unicode-trie
Advanced tools
Comparing version 0.3.1 to 1.0.0
800
builder.js
@@ -1,173 +0,302 @@ | ||
// Generated by CoffeeScript 1.7.1 | ||
var UnicodeTrie, UnicodeTrieBuilder, pako; | ||
const UnicodeTrie = require('./'); | ||
const pako = require('pako'); | ||
UnicodeTrie = require('./'); | ||
// Shift size for getting the index-1 table offset. | ||
const SHIFT_1 = 6 + 5; | ||
pako = require('pako'); | ||
// Shift size for getting the index-2 table offset. | ||
const SHIFT_2 = 5; | ||
UnicodeTrieBuilder = (function() { | ||
var BAD_UTF8_DATA_OFFSET, CP_PER_INDEX_1_ENTRY, DATA_0800_OFFSET, DATA_BLOCK_LENGTH, DATA_GRANULARITY, DATA_MASK, DATA_NULL_OFFSET, DATA_START_OFFSET, INDEX_1_LENGTH, INDEX_1_OFFSET, INDEX_2_BLOCK_LENGTH, INDEX_2_BMP_LENGTH, INDEX_2_MASK, INDEX_2_NULL_OFFSET, INDEX_2_OFFSET, INDEX_2_START_OFFSET, INDEX_GAP_LENGTH, INDEX_GAP_OFFSET, INDEX_SHIFT, INITIAL_DATA_LENGTH, LSCP_INDEX_2_LENGTH, LSCP_INDEX_2_OFFSET, MAX_DATA_LENGTH, MAX_INDEX_1_LENGTH, MAX_INDEX_2_LENGTH, MAX_INDEX_LENGTH, MEDIUM_DATA_LENGTH, NEW_DATA_START_OFFSET, OMITTED_BMP_INDEX_1_LENGTH, SHIFT_1, SHIFT_1_2, SHIFT_2, UTF8_2B_INDEX_2_LENGTH, UTF8_2B_INDEX_2_OFFSET, equal_int; | ||
// Difference between the two shift sizes, | ||
// for getting an index-1 offset from an index-2 offset. 6=11-5 | ||
const SHIFT_1_2 = SHIFT_1 - SHIFT_2; | ||
SHIFT_1 = 6 + 5; | ||
// Number of index-1 entries for the BMP. 32=0x20 | ||
// This part of the index-1 table is omitted from the serialized form. | ||
const OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> SHIFT_1; | ||
SHIFT_2 = 5; | ||
// Number of code points per index-1 table entry. 2048=0x800 | ||
const CP_PER_INDEX_1_ENTRY = 1 << SHIFT_1; | ||
SHIFT_1_2 = SHIFT_1 - SHIFT_2; | ||
// Number of entries in an index-2 block. 64=0x40 | ||
const INDEX_2_BLOCK_LENGTH = 1 << SHIFT_1_2; | ||
OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> SHIFT_1; | ||
// Mask for getting the lower bits for the in-index-2-block offset. */ | ||
const INDEX_2_MASK = INDEX_2_BLOCK_LENGTH - 1; | ||
CP_PER_INDEX_1_ENTRY = 1 << SHIFT_1; | ||
// Number of entries in a data block. 32=0x20 | ||
const DATA_BLOCK_LENGTH = 1 << SHIFT_2; | ||
INDEX_2_BLOCK_LENGTH = 1 << SHIFT_1_2; | ||
// Mask for getting the lower bits for the in-data-block offset. | ||
const DATA_MASK = DATA_BLOCK_LENGTH - 1; | ||
INDEX_2_MASK = INDEX_2_BLOCK_LENGTH - 1; | ||
// Shift size for shifting left the index array values. | ||
// Increases possible data size with 16-bit index values at the cost | ||
// of compactability. | ||
// This requires data blocks to be aligned by DATA_GRANULARITY. | ||
const INDEX_SHIFT = 2; | ||
DATA_BLOCK_LENGTH = 1 << SHIFT_2; | ||
// The alignment size of a data block. Also the granularity for compaction. | ||
const DATA_GRANULARITY = 1 << INDEX_SHIFT; | ||
DATA_MASK = DATA_BLOCK_LENGTH - 1; | ||
// The BMP part of the index-2 table is fixed and linear and starts at offset 0. | ||
// Length=2048=0x800=0x10000>>SHIFT_2. | ||
const INDEX_2_OFFSET = 0; | ||
INDEX_SHIFT = 2; | ||
// The part of the index-2 table for U+D800..U+DBFF stores values for | ||
// lead surrogate code _units_ not code _points_. | ||
// Values for lead surrogate code _points_ are indexed with this portion of the table. | ||
// Length=32=0x20=0x400>>SHIFT_2. (There are 1024=0x400 lead surrogates.) | ||
const LSCP_INDEX_2_OFFSET = 0x10000 >> SHIFT_2; | ||
const LSCP_INDEX_2_LENGTH = 0x400 >> SHIFT_2; | ||
DATA_GRANULARITY = 1 << INDEX_SHIFT; | ||
// Count the lengths of both BMP pieces. 2080=0x820 | ||
const INDEX_2_BMP_LENGTH = LSCP_INDEX_2_OFFSET + LSCP_INDEX_2_LENGTH; | ||
INDEX_2_OFFSET = 0; | ||
// The 2-byte UTF-8 version of the index-2 table follows at offset 2080=0x820. | ||
// Length 32=0x20 for lead bytes C0..DF, regardless of SHIFT_2. | ||
const UTF8_2B_INDEX_2_OFFSET = INDEX_2_BMP_LENGTH; | ||
const UTF8_2B_INDEX_2_LENGTH = 0x800 >> 6; // U+0800 is the first code point after 2-byte UTF-8 | ||
LSCP_INDEX_2_OFFSET = 0x10000 >> SHIFT_2; | ||
// The index-1 table, only used for supplementary code points, at offset 2112=0x840. | ||
// Variable length, for code points up to highStart, where the last single-value range starts. | ||
// Maximum length 512=0x200=0x100000>>SHIFT_1. | ||
// (For 0x100000 supplementary code points U+10000..U+10ffff.) | ||
// | ||
// The part of the index-2 table for supplementary code points starts | ||
// after this index-1 table. | ||
// | ||
// Both the index-1 table and the following part of the index-2 table | ||
// are omitted completely if there is only BMP data. | ||
const INDEX_1_OFFSET = UTF8_2B_INDEX_2_OFFSET + UTF8_2B_INDEX_2_LENGTH; | ||
const MAX_INDEX_1_LENGTH = 0x100000 >> SHIFT_1; | ||
LSCP_INDEX_2_LENGTH = 0x400 >> SHIFT_2; | ||
// The illegal-UTF-8 data block follows the ASCII block, at offset 128=0x80. | ||
// Used with linear access for single bytes 0..0xbf for simple error handling. | ||
// Length 64=0x40, not DATA_BLOCK_LENGTH. | ||
const BAD_UTF8_DATA_OFFSET = 0x80; | ||
INDEX_2_BMP_LENGTH = LSCP_INDEX_2_OFFSET + LSCP_INDEX_2_LENGTH; | ||
// The start of non-linear-ASCII data blocks, at offset 192=0xc0. | ||
// !!!! | ||
const DATA_START_OFFSET = 0xc0; | ||
UTF8_2B_INDEX_2_OFFSET = INDEX_2_BMP_LENGTH; | ||
// The null data block. | ||
// Length 64=0x40 even if DATA_BLOCK_LENGTH is smaller, | ||
// to work with 6-bit trail bytes from 2-byte UTF-8. | ||
const DATA_NULL_OFFSET = DATA_START_OFFSET; | ||
UTF8_2B_INDEX_2_LENGTH = 0x800 >> 6; | ||
// The start of allocated data blocks. | ||
const NEW_DATA_START_OFFSET = DATA_NULL_OFFSET + 0x40; | ||
INDEX_1_OFFSET = UTF8_2B_INDEX_2_OFFSET + UTF8_2B_INDEX_2_LENGTH; | ||
// The start of data blocks for U+0800 and above. | ||
// Below, compaction uses a block length of 64 for 2-byte UTF-8. | ||
// From here on, compaction uses DATA_BLOCK_LENGTH. | ||
// Data values for 0x780 code points beyond ASCII. | ||
const DATA_0800_OFFSET = NEW_DATA_START_OFFSET + 0x780; | ||
MAX_INDEX_1_LENGTH = 0x100000 >> SHIFT_1; | ||
// Start with allocation of 16k data entries. */ | ||
const INITIAL_DATA_LENGTH = 1 << 14; | ||
BAD_UTF8_DATA_OFFSET = 0x80; | ||
// Grow about 8x each time. | ||
const MEDIUM_DATA_LENGTH = 1 << 17; | ||
DATA_START_OFFSET = 0xc0; | ||
// Maximum length of the runtime data array. | ||
// Limited by 16-bit index values that are left-shifted by INDEX_SHIFT, | ||
// and by uint16_t UTrie2Header.shiftedDataLength. | ||
const MAX_DATA_LENGTH_RUNTIME = 0xffff << INDEX_SHIFT; | ||
DATA_NULL_OFFSET = DATA_START_OFFSET; | ||
const INDEX_1_LENGTH = 0x110000 >> SHIFT_1; | ||
NEW_DATA_START_OFFSET = DATA_NULL_OFFSET + 0x40; | ||
// Maximum length of the build-time data array. | ||
// One entry per 0x110000 code points, plus the illegal-UTF-8 block and the null block, | ||
// plus values for the 0x400 surrogate code units. | ||
const MAX_DATA_LENGTH_BUILDTIME = 0x110000 + 0x40 + 0x40 + 0x400; | ||
DATA_0800_OFFSET = NEW_DATA_START_OFFSET + 0x780; | ||
// At build time, leave a gap in the index-2 table, | ||
// at least as long as the maximum lengths of the 2-byte UTF-8 index-2 table | ||
// and the supplementary index-1 table. | ||
// Round up to INDEX_2_BLOCK_LENGTH for proper compacting. | ||
const INDEX_GAP_OFFSET = INDEX_2_BMP_LENGTH; | ||
const INDEX_GAP_LENGTH = ((UTF8_2B_INDEX_2_LENGTH + MAX_INDEX_1_LENGTH) + INDEX_2_MASK) & ~INDEX_2_MASK; | ||
INITIAL_DATA_LENGTH = 1 << 14; | ||
// Maximum length of the build-time index-2 array. | ||
// Maximum number of Unicode code points (0x110000) shifted right by SHIFT_2, | ||
// plus the part of the index-2 table for lead surrogate code points, | ||
// plus the build-time index gap, | ||
// plus the null index-2 block.) | ||
const MAX_INDEX_2_LENGTH = (0x110000 >> SHIFT_2) + LSCP_INDEX_2_LENGTH + INDEX_GAP_LENGTH + INDEX_2_BLOCK_LENGTH; | ||
MEDIUM_DATA_LENGTH = 1 << 17; | ||
// The null index-2 block, following the gap in the index-2 table. | ||
const INDEX_2_NULL_OFFSET = INDEX_GAP_OFFSET + INDEX_GAP_LENGTH; | ||
MAX_DATA_LENGTH = 0xffff << INDEX_SHIFT; | ||
// The start of allocated index-2 blocks. | ||
const INDEX_2_START_OFFSET = INDEX_2_NULL_OFFSET + INDEX_2_BLOCK_LENGTH; | ||
INDEX_1_LENGTH = 0x110000 >> SHIFT_1; | ||
// Maximum length of the runtime index array. | ||
// Limited by its own 16-bit index values, and by uint16_t UTrie2Header.indexLength. | ||
// (The actual maximum length is lower, | ||
// (0x110000>>SHIFT_2)+UTF8_2B_INDEX_2_LENGTH+MAX_INDEX_1_LENGTH.) | ||
const MAX_INDEX_LENGTH = 0xffff; | ||
MAX_DATA_LENGTH = 0x110000 + 0x40 + 0x40 + 0x400; | ||
const equal_int = (a, s, t, length) => { | ||
for (let i = 0; i < length; i++) { | ||
if (a[s + i] !== a[t + i]) { | ||
return false; | ||
} | ||
} | ||
INDEX_GAP_OFFSET = INDEX_2_BMP_LENGTH; | ||
return true; | ||
}; | ||
INDEX_GAP_LENGTH = ((UTF8_2B_INDEX_2_LENGTH + MAX_INDEX_1_LENGTH) + INDEX_2_MASK) & ~INDEX_2_MASK; | ||
MAX_INDEX_2_LENGTH = (0x110000 >> SHIFT_2) + LSCP_INDEX_2_LENGTH + INDEX_GAP_LENGTH + INDEX_2_BLOCK_LENGTH; | ||
INDEX_2_NULL_OFFSET = INDEX_GAP_OFFSET + INDEX_GAP_LENGTH; | ||
INDEX_2_START_OFFSET = INDEX_2_NULL_OFFSET + INDEX_2_BLOCK_LENGTH; | ||
MAX_INDEX_LENGTH = 0xffff; | ||
function UnicodeTrieBuilder(initialValue, errorValue) { | ||
var i, j, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _ref, _s, _t; | ||
this.initialValue = initialValue != null ? initialValue : 0; | ||
this.errorValue = errorValue != null ? errorValue : 0; | ||
class UnicodeTrieBuilder { | ||
constructor(initialValue, errorValue) { | ||
let i, j; | ||
if (initialValue == null) { | ||
initialValue = 0; | ||
} | ||
this.initialValue = initialValue; | ||
if (errorValue == null) { | ||
errorValue = 0; | ||
} | ||
this.errorValue = errorValue; | ||
this.index1 = new Int32Array(INDEX_1_LENGTH); | ||
this.index2 = new Int32Array(MAX_INDEX_2_LENGTH); | ||
this.highStart = 0x110000; | ||
this.data = new Uint32Array(INITIAL_DATA_LENGTH); | ||
this.dataCapacity = INITIAL_DATA_LENGTH; | ||
this.firstFreeBlock = 0; | ||
this.isCompacted = false; | ||
this.map = new Int32Array(MAX_DATA_LENGTH >> SHIFT_2); | ||
for (i = _i = 0; _i < 0x80; i = _i += 1) { | ||
// Multi-purpose per-data-block table. | ||
// | ||
// Before compacting: | ||
// | ||
// Per-data-block reference counters/free-block list. | ||
// 0: unused | ||
// >0: reference counter (number of index-2 entries pointing here) | ||
// <0: next free data block in free-block list | ||
// | ||
// While compacting: | ||
// | ||
// Map of adjusted indexes, used in compactData() and compactIndex2(). | ||
// Maps from original indexes to new ones. | ||
this.map = new Int32Array(MAX_DATA_LENGTH_BUILDTIME >> SHIFT_2); | ||
for (i = 0; i < 0x80; i++) { | ||
this.data[i] = this.initialValue; | ||
} | ||
for (i = _j = i; _j < 0xc0; i = _j += 1) { | ||
for (i = i; i < 0xc0; i++) { | ||
this.data[i] = this.errorValue; | ||
} | ||
for (i = _k = DATA_NULL_OFFSET; _k < NEW_DATA_START_OFFSET; i = _k += 1) { | ||
for (i = DATA_NULL_OFFSET; i < NEW_DATA_START_OFFSET; i++) { | ||
this.data[i] = this.initialValue; | ||
} | ||
this.dataNullOffset = DATA_NULL_OFFSET; | ||
this.dataLength = NEW_DATA_START_OFFSET; | ||
// set the index-2 indexes for the 2=0x80>>SHIFT_2 ASCII data blocks | ||
i = 0; | ||
for (j = _l = 0; _l < 128; j = _l += DATA_BLOCK_LENGTH) { | ||
for (j = 0; j < 0x80; j += DATA_BLOCK_LENGTH) { | ||
this.index2[i] = j; | ||
this.map[i++] = 1; | ||
} | ||
for (j = _m = j; DATA_BLOCK_LENGTH > 0 ? _m < 0xc0 : _m > 0xc0; j = _m += DATA_BLOCK_LENGTH) { | ||
// reference counts for the bad-UTF-8-data block | ||
for (j = j; j < 0xc0; j += DATA_BLOCK_LENGTH) { | ||
this.map[i++] = 0; | ||
} | ||
this.map[i++] = (0x110000 >> SHIFT_2) - (0x80 >> SHIFT_2) + 1 + LSCP_INDEX_2_LENGTH; | ||
// Reference counts for the null data block: all blocks except for the ASCII blocks. | ||
// Plus 1 so that we don't drop this block during compaction. | ||
// Plus as many as needed for lead surrogate code points. | ||
// i==newTrie->dataNullOffset | ||
this.map[i++] = ((0x110000 >> SHIFT_2) - (0x80 >> SHIFT_2)) + 1 + LSCP_INDEX_2_LENGTH; | ||
j += DATA_BLOCK_LENGTH; | ||
for (j = _n = j; DATA_BLOCK_LENGTH > 0 ? _n < NEW_DATA_START_OFFSET : _n > NEW_DATA_START_OFFSET; j = _n += DATA_BLOCK_LENGTH) { | ||
for (j = j; j < NEW_DATA_START_OFFSET; j += DATA_BLOCK_LENGTH) { | ||
this.map[i++] = 0; | ||
} | ||
for (i = _o = _ref = 0x80 >> SHIFT_2; _o < INDEX_2_BMP_LENGTH; i = _o += 1) { | ||
// set the remaining indexes in the BMP index-2 block | ||
// to the null data block | ||
for (i = 0x80 >> SHIFT_2; i < INDEX_2_BMP_LENGTH; i++) { | ||
this.index2[i] = DATA_NULL_OFFSET; | ||
} | ||
for (i = _p = 0; _p < INDEX_GAP_LENGTH; i = _p += 1) { | ||
// Fill the index gap with impossible values so that compaction | ||
// does not overlap other index-2 blocks with the gap. | ||
for (i = 0; i < INDEX_GAP_LENGTH; i++) { | ||
this.index2[INDEX_GAP_OFFSET + i] = -1; | ||
} | ||
for (i = _q = 0; _q < INDEX_2_BLOCK_LENGTH; i = _q += 1) { | ||
// set the indexes in the null index-2 block | ||
for (i = 0; i < INDEX_2_BLOCK_LENGTH; i++) { | ||
this.index2[INDEX_2_NULL_OFFSET + i] = DATA_NULL_OFFSET; | ||
} | ||
this.index2NullOffset = INDEX_2_NULL_OFFSET; | ||
this.index2Length = INDEX_2_START_OFFSET; | ||
// set the index-1 indexes for the linear index-2 block | ||
j = 0; | ||
for (i = _r = 0; _r < OMITTED_BMP_INDEX_1_LENGTH; i = _r += 1) { | ||
for (i = 0; i < OMITTED_BMP_INDEX_1_LENGTH; i++) { | ||
this.index1[i] = j; | ||
j += INDEX_2_BLOCK_LENGTH; | ||
} | ||
for (i = _s = i; _s < INDEX_1_LENGTH; i = _s += 1) { | ||
// set the remaining index-1 indexes to the null index-2 block | ||
for (i = i; i < INDEX_1_LENGTH; i++) { | ||
this.index1[i] = INDEX_2_NULL_OFFSET; | ||
} | ||
for (i = _t = 0x80; _t < 2048; i = _t += DATA_BLOCK_LENGTH) { | ||
// Preallocate and reset data for U+0080..U+07ff, | ||
// for 2-byte UTF-8 which will be compacted in 64-blocks | ||
// even if DATA_BLOCK_LENGTH is smaller. | ||
for (i = 0x80; i < 0x800; i += DATA_BLOCK_LENGTH) { | ||
this.set(i, this.initialValue); | ||
} | ||
return; | ||
} | ||
UnicodeTrieBuilder.prototype.set = function(codePoint, value) { | ||
var block; | ||
if (codePoint < 0 || codePoint > 0x10ffff) { | ||
set(codePoint, value) { | ||
if ((codePoint < 0) || (codePoint > 0x10ffff)) { | ||
throw new Error('Invalid code point'); | ||
} | ||
if (this.isCompacted) { | ||
throw new Error('Already compacted'); | ||
} | ||
block = this._getDataBlock(codePoint, true); | ||
const block = this._getDataBlock(codePoint, true); | ||
this.data[block + (codePoint & DATA_MASK)] = value; | ||
return this; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype.setRange = function(start, end, value, overwrite) { | ||
var block, i2, limit, nextStart, repeatBlock, rest, setRepeatBlock; | ||
setRange(start, end, value, overwrite) { | ||
let block, repeatBlock; | ||
if (overwrite == null) { | ||
overwrite = true; | ||
} | ||
if (start > 0x10ffff || end > 0x10ffff || start > end) { | ||
if ((start > 0x10ffff) || (end > 0x10ffff) || (start > end)) { | ||
throw new Error('Invalid code point'); | ||
} | ||
if (this.isCompacted) { | ||
throw new Error('Already compacted'); | ||
} | ||
if (!overwrite && value === this.initialValue) { | ||
return this; | ||
if (!overwrite && (value === this.initialValue)) { | ||
return this; // nothing to do | ||
} | ||
limit = end + 1; | ||
let limit = end + 1; | ||
if ((start & DATA_MASK) !== 0) { | ||
// set partial block at [start..following block boundary | ||
block = this._getDataBlock(start, true); | ||
nextStart = (start + DATA_BLOCK_LENGTH) & ~DATA_MASK; | ||
const nextStart = (start + DATA_BLOCK_LENGTH) & ~DATA_MASK; | ||
if (nextStart <= limit) { | ||
@@ -181,4 +310,10 @@ this._fillBlock(block, start & DATA_MASK, DATA_BLOCK_LENGTH, value, this.initialValue, overwrite); | ||
} | ||
rest = limit & DATA_MASK; | ||
// number of positions in the last, partial block | ||
const rest = limit & DATA_MASK; | ||
// round down limit to a block boundary | ||
limit &= ~DATA_MASK; | ||
// iterate over all-value blocks | ||
if (value === this.initialValue) { | ||
@@ -189,20 +324,46 @@ repeatBlock = this.dataNullOffset; | ||
} | ||
while (start < limit) { | ||
setRepeatBlock = false; | ||
if (value === this.initialValue && this._isInNullBlock(start, true)) { | ||
start += DATA_BLOCK_LENGTH; | ||
let setRepeatBlock = false; | ||
if ((value === this.initialValue) && this._isInNullBlock(start, true)) { | ||
start += DATA_BLOCK_LENGTH; // nothing to do | ||
continue; | ||
} | ||
i2 = this._getIndex2Block(start, true); | ||
// get index value | ||
let i2 = this._getIndex2Block(start, true); | ||
i2 += (start >> SHIFT_2) & INDEX_2_MASK; | ||
block = this.index2[i2]; | ||
if (this._isWritableBlock(block)) { | ||
if (overwrite && block >= DATA_0800_OFFSET) { | ||
// already allocated | ||
if (overwrite && (block >= DATA_0800_OFFSET)) { | ||
// We overwrite all values, and it's not a | ||
// protected (ASCII-linear or 2-byte UTF-8) block: | ||
// replace with the repeatBlock. | ||
setRepeatBlock = true; | ||
} else { | ||
// protected block: just write the values into this block | ||
this._fillBlock(block, 0, DATA_BLOCK_LENGTH, value, this.initialValue, overwrite); | ||
} | ||
} else if (this.data[block] !== value && (overwrite || block === this.dataNullOffset)) { | ||
} else if ((this.data[block] !== value) && (overwrite || (block === this.dataNullOffset))) { | ||
// Set the repeatBlock instead of the null block or previous repeat block: | ||
// | ||
// If !isWritableBlock() then all entries in the block have the same value | ||
// because it's the null block or a range block (the repeatBlock from a previous | ||
// call to utrie2_setRange32()). | ||
// No other blocks are used multiple times before compacting. | ||
// | ||
// The null block is the only non-writable block with the initialValue because | ||
// of the repeatBlock initialization above. (If value==initialValue, then | ||
// the repeatBlock will be the null data block.) | ||
// | ||
// We set our repeatBlock if the desired value differs from the block's value, | ||
// and if we overwrite any data or if the data is all initial values | ||
// (which is the same as the block being the null block, see above). | ||
setRepeatBlock = true; | ||
} | ||
if (setRepeatBlock) { | ||
@@ -212,2 +373,3 @@ if (repeatBlock >= 0) { | ||
} else { | ||
// create and set and fill the repeatBlock | ||
repeatBlock = this._getDataBlock(start, true); | ||
@@ -217,23 +379,29 @@ this._writeBlock(repeatBlock, value); | ||
} | ||
start += DATA_BLOCK_LENGTH; | ||
} | ||
if (rest > 0) { | ||
// set partial block at [last block boundary..limit | ||
block = this._getDataBlock(start, true); | ||
this._fillBlock(block, 0, rest, value, this.initialValue, overwrite); | ||
} | ||
return this; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype.get = function(c, fromLSCP) { | ||
var block, i2; | ||
get(c, fromLSCP) { | ||
let i2; | ||
if (fromLSCP == null) { | ||
fromLSCP = true; | ||
} | ||
if (c < 0 || c > 0x10ffff) { | ||
if ((c < 0) || (c > 0x10ffff)) { | ||
return this.errorValue; | ||
} | ||
if (c >= this.highStart && (!(c >= 0xd800 && c < 0xdc00) || fromLSCP)) { | ||
if ((c >= this.highStart) && (!((c >= 0xd800) && (c < 0xdc00)) || fromLSCP)) { | ||
return this.data[this.dataLength - DATA_GRANULARITY]; | ||
} | ||
if ((c >= 0xd800 && c < 0xdc00) && fromLSCP) { | ||
if (((c >= 0xd800) && (c < 0xdc00)) && fromLSCP) { | ||
i2 = (LSCP_INDEX_2_OFFSET - (0xd800 >> SHIFT_2)) + (c >> SHIFT_2); | ||
@@ -243,36 +411,42 @@ } else { | ||
} | ||
block = this.index2[i2]; | ||
const block = this.index2[i2]; | ||
return this.data[block + (c & DATA_MASK)]; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._isInNullBlock = function(c, forLSCP) { | ||
var block, i2; | ||
if ((c & 0xfffffc00) === 0xd800 && forLSCP) { | ||
i2 = LSCP_INDEX_2_OFFSET - (0xd800 >> SHIFT_2) + (c >> SHIFT_2); | ||
_isInNullBlock(c, forLSCP) { | ||
let i2; | ||
if (((c & 0xfffffc00) === 0xd800) && forLSCP) { | ||
i2 = (LSCP_INDEX_2_OFFSET - (0xd800 >> SHIFT_2)) + (c >> SHIFT_2); | ||
} else { | ||
i2 = this.index1[c >> SHIFT_1] + ((c >> SHIFT_2) & INDEX_2_MASK); | ||
} | ||
block = this.index2[i2]; | ||
const block = this.index2[i2]; | ||
return block === this.dataNullOffset; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._allocIndex2Block = function() { | ||
var newBlock, newTop; | ||
newBlock = this.index2Length; | ||
newTop = newBlock + INDEX_2_BLOCK_LENGTH; | ||
_allocIndex2Block() { | ||
const newBlock = this.index2Length; | ||
const newTop = newBlock + INDEX_2_BLOCK_LENGTH; | ||
if (newTop > this.index2.length) { | ||
// Should never occur. | ||
// Either MAX_BUILD_TIME_INDEX_LENGTH is incorrect, | ||
// or the code writes more values than should be possible. | ||
throw new Error("Internal error in Trie2 creation."); | ||
} | ||
this.index2Length = newTop; | ||
this.index2.set(this.index2.subarray(this.index2NullOffset, this.index2NullOffset + INDEX_2_BLOCK_LENGTH), newBlock); | ||
return newBlock; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._getIndex2Block = function(c, forLSCP) { | ||
var i1, i2; | ||
if (c >= 0xd800 && c < 0xdc00 && forLSCP) { | ||
_getIndex2Block(c, forLSCP) { | ||
if ((c >= 0xd800) && (c < 0xdc00) && forLSCP) { | ||
return LSCP_INDEX_2_OFFSET; | ||
} | ||
i1 = c >> SHIFT_1; | ||
i2 = this.index1[i1]; | ||
const i1 = c >> SHIFT_1; | ||
let i2 = this.index1[i1]; | ||
if (i2 === this.index2NullOffset) { | ||
@@ -282,26 +456,35 @@ i2 = this._allocIndex2Block(); | ||
} | ||
return i2; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._isWritableBlock = function(block) { | ||
return block !== this.dataNullOffset && this.map[block >> SHIFT_2] === 1; | ||
}; | ||
_isWritableBlock(block) { | ||
return (block !== this.dataNullOffset) && (this.map[block >> SHIFT_2] === 1); | ||
} | ||
UnicodeTrieBuilder.prototype._allocDataBlock = function(copyBlock) { | ||
var capacity, newBlock, newData, newTop; | ||
_allocDataBlock(copyBlock) { | ||
let newBlock; | ||
if (this.firstFreeBlock !== 0) { | ||
// get the first free block | ||
newBlock = this.firstFreeBlock; | ||
this.firstFreeBlock = -this.map[newBlock >> SHIFT_2]; | ||
} else { | ||
// get a new block from the high end | ||
newBlock = this.dataLength; | ||
newTop = newBlock + DATA_BLOCK_LENGTH; | ||
const newTop = newBlock + DATA_BLOCK_LENGTH; | ||
if (newTop > this.dataCapacity) { | ||
// out of memory in the data array | ||
let capacity; | ||
if (this.dataCapacity < MEDIUM_DATA_LENGTH) { | ||
capacity = MEDIUM_DATA_LENGTH; | ||
} else if (this.dataCapacity < MAX_DATA_LENGTH) { | ||
capacity = MAX_DATA_LENGTH; | ||
} else if (this.dataCapacity < MAX_DATA_LENGTH_BUILDTIME) { | ||
capacity = MAX_DATA_LENGTH_BUILDTIME; | ||
} else { | ||
// Should never occur. | ||
// Either MAX_DATA_LENGTH_BUILDTIME is incorrect, | ||
// or the code writes more values than should be possible. | ||
throw new Error("Internal error in Trie2 creation."); | ||
} | ||
newData = new Uint32Array(capacity); | ||
const newData = new Uint32Array(capacity); | ||
newData.set(this.data.subarray(0, this.dataLength)); | ||
@@ -311,45 +494,50 @@ this.data = newData; | ||
} | ||
this.dataLength = newTop; | ||
} | ||
this.data.set(this.data.subarray(copyBlock, copyBlock + DATA_BLOCK_LENGTH), newBlock); | ||
this.map[newBlock >> SHIFT_2] = 0; | ||
return newBlock; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._releaseDataBlock = function(block) { | ||
_releaseDataBlock(block) { | ||
// put this block at the front of the free-block chain | ||
this.map[block >> SHIFT_2] = -this.firstFreeBlock; | ||
return this.firstFreeBlock = block; | ||
}; | ||
this.firstFreeBlock = block; | ||
} | ||
UnicodeTrieBuilder.prototype._setIndex2Entry = function(i2, block) { | ||
var oldBlock; | ||
++this.map[block >> SHIFT_2]; | ||
oldBlock = this.index2[i2]; | ||
_setIndex2Entry(i2, block) { | ||
++this.map[block >> SHIFT_2]; // increment first, in case block == oldBlock! | ||
const oldBlock = this.index2[i2]; | ||
if (--this.map[oldBlock >> SHIFT_2] === 0) { | ||
this._releaseDataBlock(oldBlock); | ||
} | ||
return this.index2[i2] = block; | ||
}; | ||
UnicodeTrieBuilder.prototype._getDataBlock = function(c, forLSCP) { | ||
var i2, newBlock, oldBlock; | ||
i2 = this._getIndex2Block(c, forLSCP); | ||
this.index2[i2] = block; | ||
} | ||
_getDataBlock(c, forLSCP) { | ||
let i2 = this._getIndex2Block(c, forLSCP); | ||
i2 += (c >> SHIFT_2) & INDEX_2_MASK; | ||
oldBlock = this.index2[i2]; | ||
const oldBlock = this.index2[i2]; | ||
if (this._isWritableBlock(oldBlock)) { | ||
return oldBlock; | ||
} | ||
newBlock = this._allocDataBlock(oldBlock); | ||
// allocate a new data block | ||
const newBlock = this._allocDataBlock(oldBlock); | ||
this._setIndex2Entry(i2, newBlock); | ||
return newBlock; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._fillBlock = function(block, start, limit, value, initialValue, overwrite) { | ||
var i, _i, _j, _ref, _ref1, _ref2, _ref3; | ||
_fillBlock(block, start, limit, value, initialValue, overwrite) { | ||
let i; | ||
if (overwrite) { | ||
for (i = _i = _ref = block + start, _ref1 = block + limit; _i < _ref1; i = _i += 1) { | ||
for (i = block + start; i < block + limit; i++) { | ||
this.data[i] = value; | ||
} | ||
} else { | ||
for (i = _j = _ref2 = block + start, _ref3 = block + limit; _j < _ref3; i = _j += 1) { | ||
for (i = block + start; i < block + limit; i++) { | ||
if (this.data[i] === initialValue) { | ||
@@ -360,18 +548,19 @@ this.data[i] = value; | ||
} | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._writeBlock = function(block, value) { | ||
var limit; | ||
limit = block + DATA_BLOCK_LENGTH; | ||
_writeBlock(block, value) { | ||
const limit = block + DATA_BLOCK_LENGTH; | ||
while (block < limit) { | ||
this.data[block++] = value; | ||
} | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._findHighStart = function(highValue) { | ||
var block, c, data32, i1, i2, i2Block, index2NullOffset, initialValue, j, nullBlock, prev, prevBlock, prevI2Block, value; | ||
data32 = this.data; | ||
initialValue = this.initialValue; | ||
index2NullOffset = this.index2NullOffset; | ||
nullBlock = this.dataNullOffset; | ||
_findHighStart(highValue) { | ||
let prevBlock, prevI2Block; | ||
const data32 = this.data; | ||
const { initialValue } = this; | ||
const { index2NullOffset } = this; | ||
const nullBlock = this.dataNullOffset; | ||
// set variables for previous range | ||
if (highValue === initialValue) { | ||
@@ -384,13 +573,19 @@ prevI2Block = index2NullOffset; | ||
} | ||
prev = 0x110000; | ||
i1 = INDEX_1_LENGTH; | ||
c = prev; | ||
const prev = 0x110000; | ||
// enumerate index-2 blocks | ||
let i1 = INDEX_1_LENGTH; | ||
let c = prev; | ||
while (c > 0) { | ||
i2Block = this.index1[--i1]; | ||
const i2Block = this.index1[--i1]; | ||
if (i2Block === prevI2Block) { | ||
// the index-2 block is the same as the previous one, and filled with highValue | ||
c -= CP_PER_INDEX_1_ENTRY; | ||
continue; | ||
} | ||
prevI2Block = i2Block; | ||
if (i2Block === index2NullOffset) { | ||
// this is the null index-2 block | ||
if (highValue !== initialValue) { | ||
@@ -401,11 +596,15 @@ return c; | ||
} else { | ||
i2 = INDEX_2_BLOCK_LENGTH; | ||
// enumerate data blocks for one index-2 block | ||
let i2 = INDEX_2_BLOCK_LENGTH; | ||
while (i2 > 0) { | ||
block = this.index2[i2Block + --i2]; | ||
const block = this.index2[i2Block + --i2]; | ||
if (block === prevBlock) { | ||
// the block is the same as the previous one, and filled with highValue | ||
c -= DATA_BLOCK_LENGTH; | ||
continue; | ||
} | ||
prevBlock = block; | ||
if (block === nullBlock) { | ||
// this is the null data block | ||
if (highValue !== initialValue) { | ||
@@ -416,5 +615,5 @@ return c; | ||
} else { | ||
j = DATA_BLOCK_LENGTH; | ||
let j = DATA_BLOCK_LENGTH; | ||
while (j > 0) { | ||
value = data32[block + --j]; | ||
const value = data32[block + --j]; | ||
if (value !== highValue) { | ||
@@ -429,19 +628,11 @@ return c; | ||
} | ||
// deliver last range | ||
return 0; | ||
}; | ||
} | ||
equal_int = function(a, s, t, length) { | ||
var i, _i; | ||
for (i = _i = 0; _i < length; i = _i += 1) { | ||
if (a[s + i] !== a[t + i]) { | ||
return false; | ||
} | ||
} | ||
return true; | ||
}; | ||
UnicodeTrieBuilder.prototype._findSameDataBlock = function(dataLength, otherBlock, blockLength) { | ||
var block; | ||
_findSameDataBlock(dataLength, otherBlock, blockLength) { | ||
// ensure that we do not even partially get past dataLength | ||
dataLength -= blockLength; | ||
block = 0; | ||
let block = 0; | ||
while (block <= dataLength) { | ||
@@ -453,9 +644,10 @@ if (equal_int(this.data, block, otherBlock, blockLength)) { | ||
} | ||
return -1; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._findSameIndex2Block = function(index2Length, otherBlock) { | ||
var block, _i; | ||
_findSameIndex2Block(index2Length, otherBlock) { | ||
// ensure that we do not even partially get past index2Length | ||
index2Length -= INDEX_2_BLOCK_LENGTH; | ||
for (block = _i = 0; _i <= index2Length; block = _i += 1) { | ||
for (let block = 0; block <= index2Length; block++) { | ||
if (equal_int(this.index2, block, otherBlock, INDEX_2_BLOCK_LENGTH)) { | ||
@@ -465,10 +657,12 @@ return block; | ||
} | ||
return -1; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._compactData = function() { | ||
var blockCount, blockLength, i, mapIndex, movedStart, newStart, overlap, start, _i, _j, _k, _l, _ref; | ||
newStart = DATA_START_OFFSET; | ||
start = 0; | ||
i = 0; | ||
_compactData() { | ||
// do not compact linear-ASCII data | ||
let newStart = DATA_START_OFFSET; | ||
let start = 0; | ||
let i = 0; | ||
while (start < newStart) { | ||
@@ -478,6 +672,13 @@ this.map[i++] = start; | ||
} | ||
blockLength = 64; | ||
blockCount = blockLength >> SHIFT_2; | ||
// Start with a block length of 64 for 2-byte UTF-8, | ||
// then switch to DATA_BLOCK_LENGTH. | ||
let blockLength = 64; | ||
let blockCount = blockLength >> SHIFT_2; | ||
start = newStart; | ||
while (start < this.dataLength) { | ||
// start: index of first entry of current block | ||
// newStart: index where the current block is to be moved | ||
// (right after current end of already-compacted data) | ||
var mapIndex, movedStart; | ||
if (start === DATA_0800_OFFSET) { | ||
@@ -487,41 +688,66 @@ blockLength = DATA_BLOCK_LENGTH; | ||
} | ||
// skip blocks that are not used | ||
if (this.map[start >> SHIFT_2] <= 0) { | ||
// advance start to the next block | ||
start += blockLength; | ||
// leave newStart with the previous block! | ||
continue; | ||
} | ||
// search for an identical block | ||
if ((movedStart = this._findSameDataBlock(newStart, start, blockLength)) >= 0) { | ||
// found an identical block, set the other block's index value for the current block | ||
mapIndex = start >> SHIFT_2; | ||
for (i = _i = blockCount; _i > 0; i = _i += -1) { | ||
for (i = blockCount; i > 0; i--) { | ||
this.map[mapIndex++] = movedStart; | ||
movedStart += DATA_BLOCK_LENGTH; | ||
} | ||
// advance start to the next block | ||
start += blockLength; | ||
// leave newStart with the previous block! | ||
continue; | ||
} | ||
overlap = blockLength - DATA_GRANULARITY; | ||
while (overlap > 0 && !equal_int(this.data, newStart - overlap, start, overlap)) { | ||
// see if the beginning of this block can be overlapped with the end of the previous block | ||
// look for maximum overlap (modulo granularity) with the previous, adjacent block | ||
let overlap = blockLength - DATA_GRANULARITY; | ||
while ((overlap > 0) && !equal_int(this.data, (newStart - overlap), start, overlap)) { | ||
overlap -= DATA_GRANULARITY; | ||
} | ||
if (overlap > 0 || newStart < start) { | ||
if ((overlap > 0) || (newStart < start)) { | ||
// some overlap, or just move the whole block | ||
movedStart = newStart - overlap; | ||
mapIndex = start >> SHIFT_2; | ||
for (i = _j = blockCount; _j > 0; i = _j += -1) { | ||
for (i = blockCount; i > 0; i--) { | ||
this.map[mapIndex++] = movedStart; | ||
movedStart += DATA_BLOCK_LENGTH; | ||
} | ||
// move the non-overlapping indexes to their new positions | ||
start += overlap; | ||
for (i = _k = _ref = blockLength - overlap; _k > 0; i = _k += -1) { | ||
for (i = blockLength - overlap; i > 0; i--) { | ||
this.data[newStart++] = this.data[start++]; | ||
} | ||
} else { | ||
} else { // no overlap && newStart==start | ||
mapIndex = start >> SHIFT_2; | ||
for (i = _l = blockCount; _l > 0; i = _l += -1) { | ||
for (i = blockCount; i > 0; i--) { | ||
this.map[mapIndex++] = start; | ||
start += DATA_BLOCK_LENGTH; | ||
} | ||
newStart = start; | ||
} | ||
} | ||
// now adjust the index-2 table | ||
i = 0; | ||
while (i < this.index2Length) { | ||
// Gap indexes are invalid (-1). Skip over the gap. | ||
if (i === INDEX_GAP_OFFSET) { | ||
@@ -533,3 +759,6 @@ i += INDEX_GAP_LENGTH; | ||
} | ||
this.dataNullOffset = this.map[this.dataNullOffset >> SHIFT_2]; | ||
// ensure dataLength alignment | ||
while ((newStart & (DATA_GRANULARITY - 1)) !== 0) { | ||
@@ -539,9 +768,10 @@ this.data[newStart++] = this.initialValue; | ||
this.dataLength = newStart; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype._compactIndex2 = function() { | ||
var i, movedStart, newStart, overlap, start, _i, _j, _ref; | ||
newStart = INDEX_2_BMP_LENGTH; | ||
start = 0; | ||
i = 0; | ||
_compactIndex2() { | ||
// do not compact linear-BMP index-2 blocks | ||
let newStart = INDEX_2_BMP_LENGTH; | ||
let start = 0; | ||
let i = 0; | ||
while (start < newStart) { | ||
@@ -551,21 +781,42 @@ this.map[i++] = start; | ||
} | ||
// Reduce the index table gap to what will be needed at runtime. | ||
newStart += UTF8_2B_INDEX_2_LENGTH + ((this.highStart - 0x10000) >> SHIFT_1); | ||
start = INDEX_2_NULL_OFFSET; | ||
while (start < this.index2Length) { | ||
// start: index of first entry of current block | ||
// newStart: index where the current block is to be moved | ||
// (right after current end of already-compacted data) | ||
// search for an identical block | ||
var movedStart; | ||
if ((movedStart = this._findSameIndex2Block(newStart, start)) >= 0) { | ||
// found an identical block, set the other block's index value for the current block | ||
this.map[start >> SHIFT_1_2] = movedStart; | ||
// advance start to the next block | ||
start += INDEX_2_BLOCK_LENGTH; | ||
// leave newStart with the previous block! | ||
continue; | ||
} | ||
overlap = INDEX_2_BLOCK_LENGTH - 1; | ||
while (overlap > 0 && !equal_int(this.index2, newStart - overlap, start, overlap)) { | ||
// see if the beginning of this block can be overlapped with the end of the previous block | ||
// look for maximum overlap with the previous, adjacent block | ||
let overlap = INDEX_2_BLOCK_LENGTH - 1; | ||
while ((overlap > 0) && !equal_int(this.index2, (newStart - overlap), start, overlap)) { | ||
--overlap; | ||
} | ||
if (overlap > 0 || newStart < start) { | ||
if ((overlap > 0) || (newStart < start)) { | ||
// some overlap, or just move the whole block | ||
this.map[start >> SHIFT_1_2] = newStart - overlap; | ||
// move the non-overlapping indexes to their new positions | ||
start += overlap; | ||
for (i = _i = _ref = INDEX_2_BLOCK_LENGTH - overlap; _i > 0; i = _i += -1) { | ||
for (i = INDEX_2_BLOCK_LENGTH - overlap; i > 0; i--) { | ||
this.index2[newStart++] = this.index2[start++]; | ||
} | ||
} else { | ||
} else { // no overlap && newStart==start | ||
this.map[start >> SHIFT_1_2] = start; | ||
@@ -576,16 +827,27 @@ start += INDEX_2_BLOCK_LENGTH; | ||
} | ||
for (i = _j = 0; _j < INDEX_1_LENGTH; i = _j += 1) { | ||
// now adjust the index-1 table | ||
for (i = 0; i < INDEX_1_LENGTH; i++) { | ||
this.index1[i] = this.map[this.index1[i] >> SHIFT_1_2]; | ||
} | ||
this.index2NullOffset = this.map[this.index2NullOffset >> SHIFT_1_2]; | ||
// Ensure data table alignment: | ||
// Needs to be granularity-aligned for 16-bit trie | ||
// (so that dataMove will be down-shiftable), | ||
// and 2-aligned for uint32_t data. | ||
// Arbitrary value: 0x3fffc not possible for real data. | ||
while ((newStart & ((DATA_GRANULARITY - 1) | 1)) !== 0) { | ||
this.index2[newStart++] = 0x0000ffff << INDEX_SHIFT; | ||
} | ||
return this.index2Length = newStart; | ||
}; | ||
UnicodeTrieBuilder.prototype._compact = function() { | ||
var highStart, highValue, suppHighStart; | ||
highValue = this.get(0x10ffff); | ||
highStart = this._findHighStart(highValue); | ||
this.index2Length = newStart; | ||
} | ||
_compact() { | ||
// find highStart and round it up | ||
let highValue = this.get(0x10ffff); | ||
let highStart = this._findHighStart(highValue); | ||
highStart = (highStart + (CP_PER_INDEX_1_ENTRY - 1)) & ~(CP_PER_INDEX_1_ENTRY - 1); | ||
@@ -595,7 +857,12 @@ if (highStart === 0x110000) { | ||
} | ||
// Set trie->highStart only after utrie2_get32(trie, highStart). | ||
// Otherwise utrie2_get32(trie, highStart) would try to read the highValue. | ||
this.highStart = highStart; | ||
if (this.highStart < 0x110000) { | ||
suppHighStart = this.highStart <= 0x10000 ? 0x10000 : this.highStart; | ||
// Blank out [highStart..10ffff] to release associated data blocks. | ||
const suppHighStart = this.highStart <= 0x10000 ? 0x10000 : this.highStart; | ||
this.setRange(suppHighStart, 0x10ffff, this.initialValue, true); | ||
} | ||
this._compactData(); | ||
@@ -605,2 +872,6 @@ if (this.highStart > 0x10000) { | ||
} | ||
// Store the highValue in the data array and round up the dataLength. | ||
// Must be done after compactData() because that assumes that dataLength | ||
// is a multiple of DATA_BLOCK_LENGTH. | ||
this.data[this.dataLength++] = highValue; | ||
@@ -610,10 +881,12 @@ while ((this.dataLength & (DATA_GRANULARITY - 1)) !== 0) { | ||
} | ||
return this.isCompacted = true; | ||
}; | ||
UnicodeTrieBuilder.prototype.freeze = function() { | ||
var allIndexesLength, data, dataMove, dest, destIdx, i, index1Length, index2Offset, indexLength, _i, _j, _k, _l, _m, _n, _ref, _ref1, _ref2, _ref3; | ||
this.isCompacted = true; | ||
} | ||
freeze() { | ||
let allIndexesLength, i; | ||
if (!this.isCompacted) { | ||
this._compact(); | ||
} | ||
if (this.highStart <= 0x10000) { | ||
@@ -624,60 +897,89 @@ allIndexesLength = INDEX_1_OFFSET; | ||
} | ||
dataMove = allIndexesLength; | ||
if (allIndexesLength > MAX_INDEX_LENGTH || (dataMove + this.dataNullOffset) > 0xffff || (dataMove + DATA_0800_OFFSET) > 0xffff || (dataMove + this.dataLength) > MAX_DATA_LENGTH) { | ||
const dataMove = allIndexesLength; | ||
// are indexLength and dataLength within limits? | ||
if ((allIndexesLength > MAX_INDEX_LENGTH) || // for unshifted indexLength | ||
((dataMove + this.dataNullOffset) > 0xffff) || // for unshifted dataNullOffset | ||
((dataMove + DATA_0800_OFFSET) > 0xffff) || // for unshifted 2-byte UTF-8 index-2 values | ||
((dataMove + this.dataLength) > MAX_DATA_LENGTH_RUNTIME)) { // for shiftedDataLength | ||
throw new Error("Trie data is too large."); | ||
} | ||
indexLength = allIndexesLength + this.dataLength; | ||
data = new Int32Array(indexLength); | ||
destIdx = 0; | ||
for (i = _i = 0; _i < INDEX_2_BMP_LENGTH; i = _i += 1) { | ||
data[destIdx++] = (this.index2[i] + dataMove) >> INDEX_SHIFT; | ||
// calculate the sizes of, and allocate, the index and data arrays | ||
const indexLength = allIndexesLength + this.dataLength; | ||
const data = new Int32Array(indexLength); | ||
// write the index-2 array values shifted right by INDEX_SHIFT, after adding dataMove | ||
let destIdx = 0; | ||
for (i = 0; i < INDEX_2_BMP_LENGTH; i++) { | ||
data[destIdx++] = ((this.index2[i] + dataMove) >> INDEX_SHIFT); | ||
} | ||
for (i = _j = 0, _ref = 0xc2 - 0xc0; _j < _ref; i = _j += 1) { | ||
data[destIdx++] = dataMove + BAD_UTF8_DATA_OFFSET; | ||
// write UTF-8 2-byte index-2 values, not right-shifted | ||
for (i = 0; i < 0xc2 - 0xc0; i++) { // C0..C1 | ||
data[destIdx++] = (dataMove + BAD_UTF8_DATA_OFFSET); | ||
} | ||
for (i = _k = i, _ref1 = 0xe0 - 0xc0; _k < _ref1; i = _k += 1) { | ||
data[destIdx++] = dataMove + this.index2[i << (6 - SHIFT_2)]; | ||
for (i = i; i < 0xe0 - 0xc0; i++) { // C2..DF | ||
data[destIdx++] = (dataMove + this.index2[i << (6 - SHIFT_2)]); | ||
} | ||
if (this.highStart > 0x10000) { | ||
index1Length = (this.highStart - 0x10000) >> SHIFT_1; | ||
index2Offset = INDEX_2_BMP_LENGTH + UTF8_2B_INDEX_2_LENGTH + index1Length; | ||
for (i = _l = 0; _l < index1Length; i = _l += 1) { | ||
data[destIdx++] = INDEX_2_OFFSET + this.index1[i + OMITTED_BMP_INDEX_1_LENGTH]; | ||
const index1Length = (this.highStart - 0x10000) >> SHIFT_1; | ||
const index2Offset = INDEX_2_BMP_LENGTH + UTF8_2B_INDEX_2_LENGTH + index1Length; | ||
// write 16-bit index-1 values for supplementary code points | ||
for (i = 0; i < index1Length; i++) { | ||
data[destIdx++] = (INDEX_2_OFFSET + this.index1[i + OMITTED_BMP_INDEX_1_LENGTH]); | ||
} | ||
for (i = _m = 0, _ref2 = this.index2Length - index2Offset; _m < _ref2; i = _m += 1) { | ||
data[destIdx++] = (dataMove + this.index2[index2Offset + i]) >> INDEX_SHIFT; | ||
// write the index-2 array values for supplementary code points, | ||
// shifted right by INDEX_SHIFT, after adding dataMove | ||
for (i = 0; i < this.index2Length - index2Offset; i++) { | ||
data[destIdx++] = ((dataMove + this.index2[index2Offset + i]) >> INDEX_SHIFT); | ||
} | ||
} | ||
for (i = _n = 0, _ref3 = this.dataLength; _n < _ref3; i = _n += 1) { | ||
// write 16-bit data values | ||
for (i = 0; i < this.dataLength; i++) { | ||
data[destIdx++] = this.data[i]; | ||
} | ||
dest = new UnicodeTrie({ | ||
data: data, | ||
const dest = new UnicodeTrie({ | ||
data, | ||
highStart: this.highStart, | ||
errorValue: this.errorValue | ||
}); | ||
return dest; | ||
}; | ||
} | ||
UnicodeTrieBuilder.prototype.toBuffer = function() { | ||
var b, buf, compressed, data, i, trie, _i, _len; | ||
trie = this.freeze(); | ||
data = new Uint8Array(trie.data.buffer); | ||
compressed = pako.deflateRaw(data); | ||
// Generates a Buffer containing the serialized and compressed trie. | ||
// Trie data is compressed twice using the deflate algorithm to minimize file size. | ||
// Format: | ||
// uint32_t highStart; | ||
// uint32_t errorValue; | ||
// uint32_t uncompressedDataLength; | ||
// uint8_t trieData[dataLength]; | ||
toBuffer() { | ||
const trie = this.freeze(); | ||
const data = new Uint8Array(trie.data.buffer); | ||
let compressed = pako.deflateRaw(data); | ||
compressed = pako.deflateRaw(compressed); | ||
buf = new Buffer(compressed.length + 12); | ||
const buf = new Buffer(compressed.length + 12); | ||
buf.writeUInt32BE(trie.highStart, 0); | ||
buf.writeUInt32BE(trie.errorValue, 4); | ||
buf.writeUInt32BE(data.length, 8); | ||
for (i = _i = 0, _len = compressed.length; _i < _len; i = ++_i) { | ||
b = compressed[i]; | ||
for (let i = 0; i < compressed.length; i++) { | ||
const b = compressed[i]; | ||
buf[i + 12] = b; | ||
} | ||
return buf; | ||
}; | ||
} | ||
} | ||
return UnicodeTrieBuilder; | ||
})(); | ||
module.exports = UnicodeTrieBuilder; | ||
module.exports = UnicodeTrieBuilder; |
118
index.js
@@ -1,45 +0,72 @@ | ||
// Generated by CoffeeScript 1.7.1 | ||
var UnicodeTrie, inflate; | ||
const inflate = require('tiny-inflate'); | ||
inflate = require('tiny-inflate'); | ||
// Shift size for getting the index-1 table offset. | ||
const SHIFT_1 = 6 + 5; | ||
UnicodeTrie = (function() { | ||
var DATA_BLOCK_LENGTH, DATA_GRANULARITY, DATA_MASK, INDEX_1_OFFSET, INDEX_2_BLOCK_LENGTH, INDEX_2_BMP_LENGTH, INDEX_2_MASK, INDEX_SHIFT, LSCP_INDEX_2_LENGTH, LSCP_INDEX_2_OFFSET, OMITTED_BMP_INDEX_1_LENGTH, SHIFT_1, SHIFT_1_2, SHIFT_2, UTF8_2B_INDEX_2_LENGTH, UTF8_2B_INDEX_2_OFFSET; | ||
// Shift size for getting the index-2 table offset. | ||
const SHIFT_2 = 5; | ||
SHIFT_1 = 6 + 5; | ||
// Difference between the two shift sizes, | ||
// for getting an index-1 offset from an index-2 offset. 6=11-5 | ||
const SHIFT_1_2 = SHIFT_1 - SHIFT_2; | ||
SHIFT_2 = 5; | ||
// Number of index-1 entries for the BMP. 32=0x20 | ||
// This part of the index-1 table is omitted from the serialized form. | ||
const OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> SHIFT_1; | ||
SHIFT_1_2 = SHIFT_1 - SHIFT_2; | ||
// Number of entries in an index-2 block. 64=0x40 | ||
const INDEX_2_BLOCK_LENGTH = 1 << SHIFT_1_2; | ||
OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> SHIFT_1; | ||
// Mask for getting the lower bits for the in-index-2-block offset. */ | ||
const INDEX_2_MASK = INDEX_2_BLOCK_LENGTH - 1; | ||
INDEX_2_BLOCK_LENGTH = 1 << SHIFT_1_2; | ||
// Shift size for shifting left the index array values. | ||
// Increases possible data size with 16-bit index values at the cost | ||
// of compactability. | ||
// This requires data blocks to be aligned by DATA_GRANULARITY. | ||
const INDEX_SHIFT = 2; | ||
INDEX_2_MASK = INDEX_2_BLOCK_LENGTH - 1; | ||
// Number of entries in a data block. 32=0x20 | ||
const DATA_BLOCK_LENGTH = 1 << SHIFT_2; | ||
INDEX_SHIFT = 2; | ||
// Mask for getting the lower bits for the in-data-block offset. | ||
const DATA_MASK = DATA_BLOCK_LENGTH - 1; | ||
DATA_BLOCK_LENGTH = 1 << SHIFT_2; | ||
// The part of the index-2 table for U+D800..U+DBFF stores values for | ||
// lead surrogate code _units_ not code _points_. | ||
// Values for lead surrogate code _points_ are indexed with this portion of the table. | ||
// Length=32=0x20=0x400>>SHIFT_2. (There are 1024=0x400 lead surrogates.) | ||
const LSCP_INDEX_2_OFFSET = 0x10000 >> SHIFT_2; | ||
const LSCP_INDEX_2_LENGTH = 0x400 >> SHIFT_2; | ||
DATA_MASK = DATA_BLOCK_LENGTH - 1; | ||
// Count the lengths of both BMP pieces. 2080=0x820 | ||
const INDEX_2_BMP_LENGTH = LSCP_INDEX_2_OFFSET + LSCP_INDEX_2_LENGTH; | ||
LSCP_INDEX_2_OFFSET = 0x10000 >> SHIFT_2; | ||
// The 2-byte UTF-8 version of the index-2 table follows at offset 2080=0x820. | ||
// Length 32=0x20 for lead bytes C0..DF, regardless of SHIFT_2. | ||
const UTF8_2B_INDEX_2_OFFSET = INDEX_2_BMP_LENGTH; | ||
const UTF8_2B_INDEX_2_LENGTH = 0x800 >> 6; // U+0800 is the first code point after 2-byte UTF-8 | ||
LSCP_INDEX_2_LENGTH = 0x400 >> SHIFT_2; | ||
// The index-1 table, only used for supplementary code points, at offset 2112=0x840. | ||
// Variable length, for code points up to highStart, where the last single-value range starts. | ||
// Maximum length 512=0x200=0x100000>>SHIFT_1. | ||
// (For 0x100000 supplementary code points U+10000..U+10ffff.) | ||
// | ||
// The part of the index-2 table for supplementary code points starts | ||
// after this index-1 table. | ||
// | ||
// Both the index-1 table and the following part of the index-2 table | ||
// are omitted completely if there is only BMP data. | ||
const INDEX_1_OFFSET = UTF8_2B_INDEX_2_OFFSET + UTF8_2B_INDEX_2_LENGTH; | ||
INDEX_2_BMP_LENGTH = LSCP_INDEX_2_OFFSET + LSCP_INDEX_2_LENGTH; | ||
// The alignment size of a data block. Also the granularity for compaction. | ||
const DATA_GRANULARITY = 1 << INDEX_SHIFT; | ||
UTF8_2B_INDEX_2_OFFSET = INDEX_2_BMP_LENGTH; | ||
class UnicodeTrie { | ||
constructor(data) { | ||
const isBuffer = (typeof data.readUInt32BE === 'function') && (typeof data.slice === 'function'); | ||
UTF8_2B_INDEX_2_LENGTH = 0x800 >> 6; | ||
INDEX_1_OFFSET = UTF8_2B_INDEX_2_OFFSET + UTF8_2B_INDEX_2_LENGTH; | ||
DATA_GRANULARITY = 1 << INDEX_SHIFT; | ||
function UnicodeTrie(data) { | ||
var isBuffer, uncompressedLength, view; | ||
isBuffer = typeof data.readUInt32BE === 'function' && typeof data.slice === 'function'; | ||
if (isBuffer || data instanceof Uint8Array) { | ||
// read binary format | ||
let uncompressedLength; | ||
if (isBuffer) { | ||
@@ -51,3 +78,3 @@ this.highStart = data.readUInt32BE(0); | ||
} else { | ||
view = new DataView(data.buffer); | ||
const view = new DataView(data.buffer); | ||
this.highStart = view.getUint32(0); | ||
@@ -58,24 +85,39 @@ this.errorValue = view.getUint32(4); | ||
} | ||
// double inflate the actual trie data | ||
data = inflate(data, new Uint8Array(uncompressedLength)); | ||
data = inflate(data, new Uint8Array(uncompressedLength)); | ||
this.data = new Uint32Array(data.buffer); | ||
} else { | ||
this.data = data.data, this.highStart = data.highStart, this.errorValue = data.errorValue; | ||
// pre-parsed data | ||
({ data: this.data, highStart: this.highStart, errorValue: this.errorValue } = data); | ||
} | ||
} | ||
UnicodeTrie.prototype.get = function(codePoint) { | ||
var index; | ||
if (codePoint < 0 || codePoint > 0x10ffff) { | ||
get(codePoint) { | ||
let index; | ||
if ((codePoint < 0) || (codePoint > 0x10ffff)) { | ||
return this.errorValue; | ||
} | ||
if (codePoint < 0xd800 || (codePoint > 0xdbff && codePoint <= 0xffff)) { | ||
if ((codePoint < 0xd800) || ((codePoint > 0xdbff) && (codePoint <= 0xffff))) { | ||
// Ordinary BMP code point, excluding leading surrogates. | ||
// BMP uses a single level lookup. BMP index starts at offset 0 in the index. | ||
// data is stored in the index array itself. | ||
index = (this.data[codePoint >> SHIFT_2] << INDEX_SHIFT) + (codePoint & DATA_MASK); | ||
return this.data[index]; | ||
} | ||
if (codePoint <= 0xffff) { | ||
// Lead Surrogate Code Point. A Separate index section is stored for | ||
// lead surrogate code units and code points. | ||
// The main index has the code unit data. | ||
// For this function, we need the code point data. | ||
index = (this.data[LSCP_INDEX_2_OFFSET + ((codePoint - 0xd800) >> SHIFT_2)] << INDEX_SHIFT) + (codePoint & DATA_MASK); | ||
return this.data[index]; | ||
} | ||
if (codePoint < this.highStart) { | ||
// Supplemental code point, use two-level lookup. | ||
index = this.data[(INDEX_1_OFFSET - OMITTED_BMP_INDEX_1_LENGTH) + (codePoint >> SHIFT_1)]; | ||
@@ -86,9 +128,7 @@ index = this.data[index + ((codePoint >> SHIFT_2) & INDEX_2_MASK)]; | ||
} | ||
return this.data[this.data.length - DATA_GRANULARITY]; | ||
}; | ||
} | ||
} | ||
return UnicodeTrie; | ||
})(); | ||
module.exports = UnicodeTrie; | ||
module.exports = UnicodeTrie; |
{ | ||
"name": "unicode-trie", | ||
"version": "0.3.1", | ||
"version": "1.0.0", | ||
"description": "Unicode Trie data structure for fast character metadata lookup, ported from ICU", | ||
"devDependencies": { | ||
"mocha": "^1.20.1", | ||
"coffee-script": "^1.7.1", | ||
"coffee-coverage": "^0.4.2" | ||
"mocha": "^6.1.4", | ||
"nyc": "^14.1.1" | ||
}, | ||
"scripts": { | ||
"prepublish": "make build", | ||
"postpublish": "make clean", | ||
"test": "mocha" | ||
"test": "mocha", | ||
"coverage": "nyc mocha" | ||
}, | ||
@@ -15,0 +13,0 @@ "repository": { |
@@ -32,24 +32,24 @@ # unicode-trie | ||
```coffeescript | ||
UnicodeTrieBuilder = require 'unicode-trie/builder' | ||
```js | ||
const UnicodeTrieBuilder = require('unicode-trie/builder'); | ||
# create a trie | ||
t = new UnicodeTrieBuilder | ||
// create a trie | ||
let t = new UnicodeTrieBuilder(); | ||
# optional parameters for default value, and error value | ||
# if not provided, both are set to 0 | ||
t = new UnicodeTrieBuilder 10, 999 | ||
// optional parameters for default value, and error value | ||
// if not provided, both are set to 0 | ||
t = new UnicodeTrieBuilder(10, 999); | ||
# set individual values and ranges | ||
t.set 0x4567, 99 | ||
t.setRange 0x40, 0xe7, 0x1234 | ||
// set individual values and ranges | ||
t.set(0x4567, 99); | ||
t.setRange(0x40, 0xe7, 0x1234); | ||
# you can lookup a value if you like | ||
t.get 0x4567 # => 99 | ||
// you can lookup a value if you like | ||
t.get(0x4567); // => 99 | ||
# get a compiled trie (returns a UnicodeTrie object) | ||
trie = t.freeze() | ||
// get a compiled trie (returns a UnicodeTrie object) | ||
const trie = t.freeze(); | ||
# write compressed trie to a binary file | ||
fs.writeFile 'data.trie', t.toBuffer() | ||
// write compressed trie to a binary file | ||
fs.writeFile('data.trie', t.toBuffer()); | ||
``` | ||
@@ -63,12 +63,12 @@ | ||
```coffeescript | ||
UnicodeTrie = require 'unicode-trie' | ||
fs = require 'fs' | ||
```js | ||
const UnicodeTrie = require('unicode-trie'); | ||
const fs = require('fs'); | ||
# load serialized trie from binary file | ||
data = fs.readFileSync 'data.trie' | ||
trie = new UnicodeTrie data | ||
// load serialized trie from binary file | ||
const data = fs.readFileSync('data.trie'); | ||
const trie = new UnicodeTrie(data); | ||
# lookup a value | ||
trie.get 0x4567 # => 99 | ||
// lookup a value | ||
trie.get(0x4567); // => 99 | ||
``` | ||
@@ -75,0 +75,0 @@ |
Sorry, the diff of this file is not supported yet
License Policy Violation
LicenseThis package is not allowed per your license policy. Review the package's license to ensure compliance.
Found 1 instance in 1 package
Major refactor
Supply chain riskPackage has recently undergone a major refactor. It may be unstable or indicate significant internal changes. Use caution when updating to versions that include significant changes.
Found 1 instance in 1 package
License Policy Violation
LicenseThis package is not allowed per your license policy. Review the package's license to ensure compliance.
Found 1 instance in 1 package
No v1
QualityPackage is not semver >=1. This means it is not stable and does not support ^ ranges.
Found 1 instance in 1 package
2
1116
0
47565
7
1