commit 948c92464c825edf2e6642ffd1963b690b4ddeea
Author: Stephan Witt <[email protected]>
Date: Sun Dec 14 16:29:12 2025 +0100
#13261 fix wrong locations and lengths of misspelled text portions of Apple
speller
- change to with UCS4 docstring compatible interface
- check for strings with unicode characters encoded in UTF16 with multiple
codes
- adjust code unit locations and lengths to character locations and lengths
compatible with UCS4 strings
(cherry picked from commit 69156b01f855adc7b50744df8bfe7fb1bab6722e)
---
src/AppleSpellChecker.cpp | 10 ++---
src/support/AppleSpeller.h | 10 ++---
src/support/AppleSpeller.m | 94 ++++++++++++++++++++++++++++++++++++++--------
3 files changed, 89 insertions(+), 25 deletions(-)
diff --git a/src/AppleSpellChecker.cpp b/src/AppleSpellChecker.cpp
index 0a82d38b32..d7b801dfc9 100644
--- a/src/AppleSpellChecker.cpp
+++ b/src/AppleSpellChecker.cpp
@@ -84,7 +84,7 @@ SpellChecker::Result AppleSpellChecker::check(WordLangTuple
const & word,
if (!hasDictionary(word.lang()))
return NO_DICTIONARY;
- string const word_str = to_utf8(word.word());
+ docstring const word_str = word.word();
string const lang = d->languageMap[word.lang()->lang()];
vector<WordLangTuple>::const_iterator it = docdict.begin();
@@ -114,7 +114,7 @@ void AppleSpellChecker::advanceChangeNumber()
// add to personal dictionary
void AppleSpellChecker::insert(WordLangTuple const & word)
{
- string const word_str = to_utf8(word.word());
+ docstring const word_str = word.word();
AppleSpeller_learn(d->speller, word_str.c_str());
LYXERR(Debug::GUI, "learn word: \"" << word.word() << "\"");
advanceChangeNumber();
@@ -124,7 +124,7 @@ void AppleSpellChecker::insert(WordLangTuple const & word)
// remove from personal dictionary
void AppleSpellChecker::remove(WordLangTuple const & word)
{
- string const word_str = to_utf8(word.word());
+ docstring const word_str = word.word();
AppleSpeller_unlearn(d->speller, word_str.c_str());
LYXERR(Debug::GUI, "unlearn word: \"" << word.word() << "\"");
advanceChangeNumber();
@@ -134,7 +134,7 @@ void AppleSpellChecker::remove(WordLangTuple const & word)
// ignore for session
void AppleSpellChecker::accept(WordLangTuple const & word)
{
- string const word_str = to_utf8(word.word());
+ docstring const word_str = word.word();
AppleSpeller_ignore(d->speller, word_str.c_str());
LYXERR(Debug::GUI, "ignore word: \"" << word.word() << "\"");
advanceChangeNumber();
@@ -145,7 +145,7 @@ void AppleSpellChecker::suggest(WordLangTuple const & wl,
docstring_list & suggestions)
{
suggestions.clear();
- string const word_str = to_utf8(wl.word());
+ docstring const word_str = wl.word();
size_t num = AppleSpeller_makeSuggestion(d->speller,
word_str.c_str(),
wl.lang()->code().c_str());
for (size_t i = 0; i < num; i++) {
diff --git a/src/support/AppleSpeller.h b/src/support/AppleSpeller.h
index 5910fba142..62f3fe2de6 100644
--- a/src/support/AppleSpeller.h
+++ b/src/support/AppleSpeller.h
@@ -28,12 +28,12 @@ typedef struct AppleSpellerRec * AppleSpeller ;
AppleSpeller newAppleSpeller(void);
void freeAppleSpeller(AppleSpeller speller);
-SpellCheckResult AppleSpeller_check(AppleSpeller speller, const char * word,
const char * lang);
-void AppleSpeller_ignore(AppleSpeller speller, const char * word);
-size_t AppleSpeller_makeSuggestion(AppleSpeller speller, const char * word,
const char * lang);
+SpellCheckResult AppleSpeller_check(AppleSpeller speller, const wchar_t *
word, const char * lang);
+void AppleSpeller_ignore(AppleSpeller speller, const wchar_t * word);
+size_t AppleSpeller_makeSuggestion(AppleSpeller speller, const wchar_t * word,
const char * lang);
const char * AppleSpeller_getSuggestion(AppleSpeller speller, size_t pos);
-void AppleSpeller_learn(AppleSpeller speller, const char * word);
-void AppleSpeller_unlearn(AppleSpeller speller, const char * word);
+void AppleSpeller_learn(AppleSpeller speller, const wchar_t * word);
+void AppleSpeller_unlearn(AppleSpeller speller, const wchar_t * word);
int AppleSpeller_hasLanguage(AppleSpeller speller, const char * lang);
int AppleSpeller_numMisspelledWords(AppleSpeller speller);
void AppleSpeller_misspelledWord(AppleSpeller speller, int index, int * start,
int * length);
diff --git a/src/support/AppleSpeller.m b/src/support/AppleSpeller.m
index ce8b7a96b4..593999812b 100644
--- a/src/support/AppleSpeller.m
+++ b/src/support/AppleSpeller.m
@@ -12,6 +12,8 @@
#import <AvailabilityMacros.h>
+#include <wchar.h>
+
#include "support/AppleSpeller.h"
typedef struct AppleSpellerRec {
@@ -48,9 +50,18 @@ void freeAppleSpeller(AppleSpeller speller)
}
-static NSString * toString(const char * word)
+static NSString * toString(const char * lang)
+{
+ return [[NSString alloc] initWithBytes:lang length:strlen(lang)
encoding:NSUTF8StringEncoding];
+}
+
+
+NSString * wcharToString(const wchar_t* text, NSUInteger length)
{
- return [[NSString alloc] initWithBytes:word length:strlen(word)
encoding:NSUTF8StringEncoding];
+ BOOL lendian = NSHostByteOrder() == NS_LittleEndian;
+ NSUInteger bytes = length*sizeof(wchar_t);
+
+ return [[NSString alloc] initWithBytes:text length:bytes
encoding:(lendian ? NSUTF32LittleEndianStringEncoding :
NSUTF32BigEndianStringEncoding)];
}
@@ -77,22 +88,70 @@ static NSString * toLanguage(AppleSpeller speller, const
char * lang)
}
-SpellCheckResult AppleSpeller_check(AppleSpeller speller, const char * word,
const char * lang)
+BOOL surrorate(unichar curr, unichar next) {
+ return 0xD800 <= curr && curr <= 0xDBFF &&
+ 0xDC00 <= next && next <= 0xDFFF;
+}
+
+
+NSArray * AppleSpeller_adjustPositions(NSArray * misspelled,
+ NSString * text, NSUInteger length)
+{
+ NSUInteger r = 0;
+ NSUInteger i = 0;
+ NSUInteger rcount = [misspelled count];
+ NSRange range = [[misspelled objectAtIndex:r] rangeValue];
+ NSUInteger rstart = range.location;
+ NSUInteger rend = range.location+range.length;
+ NSUInteger asurrogates = 0;
+ NSUInteger rsurrogates = 0;
+ NSMutableArray * result = [NSMutableArray arrayWithCapacity:rcount+1];
+ unichar curr = [text characterAtIndex:i];
+
+ while (i < length-1 && r < rcount) {
+ if (i == rstart) {
+ range.location -= asurrogates;
+ rsurrogates = 0;
+ } else if (i == rend) {
+ range.length -= rsurrogates;
+ [result addObject:[NSValue valueWithRange:range]];
+ if (++r < rcount) {
+ range = [[misspelled objectAtIndex:r]
rangeValue];
+ rstart = range.location;
+ rend = range.location+range.length;
+ }
+ }
+ unichar prev = curr;
+ curr = [text characterAtIndex:++i];
+ if (surrorate(prev, curr) && i < length-1) {
+ curr = [text characterAtIndex:++i];
+ asurrogates++;
+ rsurrogates++;
+ }
+ }
+ return result;
+}
+
+
+SpellCheckResult AppleSpeller_check(AppleSpeller speller,
+ const wchar_t * word, const char * lang)
{
if (!speller->checker || !lang || !word)
return SPELL_CHECK_FAILED;
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
- NSString * word_ = toString(word);
+ NSUInteger wlength = wcslen(word);
+ NSString * word_ = wcharToString(word, wlength);
NSString * lang_ = toString(lang);
SpellCheckResult result = SPELL_CHECK_FAILED;
- int start = 0;
- int length = [word_ length];
+ NSUInteger start = 0;
+ NSUInteger ulength = [word_ length];
[speller->misspelled release];
speller->misspelled = nil;
+ BOOL surrogates = ulength > wlength;
- while (result == SPELL_CHECK_FAILED && start < length) {
+ while (result == SPELL_CHECK_FAILED && start < ulength) {
NSRange match = [speller->checker
checkSpellingOfString:word_
startingAt:start
@@ -117,6 +176,11 @@ SpellCheckResult AppleSpeller_check(AppleSpeller speller,
const char * word, con
start = match.location + match.length + 1;
}
}
+ if ([speller->misspelled count] > 0 && surrogates) {
+ NSArray * misspelled =
AppleSpeller_adjustPositions(speller->misspelled, word_, ulength);
+ [speller->misspelled release];
+ speller->misspelled = [[NSArray arrayWithArray:misspelled]
retain];
+ }
[word_ release];
[lang_ release];
@@ -126,10 +190,10 @@ SpellCheckResult AppleSpeller_check(AppleSpeller speller,
const char * word, con
}
-void AppleSpeller_ignore(AppleSpeller speller, const char * word)
+void AppleSpeller_ignore(AppleSpeller speller, const wchar_t * word)
{
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
- NSString * word_ = toString(word);
+ NSString * word_ = wcharToString(word, wcslen(word));
[speller->checker ignoreWord:word_
inSpellDocumentWithTag:(speller->doctag)];
@@ -138,13 +202,13 @@ void AppleSpeller_ignore(AppleSpeller speller, const char
* word)
}
-size_t AppleSpeller_makeSuggestion(AppleSpeller speller, const char * word,
const char * lang)
+size_t AppleSpeller_makeSuggestion(AppleSpeller speller, const wchar_t * word,
const char * lang)
{
if (!speller->checker || !word || !lang)
return 0;
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
- NSString * word_ = toString(word);
+ NSString * word_ = wcharToString(word, wcslen(word));
NSString * lang_ = toString(lang);
NSArray * result ;
@@ -177,10 +241,10 @@ const char * AppleSpeller_getSuggestion(AppleSpeller
speller, size_t pos)
}
-void AppleSpeller_learn(AppleSpeller speller, const char * word)
+void AppleSpeller_learn(AppleSpeller speller, const wchar_t * word)
{
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
- NSString * word_ = toString(word);
+ NSString * word_ = wcharToString(word, wcslen(word));
if ([NSSpellChecker instancesRespondToSelector:@selector(learnWord:)])
[speller->checker learnWord:word_];
@@ -190,10 +254,10 @@ void AppleSpeller_learn(AppleSpeller speller, const char
* word)
}
-void AppleSpeller_unlearn(AppleSpeller speller, const char * word)
+void AppleSpeller_unlearn(AppleSpeller speller, const wchar_t * word)
{
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
- NSString * word_ = toString(word);
+ NSString * word_ = wcharToString(word, wcslen(word));
if ([NSSpellChecker instancesRespondToSelector:@selector(unlearnWord:)])
[speller->checker unlearnWord:word_];
--
lyx-cvs mailing list
[email protected]
https://lists.lyx.org/mailman/listinfo/lyx-cvs
