Issue #1971 - Part 2: Update ICU source to 63.2.

This commit is contained in:
Job Bautista 2022-07-24 21:03:27 +08:00 committed by roytam1
commit 1e69214382
3160 changed files with 275815 additions and 234203 deletions

View file

@ -1,4 +1,4 @@
// Copyright (C) 2016 and later: Unicode, Inc. and others.
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
//
// rbbisetb.cpp
@ -35,7 +35,7 @@
#if !UCONFIG_NO_BREAK_ITERATION
#include "unicode/uniset.h"
#include "utrie.h"
#include "utrie2.h"
#include "uvector.h"
#include "uassert.h"
#include "cmemory.h"
@ -44,43 +44,6 @@
#include "rbbisetb.h"
#include "rbbinode.h"
//------------------------------------------------------------------------
//
// getFoldedRBBIValue Call-back function used during building of Trie table.
// Folding value: just store the offset (16 bits)
// if there is any non-0 entry.
// (It'd really be nice if the Trie builder would provide a
// simple default, so this function could go away from here.)
//
//------------------------------------------------------------------------
/* folding value: just store the offset (16 bits) if there is any non-0 entry */
U_CDECL_BEGIN
static uint32_t U_CALLCONV
getFoldedRBBIValue(UNewTrie *trie, UChar32 start, int32_t offset) {
uint32_t value;
UChar32 limit;
UBool inBlockZero;
limit=start+0x400;
while(start<limit) {
value=utrie_get32(trie, start, &inBlockZero);
if(inBlockZero) {
start+=UTRIE_DATA_BLOCK_LENGTH;
} else if(value!=0) {
return (uint32_t)(offset|0x8000);
} else {
++start;
}
}
return 0;
}
U_CDECL_END
U_NAMESPACE_BEGIN
//------------------------------------------------------------------------
@ -116,7 +79,7 @@ RBBISetBuilder::~RBBISetBuilder()
delete r;
}
utrie_close(fTrie);
utrie2_close(fTrie);
}
@ -128,7 +91,7 @@ RBBISetBuilder::~RBBISetBuilder()
// from the Unicode Sets.
//
//------------------------------------------------------------------------
void RBBISetBuilder::build() {
void RBBISetBuilder::buildRanges() {
RBBINode *usetNode;
RangeDescriptor *rlRange;
@ -282,38 +245,64 @@ void RBBISetBuilder::build() {
if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "rgroup")) {printRangeGroups();}
if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "esets")) {printSets();}
//
// Build the Trie table for mapping UChar32 values to the corresponding
// range group number
//
fTrie = utrie_open(NULL, // Pre-existing trie to be filled in
NULL, // Data array (utrie will allocate one)
100000, // Max Data Length
0, // Initial value for all code points
0, // Lead surrogate unit value
TRUE); // Keep Latin 1 in separately
}
for (rlRange = fRangeList; rlRange!=0; rlRange=rlRange->fNext) {
utrie_setRange32(fTrie, rlRange->fStartChar, rlRange->fEndChar+1, rlRange->fNum, TRUE);
//
// Build the Trie table for mapping UChar32 values to the corresponding
// range group number.
//
void RBBISetBuilder::buildTrie() {
RangeDescriptor *rlRange;
fTrie = utrie2_open(0, // Initial value for all code points.
0, // Error value for out-of-range input.
fStatus);
for (rlRange = fRangeList; rlRange!=0 && U_SUCCESS(*fStatus); rlRange=rlRange->fNext) {
utrie2_setRange32(fTrie,
rlRange->fStartChar, // Range start
rlRange->fEndChar, // Range end (inclusive)
rlRange->fNum, // value for range
TRUE, // Overwrite previously written values
fStatus);
}
}
void RBBISetBuilder::mergeCategories(IntPair categories) {
U_ASSERT(categories.first >= 1);
U_ASSERT(categories.second > categories.first);
for (RangeDescriptor *rd = fRangeList; rd != nullptr; rd = rd->fNext) {
int32_t rangeNum = rd->fNum & ~DICT_BIT;
int32_t rangeDict = rd->fNum & DICT_BIT;
if (rangeNum == categories.second) {
rd->fNum = categories.first | rangeDict;
} else if (rangeNum > categories.second) {
rd->fNum--;
}
}
--fGroupCount;
}
//-----------------------------------------------------------------------------------
//
// getTrieSize() Return the size that will be required to serialize the Trie.
//
//-----------------------------------------------------------------------------------
int32_t RBBISetBuilder::getTrieSize() /*const*/ {
fTrieSize = utrie_serialize(fTrie,
NULL, // Buffer
0, // Capacity
getFoldedRBBIValue,
TRUE, // Reduce to 16 bits
fStatus);
int32_t RBBISetBuilder::getTrieSize() {
if (U_FAILURE(*fStatus)) {
return 0;
}
utrie2_freeze(fTrie, UTRIE2_16_VALUE_BITS, fStatus);
fTrieSize = utrie2_serialize(fTrie,
NULL, // Buffer
0, // Capacity
fStatus);
if (*fStatus == U_BUFFER_OVERFLOW_ERROR) {
*fStatus = U_ZERO_ERROR;
}
// RBBIDebugPrintf("Trie table size is %d\n", trieSize);
return fTrieSize;
}
@ -327,12 +316,10 @@ int32_t RBBISetBuilder::getTrieSize() /*const*/ {
//
//-----------------------------------------------------------------------------------
void RBBISetBuilder::serializeTrie(uint8_t *where) {
utrie_serialize(fTrie,
where, // Buffer
fTrieSize, // Capacity
getFoldedRBBIValue,
TRUE, // Reduce to 16 bits
fStatus);
utrie2_serialize(fTrie,
where, // Buffer
fTrieSize, // Capacity
fStatus);
}
//------------------------------------------------------------------------
@ -480,7 +467,7 @@ void RBBISetBuilder::printRangeGroups() {
lastPrintedGroupNum = groupNum;
RBBIDebugPrintf("%2i ", groupNum);
if (rlRange->fNum & 0x4000) { RBBIDebugPrintf(" <DICT> ");}
if (rlRange->fNum & DICT_BIT) { RBBIDebugPrintf(" <DICT> ");}
for (i=0; i<rlRange->fIncludesSets->size(); i++) {
RBBINode *usetNode = (RBBINode *)rlRange->fIncludesSets->elementAt(i);
@ -673,20 +660,20 @@ void RangeDescriptor::split(UChar32 where, UErrorCode &status) {
void RangeDescriptor::setDictionaryFlag() {
int i;
for (i=0; i<this->fIncludesSets->size(); i++) {
RBBINode *usetNode = (RBBINode *)fIncludesSets->elementAt(i);
UnicodeString setName;
RBBINode *setRef = usetNode->fParent;
if (setRef != NULL) {
static const char16_t *dictionary = u"dictionary";
for (i=0; i<fIncludesSets->size(); i++) {
RBBINode *usetNode = (RBBINode *)fIncludesSets->elementAt(i);
RBBINode *setRef = usetNode->fParent;
if (setRef != nullptr) {
RBBINode *varRef = setRef->fParent;
if (varRef != NULL && varRef->fType == RBBINode::varRef) {
setName = varRef->fText;
if (varRef && varRef->fType == RBBINode::varRef) {
const UnicodeString *setName = &varRef->fText;
if (setName->compare(dictionary, -1) == 0) {
fNum |= RBBISetBuilder::DICT_BIT;
break;
}
}
}
if (setName.compare(UNICODE_STRING("dictionary", 10)) == 0) { // TODO: no string literals.
this->fNum |= 0x4000;
break;
}
}
}