AuroraMiddleware/brotli
1
0
Fork 0
mirror of https://github.com/google/brotli.git synced 2024-11-21 19:20:09 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update the dictionary and the transforms.

Browse Source
This commit is contained in:
Zoltan Szabadka 2014-03-25 16:48:25 +01:00
parent e7650080a8
commit 347781947a
9 changed files with 22402 additions and 11441 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4527
brotlispec.txt
View File

File diff suppressed because it is too large Load Diff

3
dec/decode.c
View File

@ -987,7 +987,8 @@ int BrotliDecompress(BrotliInput input, BrotliOutput output) {
copy_dst = &ringbuffer[pos & ringbuffer_mask]; copy_dst = &ringbuffer[pos & ringbuffer_mask];
if (distance > max_distance) { if (distance > max_distance) {
if (copy_length >= 3 && copy_length <= kMaxDictionaryWordLength) { if (copy_length >= kMinDictionaryWordLength &&
copy_length <= kMaxDictionaryWordLength) {
int offset = kBrotliDictionaryOffsetsByLength[copy_length]; int offset = kBrotliDictionaryOffsetsByLength[copy_length];
int word_id = distance - max_distance - 1; int word_id = distance - max_distance - 1;
int shift = kBrotliDictionarySizeBitsByLength[copy_length]; int shift = kBrotliDictionarySizeBitsByLength[copy_length];

14384
dec/dictionary.h
View File

File diff suppressed because it is too large Load Diff

257
dec/transform.h
View File

@ -28,17 +28,26 @@ extern "C" {
enum WordTransformType { enum WordTransformType {
kIdentity = 0, kIdentity = 0,
kOmit1 = 1, kOmitLast1 = 1,
kOmit2 = 2, kOmitLast2 = 2,
kOmit3 = 3, kOmitLast3 = 3,
kOmit4 = 4, kOmitLast4 = 4,
kOmit5 = 5, kOmitLast5 = 5,
kOmit6 = 6, kOmitLast6 = 6,
kOmit7 = 7, kOmitLast7 = 7,
kOmit8 = 8, kOmitLast8 = 8,
kOmit9 = 9, kOmitLast9 = 9,
kUppercaseFirst = 10, kUppercaseFirst = 10,
kUppercaseAll = 11 kUppercaseAll = 11,
kOmitFirst1 = 12,
kOmitFirst2 = 13,
kOmitFirst3 = 14,
kOmitFirst4 = 15,
kOmitFirst5 = 16,
kOmitFirst6 = 17,
kOmitFirst7 = 18,
kOmitFirst8 = 19,
kOmitFirst9 = 20
}; };
typedef struct { typedef struct {
@ -48,105 +57,127 @@ typedef struct {
} Transform; } Transform;
static const Transform kTransforms[] = { static const Transform kTransforms[] = {
{ "", kIdentity, "" }, { "", kIdentity, "" },
{ "", kIdentity, " ", }, { "", kIdentity, " " },
{ "", kIdentity, "\">" }, { " ", kIdentity, " " },
{ "", kUppercaseFirst, "" }, { "", kOmitFirst1, "" },
{ "", kIdentity, "\"" }, { "", kUppercaseFirst, " " },
{ "", kIdentity, ".", }, { "", kIdentity, " the " },
{ "", kIdentity, "=\"" }, { " ", kIdentity, "" },
{ "", kUppercaseFirst, " ", }, { "s ", kIdentity, " " },
{ " ", kIdentity, "=\"" }, { "", kIdentity, " of " },
{ " ", kIdentity, " ", }, { "", kUppercaseFirst, "" },
{ "", kIdentity, ":", }, { "", kIdentity, " and " },
{ " ", kIdentity, "" }, { "", kOmitFirst2, "" },
{ "", kIdentity, "\n" }, { "", kOmitLast1, "" },
{ "", kIdentity, "(", }, { ", ", kIdentity, " " },
{ "", kUppercaseAll, "" }, { "", kIdentity, ", " },
{ ".", kIdentity, "(", }, { " ", kUppercaseFirst, " " },
{ "", kIdentity, "'" }, { "", kIdentity, " in " },
{ "", kUppercaseFirst, "\"" }, { "", kIdentity, " to " },
{ " ", kUppercaseFirst, " ", }, { "e ", kIdentity, " " },
{ "", kOmit3, "" }, { "", kIdentity, "\"" },
{ "", kOmit4, "" }, { "", kIdentity, "." },
{ ".", kIdentity, "" }, { "", kIdentity, "\">" },
{ "", kOmit1, "" }, { "", kIdentity, "\n" },
{ "", kOmit2, "" }, { "", kOmitLast3, "" },
{ "", kUppercaseFirst, "\">" }, { "", kIdentity, "]" },
{ "", kOmit5, "" }, { "", kIdentity, " for " },
{ "", kUppercaseAll, " ", }, { "", kOmitFirst3, "" },
{ " ", kUppercaseFirst, "" }, { "", kOmitLast2, "" },
{ "", kIdentity, ", ", }, { "", kIdentity, " a " },
{ "", kUppercaseFirst, "(", }, { "", kIdentity, " that " },
{ "", kIdentity, "\n\t" }, { " ", kUppercaseFirst, "" },
{ "", kUppercaseFirst, "'" }, { "", kIdentity, ". " },
{ ".", kIdentity, " ", }, { ".", kIdentity, "" },
{ " ", kUppercaseAll, " ", }, { " ", kIdentity, ", " },
{ "", kIdentity, "='" }, { "", kOmitFirst4, "" },
{ "", kUppercaseFirst, ".", }, { "", kIdentity, " with " },
{ " ", kIdentity, ".", }, { "", kIdentity, "'" },
{ " ", kIdentity, ", ", }, { "", kIdentity, " from " },
{ " ", kUppercaseAll, "" }, { "", kIdentity, " by " },
{ "", kOmit6, "" }, { "", kOmitFirst5, "" },
{ "", kOmit9, "" }, { "", kOmitFirst6, "" },
{ "", kUppercaseAll, "\"" }, { " the ", kIdentity, "" },
{ "", kIdentity, " the " }, { "", kOmitLast4, "" },
{ "", kIdentity, " in " }, { "", kIdentity, ". The " },
{ "", kIdentity, " of " }, { "", kUppercaseAll, "" },
{ "", kIdentity, " to " }, { "", kIdentity, " on " },
{ "", kIdentity, " and " }, { "", kIdentity, " as " },
{ "", kIdentity, " is " }, { "", kIdentity, " is " },
{ "", kIdentity, " on " }, { "", kOmitLast7, "" },
{ "", kIdentity, " by " }, { "", kOmitLast1, "ing " },
{ "", kIdentity, " for " }, { "", kIdentity, "\n\t" },
{ "", kIdentity, " with " }, { "", kIdentity, ":" },
{ "", kIdentity, " from " }, { " ", kIdentity, ". " },
{ "", kIdentity, " as " }, { "", kIdentity, "ed " },
{ "", kIdentity, " at " }, { "", kOmitFirst9, "" },
{ "", kIdentity, "er " }, { "", kOmitFirst7, "" },
{ " ", kIdentity, "='" }, { "", kOmitLast6, "" },
{ "", kIdentity, " a " }, { "", kIdentity, "(" },
{ "", kOmit7, "" }, { "", kUppercaseFirst, ", " },
{ "", kOmit8, "" }, { "", kOmitLast8, "" },
{ " ", kIdentity, "(", }, { "", kIdentity, " at " },
{ " ", kIdentity, ". ", }, { "", kIdentity, "ly " },
{ "", kIdentity, ". ", }, { " the ", kIdentity, " of " },
{ "", kIdentity, ",", }, { "", kOmitLast5, "" },
{ "", kOmit1, "ing " }, { "", kOmitLast9, "" },
{ "", kIdentity, "ed " }, { " ", kUppercaseFirst, ", " },
{ "", kUppercaseFirst, ", ", }, { "", kUppercaseFirst, "\"" },
{ "", kUppercaseAll, ".", }, { ".", kIdentity, "(" },
{ "", kUppercaseAll, "=\"" }, { "", kUppercaseAll, " " },
{ "", kUppercaseAll, ", ", }, { "", kUppercaseFirst, "\">" },
{ "", kUppercaseAll, "\">" }, { "", kIdentity, "=\"" },
{ " ", kUppercaseFirst, ".", }, { " ", kIdentity, "." },
{ " ", kUppercaseAll, "=\"" }, { ".com/", kIdentity, "" },
{ " ", kUppercaseFirst, ", ", }, { " the ", kIdentity, " of the " },
{ "", kUppercaseAll, "'" }, { "", kUppercaseFirst, "'" },
{ "", kUppercaseFirst, "=\"" }, { "", kIdentity, ". This " },
{ " ", kIdentity, ",", }, { "", kIdentity, "," },
{ "", kIdentity, " that " }, { ".", kIdentity, " " },
{ "", kUppercaseFirst, "='" }, { "", kUppercaseFirst, "(" },
{ "", kUppercaseFirst, ". ", }, { "", kUppercaseFirst, "." },
{ "", kUppercaseFirst, ",", }, { "", kIdentity, " not " },
{ "", kIdentity, ". The " }, { " ", kIdentity, "=\"" },
{ "\xc2\xa0", kIdentity, "" }, { "", kIdentity, "er " },
{ " ", kUppercaseFirst, ". ", }, { " ", kUppercaseAll, " " },
{ "", kUppercaseAll, ",", }, { "", kIdentity, "al " },
{ "", kUppercaseAll, "(", }, { " ", kUppercaseAll, "" },
{ " ", kUppercaseAll, "='" }, { "", kIdentity, "='" },
{ "", kIdentity, "]" }, { "", kUppercaseAll, "\"" },
{ "", kUppercaseAll, "='" }, { "", kUppercaseFirst, ". " },
{ " ", kUppercaseAll, ".", }, { " ", kIdentity, "(" },
{ "", kUppercaseAll, ". ", }, { "", kIdentity, "ful " },
{ " ", kUppercaseFirst, "=\"" }, { " ", kUppercaseFirst, ". " },
{ " ", kUppercaseAll, ". ", }, { "", kIdentity, "ive " },
{ " ", kUppercaseFirst, ",", }, { "", kIdentity, "less " },
{ " ", kUppercaseAll, ", ", }, { "", kUppercaseAll, "'" },
{ "", kIdentity, "ize " }, { "", kIdentity, "est " },
{ " ", kUppercaseFirst, "='" }, { " ", kUppercaseFirst, "." },
{ "", kIdentity, "est " }, { "", kUppercaseAll, "\">" },
{ "", kIdentity, ". This " }, { " ", kIdentity, "='" },
{ "", kUppercaseFirst, "," },
{ "", kIdentity, "ize " },
{ "", kUppercaseAll, "." },
{ "\xc2\xa0", kIdentity, "" },
{ " ", kIdentity, "," },
{ "", kUppercaseFirst, "=\"" },
{ "", kUppercaseAll, "=\"" },
{ "", kIdentity, "ous " },
{ "", kUppercaseAll, ", " },
{ "", kUppercaseFirst, "='" },
{ " ", kUppercaseFirst, "," },
{ " ", kUppercaseAll, "=\"" },
{ " ", kUppercaseAll, ", " },
{ "", kUppercaseAll, "," },
{ "", kUppercaseAll, "(" },
{ "", kUppercaseAll, ". " },
{ " ", kUppercaseAll, "." },
{ "", kUppercaseAll, "='" },
{ " ", kUppercaseAll, ". " },
{ " ", kUppercaseFirst, "=\"" },
{ " ", kUppercaseAll, "='" },
{ " ", kUppercaseFirst, "='" },
}; };
static const int kNumTransforms = sizeof(kTransforms) / sizeof(kTransforms[0]); static const int kNumTransforms = sizeof(kTransforms) / sizeof(kTransforms[0]);
@ -173,11 +204,17 @@ static BROTLI_INLINE int TransformDictionaryWord(
const char* prefix = kTransforms[transform].prefix; const char* prefix = kTransforms[transform].prefix;
const char* suffix = kTransforms[transform].suffix; const char* suffix = kTransforms[transform].suffix;
const int t = kTransforms[transform].transform; const int t = kTransforms[transform].transform;
int skip = t < kOmitFirst1 ? 0 : t - (kOmitFirst1 - 1);
int idx = 0; int idx = 0;
int i = 0; int i = 0;
uint8_t* uppercase; uint8_t* uppercase;
if (skip > len) {
skip = len;
}
while (*prefix) { dst[idx++] = (uint8_t)*prefix++; } while (*prefix) { dst[idx++] = (uint8_t)*prefix++; }
if (t <= kOmit9) { word += skip;
len -= skip;
if (t <= kOmitLast9) {
len -= t; len -= t;
} }
while (i < len) { dst[idx++] = word[i++]; } while (i < len) { dst[idx++] = word[i++]; }

14384
enc/dictionary.h
View File

File diff suppressed because it is too large Load Diff

5
enc/encode.cc
View File

@ -909,12 +909,13 @@ void BrotliCompressor::StoreDictionaryWordHashes() {
if (static_dictionary_ == NULL) { if (static_dictionary_ == NULL) {
static_dictionary_ = new StaticDictionary; static_dictionary_ = new StaticDictionary;
for (int t = num_transforms - 1; t >= 0; --t) { for (int t = num_transforms - 1; t >= 0; --t) {
for (int i = kMaxDictionaryWordLength; i >= 3; --i) { for (int i = kMaxDictionaryWordLength;
i >= kMinDictionaryWordLength; --i) {
const int num_words = 1 << kBrotliDictionarySizeBitsByLength[i]; const int num_words = 1 << kBrotliDictionarySizeBitsByLength[i];
for (int j = num_words - 1; j >= 0; --j) { for (int j = num_words - 1; j >= 0; --j) {
int word_id = t * num_words + j; int word_id = t * num_words + j;
std::string word = GetTransformedDictionaryWord(i, word_id); std::string word = GetTransformedDictionaryWord(i, word_id);
if (word.size() >= 3) { if (word.size() >= 4) {
static_dictionary_->Insert(word, i, word_id); static_dictionary_->Insert(word, i, word_id);
} }
} }

2
enc/hash.h
View File

@ -193,7 +193,7 @@ class HashLongestMatch {
literal_cost[(cur_ix + 1) & ring_buffer_mask] + 1.2; literal_cost[(cur_ix + 1) & ring_buffer_mask] + 1.2;
bool match_found = false; bool match_found = false;
// Don't accept a short copy from far away. // Don't accept a short copy from far away.
double best_score = 8.11; double best_score = 8.115;
if (insert_length_ < 4) { if (insert_length_ < 4) {
double cost_diff[4] = { 0.10, 0.04, 0.02, 0.01 }; double cost_diff[4] = { 0.10, 0.04, 0.02, 0.01 };
best_score += cost_diff[insert_length_]; best_score += cost_diff[insert_length_];

7
enc/literal_cost.cc
View File

@ -118,6 +118,13 @@ void EstimateBitCostsForLiteralsUTF8(size_t pos, size_t len, size_t mask,
lit_cost *= 0.5; lit_cost *= 0.5;
lit_cost += 0.5; lit_cost += 0.5;
} }
// Make the first bytes more expensive -- seems to help, not sure why.
// Perhaps because the entropy source is changing its properties
// rapidly in the beginning of the file, perhaps because the beginning
// of the data is a statistical "anomaly".
if (i < 2000) {
lit_cost += 0.7 - ((2000 - i) / 2000.0 * 0.35);
}
cost[(pos + i) & cost_mask] = lit_cost; cost[(pos + i) & cost_mask] = lit_cost;
} }
} }

274
enc/transform.h
View File

@ -25,17 +25,26 @@ namespace brotli {
enum WordTransformType { enum WordTransformType {
kIdentity = 0, kIdentity = 0,
kOmit1 = 1, kOmitLast1 = 1,
kOmit2 = 2, kOmitLast2 = 2,
kOmit3 = 3, kOmitLast3 = 3,
kOmit4 = 4, kOmitLast4 = 4,
kOmit5 = 5, kOmitLast5 = 5,
kOmit6 = 6, kOmitLast6 = 6,
kOmit7 = 7, kOmitLast7 = 7,
kOmit8 = 8, kOmitLast8 = 8,
kOmit9 = 9, kOmitLast9 = 9,
kUppercaseFirst = 10, kUppercaseFirst = 10,
kUppercaseAll = 11, kUppercaseAll = 11,
kOmitFirst1 = 12,
kOmitFirst2 = 13,
kOmitFirst3 = 14,
kOmitFirst4 = 15,
kOmitFirst5 = 16,
kOmitFirst6 = 17,
kOmitFirst7 = 18,
kOmitFirst8 = 19,
kOmitFirst9 = 20,
}; };
struct Transform { struct Transform {
@ -45,105 +54,127 @@ struct Transform {
}; };
static const Transform kTransforms[] = { static const Transform kTransforms[] = {
{ "", kIdentity, "" }, { "", kIdentity, "" },
{ "", kIdentity, " ", }, { "", kIdentity, " " },
{ "", kIdentity, "\">" }, { " ", kIdentity, " " },
{ "", kUppercaseFirst, "" }, { "", kOmitFirst1, "" },
{ "", kIdentity, "\"" }, { "", kUppercaseFirst, " " },
{ "", kIdentity, ".", }, { "", kIdentity, " the " },
{ "", kIdentity, "=\"" }, { " ", kIdentity, "" },
{ "", kUppercaseFirst, " ", }, { "s ", kIdentity, " " },
{ " ", kIdentity, "=\"" }, { "", kIdentity, " of " },
{ " ", kIdentity, " ", }, { "", kUppercaseFirst, "" },
{ "", kIdentity, ":", }, { "", kIdentity, " and " },
{ " ", kIdentity, "" }, { "", kOmitFirst2, "" },
{ "", kIdentity, "\n" }, { "", kOmitLast1, "" },
{ "", kIdentity, "(", }, { ", ", kIdentity, " " },
{ "", kUppercaseAll, "" }, { "", kIdentity, ", " },
{ ".", kIdentity, "(", }, { " ", kUppercaseFirst, " " },
{ "", kIdentity, "'" }, { "", kIdentity, " in " },
{ "", kUppercaseFirst, "\"" }, { "", kIdentity, " to " },
{ " ", kUppercaseFirst, " ", }, { "e ", kIdentity, " " },
{ "", kOmit3, "" }, { "", kIdentity, "\"" },
{ "", kOmit4, "" }, { "", kIdentity, "." },
{ ".", kIdentity, "" }, { "", kIdentity, "\">" },
{ "", kOmit1, "" }, { "", kIdentity, "\n" },
{ "", kOmit2, "" }, { "", kOmitLast3, "" },
{ "", kUppercaseFirst, "\">" }, { "", kIdentity, "]" },
{ "", kOmit5, "" }, { "", kIdentity, " for " },
{ "", kUppercaseAll, " ", }, { "", kOmitFirst3, "" },
{ " ", kUppercaseFirst, "" }, { "", kOmitLast2, "" },
{ "", kIdentity, ", ", }, { "", kIdentity, " a " },
{ "", kUppercaseFirst, "(", }, { "", kIdentity, " that " },
{ "", kIdentity, "\n\t" }, { " ", kUppercaseFirst, "" },
{ "", kUppercaseFirst, "'" }, { "", kIdentity, ". " },
{ ".", kIdentity, " ", }, { ".", kIdentity, "" },
{ " ", kUppercaseAll, " ", }, { " ", kIdentity, ", " },
{ "", kIdentity, "='" }, { "", kOmitFirst4, "" },
{ "", kUppercaseFirst, ".", }, { "", kIdentity, " with " },
{ " ", kIdentity, ".", }, { "", kIdentity, "'" },
{ " ", kIdentity, ", ", }, { "", kIdentity, " from " },
{ " ", kUppercaseAll, "" }, { "", kIdentity, " by " },
{ "", kOmit6, "" }, { "", kOmitFirst5, "" },
{ "", kOmit9, "" }, { "", kOmitFirst6, "" },
{ "", kUppercaseAll, "\"" }, { " the ", kIdentity, "" },
{ "", kIdentity, " the " }, { "", kOmitLast4, "" },
{ "", kIdentity, " in " }, { "", kIdentity, ". The " },
{ "", kIdentity, " of " }, { "", kUppercaseAll, "" },
{ "", kIdentity, " to " }, { "", kIdentity, " on " },
{ "", kIdentity, " and " }, { "", kIdentity, " as " },
{ "", kIdentity, " is " }, { "", kIdentity, " is " },
{ "", kIdentity, " on " }, { "", kOmitLast7, "" },
{ "", kIdentity, " by " }, { "", kOmitLast1, "ing " },
{ "", kIdentity, " for " }, { "", kIdentity, "\n\t" },
{ "", kIdentity, " with " }, { "", kIdentity, ":" },
{ "", kIdentity, " from " }, { " ", kIdentity, ". " },
{ "", kIdentity, " as " }, { "", kIdentity, "ed " },
{ "", kIdentity, " at " }, { "", kOmitFirst9, "" },
{ "", kIdentity, "er " }, { "", kOmitFirst7, "" },
{ " ", kIdentity, "='" }, { "", kOmitLast6, "" },
{ "", kIdentity, " a " }, { "", kIdentity, "(" },
{ "", kOmit7, "" }, { "", kUppercaseFirst, ", " },
{ "", kOmit8, "" }, { "", kOmitLast8, "" },
{ " ", kIdentity, "(", }, { "", kIdentity, " at " },
{ " ", kIdentity, ". ", }, { "", kIdentity, "ly " },
{ "", kIdentity, ". ", }, { " the ", kIdentity, " of " },
{ "", kIdentity, ",", }, { "", kOmitLast5, "" },
{ "", kOmit1, "ing " }, { "", kOmitLast9, "" },
{ "", kIdentity, "ed " }, { " ", kUppercaseFirst, ", " },
{ "", kUppercaseFirst, ", ", }, { "", kUppercaseFirst, "\"" },
{ "", kUppercaseAll, ".", }, { ".", kIdentity, "(" },
{ "", kUppercaseAll, "=\"" }, { "", kUppercaseAll, " " },
{ "", kUppercaseAll, ", ", }, { "", kUppercaseFirst, "\">" },
{ "", kUppercaseAll, "\">" }, { "", kIdentity, "=\"" },
{ " ", kUppercaseFirst, ".", }, { " ", kIdentity, "." },
{ " ", kUppercaseAll, "=\"" }, { ".com/", kIdentity, "" },
{ " ", kUppercaseFirst, ", ", }, { " the ", kIdentity, " of the " },
{ "", kUppercaseAll, "'" }, { "", kUppercaseFirst, "'" },
{ "", kUppercaseFirst, "=\"" }, { "", kIdentity, ". This " },
{ " ", kIdentity, ",", }, { "", kIdentity, "," },
{ "", kIdentity, " that " }, { ".", kIdentity, " " },
{ "", kUppercaseFirst, "='" }, { "", kUppercaseFirst, "(" },
{ "", kUppercaseFirst, ". ", }, { "", kUppercaseFirst, "." },
{ "", kUppercaseFirst, ",", }, { "", kIdentity, " not " },
{ "", kIdentity, ". The " }, { " ", kIdentity, "=\"" },
{ "\xc2\xa0", kIdentity, "" }, { "", kIdentity, "er " },
{ " ", kUppercaseFirst, ". ", }, { " ", kUppercaseAll, " " },
{ "", kUppercaseAll, ",", }, { "", kIdentity, "al " },
{ "", kUppercaseAll, "(", }, { " ", kUppercaseAll, "" },
{ " ", kUppercaseAll, "='" }, { "", kIdentity, "='" },
{ "", kIdentity, "]" }, { "", kUppercaseAll, "\"" },
{ "", kUppercaseAll, "='" }, { "", kUppercaseFirst, ". " },
{ " ", kUppercaseAll, ".", }, { " ", kIdentity, "(" },
{ "", kUppercaseAll, ". ", }, { "", kIdentity, "ful " },
{ " ", kUppercaseFirst, "=\"" }, { " ", kUppercaseFirst, ". " },
{ " ", kUppercaseAll, ". ", }, { "", kIdentity, "ive " },
{ " ", kUppercaseFirst, ",", }, { "", kIdentity, "less " },
{ " ", kUppercaseAll, ", ", }, { "", kUppercaseAll, "'" },
{ "", kIdentity, "ize " }, { "", kIdentity, "est " },
{ " ", kUppercaseFirst, "='" }, { " ", kUppercaseFirst, "." },
{ "", kIdentity, "est " }, { "", kUppercaseAll, "\">" },
{ "", kIdentity, ". This " }, { " ", kIdentity, "='" },
{ "", kUppercaseFirst, "," },
{ "", kIdentity, "ize " },
{ "", kUppercaseAll, "." },
{ "\xc2\xa0", kIdentity, "" },
{ " ", kIdentity, "," },
{ "", kUppercaseFirst, "=\"" },
{ "", kUppercaseAll, "=\"" },
{ "", kIdentity, "ous " },
{ "", kUppercaseAll, ", " },
{ "", kUppercaseFirst, "='" },
{ " ", kUppercaseFirst, "," },
{ " ", kUppercaseAll, "=\"" },
{ " ", kUppercaseAll, ", " },
{ "", kUppercaseAll, "," },
{ "", kUppercaseAll, "(" },
{ "", kUppercaseAll, ". " },
{ " ", kUppercaseAll, "." },
{ "", kUppercaseAll, "='" },
{ " ", kUppercaseAll, ". " },
{ " ", kUppercaseFirst, "=\"" },
{ " ", kUppercaseAll, "='" },
{ " ", kUppercaseFirst, "='" },
}; };
static const int kNumTransforms = sizeof(kTransforms) / sizeof(kTransforms[0]); static const int kNumTransforms = sizeof(kTransforms) / sizeof(kTransforms[0]);
@ -169,19 +200,26 @@ static int ToUpperCase(uint8_t *p, int len) {
inline std::string ApplyTransform( inline std::string ApplyTransform(
const Transform& t, const uint8_t* word, int len) { const Transform& t, const uint8_t* word, int len) {
std::string ret(t.prefix); std::string ret(t.prefix);
if (t.word_transform <= kOmit9) { if (t.word_transform <= kOmitLast9) {
len -= t.word_transform; len -= t.word_transform;
} }
if (len > 0) { if (len > 0) {
ret += std::string(word, word + len); if (t.word_transform >= kOmitFirst1) {
uint8_t *uppercase = reinterpret_cast<uint8_t*>(&ret[ret.size() - len]); const int skip = t.word_transform - (kOmitFirst1 - 1);
if (t.word_transform == kUppercaseFirst) { if (len > skip) {
ToUpperCase(uppercase, len); ret += std::string(word + skip, word + len);
} else if (t.word_transform == kUppercaseAll) { }
while (len > 0) { } else {
int step = ToUpperCase(uppercase, len); ret += std::string(word, word + len);
uppercase += step; uint8_t *uppercase = reinterpret_cast<uint8_t*>(&ret[ret.size() - len]);
len -= step; if (t.word_transform == kUppercaseFirst) {
ToUpperCase(uppercase, len);
} else if (t.word_transform == kUppercaseAll) {
while (len > 0) {
int step = ToUpperCase(uppercase, len);
uppercase += step;
len -= step;
}
} }
} }
} }