feat: use PG text for MoonBit String

Makefile

@@ -3,18 +3,19 @@ DIST_DIR = $(shell pwd)/dist
 DEV_DATA_DIR = $(shell pwd)/data
 DEV_LLVM_DIR = /usr/lib/llvm18
 
-# Default target: build for development
-.PHONY: build
-build: $(BUILD_DIR)/install
+$(BUILD_DIR)/install/.stamp: $(BUILD_DIR)/build/.stamp meson.build meson.py $(shell find src -name '*.c' -o -name '*.h')
+	uv run meson.py install -C $(BUILD_DIR)/build --destdir=$(BUILD_DIR)/install
+	touch $@
+
+.PHONY: rebuild
+rebuild:
+	touch $(BUILD_DIR)/build/.stamp
+	$(MAKE) $(BUILD_DIR)/install/.stamp
 
 $(BUILD_DIR)/build/.stamp:
 	uv run meson.py setup $(BUILD_DIR)/build --prefix=/ -Dllvm_dir=$(DEV_LLVM_DIR)
 	touch $@
 
-$(BUILD_DIR)/install/.stamp: $(BUILD_DIR)/build/.stamp
-	uv run meson.py install -C $(BUILD_DIR)/build --destdir=$(BUILD_DIR)/install
-	touch $@
-
 # Run a development PostgreSQL instance
 .PHONY: run
 run: $(DEV_DATA_DIR)/.stamp
@@ -101,4 +102,4 @@ clean:
 # Delete all build artifacts, development files, and distribution files
 .PHONY: distclean
 distclean: clean
-	rm -rf $(BUILD_DIR) $(DIST_DIR) $(DEV_DATA_DIR)
+	rm -rf $(BUILD_DIR) $(DIST_DIR) $(DEV_DATA_DIR) subprojects/wasm-micro-runtime-WAMR-*

meson.build

@@ -219,7 +219,11 @@ postgres_pure_lib = static_library('postgres_lib',
 
 cc = meson.get_compiler('c')
 llvm_inc_flag = '-I' + (get_option('llvm_dir') / 'include')
-wamr_c_flags = ['-Wno-incompatible-pointer-types', llvm_inc_flag]
+wamr_c_flags = [
+    '-DWASM_SUPPORT_NUL_IN_STRING=1',
+    '-Wno-incompatible-pointer-types',
+    llvm_inc_flag,
+]
 # https://github.com/bytecodealliance/wasm-micro-runtime/issues/4640
 if cc.has_argument('-fzero-init-padding-bits=unions')
   wamr_c_flags += ['-fzero-init-padding-bits=unions']
@@ -269,7 +273,11 @@ uncommon_shared_lib = shared_library('uncommon_shared',
 # Build the Rustica Engine
 
 rustica_shared = files(
-    'src/rustica/env.c',
+    'src/rustica/bh_log_to_pg.c',
+    'src/rustica/datatypes.c',
+    'src/rustica/adt/clock.c',
+    'src/rustica/adt/stringbuilder.c',
+    'src/rustica/adt/text.c',
 )
 rustica_deps = [vmlib, pg.get_variable('uuid')]
 rustica_cargs = wamr.get_cmake_definitions('-DWASM')
@@ -301,7 +309,7 @@ executable('rustica-engine',
 
 pg_mod_args = pg.get_variable('pg_mod_args')
 shared_module(
-    'rustica-engine',
+    'rustica-engine.so',
     'src/rustica/extension.c',
     sources: rustica_shared,
     include_directories: 'src',

src/rustica/adt/clock.c

@@ -0,0 +1,42 @@
+// SPDX-FileCopyrightText: 2025 燕几（北京）科技有限公司
+// SPDX-License-Identifier: Apache-2.0 OR MulanPSL-2.0
+
+#include "postgres.h"
+
+#include "rustica/datatypes.h"
+
+static uint64_t
+realtime_micros_since_unix_epoch(wasm_exec_env_t exec_env) {
+    struct timespec ts;
+    clock_gettime(CLOCK_REALTIME, &ts);
+    return (uint64_t)ts.tv_sec * 1000000 + ts.tv_nsec / 1000;
+}
+
+static wasm_externref_obj_t
+monotonic_now(wasm_exec_env_t exec_env) {
+    obj_t mono = rst_obj_new(exec_env, OBJ_CLOCK_MONOTONIC, NULL, 0);
+    INSTR_TIME_SET_CURRENT(mono->body.instr_time);
+    return rst_externref_of_obj(exec_env, mono);
+}
+
+static uint64_t
+monotonic_nanos_since(wasm_exec_env_t exec_env, wasm_obj_t ref) {
+    obj_t mono = wasm_externref_obj_get_obj(ref, OBJ_CLOCK_MONOTONIC);
+    instr_time it;
+    INSTR_TIME_SET_CURRENT(it);
+    INSTR_TIME_SUBTRACT(it, mono->body.instr_time);
+    return INSTR_TIME_GET_NANOSEC(it);
+}
+
+static NativeSymbol clock_natives[] = {
+    { "realtime_micros_since_unix_epoch",
+      realtime_micros_since_unix_epoch,
+      "()I" },
+    { "monotonic_now", monotonic_now, "()r" },
+    { "monotonic_nanos_since", monotonic_nanos_since, "(r)I" },
+};
+
+void
+rst_register_natives_clock() {
+    REGISTER_WASM_NATIVES("env", clock_natives);
+}

src/rustica/adt/stringbuilder.c

@@ -8,7 +8,6 @@
 
 #include "wasm_runtime_common.h"
 #include "rustica/datatypes.h"
-#include "rustica/module.h"
 
 static wasm_externref_obj_t
 sb_new(wasm_exec_env_t exec_env, int32_t size_hint) {
@@ -23,21 +22,41 @@ sb_new(wasm_exec_env_t exec_env, int32_t size_hint) {
     return rst_externref_of_obj(exec_env, obj);
 }
 
+static wasm_externref_obj_t
+sb_read_text(wasm_exec_env_t exec_env, wasm_obj_t ref) {
+    Datum txt_datum = wasm_externref_obj_get_datum(ref, TEXTOID);
+    text *txt = DatumGetTextPP(txt_datum);
+    char *start = VARDATA_ANY(txt);
+    obj_t obj =
+        rst_obj_new(exec_env, OBJ_STRING_INFO, ref, sizeof(StringInfoData));
+    if (txt_datum != PointerGetDatum(txt))
+        obj->flags |= OBJ_OWNS_BODY_MEMBERS;
+    obj->body.sb->data = (char *)txt;
+    obj->body.sb->cursor = start - (char *)txt;
+    obj->body.sb->len = VARSIZE_ANY_EXHDR(txt) + obj->body.sb->cursor;
+    obj->body.sb->maxlen = 0; // read-only
+    return rst_externref_of_obj(exec_env, obj);
+}
+
 static inline StringInfo
-sb_ensure_string_info(wasm_obj_t refobj) {
+sb_ensure_string_info(wasm_obj_t refobj, bool readonly) {
     obj_t obj = wasm_externref_obj_get_obj(refobj, OBJ_STRING_INFO);
+    if (!readonly && obj->body.sb->maxlen == 0)
+        ereport(ERROR, errmsg("StringInfo is read-only"));
+    if (readonly && obj->body.sb->maxlen != 0)
+        ereport(ERROR, errmsg("StringInfo is read-write"));
     return obj->body.sb;
 }
 
 static int32_t
 sb_mblength(wasm_exec_env_t exec_env, wasm_obj_t refobj) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     return sb->len;
 }
 
 static int32_t
 sb_write_string(wasm_exec_env_t exec_env, wasm_obj_t refobj, wasm_obj_t str) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     Datum jsstr = wasm_externref_obj_get_datum(str, TEXTOID);
     text *txt = DatumGetTextPP(jsstr);
     appendBinaryStringInfoNT(sb, VARDATA_ANY(txt), VARSIZE_ANY_EXHDR(txt));
@@ -52,7 +71,7 @@ sb_write_substring(wasm_exec_env_t exec_env,
                    wasm_obj_t str,
                    int32_t start,
                    int32_t len) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     Datum jsstr = wasm_externref_obj_get_datum(str, TEXTOID);
     text *txt = DatumGetTextPP(jsstr);
     char *data = VARDATA_ANY(txt);
@@ -69,7 +88,7 @@ sb_write_substring(wasm_exec_env_t exec_env,
 
 static int32_t
 sb_write_char(wasm_exec_env_t exec_env, wasm_obj_t refobj, int32_t ch) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     enlargeStringInfo(sb, MAX_UNICODE_EQUIVALENT_STRING);
     pg_unicode_to_server(ch, (unsigned char *)sb->data + sb->len);
     sb->len += (int)strlen(sb->data + sb->len);
@@ -82,7 +101,7 @@ sb_write_bytes(wasm_exec_env_t exec_env,
                wasm_obj_t bytes,
                int32_t start,
                int32_t len) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     bytea *b = DatumGetByteaP(wasm_externref_obj_get_datum(bytes, BYTEAOID));
     if (start < 0 || start + len > VARSIZE_ANY_EXHDR(b))
         ereport(ERROR, errmsg("sb_write_bytes: index out of bound"));
@@ -92,26 +111,46 @@ sb_write_bytes(wasm_exec_env_t exec_env,
 
 static int32_t
 sb_write_byte(wasm_exec_env_t exec_env, wasm_obj_t refobj, int32_t byte) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     enlargeStringInfo(sb, 1);
     sb->data[sb->len++] = (char)byte;
     return 0;
 }
 
 static wasm_externref_obj_t
 sb_to_string(wasm_exec_env_t exec_env, wasm_obj_t refobj) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     return cstring_into_varatt_obj(exec_env, sb->data, sb->len, TEXTOID);
 }
 
 static wasm_externref_obj_t
 sb_to_bytes(wasm_exec_env_t exec_env, wasm_obj_t refobj) {
-    StringInfo sb = sb_ensure_string_info(refobj);
+    StringInfo sb = sb_ensure_string_info(refobj, false);
     return cstring_into_varatt_obj(exec_env, sb->data, sb->len, BYTEAOID);
 }
 
+static int32_t
+sb_read_char(wasm_exec_env_t exec_env, wasm_obj_t refobj) {
+    int ch;
+    pg_wchar rv[2] = { '?', 0 };
+    StringInfo sb = sb_ensure_string_info(refobj, true);
+    if (sb->cursor >= sb->len)
+        return -1; // EOF
+    ch = pg_mblen(sb->data + sb->cursor);
+    if (ch > 0 && sb->cursor + ch <= sb->len) {
+        pg_mb2wchar_with_len(sb->data + sb->cursor, rv, ch);
+        sb->cursor += ch;
+    }
+    else {
+        // Invalid UTF-8 sequence, skip one byte
+        sb->cursor += 1;
+    }
+    return (int32_t)rv[0];
+}
+
 static NativeSymbol sb_symbols[] = {
     { "sb_new", sb_new, "(i)r" },
+    { "si_read_text", sb_read_text, "(r)r" },
     { "sb_mblength", sb_mblength, "(r)i" },
     { "sb_write_string", sb_write_string, "(rr)i" },
     { "sb_write_substring", sb_write_substring, "(rrii)i" },
@@ -120,6 +159,7 @@ static NativeSymbol sb_symbols[] = {
     { "sb_write_byte", sb_write_byte, "(ri)i" },
     { "sb_to_string", sb_to_string, "(r)r" },
     { "sb_to_bytes", sb_to_bytes, "(r)r" },
+    { "si_read_char", sb_read_char, "(r)i" },
 };
 
 void