Add strip option in qjsc to reduce object size (#388)

- `-s` strips the source code
- `-ss` strips source and line/column numbers information
- `qjsc repl.js` generates an object size of **105726** bytes
- `qjsc -s repl.js` generates an object size of **20853** bytes
- `qjsc -ss repl.js` generates an object size of only **16147** bytes
- compile repl.js with `-ss`
- bump byte code version to 12

Makefile

@@ -61,7 +61,7 @@ clean:
 	cmake --build $(BUILD_DIR) --target clean
 
 codegen: $(QJSC)
-	$(QJSC) -o gen/repl.c -m repl.js
+	$(QJSC) -ss -o gen/repl.c -m repl.js
 	$(QJSC) -e -o gen/function_source.c tests/function_source.js
 	$(QJSC) -e -o gen/hello.c examples/hello.js
 	$(QJSC) -e -o gen/hello_module.c -m examples/hello_module.js

gen/function_source.c

@@ -5,7 +5,7 @@
 const uint32_t qjsc_function_source_size = 384;
 
 const uint8_t qjsc_function_source[384] = {
- 0x0b, 0x06, 0x0c, 0x61, 0x63, 0x74, 0x75, 0x61,
+ 0x0c, 0x06, 0x0c, 0x61, 0x63, 0x74, 0x75, 0x61,
  0x6c, 0x02, 0x66, 0x30, 0x74, 0x65, 0x73, 0x74,
  0x73, 0x2f, 0x66, 0x75, 0x6e, 0x63, 0x74, 0x69,
  0x6f, 0x6e, 0x5f, 0x73, 0x6f, 0x75, 0x72, 0x63,
@@ -15,12 +15,12 @@ const uint8_t qjsc_function_source[384] = {
  0x75, 0x6e, 0x63, 0x74, 0x69, 0x6f, 0x6e, 0x20,
  0x66, 0x28, 0x29, 0x20, 0x7b, 0x20, 0x72, 0x65,
  0x74, 0x75, 0x72, 0x6e, 0x20, 0x34, 0x32, 0x20,
- 0x7d, 0x0c, 0x00, 0x02, 0x01, 0x9e, 0x01, 0x00,
+ 0x7d, 0x0c, 0x00, 0xfa, 0x01, 0x9e, 0x01, 0x00,
  0x06, 0x00, 0x03, 0x00, 0x01, 0xa0, 0x01, 0x06,
  0xa0, 0x01, 0x00, 0x00, 0x00, 0xb2, 0x03, 0x02,
  0x00, 0x30, 0xb4, 0x03, 0x04, 0x00, 0x70, 0xb2,
  0x03, 0x04, 0x03, 0x70, 0x10, 0x00, 0x01, 0x00,
- 0xe0, 0x01, 0x00, 0x01, 0x00, 0x0c, 0x43, 0x02,
+ 0xe0, 0x01, 0x00, 0x01, 0x00, 0x0c, 0x43, 0xfa,
  0x01, 0xb4, 0x03, 0x00, 0x00, 0x00, 0x01, 0x00,
  0x00, 0x03, 0x00, 0xbb, 0x2a, 0x28, 0xb6, 0x03,
  0x03, 0x01, 0x04, 0x02, 0x1e, 0x0c, 0x0e, 0x1a,

gen/hello.c

@@ -5,12 +5,12 @@
 const uint32_t qjsc_hello_size = 89;
 
 const uint8_t qjsc_hello[89] = {
- 0x0b, 0x04, 0x0e, 0x63, 0x6f, 0x6e, 0x73, 0x6f,
+ 0x0c, 0x04, 0x0e, 0x63, 0x6f, 0x6e, 0x73, 0x6f,
  0x6c, 0x65, 0x06, 0x6c, 0x6f, 0x67, 0x16, 0x48,
  0x65, 0x6c, 0x6c, 0x6f, 0x20, 0x57, 0x6f, 0x72,
  0x6c, 0x64, 0x22, 0x65, 0x78, 0x61, 0x6d, 0x70,
  0x6c, 0x65, 0x73, 0x2f, 0x68, 0x65, 0x6c, 0x6c,
- 0x6f, 0x2e, 0x6a, 0x73, 0x0c, 0x00, 0x02, 0x00,
+ 0x6f, 0x2e, 0x6a, 0x73, 0x0c, 0x00, 0xfa, 0x00,
  0x9e, 0x01, 0x00, 0x01, 0x00, 0x03, 0x00, 0x00,
  0x14, 0x01, 0xa0, 0x01, 0x00, 0x00, 0x00, 0x38,
  0xd9, 0x00, 0x00, 0x00, 0x42, 0xda, 0x00, 0x00,

gen/hello_module.c

@@ -5,14 +5,14 @@
 const uint32_t qjsc_fib_module_size = 310;
 
 const uint8_t qjsc_fib_module[310] = {
- 0x0b, 0x03, 0x2c, 0x65, 0x78, 0x61, 0x6d, 0x70,
+ 0x0c, 0x03, 0x2c, 0x65, 0x78, 0x61, 0x6d, 0x70,
  0x6c, 0x65, 0x73, 0x2f, 0x66, 0x69, 0x62, 0x5f,
  0x6d, 0x6f, 0x64, 0x75, 0x6c, 0x65, 0x2e, 0x6a,
  0x73, 0x06, 0x66, 0x69, 0x62, 0x02, 0x6e, 0x0d,
  0xb2, 0x03, 0x00, 0x01, 0x00, 0x00, 0xb4, 0x03,
- 0x00, 0x00, 0x0c, 0x20, 0x02, 0x01, 0x9e, 0x01,
+ 0x00, 0x00, 0x0c, 0x20, 0xfa, 0x01, 0x9e, 0x01,
  0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x09, 0x00,
- 0xb4, 0x03, 0x00, 0x01, 0x0c, 0x43, 0x02, 0x01,
+ 0xb4, 0x03, 0x00, 0x01, 0x0c, 0x43, 0xfa, 0x01,
  0xb4, 0x03, 0x01, 0x00, 0x01, 0x04, 0x01, 0x00,
  0x1a, 0x01, 0xb6, 0x03, 0x00, 0x01, 0x00, 0xb4,
  0x03, 0x00, 0x00, 0xd0, 0xb3, 0xa7, 0xe9, 0x03,
@@ -49,7 +49,7 @@ const uint8_t qjsc_fib_module[310] = {
 const uint32_t qjsc_hello_module_size = 177;
 
 const uint8_t qjsc_hello_module[177] = {
- 0x0b, 0x07, 0x30, 0x65, 0x78, 0x61, 0x6d, 0x70,
+ 0x0c, 0x07, 0x30, 0x65, 0x78, 0x61, 0x6d, 0x70,
  0x6c, 0x65, 0x73, 0x2f, 0x68, 0x65, 0x6c, 0x6c,
  0x6f, 0x5f, 0x6d, 0x6f, 0x64, 0x75, 0x6c, 0x65,
  0x2e, 0x6a, 0x73, 0x1e, 0x2e, 0x2f, 0x66, 0x69,
@@ -60,7 +60,7 @@ const uint8_t qjsc_hello_module[177] = {
  0x6f, 0x20, 0x57, 0x6f, 0x72, 0x6c, 0x64, 0x10,
  0x66, 0x69, 0x62, 0x28, 0x31, 0x30, 0x29, 0x3d,
  0x0d, 0xb2, 0x03, 0x01, 0xb4, 0x03, 0x00, 0x00,
- 0x01, 0x00, 0xb6, 0x03, 0x00, 0x0c, 0x20, 0x02,
+ 0x01, 0x00, 0xb6, 0x03, 0x00, 0x0c, 0x20, 0xfa,
  0x01, 0x9e, 0x01, 0x00, 0x00, 0x00, 0x05, 0x01,
  0x00, 0x32, 0x00, 0xb6, 0x03, 0x00, 0x0c, 0x08,
  0xe9, 0x02, 0x29, 0x38, 0xdc, 0x00, 0x00, 0x00,

gen/test_fib.c

@@ -5,7 +5,7 @@
 const uint32_t qjsc_test_fib_size = 166;
 
 const uint8_t qjsc_test_fib[166] = {
- 0x0b, 0x07, 0x28, 0x65, 0x78, 0x61, 0x6d, 0x70,
+ 0x0c, 0x07, 0x28, 0x65, 0x78, 0x61, 0x6d, 0x70,
  0x6c, 0x65, 0x73, 0x2f, 0x74, 0x65, 0x73, 0x74,
  0x5f, 0x66, 0x69, 0x62, 0x2e, 0x6a, 0x73, 0x10,
  0x2e, 0x2f, 0x66, 0x69, 0x62, 0x2e, 0x73, 0x6f,
@@ -15,7 +15,7 @@ const uint8_t qjsc_test_fib[166] = {
  0x6f, 0x72, 0x6c, 0x64, 0x10, 0x66, 0x69, 0x62,
  0x28, 0x31, 0x30, 0x29, 0x3d, 0x0d, 0xb2, 0x03,
  0x01, 0xb4, 0x03, 0x00, 0x00, 0x01, 0x00, 0xb6,
- 0x03, 0x00, 0x0c, 0x20, 0x02, 0x01, 0x9e, 0x01,
+ 0x03, 0x00, 0x0c, 0x20, 0xfa, 0x01, 0x9e, 0x01,
  0x00, 0x00, 0x00, 0x05, 0x01, 0x00, 0x32, 0x00,
  0xb6, 0x03, 0x00, 0x0c, 0x08, 0xe9, 0x02, 0x29,
  0x38, 0xdc, 0x00, 0x00, 0x00, 0x42, 0xdd, 0x00,

qjsc.c

@@ -56,7 +56,7 @@ static namelist_t cmodule_list;
 static namelist_t init_module_list;
 static FILE *outfile;
 static const char *c_ident_prefix = "qjsc_";
-
+static int strip;
 
 void namelist_add(namelist_t *lp, const char *name, const char *short_name,
                   int flags)
@@ -155,8 +155,15 @@ static void output_object_code(JSContext *ctx,
 {
     uint8_t *out_buf;
     size_t out_buf_len;
+    int flags = JS_WRITE_OBJ_BYTECODE;
 
-    out_buf = JS_WriteObject(ctx, &out_buf_len, obj, JS_WRITE_OBJ_BYTECODE);
+    if (strip) {
+        flags |= JS_WRITE_OBJ_STRIP_SOURCE;
+        if (strip > 1)
+            flags |= JS_WRITE_OBJ_STRIP_DEBUG;
+    }
+
+    out_buf = JS_WriteObject(ctx, &out_buf_len, obj, flags);
     if (!out_buf) {
         js_std_dump_error(ctx);
         exit(1);
@@ -354,6 +361,7 @@ int main(int argc, char **argv)
     cname = NULL;
     module = -1;
     verbose = 0;
+    strip = 0;
     stack_size = 0;
     memset(&dynamic_module_list, 0, sizeof(dynamic_module_list));
 
@@ -362,7 +370,7 @@ int main(int argc, char **argv)
     namelist_add(&cmodule_list, "os", "os", 0);
 
     for(;;) {
-        c = getopt(argc, argv, "ho:N:mxevM:p:S:D:");
+        c = getopt(argc, argv, "ho:N:mxesvM:p:S:D:");
         if (c == -1)
             break;
         switch(c) {
@@ -399,6 +407,9 @@ int main(int argc, char **argv)
         case 'D':
             namelist_add(&dynamic_module_list, optarg, NULL, 0);
             break;
+        case 's':
+            strip++;
+            break;
         case 'v':
             verbose++;
             break;

quickjs.c

@@ -32601,7 +32601,7 @@ typedef enum BCTagEnum {
     BC_TAG_OBJECT_REFERENCE,
 } BCTagEnum;
 
-#define BC_VERSION 11
+#define BC_VERSION 12
 
 typedef struct BCWriterState {
     JSContext *ctx;
@@ -32609,6 +32609,8 @@ typedef struct BCWriterState {
     BOOL allow_bytecode : 8;
     BOOL allow_sab : 8;
     BOOL allow_reference : 8;
+    BOOL allow_source : 1;
+    BOOL allow_debug : 1;
     uint32_t first_atom;
     uint32_t *atom_to_idx;
     int atom_to_idx_size;
@@ -32971,6 +32973,7 @@ static int JS_WriteFunctionTag(BCWriterState *s, JSValue obj)
     bc_set_flags(&flags, &idx, b->super_allowed, 1);
     bc_set_flags(&flags, &idx, b->arguments_allowed, 1);
     bc_set_flags(&flags, &idx, b->backtrace_barrier, 1);
+    bc_set_flags(&flags, &idx, s->allow_debug, 1);
     assert(idx <= 16);
     bc_put_u16(s, flags);
     bc_put_u8(s, b->js_mode);
@@ -33027,13 +33030,19 @@ static int JS_WriteFunctionTag(BCWriterState *s, JSValue obj)
     if (JS_WriteFunctionBytecode(s, b))
         goto fail;
 
+    if (s->allow_debug) {
         bc_put_atom(s, b->filename);
         bc_put_leb128(s, b->line_num);
         bc_put_leb128(s, b->col_num);
         bc_put_leb128(s, b->pc2line_len);
         dbuf_put(&s->dbuf, b->pc2line_buf, b->pc2line_len);
+        if (s->allow_source && b->source) {
             bc_put_leb128(s, b->source_len);
             dbuf_put(&s->dbuf, b->source, b->source_len);
+        } else {
+            bc_put_leb128(s, 0);
+        }
+    }
     return 0;
  fail:
     return -1;
@@ -33409,6 +33418,8 @@ uint8_t *JS_WriteObject2(JSContext *ctx, size_t *psize, JSValue obj,
     s->allow_bytecode = ((flags & JS_WRITE_OBJ_BYTECODE) != 0);
     s->allow_sab = ((flags & JS_WRITE_OBJ_SAB) != 0);
     s->allow_reference = ((flags & JS_WRITE_OBJ_REFERENCE) != 0);
+    s->allow_source = ((flags & JS_WRITE_OBJ_STRIP_SOURCE) == 0);
+    s->allow_debug = ((flags & JS_WRITE_OBJ_STRIP_DEBUG) == 0);
     /* XXX: could use a different version when bytecode is included */
     if (s->allow_bytecode)
         s->first_atom = JS_ATOM_END;
@@ -33864,7 +33875,7 @@ static JSValue JS_ReadFunctionTag(BCReaderState *s)
     JSValue obj = JS_UNDEFINED;
     uint16_t v16;
     uint8_t v8;
-    int idx, i, local_count;
+    int idx, i, local_count, has_debug_info;
     int function_size, cpool_offset, byte_code_offset;
     int closure_var_offset, vardefs_offset;
     uint32_t ic_len;
@@ -33887,6 +33898,7 @@ static JSValue JS_ReadFunctionTag(BCReaderState *s)
     bc.super_allowed = bc_get_flags(v16, &idx, 1);
     bc.arguments_allowed = bc_get_flags(v16, &idx, 1);
     bc.backtrace_barrier = bc_get_flags(v16, &idx, 1);
+    has_debug_info = bc_get_flags(v16, &idx, 1);
     if (bc_get_u8(s, &v8))
         goto fail;
     bc.js_mode = v8;
@@ -34041,6 +34053,9 @@ static JSValue JS_ReadFunctionTag(BCReaderState *s)
             goto fail;
         bc_read_trace(s, "}\n");
     }
+    if (!has_debug_info)
+        goto nodebug;
+
     /* read optional debug information */
     bc_read_trace(s, "debug {\n");
     if (bc_get_atom(s, &b->filename))
@@ -34078,8 +34093,11 @@ static JSValue JS_ReadFunctionTag(BCReaderState *s)
             goto fail;
     }
     bc_read_trace(s, "}\n");
+
+ nodebug:
     b->realm = JS_DupContext(ctx);
     return obj;
+
  fail:
     JS_FreeAtom(ctx, bc.func_name);
     JS_FreeValue(ctx, obj);
@@ -36510,6 +36528,7 @@ static JSValue js_function_toString(JSContext *ctx, JSValue this_val,
     p = JS_VALUE_GET_OBJ(this_val);
     if (js_class_has_bytecode(p->class_id)) {
         JSFunctionBytecode *b = p->u.func.function_bytecode;
+        if (b->source)
             return JS_NewStringLen(ctx, b->source, b->source_len);
     }
     {

quickjs.h

@@ -816,9 +816,9 @@ JS_EXTERN int JS_ExecutePendingJob(JSRuntime *rt, JSContext **pctx);
 #define JS_WRITE_OBJ_BYTECODE  (1 << 0) /* allow function/module */
 #define JS_WRITE_OBJ_BSWAP     (0)      /* byte swapped output (obsolete, handled transparently) */
 #define JS_WRITE_OBJ_SAB       (1 << 2) /* allow SharedArrayBuffer */
-#define JS_WRITE_OBJ_REFERENCE (1 << 3) /* allow object references to
+#define JS_WRITE_OBJ_REFERENCE (1 << 3) /* allow object references to encode arbitrary object graph */
-                                           encode arbitrary object
+#define JS_WRITE_OBJ_STRIP_SOURCE  (1 << 4) /* do not write source code information */
-                                           graph */
+#define JS_WRITE_OBJ_STRIP_DEBUG   (1 << 5) /* do not write debug information */
 JS_EXTERN uint8_t *JS_WriteObject(JSContext *ctx, size_t *psize, JSValue obj, int flags);
 JS_EXTERN uint8_t *JS_WriteObject2(JSContext *ctx, size_t *psize, JSValue obj,
                                    int flags, uint8_t ***psab_tab, size_t *psab_tab_len);