Wed, 31 Dec 2014 06:09:35 +0100
Cloned upstream origin tor-browser at tor-browser-31.3.0esr-4.5-1-build1
revision ID fc1c9ff7c1b2defdbc039f12214767608f46423f for hacking purpose.
michael@0 | 1 | /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */ |
michael@0 | 2 | /* This Source Code Form is subject to the terms of the Mozilla Public |
michael@0 | 3 | * License, v. 2.0. If a copy of the MPL was not distributed with this |
michael@0 | 4 | * file, You can obtain one at http://mozilla.org/MPL/2.0/. */ |
michael@0 | 5 | |
michael@0 | 6 | #ifndef CharDistribution_h__ |
michael@0 | 7 | #define CharDistribution_h__ |
michael@0 | 8 | |
michael@0 | 9 | #include "nscore.h" |
michael@0 | 10 | |
michael@0 | 11 | #define ENOUGH_DATA_THRESHOLD 1024 |
michael@0 | 12 | |
michael@0 | 13 | #define MINIMUM_DATA_THRESHOLD 4 |
michael@0 | 14 | |
michael@0 | 15 | class CharDistributionAnalysis |
michael@0 | 16 | { |
michael@0 | 17 | public: |
michael@0 | 18 | CharDistributionAnalysis() {Reset(false);} |
michael@0 | 19 | |
michael@0 | 20 | //feed a block of data and do distribution analysis |
michael@0 | 21 | void HandleData(const char* aBuf, uint32_t aLen) {} |
michael@0 | 22 | |
michael@0 | 23 | //Feed a character with known length |
michael@0 | 24 | void HandleOneChar(const char* aStr, uint32_t aCharLen) |
michael@0 | 25 | { |
michael@0 | 26 | int32_t order; |
michael@0 | 27 | |
michael@0 | 28 | //we only care about 2-bytes character in our distribution analysis |
michael@0 | 29 | order = (aCharLen == 2) ? GetOrder(aStr) : -1; |
michael@0 | 30 | |
michael@0 | 31 | if (order >= 0) |
michael@0 | 32 | { |
michael@0 | 33 | mTotalChars++; |
michael@0 | 34 | //order is valid |
michael@0 | 35 | if ((uint32_t)order < mTableSize) |
michael@0 | 36 | { |
michael@0 | 37 | if (512 > mCharToFreqOrder[order]) |
michael@0 | 38 | mFreqChars++; |
michael@0 | 39 | } |
michael@0 | 40 | } |
michael@0 | 41 | } |
michael@0 | 42 | |
michael@0 | 43 | //return confidence base on existing data |
michael@0 | 44 | float GetConfidence(void); |
michael@0 | 45 | |
michael@0 | 46 | //Reset analyser, clear any state |
michael@0 | 47 | void Reset(bool aIsPreferredLanguage) |
michael@0 | 48 | { |
michael@0 | 49 | mDone = false; |
michael@0 | 50 | mTotalChars = 0; |
michael@0 | 51 | mFreqChars = 0; |
michael@0 | 52 | mDataThreshold = aIsPreferredLanguage ? 0 : MINIMUM_DATA_THRESHOLD; |
michael@0 | 53 | } |
michael@0 | 54 | |
michael@0 | 55 | //It is not necessary to receive all data to draw conclusion. For charset detection, |
michael@0 | 56 | // certain amount of data is enough |
michael@0 | 57 | bool GotEnoughData() {return mTotalChars > ENOUGH_DATA_THRESHOLD;} |
michael@0 | 58 | |
michael@0 | 59 | protected: |
michael@0 | 60 | //we do not handle character base on its original encoding string, but |
michael@0 | 61 | //convert this encoding string to a number, here called order. |
michael@0 | 62 | //This allow multiple encoding of a language to share one frequency table |
michael@0 | 63 | virtual int32_t GetOrder(const char* str) {return -1;} |
michael@0 | 64 | |
michael@0 | 65 | //If this flag is set to true, detection is done and conclusion has been made |
michael@0 | 66 | bool mDone; |
michael@0 | 67 | |
michael@0 | 68 | //The number of characters whose frequency order is less than 512 |
michael@0 | 69 | uint32_t mFreqChars; |
michael@0 | 70 | |
michael@0 | 71 | //Total character encounted. |
michael@0 | 72 | uint32_t mTotalChars; |
michael@0 | 73 | |
michael@0 | 74 | //Number of hi-byte characters needed to trigger detection |
michael@0 | 75 | uint32_t mDataThreshold; |
michael@0 | 76 | |
michael@0 | 77 | //Mapping table to get frequency order from char order (get from GetOrder()) |
michael@0 | 78 | const int16_t *mCharToFreqOrder; |
michael@0 | 79 | |
michael@0 | 80 | //Size of above table |
michael@0 | 81 | uint32_t mTableSize; |
michael@0 | 82 | |
michael@0 | 83 | //This is a constant value varies from language to language, it is used in |
michael@0 | 84 | //calculating confidence. See my paper for further detail. |
michael@0 | 85 | float mTypicalDistributionRatio; |
michael@0 | 86 | }; |
michael@0 | 87 | |
michael@0 | 88 | |
michael@0 | 89 | class EUCTWDistributionAnalysis: public CharDistributionAnalysis |
michael@0 | 90 | { |
michael@0 | 91 | public: |
michael@0 | 92 | EUCTWDistributionAnalysis(); |
michael@0 | 93 | protected: |
michael@0 | 94 | |
michael@0 | 95 | //for euc-TW encoding, we are interested |
michael@0 | 96 | // first byte range: 0xc4 -- 0xfe |
michael@0 | 97 | // second byte range: 0xa1 -- 0xfe |
michael@0 | 98 | //no validation needed here. State machine has done that |
michael@0 | 99 | int32_t GetOrder(const char* str) |
michael@0 | 100 | { if ((unsigned char)*str >= (unsigned char)0xc4) |
michael@0 | 101 | return 94*((unsigned char)str[0]-(unsigned char)0xc4) + (unsigned char)str[1] - (unsigned char)0xa1; |
michael@0 | 102 | else |
michael@0 | 103 | return -1; |
michael@0 | 104 | } |
michael@0 | 105 | }; |
michael@0 | 106 | |
michael@0 | 107 | |
michael@0 | 108 | class EUCKRDistributionAnalysis : public CharDistributionAnalysis |
michael@0 | 109 | { |
michael@0 | 110 | public: |
michael@0 | 111 | EUCKRDistributionAnalysis(); |
michael@0 | 112 | protected: |
michael@0 | 113 | //for euc-KR encoding, we are interested |
michael@0 | 114 | // first byte range: 0xb0 -- 0xfe |
michael@0 | 115 | // second byte range: 0xa1 -- 0xfe |
michael@0 | 116 | //no validation needed here. State machine has done that |
michael@0 | 117 | int32_t GetOrder(const char* str) |
michael@0 | 118 | { if ((unsigned char)*str >= (unsigned char)0xb0) |
michael@0 | 119 | return 94*((unsigned char)str[0]-(unsigned char)0xb0) + (unsigned char)str[1] - (unsigned char)0xa1; |
michael@0 | 120 | else |
michael@0 | 121 | return -1; |
michael@0 | 122 | } |
michael@0 | 123 | }; |
michael@0 | 124 | |
michael@0 | 125 | class GB2312DistributionAnalysis : public CharDistributionAnalysis |
michael@0 | 126 | { |
michael@0 | 127 | public: |
michael@0 | 128 | GB2312DistributionAnalysis(); |
michael@0 | 129 | protected: |
michael@0 | 130 | //for GB2312 encoding, we are interested |
michael@0 | 131 | // first byte range: 0xb0 -- 0xfe |
michael@0 | 132 | // second byte range: 0xa1 -- 0xfe |
michael@0 | 133 | //no validation needed here. State machine has done that |
michael@0 | 134 | int32_t GetOrder(const char* str) |
michael@0 | 135 | { if ((unsigned char)*str >= (unsigned char)0xb0 && (unsigned char)str[1] >= (unsigned char)0xa1) |
michael@0 | 136 | return 94*((unsigned char)str[0]-(unsigned char)0xb0) + (unsigned char)str[1] - (unsigned char)0xa1; |
michael@0 | 137 | else |
michael@0 | 138 | return -1; |
michael@0 | 139 | } |
michael@0 | 140 | }; |
michael@0 | 141 | |
michael@0 | 142 | |
michael@0 | 143 | class Big5DistributionAnalysis : public CharDistributionAnalysis |
michael@0 | 144 | { |
michael@0 | 145 | public: |
michael@0 | 146 | Big5DistributionAnalysis(); |
michael@0 | 147 | protected: |
michael@0 | 148 | //for big5 encoding, we are interested |
michael@0 | 149 | // first byte range: 0xa4 -- 0xfe |
michael@0 | 150 | // second byte range: 0x40 -- 0x7e , 0xa1 -- 0xfe |
michael@0 | 151 | //no validation needed here. State machine has done that |
michael@0 | 152 | int32_t GetOrder(const char* str) |
michael@0 | 153 | { if ((unsigned char)*str >= (unsigned char)0xa4) |
michael@0 | 154 | if ((unsigned char)str[1] >= (unsigned char)0xa1) |
michael@0 | 155 | return 157*((unsigned char)str[0]-(unsigned char)0xa4) + (unsigned char)str[1] - (unsigned char)0xa1 +63; |
michael@0 | 156 | else |
michael@0 | 157 | return 157*((unsigned char)str[0]-(unsigned char)0xa4) + (unsigned char)str[1] - (unsigned char)0x40; |
michael@0 | 158 | else |
michael@0 | 159 | return -1; |
michael@0 | 160 | } |
michael@0 | 161 | }; |
michael@0 | 162 | |
michael@0 | 163 | class SJISDistributionAnalysis : public CharDistributionAnalysis |
michael@0 | 164 | { |
michael@0 | 165 | public: |
michael@0 | 166 | SJISDistributionAnalysis(); |
michael@0 | 167 | protected: |
michael@0 | 168 | //for sjis encoding, we are interested |
michael@0 | 169 | // first byte range: 0x81 -- 0x9f , 0xe0 -- 0xfe |
michael@0 | 170 | // second byte range: 0x40 -- 0x7e, 0x81 -- oxfe |
michael@0 | 171 | //no validation needed here. State machine has done that |
michael@0 | 172 | int32_t GetOrder(const char* str) |
michael@0 | 173 | { |
michael@0 | 174 | int32_t order; |
michael@0 | 175 | if ((unsigned char)*str >= (unsigned char)0x81 && (unsigned char)*str <= (unsigned char)0x9f) |
michael@0 | 176 | order = 188 * ((unsigned char)str[0]-(unsigned char)0x81); |
michael@0 | 177 | else if ((unsigned char)*str >= (unsigned char)0xe0 && (unsigned char)*str <= (unsigned char)0xef) |
michael@0 | 178 | order = 188 * ((unsigned char)str[0]-(unsigned char)0xe0 + 31); |
michael@0 | 179 | else |
michael@0 | 180 | return -1; |
michael@0 | 181 | order += (unsigned char)*(str+1) - 0x40; |
michael@0 | 182 | if ((unsigned char)str[1] > (unsigned char)0x7f) |
michael@0 | 183 | order--; |
michael@0 | 184 | return order; |
michael@0 | 185 | } |
michael@0 | 186 | }; |
michael@0 | 187 | |
michael@0 | 188 | class EUCJPDistributionAnalysis : public CharDistributionAnalysis |
michael@0 | 189 | { |
michael@0 | 190 | public: |
michael@0 | 191 | EUCJPDistributionAnalysis(); |
michael@0 | 192 | protected: |
michael@0 | 193 | //for euc-JP encoding, we are interested |
michael@0 | 194 | // first byte range: 0xa0 -- 0xfe |
michael@0 | 195 | // second byte range: 0xa1 -- 0xfe |
michael@0 | 196 | //no validation needed here. State machine has done that |
michael@0 | 197 | int32_t GetOrder(const char* str) |
michael@0 | 198 | { if ((unsigned char)*str >= (unsigned char)0xa0) |
michael@0 | 199 | return 94*((unsigned char)str[0]-(unsigned char)0xa1) + (unsigned char)str[1] - (unsigned char)0xa1; |
michael@0 | 200 | else |
michael@0 | 201 | return -1; |
michael@0 | 202 | } |
michael@0 | 203 | }; |
michael@0 | 204 | |
michael@0 | 205 | #endif //CharDistribution_h__ |
michael@0 | 206 |