Path: blob/master/thirdparty/brotli/common/transform.c
9906 views
/* Copyright 2013 Google Inc. All Rights Reserved.12Distributed under MIT license.3See file LICENSE for detail or copy at https://opensource.org/licenses/MIT4*/56#include "transform.h"78#if defined(__cplusplus) || defined(c_plusplus)9extern "C" {10#endif1112/* RFC 7932 transforms string data */13static const char kPrefixSuffix[217] =14"\1 \2, \10 of the \4 of \2s \1.\5 and \4 "15/* 0x _0 _2 __5 _E _3 _6 _8 _E */16"in \1\"\4 to \2\">\1\n\2. \1]\5 for \3 a \6 "17/* 2x _3_ _5 _A_ _D_ _F _2 _4 _A _E */18"that \1\'\6 with \6 from \4 by \1(\6. T"19/* 4x _5_ _7 _E _5 _A _C */20"he \4 on \4 as \4 is \4ing \2\n\t\1:\3ed "21/* 6x _3 _8 _D _2 _7_ _ _A _C */22"\2=\"\4 at \3ly \1,\2=\'\5.com/\7. This \5"23/* 8x _0 _ _3 _8 _C _E _ _1 _7 _F */24" not \3er \3al \4ful \4ive \5less \4es"25/* Ax _5 _9 _D _2 _7 _D */26"t \4ize \2\xc2\xa0\4ous \5 the \2e "; /* \0 - implicit trailing zero. */27/* Cx _2 _7___ ___ _A _F _5 _8 */2829static const uint16_t kPrefixSuffixMap[50] = {300x00, 0x02, 0x05, 0x0E, 0x13, 0x16, 0x18, 0x1E, 0x23, 0x25,310x2A, 0x2D, 0x2F, 0x32, 0x34, 0x3A, 0x3E, 0x45, 0x47, 0x4E,320x55, 0x5A, 0x5C, 0x63, 0x68, 0x6D, 0x72, 0x77, 0x7A, 0x7C,330x80, 0x83, 0x88, 0x8C, 0x8E, 0x91, 0x97, 0x9F, 0xA5, 0xA9,340xAD, 0xB2, 0xB7, 0xBD, 0xC2, 0xC7, 0xCA, 0xCF, 0xD5, 0xD835};3637/* RFC 7932 transforms */38static const uint8_t kTransformsData[] = {3949, BROTLI_TRANSFORM_IDENTITY, 49,4049, BROTLI_TRANSFORM_IDENTITY, 0,410, BROTLI_TRANSFORM_IDENTITY, 0,4249, BROTLI_TRANSFORM_OMIT_FIRST_1, 49,4349, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,4449, BROTLI_TRANSFORM_IDENTITY, 47,450, BROTLI_TRANSFORM_IDENTITY, 49,464, BROTLI_TRANSFORM_IDENTITY, 0,4749, BROTLI_TRANSFORM_IDENTITY, 3,4849, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,4949, BROTLI_TRANSFORM_IDENTITY, 6,5049, BROTLI_TRANSFORM_OMIT_FIRST_2, 49,5149, BROTLI_TRANSFORM_OMIT_LAST_1, 49,521, BROTLI_TRANSFORM_IDENTITY, 0,5349, BROTLI_TRANSFORM_IDENTITY, 1,540, BROTLI_TRANSFORM_UPPERCASE_FIRST, 0,5549, BROTLI_TRANSFORM_IDENTITY, 7,5649, BROTLI_TRANSFORM_IDENTITY, 9,5748, BROTLI_TRANSFORM_IDENTITY, 0,5849, BROTLI_TRANSFORM_IDENTITY, 8,5949, BROTLI_TRANSFORM_IDENTITY, 5,6049, BROTLI_TRANSFORM_IDENTITY, 10,6149, BROTLI_TRANSFORM_IDENTITY, 11,6249, BROTLI_TRANSFORM_OMIT_LAST_3, 49,6349, BROTLI_TRANSFORM_IDENTITY, 13,6449, BROTLI_TRANSFORM_IDENTITY, 14,6549, BROTLI_TRANSFORM_OMIT_FIRST_3, 49,6649, BROTLI_TRANSFORM_OMIT_LAST_2, 49,6749, BROTLI_TRANSFORM_IDENTITY, 15,6849, BROTLI_TRANSFORM_IDENTITY, 16,690, BROTLI_TRANSFORM_UPPERCASE_FIRST, 49,7049, BROTLI_TRANSFORM_IDENTITY, 12,715, BROTLI_TRANSFORM_IDENTITY, 49,720, BROTLI_TRANSFORM_IDENTITY, 1,7349, BROTLI_TRANSFORM_OMIT_FIRST_4, 49,7449, BROTLI_TRANSFORM_IDENTITY, 18,7549, BROTLI_TRANSFORM_IDENTITY, 17,7649, BROTLI_TRANSFORM_IDENTITY, 19,7749, BROTLI_TRANSFORM_IDENTITY, 20,7849, BROTLI_TRANSFORM_OMIT_FIRST_5, 49,7949, BROTLI_TRANSFORM_OMIT_FIRST_6, 49,8047, BROTLI_TRANSFORM_IDENTITY, 49,8149, BROTLI_TRANSFORM_OMIT_LAST_4, 49,8249, BROTLI_TRANSFORM_IDENTITY, 22,8349, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,8449, BROTLI_TRANSFORM_IDENTITY, 23,8549, BROTLI_TRANSFORM_IDENTITY, 24,8649, BROTLI_TRANSFORM_IDENTITY, 25,8749, BROTLI_TRANSFORM_OMIT_LAST_7, 49,8849, BROTLI_TRANSFORM_OMIT_LAST_1, 26,8949, BROTLI_TRANSFORM_IDENTITY, 27,9049, BROTLI_TRANSFORM_IDENTITY, 28,910, BROTLI_TRANSFORM_IDENTITY, 12,9249, BROTLI_TRANSFORM_IDENTITY, 29,9349, BROTLI_TRANSFORM_OMIT_FIRST_9, 49,9449, BROTLI_TRANSFORM_OMIT_FIRST_7, 49,9549, BROTLI_TRANSFORM_OMIT_LAST_6, 49,9649, BROTLI_TRANSFORM_IDENTITY, 21,9749, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,9849, BROTLI_TRANSFORM_OMIT_LAST_8, 49,9949, BROTLI_TRANSFORM_IDENTITY, 31,10049, BROTLI_TRANSFORM_IDENTITY, 32,10147, BROTLI_TRANSFORM_IDENTITY, 3,10249, BROTLI_TRANSFORM_OMIT_LAST_5, 49,10349, BROTLI_TRANSFORM_OMIT_LAST_9, 49,1040, BROTLI_TRANSFORM_UPPERCASE_FIRST, 1,10549, BROTLI_TRANSFORM_UPPERCASE_FIRST, 8,1065, BROTLI_TRANSFORM_IDENTITY, 21,10749, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,10849, BROTLI_TRANSFORM_UPPERCASE_FIRST, 10,10949, BROTLI_TRANSFORM_IDENTITY, 30,1100, BROTLI_TRANSFORM_IDENTITY, 5,11135, BROTLI_TRANSFORM_IDENTITY, 49,11247, BROTLI_TRANSFORM_IDENTITY, 2,11349, BROTLI_TRANSFORM_UPPERCASE_FIRST, 17,11449, BROTLI_TRANSFORM_IDENTITY, 36,11549, BROTLI_TRANSFORM_IDENTITY, 33,1165, BROTLI_TRANSFORM_IDENTITY, 0,11749, BROTLI_TRANSFORM_UPPERCASE_FIRST, 21,11849, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,11949, BROTLI_TRANSFORM_IDENTITY, 37,1200, BROTLI_TRANSFORM_IDENTITY, 30,12149, BROTLI_TRANSFORM_IDENTITY, 38,1220, BROTLI_TRANSFORM_UPPERCASE_ALL, 0,12349, BROTLI_TRANSFORM_IDENTITY, 39,1240, BROTLI_TRANSFORM_UPPERCASE_ALL, 49,12549, BROTLI_TRANSFORM_IDENTITY, 34,12649, BROTLI_TRANSFORM_UPPERCASE_ALL, 8,12749, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,1280, BROTLI_TRANSFORM_IDENTITY, 21,12949, BROTLI_TRANSFORM_IDENTITY, 40,1300, BROTLI_TRANSFORM_UPPERCASE_FIRST, 12,13149, BROTLI_TRANSFORM_IDENTITY, 41,13249, BROTLI_TRANSFORM_IDENTITY, 42,13349, BROTLI_TRANSFORM_UPPERCASE_ALL, 17,13449, BROTLI_TRANSFORM_IDENTITY, 43,1350, BROTLI_TRANSFORM_UPPERCASE_FIRST, 5,13649, BROTLI_TRANSFORM_UPPERCASE_ALL, 10,1370, BROTLI_TRANSFORM_IDENTITY, 34,13849, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,13949, BROTLI_TRANSFORM_IDENTITY, 44,14049, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,14145, BROTLI_TRANSFORM_IDENTITY, 49,1420, BROTLI_TRANSFORM_IDENTITY, 33,14349, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,14449, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,14549, BROTLI_TRANSFORM_IDENTITY, 46,14649, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,14749, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,1480, BROTLI_TRANSFORM_UPPERCASE_FIRST, 33,1490, BROTLI_TRANSFORM_UPPERCASE_ALL, 30,1500, BROTLI_TRANSFORM_UPPERCASE_ALL, 1,15149, BROTLI_TRANSFORM_UPPERCASE_ALL, 33,15249, BROTLI_TRANSFORM_UPPERCASE_ALL, 21,15349, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,1540, BROTLI_TRANSFORM_UPPERCASE_ALL, 5,15549, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,1560, BROTLI_TRANSFORM_UPPERCASE_ALL, 12,1570, BROTLI_TRANSFORM_UPPERCASE_FIRST, 30,1580, BROTLI_TRANSFORM_UPPERCASE_ALL, 34,1590, BROTLI_TRANSFORM_UPPERCASE_FIRST, 34,160};161162static const BrotliTransforms kBrotliTransforms = {163sizeof(kPrefixSuffix),164(const uint8_t*)kPrefixSuffix,165kPrefixSuffixMap,166sizeof(kTransformsData) / (3 * sizeof(kTransformsData[0])),167kTransformsData,168NULL, /* no extra parameters */169{0, 12, 27, 23, 42, 63, 56, 48, 59, 64}170};171172const BrotliTransforms* BrotliGetTransforms(void) {173return &kBrotliTransforms;174}175176static int ToUpperCase(uint8_t* p) {177if (p[0] < 0xC0) {178if (p[0] >= 'a' && p[0] <= 'z') {179p[0] ^= 32;180}181return 1;182}183/* An overly simplified uppercasing model for UTF-8. */184if (p[0] < 0xE0) {185p[1] ^= 32;186return 2;187}188/* An arbitrary transform for three byte characters. */189p[2] ^= 5;190return 3;191}192193static int Shift(uint8_t* word, int word_len, uint16_t parameter) {194/* Limited sign extension: scalar < (1 << 24). */195uint32_t scalar =196(parameter & 0x7FFFu) + (0x1000000u - (parameter & 0x8000u));197if (word[0] < 0x80) {198/* 1-byte rune / 0sssssss / 7 bit scalar (ASCII). */199scalar += (uint32_t)word[0];200word[0] = (uint8_t)(scalar & 0x7Fu);201return 1;202} else if (word[0] < 0xC0) {203/* Continuation / 10AAAAAA. */204return 1;205} else if (word[0] < 0xE0) {206/* 2-byte rune / 110sssss AAssssss / 11 bit scalar. */207if (word_len < 2) return 1;208scalar += (uint32_t)((word[1] & 0x3Fu) | ((word[0] & 0x1Fu) << 6u));209word[0] = (uint8_t)(0xC0 | ((scalar >> 6u) & 0x1F));210word[1] = (uint8_t)((word[1] & 0xC0) | (scalar & 0x3F));211return 2;212} else if (word[0] < 0xF0) {213/* 3-byte rune / 1110ssss AAssssss BBssssss / 16 bit scalar. */214if (word_len < 3) return word_len;215scalar += (uint32_t)((word[2] & 0x3Fu) | ((word[1] & 0x3Fu) << 6u) |216((word[0] & 0x0Fu) << 12u));217word[0] = (uint8_t)(0xE0 | ((scalar >> 12u) & 0x0F));218word[1] = (uint8_t)((word[1] & 0xC0) | ((scalar >> 6u) & 0x3F));219word[2] = (uint8_t)((word[2] & 0xC0) | (scalar & 0x3F));220return 3;221} else if (word[0] < 0xF8) {222/* 4-byte rune / 11110sss AAssssss BBssssss CCssssss / 21 bit scalar. */223if (word_len < 4) return word_len;224scalar += (uint32_t)((word[3] & 0x3Fu) | ((word[2] & 0x3Fu) << 6u) |225((word[1] & 0x3Fu) << 12u) | ((word[0] & 0x07u) << 18u));226word[0] = (uint8_t)(0xF0 | ((scalar >> 18u) & 0x07));227word[1] = (uint8_t)((word[1] & 0xC0) | ((scalar >> 12u) & 0x3F));228word[2] = (uint8_t)((word[2] & 0xC0) | ((scalar >> 6u) & 0x3F));229word[3] = (uint8_t)((word[3] & 0xC0) | (scalar & 0x3F));230return 4;231}232return 1;233}234235int BrotliTransformDictionaryWord(uint8_t* dst, const uint8_t* word, int len,236const BrotliTransforms* transforms, int transform_idx) {237int idx = 0;238const uint8_t* prefix = BROTLI_TRANSFORM_PREFIX(transforms, transform_idx);239uint8_t type = BROTLI_TRANSFORM_TYPE(transforms, transform_idx);240const uint8_t* suffix = BROTLI_TRANSFORM_SUFFIX(transforms, transform_idx);241{242int prefix_len = *prefix++;243while (prefix_len--) { dst[idx++] = *prefix++; }244}245{246const int t = type;247int i = 0;248if (t <= BROTLI_TRANSFORM_OMIT_LAST_9) {249len -= t;250} else if (t >= BROTLI_TRANSFORM_OMIT_FIRST_1251&& t <= BROTLI_TRANSFORM_OMIT_FIRST_9) {252int skip = t - (BROTLI_TRANSFORM_OMIT_FIRST_1 - 1);253word += skip;254len -= skip;255}256while (i < len) { dst[idx++] = word[i++]; }257if (t == BROTLI_TRANSFORM_UPPERCASE_FIRST) {258ToUpperCase(&dst[idx - len]);259} else if (t == BROTLI_TRANSFORM_UPPERCASE_ALL) {260uint8_t* uppercase = &dst[idx - len];261while (len > 0) {262int step = ToUpperCase(uppercase);263uppercase += step;264len -= step;265}266} else if (t == BROTLI_TRANSFORM_SHIFT_FIRST) {267uint16_t param = (uint16_t)(transforms->params[transform_idx * 2]268+ (transforms->params[transform_idx * 2 + 1] << 8u));269Shift(&dst[idx - len], len, param);270} else if (t == BROTLI_TRANSFORM_SHIFT_ALL) {271uint16_t param = (uint16_t)(transforms->params[transform_idx * 2]272+ (transforms->params[transform_idx * 2 + 1] << 8u));273uint8_t* shift = &dst[idx - len];274while (len > 0) {275int step = Shift(shift, len, param);276shift += step;277len -= step;278}279}280}281{282int suffix_len = *suffix++;283while (suffix_len--) { dst[idx++] = *suffix++; }284return idx;285}286}287288#if defined(__cplusplus) || defined(c_plusplus)289} /* extern "C" */290#endif291292293