_SRCS) $(TEST_MEM_SRCS) $(TEST_UI_SRCS) $(TEST_SECURITY_SRCS) $(TEST_YAML_SRCS) $(TEST_SIMHASH_SRCS) $(TEST_STACK_OVERFLOW_SRCS) $(TEST_INTEGRATION_SRCS)


 # ── Build directories ────────────────────────────────────────────
@@ -9,6 +9,7 @@
 #include "lsp/py_lsp.h"
 #include "lsp/ts_lsp.h"
 #include "lsp/cs_lsp.h"
+#include "lsp/java_lsp.h"
 #include "preprocessor.h"
 #include "foundation/compat.h"
 #include "tree_sitter/api.h" // TSParser, TSNode, TSTree, TSInput, TSLanguage, TSPoint, TSParseOptions, TSParseState
@@ -503,6 +504,9 @@ CBMFileResult *cbm_extract_file(const char *source, int source_len, CBMLanguage
            cbm_run_cs_lsp(a, result, source, source_len, root);
        }
    }
+    if (language == CBM_LANG_JAVA) {
+        cbm_run_java_lsp(a, result, source, source_len, root);
+    }
    atomic_fetch_add(&total_lsp_ns, now_ns() - lsp_start);

    // Calls extracted so far all carry ORIGINAL-source line numbers; the C/C++
@@ -0,0 +1,167 @@
+#ifndef CBM_LSP_JAVA_LSP_H
+#define CBM_LSP_JAVA_LSP_H
+
+/*
+ * java_lsp.h — Pure-C Java semantic resolver.
+ *
+ * Reverse-engineered from JLS §6 (Names) and §15 (Expressions) plus the
+ * algorithm shape used by Eclipse JDT-LS / java-language-server (which both
+ * delegate to javac.com.sun.source). The goal is parity with what JDT-LS
+ * exposes through textDocument/definition + textDocument/references for
+ * call-site resolution, *without* shelling out to javac.
+ *
+ * Mirrors the structure of go_lsp.h / c_lsp.h / php_lsp.h / py_lsp.h.
+ *
+ * Resolution scheme (single file):
+ *   1. Tree-sitter parses Java source into AST.
+ *   2. Build a CBMTypeRegistry from this file's CBMDefinitions + Java
+ *      stdlib (java.lang.*, java.util.*, java.io.*, java.util.function.*,
+ *      java.util.stream.*).
+ *   3. Walk the AST and for every method_invocation / object_creation /
+ *      field_access expression, evaluate the expression's type using
+ *      JLS-style scope chains (block → method params → class members →
+ *      superclass chain → outer class → import single → import on-demand →
+ *      java.lang → same package).
+ *   4. Match the textual call (callee name + arity) against the resolved
+ *      receiver type's method set, walking superclasses and interfaces.
+ *      Best-overload resolution falls back to argument count match.
+ *   5. Emit CBMResolvedCall entries with confidence ≥ 0.6 (the LSP floor
+ *      enforced by lsp_resolve.h).
+ */
+
+#include "type_rep.h"
+#include "scope.h"
+#include "type_registry.h"
+#include "../cbm.h"
+#include "go_lsp.h"  /* CBMLSPDef, CBMResolvedCallArray reused across languages */
+
+/* Java `use`-style import kinds. Mirrors PHP's enum with Java semantics. */
+enum {
+    CBM_JAVA_IMPORT_TYPE      = 0, /* import com.foo.Bar; — type import */
+    CBM_JAVA_IMPORT_STATIC    = 1, /* import static com.foo.Bar.method; */
+    CBM_JAVA_IMPORT_ON_DEMAND = 2, /* import com.foo.*; — package on-demand */
+    CBM_JAVA_IMPORT_STATIC_OD = 3, /* import static com.foo.Bar.*; */
+};
+
+/* Per-file resolution context. */
+typedef struct {
+    CBMArena *arena;
+    const char *source;
+    int source_len;
+    const CBMTypeRegistry *registry;
+    CBMScope *current_scope;
+
+    /* Java package — the package declared in the file ("com.example"), or
+     * empty string for the unnamed package. Stored in dotted form. */
+    const char *package_name;
+
+    /* The path-derived module QN for this file (passed in from the caller),
+     * used as the QN prefix for types defined here. */
+    const char *module_qn;
+
+    /* Import map. Each entry is one of CBM_JAVA_IMPORT_*.
+     *   - TYPE:      local_name="Bar",   target_qn="com.foo.Bar"
+     *   - STATIC:    local_name="sqrt",  target_qn="java.lang.Math.sqrt"
+     *   - ON_DEMAND: local_name="*",     target_qn="com.foo"
+     *   - STATIC_OD: local_name="*",     target_qn="java.lang.Math"
+     */
+    const char **import_local_names;
+    const char **import_target_qns;
+    int *import_kinds;
+    int import_count;
+    int import_cap;
+
+    /* Current enclosing context. */
+    const char *enclosing_method_qn;  /* QN of the nearest method/ctor */
+    const char *enclosing_class_qn;   /* QN of the nearest class/interface/enum */
+    const char *enclosing_super_qn;   /* QN of the immediate superclass (NULL ⇒ Object) */
+    const char *enclosing_class_short; /* short name of enclosing class — for "this" + ctor */
+    const char **enclosing_class_stack; /* nested-class stack (enclosing_class_qn at each depth) */
+    int enclosing_class_depth;
+    int enclosing_class_cap;
+
+    /* Output: resolved + diagnostic call edges. */
+    CBMResolvedCallArray *resolved_calls;
+
+    /* Recursion guards. */
+    int eval_depth;
+    int statement_depth;
+
+    /* Debug mode (CBM_LSP_DEBUG env). */
+    bool debug;
+} JavaLSPContext;
+
+/* ── Initialization / configuration ───────────────────────────────── */
+
+void java_lsp_init(JavaLSPContext *ctx, CBMArena *arena, const char *source, int source_len,
+                   const CBMTypeRegistry *registry, const char *package_name,
+                   const char *module_qn, CBMResolvedCallArray *out);
+
+void java_lsp_add_import(JavaLSPContext *ctx, const char *local_name, const char *target_qn,
+                         int kind);
+
+/* ── Walking / resolution ─────────────────────────────────────────── */
+
+void java_lsp_process_file(JavaLSPContext *ctx, TSNode root);
+
+/* Evaluate the type of an arbitrary Java expression node. May return
+ * cbm_type_unknown(); never returns NULL. */
+const CBMType *java_eval_expr_type(JavaLSPContext *ctx, TSNode node);
+
+/* Convert a Java type AST node (type_identifier, generic_type, array_type,
+ * scoped_type_identifier, void_type, integral_type, ...) into a CBMType. */
+const CBMType *java_parse_type_node(JavaLSPContext *ctx, TSNode node);
+
+/* Process a Java statement, binding any declared variables/parameters into
+ * the current scope. */
+void java_process_statement(JavaLSPContext *ctx, TSNode node);
+
+/* Resolve a Java type name (bare or qualified) against the current scope
+ * (imports + java.lang + same package). Returns the registered FQN or NULL. */
+const char *java_resolve_type_name(JavaLSPContext *ctx, const char *name);
+
+/* Lookup a method on a class, walking the super-chain and implemented
+ * interfaces. Returns the matched function or NULL. */
+const CBMRegisteredFunc *java_lookup_method(JavaLSPContext *ctx, const char *class_qn,
+                                            const char *method_name, int arg_count);
+
+/* Lookup a field on a class, walking the super-chain. Returns the field's
+ * type, or cbm_type_unknown() on miss. */
+const CBMType *java_lookup_field_type(JavaLSPContext *ctx, const char *class_qn,
+                                       const char *field_name);
+
+/* ── Top-level entry points ───────────────────────────────────────── */
+
+/* Single-file LSP: build registry from file defs + stdlib, walk and resolve. */
+void cbm_run_java_lsp(CBMArena *arena, CBMFileResult *result, const char *source, int source_len,
+                      TSNode root);
+
+/* Cross-file LSP: build registry from defs + stdlib, re-parse if needed,
+ * walk and resolve. defs include both local + cross-file definitions. */
+void cbm_run_java_lsp_cross(CBMArena *arena, const char *source, int source_len,
+                            const char *module_qn, CBMLSPDef *defs, int def_count,
+                            const char **import_names, const char **import_qns, int import_count,
+                            TSTree *cached_tree, CBMResolvedCallArray *out);
+
+/* Register the Java standard library (java.lang/util/io/etc.) into reg.
+ * Implementation lives in generated/java_stdlib_data.c. */
+void cbm_java_stdlib_register(CBMTypeRegistry *reg, CBMArena *arena);
+
+/* ── Batch cross-file LSP ─────────────────────────────────────────── */
+
+typedef struct {
+    const char *source;
+    int source_len;
+    const char *module_qn;
+    TSTree *cached_tree;
+    CBMLSPDef *defs;
+    int def_count;
+    const char **import_names;
+    const char **import_qns;
+    int import_count;
+} CBMBatchJavaLSPFile;
+
+void cbm_batch_java_lsp_cross(CBMArena *arena, CBMBatchJavaLSPFile *files, int file_count,
+                              CBMResolvedCallArray *out);
+
+#endif /* CBM_LSP_JAVA_LSP_H */
@@ -17,3 +17,5 @@
 #include "lsp/ts_lsp.c"
 #include "lsp/cs_lsp.c"
 #include "lsp/generated/cs_stdlib_data.c"
+#include "lsp/java_lsp.c"
+#include "lsp/generated/java_stdlib_data.c"
@@ -52,6 +52,8 @@ extern void suite_py_lsp_bench(void);
 extern void suite_py_lsp_stress(void);
 extern void suite_py_lsp_scale(void);
 extern void suite_ts_lsp(void);
+extern void suite_java_lsp(void);
+extern void suite_java_lsp_coverage(void);
 extern void suite_store_arch(void);
 extern void suite_store_bulk(void);
 extern void suite_store_pragmas(void);
@@ -141,6 +143,8 @@ int main(void) {
    RUN_SUITE(py_lsp_stress);
    RUN_SUITE(py_lsp_scale);
    RUN_SUITE(ts_lsp);
+    RUN_SUITE(java_lsp);
+    RUN_SUITE(java_lsp_coverage);

    /* Architecture + ADR + Louvain */
    RUN_SUITE(store_arch);