// Copyright 2011 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. #ifndef V8_JSON_PARSER_H_ #define V8_JSON_PARSER_H_ #include "src/factory.h" #include "src/objects.h" namespace v8 { namespace internal { enum ParseElementResult { kElementFound, kElementNotFound, kNullHandle }; class JsonParseInternalizer BASE_EMBEDDED { public: static MaybeHandle Internalize(Isolate* isolate, Handle object, Handle reviver); private: JsonParseInternalizer(Isolate* isolate, Handle reviver) : isolate_(isolate), reviver_(reviver) {} MaybeHandle InternalizeJsonProperty(Handle holder, Handle key); bool RecurseAndApply(Handle holder, Handle name); Isolate* isolate_; Handle reviver_; }; // A simple json parser. template class JsonParser BASE_EMBEDDED { public: MUST_USE_RESULT static MaybeHandle Parse(Isolate* isolate, Handle source, Handle reviver) { Handle result; ASSIGN_RETURN_ON_EXCEPTION(isolate, result, JsonParser(isolate, source).ParseJson(), Object); if (reviver->IsCallable()) { return JsonParseInternalizer::Internalize(isolate, result, reviver); } return result; } static const int kEndOfString = -1; private: JsonParser(Isolate* isolate, Handle source); // Parse a string containing a single JSON value. MaybeHandle ParseJson(); INLINE(void Advance()); // The JSON lexical grammar is specified in the ECMAScript 5 standard, // section 15.12.1.1. The only allowed whitespace characters between tokens // are tab, carriage-return, newline and space. INLINE(void AdvanceSkipWhitespace()); INLINE(void SkipWhitespace()); INLINE(uc32 AdvanceGetChar()); // Checks that current charater is c. // If so, then consume c and skip whitespace. INLINE(bool MatchSkipWhiteSpace(uc32 c)); // A JSON string (production JSONString) is subset of valid JavaScript string // literals. The string must only be double-quoted (not single-quoted), and // the only allowed backslash-escapes are ", /, \, b, f, n, r, t and // four-digit hex escapes (uXXXX). Any other use of backslashes is invalid. Handle ParseJsonString() { return ScanJsonString(); } bool ParseJsonString(Handle expected); Handle ParseJsonInternalizedString() { Handle result = ScanJsonString(); if (result.is_null()) return result; return factory()->InternalizeString(result); } template Handle ScanJsonString(); // Creates a new string and copies prefix[start..end] into the beginning // of it. Then scans the rest of the string, adding characters after the // prefix. Called by ScanJsonString when reaching a '\' or non-Latin1 char. template Handle SlowScanJsonString(Handle prefix, int start, int end); // A JSON number (production JSONNumber) is a subset of the valid JavaScript // decimal number literals. // It includes an optional minus sign, must have at least one // digit before and after a decimal point, may not have prefixed zeros (unless // the integer part is zero), and may include an exponent part (e.g., "e-10"). // Hexadecimal and octal numbers are not allowed. Handle ParseJsonNumber(); // Parse a single JSON value from input (grammar production JSONValue). // A JSON value is either a (double-quoted) string literal, a number literal, // one of "true", "false", or "null", or an object or array literal. Handle ParseJsonValue(); // Parse a JSON object literal (grammar production JSONObject). // An object literal is a squiggly-braced and comma separated sequence // (possibly empty) of key/value pairs, where the key is a JSON string // literal, the value is a JSON value, and the two are separated by a colon. // A JSON array doesn't allow numbers and identifiers as keys, like a // JavaScript array. Handle ParseJsonObject(); // Helper for ParseJsonObject. Parses the form "123": obj, which is recorded // as an element, not a property. ParseElementResult ParseElement(Handle json_object); // Parses a JSON array literal (grammar production JSONArray). An array // literal is a square-bracketed and comma separated sequence (possibly empty) // of JSON values. // A JSON array doesn't allow leaving out values from the sequence, nor does // it allow a terminal comma, like a JavaScript array does. Handle ParseJsonArray(); // Mark that a parsing error has happened at the current token, and // return a null handle. Primarily for readability. inline Handle ReportUnexpectedCharacter() { return Handle::null(); } inline Isolate* isolate() { return isolate_; } inline Factory* factory() { return factory_; } inline Handle object_constructor() { return object_constructor_; } static const int kInitialSpecialStringLength = 32; static const int kPretenureTreshold = 100 * 1024; private: Zone* zone() { return &zone_; } void CommitStateToJsonObject(Handle json_object, Handle map, ZoneList >* properties); Handle source_; int source_length_; Handle seq_source_; PretenureFlag pretenure_; Isolate* isolate_; Factory* factory_; Zone zone_; Handle object_constructor_; uc32 c0_; int position_; }; } // namespace internal } // namespace v8 #endif // V8_JSON_PARSER_H_