Update the dictionary and the transforms.

This commit is contained in:
Zoltan Szabadka
2014-03-25 16:48:25 +01:00
parent 494c85cebb
commit 0829e37293
9 changed files with 22402 additions and 11441 deletions
+156 -118
View File
@@ -25,17 +25,26 @@ namespace brotli {
enum WordTransformType {
kIdentity = 0,
kOmit1 = 1,
kOmit2 = 2,
kOmit3 = 3,
kOmit4 = 4,
kOmit5 = 5,
kOmit6 = 6,
kOmit7 = 7,
kOmit8 = 8,
kOmit9 = 9,
kOmitLast1 = 1,
kOmitLast2 = 2,
kOmitLast3 = 3,
kOmitLast4 = 4,
kOmitLast5 = 5,
kOmitLast6 = 6,
kOmitLast7 = 7,
kOmitLast8 = 8,
kOmitLast9 = 9,
kUppercaseFirst = 10,
kUppercaseAll = 11,
kOmitFirst1 = 12,
kOmitFirst2 = 13,
kOmitFirst3 = 14,
kOmitFirst4 = 15,
kOmitFirst5 = 16,
kOmitFirst6 = 17,
kOmitFirst7 = 18,
kOmitFirst8 = 19,
kOmitFirst9 = 20,
};
struct Transform {
@@ -45,105 +54,127 @@ struct Transform {
};
static const Transform kTransforms[] = {
{ "", kIdentity, "" },
{ "", kIdentity, " ", },
{ "", kIdentity, "\">" },
{ "", kUppercaseFirst, "" },
{ "", kIdentity, "\"" },
{ "", kIdentity, ".", },
{ "", kIdentity, "=\"" },
{ "", kUppercaseFirst, " ", },
{ " ", kIdentity, "=\"" },
{ " ", kIdentity, " ", },
{ "", kIdentity, ":", },
{ " ", kIdentity, "" },
{ "", kIdentity, "\n" },
{ "", kIdentity, "(", },
{ "", kUppercaseAll, "" },
{ ".", kIdentity, "(", },
{ "", kIdentity, "'" },
{ "", kUppercaseFirst, "\"" },
{ " ", kUppercaseFirst, " ", },
{ "", kOmit3, "" },
{ "", kOmit4, "" },
{ ".", kIdentity, "" },
{ "", kOmit1, "" },
{ "", kOmit2, "" },
{ "", kUppercaseFirst, "\">" },
{ "", kOmit5, "" },
{ "", kUppercaseAll, " ", },
{ " ", kUppercaseFirst, "" },
{ "", kIdentity, ", ", },
{ "", kUppercaseFirst, "(", },
{ "", kIdentity, "\n\t" },
{ "", kUppercaseFirst, "'" },
{ ".", kIdentity, " ", },
{ " ", kUppercaseAll, " ", },
{ "", kIdentity, "='" },
{ "", kUppercaseFirst, ".", },
{ " ", kIdentity, ".", },
{ " ", kIdentity, ", ", },
{ " ", kUppercaseAll, "" },
{ "", kOmit6, "" },
{ "", kOmit9, "" },
{ "", kUppercaseAll, "\"" },
{ "", kIdentity, " the " },
{ "", kIdentity, " in " },
{ "", kIdentity, " of " },
{ "", kIdentity, " to " },
{ "", kIdentity, " and " },
{ "", kIdentity, " is " },
{ "", kIdentity, " on " },
{ "", kIdentity, " by " },
{ "", kIdentity, " for " },
{ "", kIdentity, " with " },
{ "", kIdentity, " from " },
{ "", kIdentity, " as " },
{ "", kIdentity, " at " },
{ "", kIdentity, "er " },
{ " ", kIdentity, "='" },
{ "", kIdentity, " a " },
{ "", kOmit7, "" },
{ "", kOmit8, "" },
{ " ", kIdentity, "(", },
{ " ", kIdentity, ". ", },
{ "", kIdentity, ". ", },
{ "", kIdentity, ",", },
{ "", kOmit1, "ing " },
{ "", kIdentity, "ed " },
{ "", kUppercaseFirst, ", ", },
{ "", kUppercaseAll, ".", },
{ "", kUppercaseAll, "=\"" },
{ "", kUppercaseAll, ", ", },
{ "", kUppercaseAll, "\">" },
{ " ", kUppercaseFirst, ".", },
{ " ", kUppercaseAll, "=\"" },
{ " ", kUppercaseFirst, ", ", },
{ "", kUppercaseAll, "'" },
{ "", kUppercaseFirst, "=\"" },
{ " ", kIdentity, ",", },
{ "", kIdentity, " that " },
{ "", kUppercaseFirst, "='" },
{ "", kUppercaseFirst, ". ", },
{ "", kUppercaseFirst, ",", },
{ "", kIdentity, ". The " },
{ "\xc2\xa0", kIdentity, "" },
{ " ", kUppercaseFirst, ". ", },
{ "", kUppercaseAll, ",", },
{ "", kUppercaseAll, "(", },
{ " ", kUppercaseAll, "='" },
{ "", kIdentity, "]" },
{ "", kUppercaseAll, "='" },
{ " ", kUppercaseAll, ".", },
{ "", kUppercaseAll, ". ", },
{ " ", kUppercaseFirst, "=\"" },
{ " ", kUppercaseAll, ". ", },
{ " ", kUppercaseFirst, ",", },
{ " ", kUppercaseAll, ", ", },
{ "", kIdentity, "ize " },
{ " ", kUppercaseFirst, "='" },
{ "", kIdentity, "est " },
{ "", kIdentity, ". This " },
{ "", kIdentity, "" },
{ "", kIdentity, " " },
{ " ", kIdentity, " " },
{ "", kOmitFirst1, "" },
{ "", kUppercaseFirst, " " },
{ "", kIdentity, " the " },
{ " ", kIdentity, "" },
{ "s ", kIdentity, " " },
{ "", kIdentity, " of " },
{ "", kUppercaseFirst, "" },
{ "", kIdentity, " and " },
{ "", kOmitFirst2, "" },
{ "", kOmitLast1, "" },
{ ", ", kIdentity, " " },
{ "", kIdentity, ", " },
{ " ", kUppercaseFirst, " " },
{ "", kIdentity, " in " },
{ "", kIdentity, " to " },
{ "e ", kIdentity, " " },
{ "", kIdentity, "\"" },
{ "", kIdentity, "." },
{ "", kIdentity, "\">" },
{ "", kIdentity, "\n" },
{ "", kOmitLast3, "" },
{ "", kIdentity, "]" },
{ "", kIdentity, " for " },
{ "", kOmitFirst3, "" },
{ "", kOmitLast2, "" },
{ "", kIdentity, " a " },
{ "", kIdentity, " that " },
{ " ", kUppercaseFirst, "" },
{ "", kIdentity, ". " },
{ ".", kIdentity, "" },
{ " ", kIdentity, ", " },
{ "", kOmitFirst4, "" },
{ "", kIdentity, " with " },
{ "", kIdentity, "'" },
{ "", kIdentity, " from " },
{ "", kIdentity, " by " },
{ "", kOmitFirst5, "" },
{ "", kOmitFirst6, "" },
{ " the ", kIdentity, "" },
{ "", kOmitLast4, "" },
{ "", kIdentity, ". The " },
{ "", kUppercaseAll, "" },
{ "", kIdentity, " on " },
{ "", kIdentity, " as " },
{ "", kIdentity, " is " },
{ "", kOmitLast7, "" },
{ "", kOmitLast1, "ing " },
{ "", kIdentity, "\n\t" },
{ "", kIdentity, ":" },
{ " ", kIdentity, ". " },
{ "", kIdentity, "ed " },
{ "", kOmitFirst9, "" },
{ "", kOmitFirst7, "" },
{ "", kOmitLast6, "" },
{ "", kIdentity, "(" },
{ "", kUppercaseFirst, ", " },
{ "", kOmitLast8, "" },
{ "", kIdentity, " at " },
{ "", kIdentity, "ly " },
{ " the ", kIdentity, " of " },
{ "", kOmitLast5, "" },
{ "", kOmitLast9, "" },
{ " ", kUppercaseFirst, ", " },
{ "", kUppercaseFirst, "\"" },
{ ".", kIdentity, "(" },
{ "", kUppercaseAll, " " },
{ "", kUppercaseFirst, "\">" },
{ "", kIdentity, "=\"" },
{ " ", kIdentity, "." },
{ ".com/", kIdentity, "" },
{ " the ", kIdentity, " of the " },
{ "", kUppercaseFirst, "'" },
{ "", kIdentity, ". This " },
{ "", kIdentity, "," },
{ ".", kIdentity, " " },
{ "", kUppercaseFirst, "(" },
{ "", kUppercaseFirst, "." },
{ "", kIdentity, " not " },
{ " ", kIdentity, "=\"" },
{ "", kIdentity, "er " },
{ " ", kUppercaseAll, " " },
{ "", kIdentity, "al " },
{ " ", kUppercaseAll, "" },
{ "", kIdentity, "='" },
{ "", kUppercaseAll, "\"" },
{ "", kUppercaseFirst, ". " },
{ " ", kIdentity, "(" },
{ "", kIdentity, "ful " },
{ " ", kUppercaseFirst, ". " },
{ "", kIdentity, "ive " },
{ "", kIdentity, "less " },
{ "", kUppercaseAll, "'" },
{ "", kIdentity, "est " },
{ " ", kUppercaseFirst, "." },
{ "", kUppercaseAll, "\">" },
{ " ", 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]);
@@ -169,19 +200,26 @@ static int ToUpperCase(uint8_t *p, int len) {
inline std::string ApplyTransform(
const Transform& t, const uint8_t* word, int len) {
std::string ret(t.prefix);
if (t.word_transform <= kOmit9) {
if (t.word_transform <= kOmitLast9) {
len -= t.word_transform;
}
if (len > 0) {
ret += std::string(word, word + len);
uint8_t *uppercase = reinterpret_cast<uint8_t*>(&ret[ret.size() - len]);
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;
if (t.word_transform >= kOmitFirst1) {
const int skip = t.word_transform - (kOmitFirst1 - 1);
if (len > skip) {
ret += std::string(word + skip, word + len);
}
} else {
ret += std::string(word, word + len);
uint8_t *uppercase = reinterpret_cast<uint8_t*>(&ret[ret.size() - len]);
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;
}
}
}
}