added JSON.parse source text access

TODO

@@ -63,4 +63,4 @@ Test262o:   0/11262 errors, 463 excluded
 Test262o commit: 7da91bceb9ce7613f87db47ddd1292a2dda58b42 (es5-tests branch)
 
 Test262:
-Result: 60/83394 errors, 3348 excluded, 6090 skipped
+Result: 60/83436 errors, 3348 excluded, 6069 skipped

quickjs-atom.h

@@ -147,6 +147,7 @@ DEF(flags, "flags")
 DEF(global, "global")
 DEF(unicode, "unicode")
 DEF(raw, "raw")
+DEF(rawJSON, "rawJSON")
 DEF(new_target, "new.target")
 DEF(this_active_func, "this.active_func")
 DEF(home_object, "<home_object>")

quickjs.c

@@ -168,6 +168,7 @@ enum {
     JS_CLASS_REGEXP_STRING_ITERATOR,   /* u.regexp_string_iterator_data */
     JS_CLASS_GENERATOR,         /* u.generator_data */
     JS_CLASS_GLOBAL_OBJECT,     /* u.global_object */
+    JS_CLASS_RAWJSON,
     JS_CLASS_PROXY,             /* u.proxy_data */
     JS_CLASS_PROMISE,           /* u.promise_data */
     JS_CLASS_PROMISE_RESOLVE_FUNCTION,  /* u.promise_function_data */
@@ -1597,6 +1598,7 @@ static JSClassShortDef const js_std_class_def[] = {
     { JS_ATOM_RegExp_String_Iterator, js_regexp_string_iterator_finalizer, js_regexp_string_iterator_mark }, /* JS_CLASS_REGEXP_STRING_ITERATOR */
     { JS_ATOM_Generator, js_generator_finalizer, js_generator_mark }, /* JS_CLASS_GENERATOR */
     { JS_ATOM_Object, js_global_object_finalizer, js_global_object_mark }, /* JS_CLASS_GLOBAL_OBJECT */
+    { JS_ATOM_Object, NULL, NULL }, /* JS_CLASS_RAWJSON */
 };
 
 static int init_class_range(JSRuntime *rt, JSClassShortDef const *tab,
@@ -21571,7 +21573,7 @@ typedef struct JSParseState {
     JSFunctionDef *cur_func;
     BOOL is_module; /* parsing a module */
     BOOL allow_html_comments;
-    BOOL ext_json; /* true if accepting JSON superset */
+    BOOL ext_json; /* JSON parsing: true if accepting JSON superset */
     GetLineColCache get_line_col_cache;
 } JSParseState;
 
@@ -48629,23 +48631,189 @@ static int json_parse_expect(JSParseState *s, int tok)
     return json_next_token(s);
 }
 
-static JSValue json_parse_value(JSParseState *s)
+
+typedef struct {
+    int count;
+    uint32_t hash_size;
+    struct JSONParseRecordEntry *entries;
+    uint32_t *hash_table;
+} JSONParseRecordObject;
+    
+typedef struct JSONParseRecord {
+    JSValue value;
+    union {
+        JSONParseRecordObject obj;
+        struct {
+            int count;
+            struct JSONParseRecord *elements;
+        } array;
+        struct {
+            uint32_t source_pos;
+            uint32_t source_len;
+        } primitive;
+    } u;
+} JSONParseRecord;
+
+typedef struct JSONParseRecordEntry {
+    JSAtom atom;
+    uint32_t hash_next;
+    JSONParseRecord parse_record;
+} JSONParseRecordEntry;
+
+static void json_parse_record_init_obj(JSContext *ctx, JSONParseRecord *pr, JSValueConst val)
+{
+    pr->value = JS_DupValue(ctx, val);
+    pr->u.obj.count = 0;
+    pr->u.obj.entries = NULL;
+    pr->u.obj.hash_table = NULL;
+    pr->u.obj.hash_size = 0;
+}
+
+static void json_parse_record_init_array(JSContext *ctx, JSONParseRecord *pr, JSValueConst val)
+{
+    pr->value = JS_DupValue(ctx, val);
+    pr->u.array.count = 0;
+    pr->u.array.elements = NULL;
+}
+
+static void json_parse_record_init_primitive(JSContext *ctx, JSONParseRecord *pr, JSValueConst val,
+                                             uint32_t source_pos, uint32_t source_len)
+{
+    pr->value = JS_DupValue(ctx, val);
+    pr->u.primitive.source_pos = source_pos;
+    pr->u.primitive.source_len = source_len;
+}
+
+static int json_parse_record_resize_hash(JSContext *ctx, JSONParseRecordObject *po, uint32_t new_hash_size)
+{
+    uint32_t i, h, *new_hash_table;
+    JSONParseRecordEntry *e;
+
+    new_hash_table = js_malloc(ctx, sizeof(new_hash_table[0]) * new_hash_size);
+    if (!new_hash_table)
+        return -1;
+    js_free(ctx, po->hash_table);
+    po->hash_table = new_hash_table;
+    po->hash_size = new_hash_size;
+
+    for(i = 0; i < po->hash_size; i++) {
+        po->hash_table[i] = -1;
+    }
+    for(i = 0; i < po->count; i++) {
+        e = &po->entries[i];
+        h = e->atom & (po->hash_size - 1);
+        e->hash_next = po->hash_table[h];
+        po->hash_table[h] = i;
+    }
+    return 0;
+}
+
+static JSONParseRecord *json_parse_record_add(JSContext *ctx, JSONParseRecord *pr, JSAtom key, int *psize)
+{
+    JSONParseRecordObject *po = &pr->u.obj;
+    JSONParseRecordEntry *e;
+    JSONParseRecord *pr1;
+    uint32_t h;
+    
+    if (js_resize_array(ctx, (void **)&po->entries, sizeof(po->entries[0]),
+                        psize, po->count + 1)) {
+        return NULL;
+    }
+    /* don't use a hash table when the number of entries is small */
+    if (po->count >= 8 && (po->count + 1) > po->hash_size) {
+        int hash_bits = 32 - clz32(po->count);
+        if (json_parse_record_resize_hash(ctx, po, 1 << hash_bits))
+            return NULL;
+    }
+
+    e = &po->entries[po->count++];
+    e->atom = JS_DupAtom(ctx, key);
+    pr1 = &e->parse_record;
+    pr1->value = JS_UNDEFINED;
+    if (po->hash_size != 0) {
+        h = key & (po->hash_size - 1);
+        e->hash_next = po->hash_table[h];
+        po->hash_table[h] = po->count - 1;
+    }
+    return pr1;
+}
+
+static JSONParseRecord *json_parse_record_find(JSONParseRecord *pr, JSAtom key)
+{
+    JSONParseRecordObject *po = &pr->u.obj;
+    JSONParseRecordEntry *e;
+    uint32_t h, i;
+    
+    if (po->hash_size == 0) {
+        for(i = 0; i < po->count; i++) {
+            if (po->entries[i].atom == key)
+                return &po->entries[i].parse_record;
+        }
+    } else {
+        h = key & (po->hash_size - 1);
+        i = po->hash_table[h];
+        while (i != -1) {
+            e = &po->entries[i];
+            if (e->atom == key)
+                return &e->parse_record;
+            i = e->hash_next;
+        }
+    }
+    return NULL;
+}
+
+static void json_free_parse_record(JSContext *ctx, JSONParseRecord *pr)
+{
+    int i;
+    if (!pr)
+        return;
+    if (JS_IsObject(pr->value)) {
+        if (JS_IsArray(ctx, pr->value)) {
+            for(i = 0; i < pr->u.array.count; i++) {
+                json_free_parse_record(ctx, &pr->u.array.elements[i]);
+            }
+            js_free(ctx, pr->u.array.elements);
+        } else {
+            for(i = 0; i < pr->u.obj.count; i++) {
+                JS_FreeAtom(ctx, pr->u.obj.entries[i].atom);
+                json_free_parse_record(ctx, &pr->u.obj.entries[i].parse_record);
+            }
+            js_free(ctx, pr->u.obj.entries);
+            js_free(ctx, pr->u.obj.hash_table);
+        }
+    }
+    JS_FreeValue(ctx, pr->value);
+    pr->value = JS_UNDEFINED; /* fail safe */
+}
+
+/* 'pr' can be NULL */
+static JSValue json_parse_value(JSParseState *s, JSONParseRecord *pr)
 {
     JSContext *ctx = s->ctx;
     JSValue val = JS_NULL;
     int ret;
 
+    if (pr) {
+        pr->value = JS_UNDEFINED;
+    }
+    
     switch(s->token.val) {
     case '{':
         {
             JSValue prop_val;
             JSAtom prop_name;
-
+            JSONParseRecord *pr1;
+            int pr_size;
+            
             if (json_next_token(s))
                 goto fail;
             val = JS_NewObject(ctx);
             if (JS_IsException(val))
                 goto fail;
+            if (pr) {
+                json_parse_record_init_obj(ctx, pr, val);
+                pr_size = 0;
+            }
             if (s->token.val != '}') {
                 for(;;) {
                     if (s->token.val == TOK_STRING) {
@@ -48662,7 +48830,14 @@ static JSValue json_parse_value(JSParseState *s)
                         goto fail1;
                     if (json_parse_expect(s, ':'))
                         goto fail1;
-                    prop_val = json_parse_value(s);
+                    if (pr) {
+                        pr1 = json_parse_record_add(ctx, pr, prop_name, &pr_size);
+                        if (!pr1)
+                            goto fail1;
+                    } else {
+                        pr1 = NULL;
+                    }
+                    prop_val = json_parse_value(s, pr1);
                     if (JS_IsException(prop_val)) {
                     fail1:
                         JS_FreeAtom(ctx, prop_name);
@@ -48690,16 +48865,31 @@ static JSValue json_parse_value(JSParseState *s)
         {
             JSValue el;
             uint32_t idx;
-
+            JSONParseRecord *pr1;
+            int pr_size;
+            
             if (json_next_token(s))
                 goto fail;
             val = JS_NewArray(ctx);
             if (JS_IsException(val))
                 goto fail;
+            if (pr) {
+                json_parse_record_init_array(ctx, pr, val);
+                pr_size = 0;
+            }
             if (s->token.val != ']') {
                 idx = 0;
                 for(;;) {
-                    el = json_parse_value(s);
+                    if (pr) {
+                        if (js_resize_array(ctx, (void **)&pr->u.array.elements, sizeof(pr->u.array.elements[0]),
+                                            &pr_size, pr->u.array.count + 1))
+                            goto fail;
+                        pr1 = &pr->u.array.elements[pr->u.array.count++];
+                        pr1->value = JS_UNDEFINED;
+                    } else {
+                        pr1 = NULL;
+                    }
+                    el = json_parse_value(s, pr1);
                     if (JS_IsException(el))
                         goto fail;
                     ret = JS_DefinePropertyValueUint32(ctx, val, idx, el, JS_PROP_C_W_E);
@@ -48720,11 +48910,21 @@ static JSValue json_parse_value(JSParseState *s)
         break;
     case TOK_STRING:
         val = JS_DupValue(ctx, s->token.u.str.str);
+        if (pr) {
+            json_parse_record_init_primitive(ctx, pr, val, 
+                                             s->token.ptr - s->buf_start,
+                                             s->buf_ptr - s->token.ptr);
+        }
         if (json_next_token(s))
             goto fail;
         break;
     case TOK_NUMBER:
         val = s->token.u.num.val;
+        if (pr) {
+            json_parse_record_init_primitive(ctx, pr, val, 
+                                             s->token.ptr - s->buf_start,
+                                             s->buf_ptr - s->token.ptr);
+        }
         if (json_next_token(s))
             goto fail;
         break;
@@ -48732,8 +48932,18 @@ static JSValue json_parse_value(JSParseState *s)
         if (s->token.u.ident.atom == JS_ATOM_false ||
             s->token.u.ident.atom == JS_ATOM_true) {
             val = JS_NewBool(ctx, s->token.u.ident.atom == JS_ATOM_true);
+            if (pr) {
+                json_parse_record_init_primitive(ctx, pr, val, 
+                                                 s->token.ptr - s->buf_start,
+                                                 s->buf_ptr - s->token.ptr);
+            }
         } else if (s->token.u.ident.atom == JS_ATOM_null) {
             val = JS_NULL;
+            if (pr) {
+                json_parse_record_init_primitive(ctx, pr, val, 
+                                                 s->token.ptr - s->buf_start,
+                                                 s->buf_ptr - s->token.ptr);
+            }
         } else if (s->token.u.ident.atom == JS_ATOM_NaN && s->ext_json) {
             /* Note: json5 identifier handling is ambiguous e.g. is 
                '{ NaN: 1 }' a valid JSON5 production ? */ 
@@ -48758,12 +48968,13 @@ static JSValue json_parse_value(JSParseState *s)
     }
     return val;
  fail:
+    json_free_parse_record(ctx, pr);
     JS_FreeValue(ctx, val);
     return JS_EXCEPTION;
 }
 
-JSValue JS_ParseJSON2(JSContext *ctx, const char *buf, size_t buf_len,
-                      const char *filename, int flags)
+JSValue JS_ParseJSON3(JSContext *ctx, const char *buf, size_t buf_len,
+                      const char *filename, int flags, JSONParseRecord *pr)
 {
     JSParseState s1, *s = &s1;
     JSValue val = JS_UNDEFINED;
@@ -48772,12 +48983,14 @@ JSValue JS_ParseJSON2(JSContext *ctx, const char *buf, size_t buf_len,
     s->ext_json = ((flags & JS_PARSE_JSON_EXT) != 0);
     if (json_next_token(s))
         goto fail;
-    val = json_parse_value(s);
+    val = json_parse_value(s, pr);
     if (JS_IsException(val))
         goto fail;
     if (s->token.val != TOK_EOF) {
-        if (js_parse_error(s, "unexpected data at the end"))
+        if (js_parse_error(s, "unexpected data at the end")) {
+            json_free_parse_record(ctx, pr);
             goto fail;
+        }
     }
     return val;
  fail:
@@ -48786,17 +48999,25 @@ JSValue JS_ParseJSON2(JSContext *ctx, const char *buf, size_t buf_len,
     return JS_EXCEPTION;
 }
 
+JSValue JS_ParseJSON2(JSContext *ctx, const char *buf, size_t buf_len,
+                      const char *filename, int flags)
+{
+    return JS_ParseJSON3(ctx, buf, buf_len, filename, flags, NULL);
+}
+
 JSValue JS_ParseJSON(JSContext *ctx, const char *buf, size_t buf_len,
                      const char *filename)
 {
-    return JS_ParseJSON2(ctx, buf, buf_len, filename, 0);
+    return JS_ParseJSON3(ctx, buf, buf_len, filename, 0, NULL);
 }
 
+/* if pr != NULL, then pr->value = holder by construction */
 static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
-                                         JSAtom name, JSValueConst reviver)
+                                         JSAtom name, JSValueConst reviver,
+                                         const char *text_str, JSONParseRecord *pr)
 {
-    JSValue val, new_el, name_val, res;
-    JSValueConst args[2];
+    JSValue val, new_el, name_val, res, context;
+    JSValueConst args[3];
     int ret, is_array;
     uint32_t i, len = 0;
     JSAtom prop;
@@ -48809,6 +49030,29 @@ static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
     val = JS_GetProperty(ctx, holder, name);
     if (JS_IsException(val))
         return val;
+
+    if (pr) {
+        if (JS_IsArray(ctx, pr->value)) {
+            if (__JS_AtomIsTaggedInt(name)) {
+                uint32_t idx = __JS_AtomToUInt32(name);
+                if (idx < pr->u.array.count) {
+                    pr = &pr->u.array.elements[idx];
+                } else {
+                    pr = NULL;
+                }
+            }
+        } else {
+            pr = json_parse_record_find(pr, name);
+        }
+        if (pr && !js_same_value(ctx, pr->value, val)) {
+            pr = NULL;
+        }
+    }
+
+    context = JS_NewObject(ctx);
+    if (JS_IsException(context))
+        goto fail;
+    
     if (JS_IsObject(val)) {
         is_array = JS_IsArray(ctx, val);
         if (is_array < 0)
@@ -48829,7 +49073,7 @@ static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
             } else {
                 prop = JS_DupAtom(ctx, atoms[i].atom);
             }
-            new_el = internalize_json_property(ctx, val, prop, reviver);
+            new_el = internalize_json_property(ctx, val, prop, reviver, text_str, pr);
             if (JS_IsException(new_el)) {
                 JS_FreeAtom(ctx, prop);
                 goto fail;
@@ -48843,6 +49087,15 @@ static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
             if (ret < 0)
                 goto fail;
         }
+    } else {
+        if (pr) {
+            new_el = JS_NewStringLen(ctx, text_str + pr->u.primitive.source_pos,
+                                     pr->u.primitive.source_len);
+            if (JS_IsException(new_el))
+                goto fail;
+            if (JS_DefinePropertyValue(ctx, context, JS_ATOM_source, new_el, JS_PROP_C_W_E) < 0)
+                goto fail;
+        }
     }
     JS_FreePropertyEnum(ctx, atoms, len);
     atoms = NULL;
@@ -48851,12 +49104,15 @@ static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
         goto fail;
     args[0] = name_val;
     args[1] = val;
-    res = JS_Call(ctx, reviver, holder, 2, args);
+    args[2] = context;
+    res = JS_Call(ctx, reviver, holder, 3, args);
     JS_FreeValue(ctx, name_val);
     JS_FreeValue(ctx, val);
+    JS_FreeValue(ctx, context);
     return res;
  fail:
     JS_FreePropertyEnum(ctx, atoms, len);
+    JS_FreeValue(ctx, context);
     JS_FreeValue(ctx, val);
     return JS_EXCEPTION;
 }
@@ -48864,37 +49120,113 @@ static JSValue internalize_json_property(JSContext *ctx, JSValueConst holder,
 static JSValue js_json_parse(JSContext *ctx, JSValueConst this_val,
                              int argc, JSValueConst *argv)
 {
-    JSValue obj, root;
-    JSValueConst reviver;
+    JSValue obj;
     const char *str;
     size_t len;
-
+    
     str = JS_ToCStringLen(ctx, &len, argv[0]);
     if (!str)
         return JS_EXCEPTION;
-    obj = JS_ParseJSON(ctx, str, len, "<input>");
-    JS_FreeCString(ctx, str);
-    if (JS_IsException(obj))
-        return obj;
     if (argc > 1 && JS_IsFunction(ctx, argv[1])) {
+        JSONParseRecord pr_s, *pr = &pr_s, *pr1;
+        JSValue root;
+        JSValueConst reviver;
+        int size;
+        
         reviver = argv[1];
         root = JS_NewObject(ctx);
-        if (JS_IsException(root)) {
-            JS_FreeValue(ctx, obj);
-            return JS_EXCEPTION;
-        }
+        if (JS_IsException(root))
+            goto fail;
+        json_parse_record_init_obj(ctx, pr, root);
+        size = 0;
+        pr1 = json_parse_record_add(ctx, pr, JS_ATOM_empty_string, &size);
+        if (!pr1)
+            goto fail1;
+
+        obj = JS_ParseJSON3(ctx, str, len, "<input>", 0, pr1);
+        if (JS_IsException(obj))
+            goto fail1;
+        
         if (JS_DefinePropertyValue(ctx, root, JS_ATOM_empty_string, obj,
                                    JS_PROP_C_W_E) < 0) {
+            JS_FreeValue(ctx, obj);
+        fail1:
+            json_free_parse_record(ctx, pr);
             JS_FreeValue(ctx, root);
-            return JS_EXCEPTION;
+            goto fail;
         }
+        
         obj = internalize_json_property(ctx, root, JS_ATOM_empty_string,
-                                        reviver);
+                                        reviver, str, pr);
+        json_free_parse_record(ctx, pr);
         JS_FreeValue(ctx, root);
+    } else {
+        obj = JS_ParseJSON3(ctx, str, len, "<input>", 0, NULL);
     }
+    JS_FreeCString(ctx, str);
     return obj;
+ fail:
+    JS_FreeCString(ctx, str);
+    return JS_EXCEPTION;
 }
 
+static JSValue js_json_isRawJSON(JSContext *ctx, JSValueConst this_val,
+                                 int argc, JSValueConst *argv)
+{
+    JSValueConst obj = argv[0];
+    if (JS_VALUE_GET_TAG(obj) == JS_TAG_OBJECT) {
+        JSObject *p = JS_VALUE_GET_OBJ(obj);
+        return JS_NewBool(ctx, p->class_id == JS_CLASS_RAWJSON);
+    } else {
+        return JS_FALSE;
+    }
+}
+
+static BOOL is_valid_raw_json_char(int c)
+{
+    return ((c >= 'a' && c <= 'z') ||
+            (c >= '0' && c <= '9') ||
+            c == '-' || 
+            c == '"');
+}
+
+static JSValue js_json_rawJSON(JSContext *ctx, JSValueConst this_val,
+                                 int argc, JSValueConst *argv)
+{
+    JSValue str, res, obj;
+    JSString *p;
+    str = JS_ToString(ctx, argv[0]);
+    if (JS_IsException(str))
+        return str;
+    p = JS_VALUE_GET_STRING(str);
+    if (p->len == 0 ||
+        !is_valid_raw_json_char(string_get(p, 0)) ||
+        !is_valid_raw_json_char(string_get(p, p->len - 1))) {
+        goto syntax_error;
+    }
+    res = js_json_parse(ctx, JS_UNDEFINED, 1, (JSValueConst *)&str);
+    if (JS_IsException(res)) {
+    syntax_error:
+        JS_ThrowSyntaxError(ctx, "invalid rawJSON string");
+        goto fail;
+    }
+    JS_FreeValue(ctx, res);
+    
+    obj = JS_NewObjectProtoClass(ctx, JS_NULL, JS_CLASS_RAWJSON);
+    if (JS_IsException(obj))
+        goto fail;
+    if (JS_DefinePropertyValue(ctx, obj, JS_ATOM_rawJSON, str, JS_PROP_ENUMERABLE) < 0) {
+        JS_FreeValue(ctx, obj);
+        return JS_EXCEPTION;
+    }
+    JS_PreventExtensions(ctx, obj);
+    return obj;
+ fail:
+    JS_FreeValue(ctx, str);
+    return JS_EXCEPTION;
+}
+
+
 typedef struct JSONStringifyContext {
     JSValueConst replacer_func;
     JSValue stack;
@@ -49064,11 +49396,18 @@ static int js_json_to_str(JSContext *ctx, JSONStringifyContext *jsc,
             if (JS_IsException(val))
                 goto exception;
             goto concat_primitive;
-        } else if (cl == JS_CLASS_BOOLEAN || cl == JS_CLASS_BIG_INT)
-        {
+        } else if (cl == JS_CLASS_BOOLEAN || cl == JS_CLASS_BIG_INT) {
             /* This will thow the same error as for the primitive object */
             set_value(ctx, &val, JS_DupValue(ctx, p->u.object_data));
             goto concat_primitive;
+        } else if (cl == JS_CLASS_RAWJSON) {
+            JSValue val1;
+            val1 = JS_GetProperty(ctx, val, JS_ATOM_rawJSON);
+            if (JS_IsException(val1))
+                goto exception;
+            JS_FreeValue(ctx, val);
+            val = val1;
+            goto concat_value;
         }
         v = js_array_includes(ctx, jsc->stack, 1, (JSValueConst *)&val);
         if (JS_IsException(v))
@@ -49357,7 +49696,9 @@ static JSValue js_json_stringify(JSContext *ctx, JSValueConst this_val,
 }
 
 static const JSCFunctionListEntry js_json_funcs[] = {
+    JS_CFUNC_DEF("isRawJSON", 1, js_json_isRawJSON ),
     JS_CFUNC_DEF("parse", 2, js_json_parse ),
+    JS_CFUNC_DEF("rawJSON", 1, js_json_rawJSON ),
     JS_CFUNC_DEF("stringify", 3, js_json_stringify ),
     JS_PROP_STRING_DEF("[Symbol.toStringTag]", "JSON", JS_PROP_CONFIGURABLE ),
 };

test262.conf

@@ -150,7 +150,7 @@ IsHTMLDDA
 iterator-helpers
 iterator-sequencing
 json-modules
-json-parse-with-source=skip
+json-parse-with-source
 json-superset
 legacy-regexp=skip
 let