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
5#ifndef V8_JSON_PARSER_H_
6#define V8_JSON_PARSER_H_
7
8#include "src/heap/factory.h"
9#include "src/isolate.h"
10#include "src/objects.h"
11#include "src/zone/zone-containers.h"
12
13namespace v8 {
14namespace internal {
15
16enum ParseElementResult { kElementFound, kElementNotFound, kNullHandle };
17
18class JsonParseInternalizer {
19 public:
20 static MaybeHandle<Object> Internalize(Isolate* isolate,
21 Handle<Object> object,
22 Handle<Object> reviver);
23
24 private:
25 JsonParseInternalizer(Isolate* isolate, Handle<JSReceiver> reviver)
26 : isolate_(isolate), reviver_(reviver) {}
27
28 MaybeHandle<Object> InternalizeJsonProperty(Handle<JSReceiver> holder,
29 Handle<String> key);
30
31 bool RecurseAndApply(Handle<JSReceiver> holder, Handle<String> name);
32
33 Isolate* isolate_;
34 Handle<JSReceiver> reviver_;
35};
36
37// A simple json parser.
38template <bool seq_one_byte>
39class JsonParser {
40 public:
41 V8_WARN_UNUSED_RESULT static MaybeHandle<Object> Parse(
42 Isolate* isolate, Handle<String> source, Handle<Object> reviver) {
43 Handle<Object> result;
44 ASSIGN_RETURN_ON_EXCEPTION(isolate, result,
45 JsonParser(isolate, source).ParseJson(), Object);
46 if (reviver->IsCallable()) {
47 return JsonParseInternalizer::Internalize(isolate, result, reviver);
48 }
49 return result;
50 }
51
52 static const int kEndOfString = -1;
53
54 private:
55 JsonParser(Isolate* isolate, Handle<String> source);
56
57 // Parse a string containing a single JSON value.
58 MaybeHandle<Object> ParseJson();
59
60 V8_INLINE void Advance();
61
62 // The JSON lexical grammar is specified in the ECMAScript 5 standard,
63 // section 15.12.1.1. The only allowed whitespace characters between tokens
64 // are tab, carriage-return, newline and space.
65
66 V8_INLINE void AdvanceSkipWhitespace();
67 V8_INLINE void SkipWhitespace();
68 V8_INLINE uc32 AdvanceGetChar();
69
70 // Checks that current charater is c.
71 // If so, then consume c and skip whitespace.
72 V8_INLINE bool MatchSkipWhiteSpace(uc32 c);
73
74 // A JSON string (production JSONString) is subset of valid JavaScript string
75 // literals. The string must only be double-quoted (not single-quoted), and
76 // the only allowed backslash-escapes are ", /, \, b, f, n, r, t and
77 // four-digit hex escapes (uXXXX). Any other use of backslashes is invalid.
78 bool ParseJsonString(Handle<String> expected);
79
80 Handle<String> ParseJsonString() {
81 Handle<String> result = ScanJsonString();
82 if (result.is_null()) return result;
83 return factory()->InternalizeString(result);
84 }
85
86 Handle<String> ScanJsonString();
87 // Creates a new string and copies prefix[start..end] into the beginning
88 // of it. Then scans the rest of the string, adding characters after the
89 // prefix. Called by ScanJsonString when reaching a '\' or non-Latin1 char.
90 template <typename StringType, typename SinkChar>
91 Handle<String> SlowScanJsonString(Handle<String> prefix, int start, int end);
92
93 // A JSON number (production JSONNumber) is a subset of the valid JavaScript
94 // decimal number literals.
95 // It includes an optional minus sign, must have at least one
96 // digit before and after a decimal point, may not have prefixed zeros (unless
97 // the integer part is zero), and may include an exponent part (e.g., "e-10").
98 // Hexadecimal and octal numbers are not allowed.
99 Handle<Object> ParseJsonNumber();
100
101 // Parse a single JSON value from input (grammar production JSONValue).
102 // A JSON value is either a (double-quoted) string literal, a number literal,
103 // one of "true", "false", or "null", or an object or array literal.
104 Handle<Object> ParseJsonValue();
105
106 // Parse a JSON object literal (grammar production JSONObject).
107 // An object literal is a squiggly-braced and comma separated sequence
108 // (possibly empty) of key/value pairs, where the key is a JSON string
109 // literal, the value is a JSON value, and the two are separated by a colon.
110 // A JSON array doesn't allow numbers and identifiers as keys, like a
111 // JavaScript array.
112 Handle<Object> ParseJsonObject();
113
114 // Helper for ParseJsonObject. Parses the form "123": obj, which is recorded
115 // as an element, not a property.
116 ParseElementResult ParseElement(Handle<JSObject> json_object);
117
118 // Parses a JSON array literal (grammar production JSONArray). An array
119 // literal is a square-bracketed and comma separated sequence (possibly empty)
120 // of JSON values.
121 // A JSON array doesn't allow leaving out values from the sequence, nor does
122 // it allow a terminal comma, like a JavaScript array does.
123 Handle<Object> ParseJsonArray();
124
125
126 // Mark that a parsing error has happened at the current token, and
127 // return a null handle. Primarily for readability.
128 inline Handle<Object> ReportUnexpectedCharacter() {
129 return Handle<Object>::null();
130 }
131
132 inline Isolate* isolate() { return isolate_; }
133 inline Factory* factory() { return isolate_->factory(); }
134 inline Handle<JSFunction> object_constructor() { return object_constructor_; }
135
136 static const int kInitialSpecialStringLength = 32;
137 static const int kPretenureTreshold = 100 * 1024;
138
139 private:
140 Zone* zone() { return &zone_; }
141
142 void CommitStateToJsonObject(Handle<JSObject> json_object, Handle<Map> map,
143 Vector<const Handle<Object>> properties);
144
145 Handle<String> source_;
146 int source_length_;
147 Handle<SeqOneByteString> seq_source_;
148
149 AllocationType allocation_;
150 Isolate* isolate_;
151 Zone zone_;
152 Handle<JSFunction> object_constructor_;
153 uc32 c0_;
154 int position_;
155
156 // Property handles are stored here inside ParseJsonObject.
157 ZoneVector<Handle<Object>> properties_;
158};
159
160// Explicit instantiation declarations.
161extern template class JsonParser<true>;
162extern template class JsonParser<false>;
163
164} // namespace internal
165} // namespace v8
166
167#endif // V8_JSON_PARSER_H_
168