Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 1 | // Copyright 2011 the V8 project authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 5 | #ifndef V8_JSON_PARSER_H_ |
| 6 | #define V8_JSON_PARSER_H_ |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 7 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 8 | #include "src/factory.h" |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 9 | #include "src/objects.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 10 | |
| 11 | namespace v8 { |
| 12 | namespace internal { |
| 13 | |
| 14 | enum ParseElementResult { kElementFound, kElementNotFound, kNullHandle }; |
| 15 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 16 | class JsonParseInternalizer BASE_EMBEDDED { |
| 17 | public: |
| 18 | static MaybeHandle<Object> Internalize(Isolate* isolate, |
| 19 | Handle<Object> object, |
| 20 | Handle<Object> reviver); |
| 21 | |
| 22 | private: |
| 23 | JsonParseInternalizer(Isolate* isolate, Handle<JSReceiver> reviver) |
| 24 | : isolate_(isolate), reviver_(reviver) {} |
| 25 | |
| 26 | MaybeHandle<Object> InternalizeJsonProperty(Handle<JSReceiver> holder, |
| 27 | Handle<String> key); |
| 28 | |
| 29 | bool RecurseAndApply(Handle<JSReceiver> holder, Handle<String> name); |
| 30 | |
| 31 | Isolate* isolate_; |
| 32 | Handle<JSReceiver> reviver_; |
| 33 | }; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 34 | |
| 35 | // A simple json parser. |
| 36 | template <bool seq_one_byte> |
| 37 | class JsonParser BASE_EMBEDDED { |
| 38 | public: |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 39 | MUST_USE_RESULT static MaybeHandle<Object> Parse(Isolate* isolate, |
| 40 | Handle<String> source, |
| 41 | Handle<Object> reviver) { |
| 42 | Handle<Object> result; |
| 43 | ASSIGN_RETURN_ON_EXCEPTION(isolate, result, |
| 44 | JsonParser(isolate, source).ParseJson(), Object); |
| 45 | if (reviver->IsCallable()) { |
| 46 | return JsonParseInternalizer::Internalize(isolate, result, reviver); |
| 47 | } |
| 48 | return result; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 49 | } |
| 50 | |
| 51 | static const int kEndOfString = -1; |
| 52 | |
| 53 | private: |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 54 | JsonParser(Isolate* isolate, Handle<String> source); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 55 | |
| 56 | // Parse a string containing a single JSON value. |
| 57 | MaybeHandle<Object> ParseJson(); |
| 58 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 59 | INLINE(void Advance()); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 60 | |
| 61 | // The JSON lexical grammar is specified in the ECMAScript 5 standard, |
| 62 | // section 15.12.1.1. The only allowed whitespace characters between tokens |
| 63 | // are tab, carriage-return, newline and space. |
| 64 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 65 | INLINE(void AdvanceSkipWhitespace()); |
| 66 | INLINE(void SkipWhitespace()); |
| 67 | INLINE(uc32 AdvanceGetChar()); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 68 | |
| 69 | // Checks that current charater is c. |
| 70 | // If so, then consume c and skip whitespace. |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 71 | INLINE(bool MatchSkipWhiteSpace(uc32 c)); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 72 | |
| 73 | // A JSON string (production JSONString) is subset of valid JavaScript string |
| 74 | // literals. The string must only be double-quoted (not single-quoted), and |
| 75 | // the only allowed backslash-escapes are ", /, \, b, f, n, r, t and |
| 76 | // four-digit hex escapes (uXXXX). Any other use of backslashes is invalid. |
| 77 | Handle<String> ParseJsonString() { |
| 78 | return ScanJsonString<false>(); |
| 79 | } |
| 80 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 81 | bool ParseJsonString(Handle<String> expected); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 82 | |
| 83 | Handle<String> ParseJsonInternalizedString() { |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 84 | Handle<String> result = ScanJsonString<true>(); |
| 85 | if (result.is_null()) return result; |
| 86 | return factory()->InternalizeString(result); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 87 | } |
| 88 | |
| 89 | template <bool is_internalized> |
| 90 | Handle<String> ScanJsonString(); |
| 91 | // Creates a new string and copies prefix[start..end] into the beginning |
| 92 | // of it. Then scans the rest of the string, adding characters after the |
| 93 | // prefix. Called by ScanJsonString when reaching a '\' or non-Latin1 char. |
| 94 | template <typename StringType, typename SinkChar> |
| 95 | Handle<String> SlowScanJsonString(Handle<String> prefix, int start, int end); |
| 96 | |
| 97 | // A JSON number (production JSONNumber) is a subset of the valid JavaScript |
| 98 | // decimal number literals. |
| 99 | // It includes an optional minus sign, must have at least one |
| 100 | // digit before and after a decimal point, may not have prefixed zeros (unless |
| 101 | // the integer part is zero), and may include an exponent part (e.g., "e-10"). |
| 102 | // Hexadecimal and octal numbers are not allowed. |
| 103 | Handle<Object> ParseJsonNumber(); |
| 104 | |
| 105 | // Parse a single JSON value from input (grammar production JSONValue). |
| 106 | // A JSON value is either a (double-quoted) string literal, a number literal, |
| 107 | // one of "true", "false", or "null", or an object or array literal. |
| 108 | Handle<Object> ParseJsonValue(); |
| 109 | |
| 110 | // Parse a JSON object literal (grammar production JSONObject). |
| 111 | // An object literal is a squiggly-braced and comma separated sequence |
| 112 | // (possibly empty) of key/value pairs, where the key is a JSON string |
| 113 | // literal, the value is a JSON value, and the two are separated by a colon. |
| 114 | // A JSON array doesn't allow numbers and identifiers as keys, like a |
| 115 | // JavaScript array. |
| 116 | Handle<Object> ParseJsonObject(); |
| 117 | |
| 118 | // Helper for ParseJsonObject. Parses the form "123": obj, which is recorded |
| 119 | // as an element, not a property. |
| 120 | ParseElementResult ParseElement(Handle<JSObject> json_object); |
| 121 | |
| 122 | // Parses a JSON array literal (grammar production JSONArray). An array |
| 123 | // literal is a square-bracketed and comma separated sequence (possibly empty) |
| 124 | // of JSON values. |
| 125 | // A JSON array doesn't allow leaving out values from the sequence, nor does |
| 126 | // it allow a terminal comma, like a JavaScript array does. |
| 127 | Handle<Object> ParseJsonArray(); |
| 128 | |
| 129 | |
| 130 | // Mark that a parsing error has happened at the current token, and |
| 131 | // return a null handle. Primarily for readability. |
| 132 | inline Handle<Object> ReportUnexpectedCharacter() { |
| 133 | return Handle<Object>::null(); |
| 134 | } |
| 135 | |
| 136 | inline Isolate* isolate() { return isolate_; } |
| 137 | inline Factory* factory() { return factory_; } |
| 138 | inline Handle<JSFunction> object_constructor() { return object_constructor_; } |
| 139 | |
| 140 | static const int kInitialSpecialStringLength = 32; |
| 141 | static const int kPretenureTreshold = 100 * 1024; |
| 142 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 143 | private: |
| 144 | Zone* zone() { return &zone_; } |
| 145 | |
| 146 | void CommitStateToJsonObject(Handle<JSObject> json_object, Handle<Map> map, |
| 147 | ZoneList<Handle<Object> >* properties); |
| 148 | |
| 149 | Handle<String> source_; |
| 150 | int source_length_; |
| 151 | Handle<SeqOneByteString> seq_source_; |
| 152 | |
| 153 | PretenureFlag pretenure_; |
| 154 | Isolate* isolate_; |
| 155 | Factory* factory_; |
| 156 | Zone zone_; |
| 157 | Handle<JSFunction> object_constructor_; |
| 158 | uc32 c0_; |
| 159 | int position_; |
| 160 | }; |
| 161 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 162 | } // namespace internal |
| 163 | } // namespace v8 |
| 164 | |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 165 | #endif // V8_JSON_PARSER_H_ |