975 lines
38 KiB
Objective-C
975 lines
38 KiB
Objective-C
//
|
|
// KBSuggestionEngine.m
|
|
// CustomKeyboard
|
|
//
|
|
|
|
#import "KBSuggestionEngine.h"
|
|
#import "KBConfig.h"
|
|
|
|
@interface KBSuggestionEngine ()
|
|
@property (nonatomic, copy) NSArray<NSString *> *words;
|
|
@property (nonatomic, strong) NSMutableDictionary<NSString *, NSNumber *> *selectionCounts;
|
|
@property (nonatomic, strong) NSSet<NSString *> *priorityWords;
|
|
@property (nonatomic, copy) NSArray<NSString *> *traditionalChineseWords;
|
|
@property (nonatomic, copy) NSArray<NSString *> *simplifiedChineseWords;
|
|
@property (nonatomic, strong) NSDictionary<NSString *, NSArray<NSString *> *> *pinyinToTraditionalMap;
|
|
@property (nonatomic, strong) NSDictionary<NSString *, NSArray<NSString *> *> *bopomofoToChineseMap;
|
|
@property (nonatomic, copy) NSArray<NSString *> *spanishWords;
|
|
@property (nonatomic, copy) NSArray<NSString *> *englishWords;
|
|
@property (nonatomic, copy) NSArray<NSString *> *portugueseWords;
|
|
@property (nonatomic, copy) NSArray<NSString *> *indonesianWords;
|
|
@end
|
|
|
|
@implementation KBSuggestionEngine
|
|
|
|
+ (instancetype)shared {
|
|
static KBSuggestionEngine *engine;
|
|
static dispatch_once_t onceToken;
|
|
dispatch_once(&onceToken, ^{
|
|
engine = [[KBSuggestionEngine alloc] init];
|
|
});
|
|
return engine;
|
|
}
|
|
|
|
- (instancetype)init {
|
|
if (self = [super init]) {
|
|
_engineType = KBSuggestionEngineTypeLatin;
|
|
_selectionCounts = [NSMutableDictionary dictionary];
|
|
NSArray<NSString *> *defaults = [self.class kb_defaultWords];
|
|
_priorityWords = [NSSet setWithArray:defaults];
|
|
}
|
|
return self;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)suggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (prefix.length == 0 || limit == 0) { return @[]; }
|
|
// 为过滤留出候选空间，避免过滤后数量过少。
|
|
NSUInteger fetchLimit = limit;
|
|
if (fetchLimit < 80) {
|
|
fetchLimit = MIN((NSUInteger)80, MAX(fetchLimit * 4, fetchLimit));
|
|
}
|
|
NSArray<NSString *> *raw = nil;
|
|
|
|
switch (self.engineType) {
|
|
case KBSuggestionEngineTypeEnglish:
|
|
raw = [self kb_englishSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypeSpanish:
|
|
raw = [self kb_spanishSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypePortuguese:
|
|
raw = [self kb_portugueseSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypeIndonesian:
|
|
raw = [self kb_indonesianSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypePinyinTraditional:
|
|
raw = [self kb_traditionalPinyinSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypePinyinSimplified:
|
|
raw = [self kb_simplifiedPinyinSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypeBopomofo:
|
|
raw = [self kb_bopomofoSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
case KBSuggestionEngineTypeLatin:
|
|
default:
|
|
raw = [self kb_latinSuggestionsForPrefix:prefix limit:fetchLimit];
|
|
break;
|
|
}
|
|
return [self kb_filterSensitiveSuggestions:raw limit:limit];
|
|
}
|
|
|
|
- (void)recordSelection:(NSString *)word {
|
|
if (word.length == 0) { return; }
|
|
NSString *key = word.lowercaseString;
|
|
NSInteger count = self.selectionCounts[key].integerValue + 1;
|
|
self.selectionCounts[key] = @(count);
|
|
}
|
|
|
|
#pragma mark - Defaults
|
|
|
|
- (NSArray<NSString *> *)kb_loadWords {
|
|
NSMutableOrderedSet<NSString *> *set = [[NSMutableOrderedSet alloc] init];
|
|
[set addObjectsFromArray:[self.class kb_defaultWords]];
|
|
|
|
NSArray<NSString *> *paths = [self kb_wordListPaths];
|
|
for (NSString *path in paths) {
|
|
if (path.length == 0) { continue; }
|
|
NSString *content = [NSString stringWithContentsOfFile:path encoding:NSUTF8StringEncoding error:nil];
|
|
if (content.length == 0) { continue; }
|
|
NSArray<NSString *> *lines = [content componentsSeparatedByCharactersInSet:[NSCharacterSet newlineCharacterSet]];
|
|
for (NSString *line in lines) {
|
|
NSString *word = [self kb_sanitizedWordFromLine:line];
|
|
if (word.length == 0) { continue; }
|
|
[set addObject:word];
|
|
}
|
|
}
|
|
|
|
NSArray<NSString *> *result = set.array ?: @[];
|
|
return result;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_wordListPaths {
|
|
NSMutableArray<NSString *> *paths = [NSMutableArray array];
|
|
// 1) App Group override (allows server-downloaded large list).
|
|
NSURL *containerURL = [[NSFileManager defaultManager] containerURLForSecurityApplicationGroupIdentifier:AppGroup];
|
|
if (containerURL.path.length > 0) {
|
|
NSString *groupPath = [[containerURL path] stringByAppendingPathComponent:@"kb_words.txt"];
|
|
[paths addObject:groupPath];
|
|
}
|
|
// 2) Bundle fallback.
|
|
NSString *bundlePath = [[NSBundle mainBundle] pathForResource:@"kb_words" ofType:@"txt"];
|
|
if (bundlePath.length > 0) {
|
|
[paths addObject:bundlePath];
|
|
}
|
|
return paths;
|
|
}
|
|
|
|
- (NSString *)kb_sanitizedWordFromLine:(NSString *)line {
|
|
NSString *trimmed = [[line stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]] lowercaseString];
|
|
if (trimmed.length == 0) { return @""; }
|
|
static NSCharacterSet *letters = nil;
|
|
static dispatch_once_t onceToken;
|
|
dispatch_once(&onceToken, ^{
|
|
letters = [NSCharacterSet characterSetWithCharactersInString:@"abcdefghijklmnopqrstuvwxyz"];
|
|
});
|
|
for (NSUInteger i = 0; i < trimmed.length; i++) {
|
|
if (![letters characterIsMember:[trimmed characterAtIndex:i]]) {
|
|
return @"";
|
|
}
|
|
}
|
|
return trimmed;
|
|
}
|
|
|
|
+ (NSArray<NSString *> *)kb_defaultWords {
|
|
return @[
|
|
@"a", @"an", @"and", @"are", @"as", @"at",
|
|
@"app", @"ap", @"apple", @"apply", @"april", @"application",
|
|
@"about", @"above", @"after", @"again", @"against", @"all",
|
|
@"am", @"among", @"amount", @"any", @"around",
|
|
@"be", @"because", @"been", @"before", @"being", @"below",
|
|
@"best", @"between", @"both", @"but", @"by",
|
|
@"can", @"could", @"come", @"common", @"case",
|
|
@"do", @"does", @"down", @"day",
|
|
@"each", @"early", @"end", @"even", @"every",
|
|
@"for", @"from", @"first", @"found", @"free",
|
|
@"get", @"good", @"great", @"go",
|
|
@"have", @"has", @"had", @"help", @"how",
|
|
@"in", @"is", @"it", @"if", @"into",
|
|
@"just", @"keep", @"kind", @"know",
|
|
@"like", @"look", @"long", @"last",
|
|
@"make", @"more", @"most", @"my",
|
|
@"new", @"no", @"not", @"now",
|
|
@"of", @"on", @"one", @"or", @"other", @"our", @"out",
|
|
@"people", @"place", @"please",
|
|
@"quick", @"quite",
|
|
@"right", @"read", @"real",
|
|
@"see", @"say", @"some", @"such", @"so",
|
|
@"the", @"to", @"this", @"that", @"them", @"then", @"there", @"they", @"these", @"time",
|
|
@"use", @"up", @"under",
|
|
@"very",
|
|
@"we", @"with", @"what", @"when", @"where", @"who", @"why", @"will", @"would",
|
|
@"you", @"your"
|
|
];
|
|
}
|
|
|
|
#pragma mark - Engine Type Management
|
|
|
|
- (void)setEngineTypeFromString:(NSString *)engineTypeString {
|
|
if ([engineTypeString isEqualToString:@"latin"]) {
|
|
self.engineType = KBSuggestionEngineTypeLatin;
|
|
} else if ([engineTypeString isEqualToString:@"spanish"]) {
|
|
self.engineType = KBSuggestionEngineTypeSpanish;
|
|
} else if ([engineTypeString isEqualToString:@"english"]) {
|
|
self.engineType = KBSuggestionEngineTypeEnglish;
|
|
} else if ([engineTypeString isEqualToString:@"portuguese"]) {
|
|
self.engineType = KBSuggestionEngineTypePortuguese;
|
|
} else if ([engineTypeString isEqualToString:@"indonesian"]) {
|
|
self.engineType = KBSuggestionEngineTypeIndonesian;
|
|
} else if ([engineTypeString isEqualToString:@"pinyin_traditional"]) {
|
|
self.engineType = KBSuggestionEngineTypePinyinTraditional;
|
|
} else if ([engineTypeString isEqualToString:@"pinyin_simplified"]) {
|
|
self.engineType = KBSuggestionEngineTypePinyinSimplified;
|
|
} else if ([engineTypeString isEqualToString:@"bopomofo"]) {
|
|
self.engineType = KBSuggestionEngineTypeBopomofo;
|
|
} else {
|
|
self.engineType = KBSuggestionEngineTypeLatin;
|
|
}
|
|
[self kb_trimCachesForEngineType:self.engineType];
|
|
NSLog(@"[KBSuggestionEngine] Engine type set to: %@", engineTypeString);
|
|
}
|
|
|
|
#pragma mark - English Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_englishSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.englishWords) {
|
|
self.englishWords = [self kb_loadEnglishWords];
|
|
}
|
|
NSArray<NSString *> *matches = [self kb_suggestionsFromWordList:self.englishWords
|
|
prefix:prefix
|
|
limit:limit];
|
|
if (matches.count == 0) {
|
|
return [self kb_latinSuggestionsForPrefix:prefix limit:limit];
|
|
}
|
|
return matches;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_loadEnglishWords {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"english_words" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] english_words.json not found, using default words");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read english_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse english_words.json: %@", error);
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSArray *wordsArray = json[@"words"];
|
|
if (![wordsArray isKindOfClass:NSArray.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid words array in english_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSMutableArray<NSString *> *result = [NSMutableArray array];
|
|
for (id item in wordsArray) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[result addObject:item];
|
|
}
|
|
}
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu English words", (unsigned long)result.count);
|
|
return result.count > 0 ? [result copy] : [self.class kb_defaultWords];
|
|
}
|
|
|
|
#pragma mark - Latin Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_latinSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.words) {
|
|
self.words = [self kb_loadWords];
|
|
}
|
|
NSString *lower = prefix.lowercaseString;
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
|
|
for (NSString *word in self.words) {
|
|
if ([word hasPrefix:lower]) {
|
|
[matches addObject:word];
|
|
if (matches.count >= limit * 3) {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (matches.count == 0) { return @[]; }
|
|
|
|
[matches sortUsingComparator:^NSComparisonResult(NSString *a, NSString *b) {
|
|
NSInteger ca = self.selectionCounts[a].integerValue;
|
|
NSInteger cb = self.selectionCounts[b].integerValue;
|
|
if (ca != cb) {
|
|
return (cb > ca) ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
BOOL pa = [self.priorityWords containsObject:a];
|
|
BOOL pb = [self.priorityWords containsObject:b];
|
|
if (pa != pb) {
|
|
return pa ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
return [a compare:b];
|
|
}];
|
|
|
|
if (matches.count > limit) {
|
|
return [matches subarrayWithRange:NSMakeRange(0, limit)];
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
#pragma mark - Traditional Chinese Pinyin Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_traditionalPinyinSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.pinyinToTraditionalMap) {
|
|
self.pinyinToTraditionalMap = [self kb_loadPinyinToTraditionalMap];
|
|
}
|
|
NSString *lower = prefix.lowercaseString;
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
|
|
NSArray<NSString *> *directMatches = self.pinyinToTraditionalMap[lower];
|
|
if (directMatches.count > 0) {
|
|
[matches addObjectsFromArray:directMatches];
|
|
}
|
|
|
|
for (NSString *key in self.pinyinToTraditionalMap) {
|
|
if ([key hasPrefix:lower] && ![key isEqualToString:lower]) {
|
|
NSArray<NSString *> *candidates = self.pinyinToTraditionalMap[key];
|
|
[matches addObjectsFromArray:candidates];
|
|
if (matches.count >= limit * 2) {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (matches.count == 0) {
|
|
return [self kb_fallbackTraditionalSuggestions:lower limit:limit];
|
|
}
|
|
|
|
[matches sortUsingComparator:^NSComparisonResult(NSString *a, NSString *b) {
|
|
NSInteger ca = self.selectionCounts[a].integerValue;
|
|
NSInteger cb = self.selectionCounts[b].integerValue;
|
|
if (ca != cb) {
|
|
return (cb > ca) ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
return [a compare:b];
|
|
}];
|
|
|
|
if (matches.count > limit) {
|
|
return [matches subarrayWithRange:NSMakeRange(0, limit)];
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_fallbackTraditionalSuggestions:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.traditionalChineseWords) {
|
|
self.traditionalChineseWords = [self kb_loadTraditionalChineseWords];
|
|
}
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
for (NSString *word in self.traditionalChineseWords) {
|
|
[matches addObject:word];
|
|
if (matches.count >= limit) {
|
|
break;
|
|
}
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
#pragma mark - Simplified Chinese Pinyin Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_simplifiedPinyinSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.pinyinToTraditionalMap) {
|
|
self.pinyinToTraditionalMap = [self kb_loadPinyinToTraditionalMap];
|
|
}
|
|
NSString *lower = prefix.lowercaseString;
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
|
|
NSArray<NSString *> *directMatches = self.pinyinToTraditionalMap[lower];
|
|
if (directMatches.count > 0) {
|
|
for (NSString *tradChar in directMatches) {
|
|
NSString *simplified = [self kb_toSimplified:tradChar];
|
|
if (simplified.length > 0) {
|
|
[matches addObject:simplified];
|
|
}
|
|
}
|
|
}
|
|
|
|
for (NSString *key in self.pinyinToTraditionalMap) {
|
|
if ([key hasPrefix:lower] && ![key isEqualToString:lower]) {
|
|
NSArray<NSString *> *candidates = self.pinyinToTraditionalMap[key];
|
|
for (NSString *tradChar in candidates) {
|
|
NSString *simplified = [self kb_toSimplified:tradChar];
|
|
if (simplified.length > 0) {
|
|
[matches addObject:simplified];
|
|
}
|
|
}
|
|
if (matches.count >= limit * 2) {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (matches.count == 0) {
|
|
return [self kb_fallbackSimplifiedSuggestions:lower limit:limit];
|
|
}
|
|
|
|
[matches sortUsingComparator:^NSComparisonResult(NSString *a, NSString *b) {
|
|
NSInteger ca = self.selectionCounts[a].integerValue;
|
|
NSInteger cb = self.selectionCounts[b].integerValue;
|
|
if (ca != cb) {
|
|
return (cb > ca) ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
return [a compare:b];
|
|
}];
|
|
|
|
if (matches.count > limit) {
|
|
return [matches subarrayWithRange:NSMakeRange(0, limit)];
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_fallbackSimplifiedSuggestions:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.simplifiedChineseWords) {
|
|
self.simplifiedChineseWords = [self kb_loadSimplifiedChineseWords];
|
|
}
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
for (NSString *word in self.simplifiedChineseWords) {
|
|
[matches addObject:word];
|
|
if (matches.count >= limit) {
|
|
break;
|
|
}
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
- (NSString *)kb_toSimplified:(NSString *)traditional {
|
|
static NSDictionary<NSString *, NSString *> *tradToSimpMap = nil;
|
|
static dispatch_once_t onceToken;
|
|
dispatch_once(&onceToken, ^{
|
|
tradToSimpMap = @{
|
|
@"臺": @"台", @"臺": @"台", @"灣": @"湾", @"語": @"语", @"體": @"体",
|
|
@"國": @"国", @"學": @"学", @"時": @"时", @"問": @"问", @"見": @"见",
|
|
@"經": @"经", @"動": @"动", @"長": @"长", @"開": @"开", @"關": @"关",
|
|
@"無": @"无", @"說": @"说", @"書": @"书", @"電": @"电", @"機": @"机",
|
|
@"氣": @"气", @"這": @"这", @"們": @"们", @"個": @"个", @"對": @"对",
|
|
@"來": @"来", @"還": @"还", @"過": @"过", @"會": @"会", @"進": @"进",
|
|
@"開": @"开", @"頭": @"头", @"點": @"点", @"問": @"问", @"題": @"题",
|
|
@"變": @"变", @"條": @"条", @"東": @"东", @"車": @"车", @"錢": @"钱",
|
|
@"門": @"门", @"聽": @"听", @"聲": @"声", @"醫": @"医", @"讓": @"让",
|
|
@"識": @"识", @"務": @"务", @"農": @"农", @"業": @"业", @"產": @"产",
|
|
@"黨": @"党", @"歷": @"历", @"史": @"史", @"後": @"后", @"前": @"前",
|
|
@"強": @"强", @"當": @"当", @"應": @"应", @"從": @"从", @"優": @"优",
|
|
@"兒": @"儿", @"兩": @"两", @"幾": @"几", @"廣": @"广", @"場": @"场",
|
|
@"決": @"决", @"許": @"许", @"設": @"设", @"請": @"请", @"論": @"论",
|
|
@"認": @"认", @"斷": @"断", @"離": @"离", @"須": @"须", @"導": @"导",
|
|
@"爭": @"争", @"重": @"重", @"輕": @"轻", @"難": @"难", @"極": @"极",
|
|
@"據": @"据", @"實": @"实", @"際": @"际", @"標": @"标", @"準": @"准",
|
|
@"確": @"确", @"證": @"证", @"驗": @"验", @"權": @"权", @"規": @"规",
|
|
@"則": @"则", @"劃": @"划", @"計": @"计", @"劃": @"划", @"術": @"术",
|
|
@"藝": @"艺", @"術": @"术", @"選": @"选", @"舉": @"举", @"團": @"团",
|
|
@"結": @"结", @"組": @"组", @"織": @"织", @"義": @"义", @"務": @"务",
|
|
@"親": @"亲", @"愛": @"爱", @"情": @"情", @"懷": @"怀", @"家": @"家",
|
|
@"屬": @"属", @"幫": @"帮", @"助": @"助", @"友": @"友", @"誼": @"谊",
|
|
@"謝": @"谢", @"謝": @"谢", @"對": @"对", @"起": @"起", @"早": @"早",
|
|
@"安": @"安", @"晚": @"晚", @"請": @"请", @"問": @"问", @"沒": @"没",
|
|
@"關": @"关", @"係": @"系", @"加": @"加", @"油": @"油", @"台": @"台",
|
|
@"北": @"北", @"高": @"高", @"雄": @"雄", @"中": @"中", @"南": @"南",
|
|
@"朋": @"朋", @"友": @"友", @"人": @"人", @"工": @"工", @"作": @"作",
|
|
@"習": @"习", @"生": @"生", @"活": @"活", @"地": @"地", @"方": @"方",
|
|
@"法": @"法", @"答": @"答", @"喜": @"喜", @"歡": @"欢", @"想": @"想",
|
|
@"念": @"念", @"開": @"开", @"心": @"心", @"快": @"快", @"樂": @"乐",
|
|
@"美": @"美", @"麗": @"丽", @"漂": @"漂", @"亮": @"亮", @"帥": @"帅",
|
|
@"氣": @"气", @"可": @"可", @"愛": @"爱", @"溫": @"温", @"柔": @"柔"
|
|
};
|
|
});
|
|
|
|
if (tradToSimpMap[traditional]) {
|
|
return tradToSimpMap[traditional];
|
|
}
|
|
|
|
NSMutableString *result = [traditional mutableCopy];
|
|
[tradToSimpMap enumerateKeysAndObjectsUsingBlock:^(NSString *trad, NSString *simp, BOOL *stop) {
|
|
[result replaceOccurrencesOfString:trad withString:simp options:0 range:NSMakeRange(0, result.length)];
|
|
}];
|
|
|
|
return result.length > 0 ? [result copy] : traditional;
|
|
}
|
|
|
|
#pragma mark - Bopomofo (Zhuyin) Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_bopomofoSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.bopomofoToChineseMap) {
|
|
self.bopomofoToChineseMap = [self kb_loadBopomofoToChineseMap];
|
|
}
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
|
|
NSArray<NSString *> *directMatches = self.bopomofoToChineseMap[prefix];
|
|
if (directMatches.count > 0) {
|
|
[matches addObjectsFromArray:directMatches];
|
|
}
|
|
|
|
for (NSString *key in self.bopomofoToChineseMap) {
|
|
if ([key hasPrefix:prefix] && ![key isEqualToString:prefix]) {
|
|
NSArray<NSString *> *candidates = self.bopomofoToChineseMap[key];
|
|
[matches addObjectsFromArray:candidates];
|
|
if (matches.count >= limit * 2) {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (matches.count == 0) {
|
|
return [self kb_fallbackTraditionalSuggestions:prefix limit:limit];
|
|
}
|
|
|
|
[matches sortUsingComparator:^NSComparisonResult(NSString *a, NSString *b) {
|
|
NSInteger ca = self.selectionCounts[a].integerValue;
|
|
NSInteger cb = self.selectionCounts[b].integerValue;
|
|
if (ca != cb) {
|
|
return (cb > ca) ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
return [a compare:b];
|
|
}];
|
|
|
|
if (matches.count > limit) {
|
|
return [matches subarrayWithRange:NSMakeRange(0, limit)];
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
#pragma mark - Chinese Word Loading
|
|
|
|
- (NSArray<NSString *> *)kb_loadTraditionalChineseWords {
|
|
// 加载繁体中文常用词
|
|
// 这里先返回一些示例词，实际应该从文件或数据库加载
|
|
return @[
|
|
@"你好", @"謝謝", @"對不起", @"再見", @"早安",
|
|
@"晚安", @"請問", @"不好意思", @"沒關係", @"加油",
|
|
@"台灣", @"台北", @"高雄", @"台中", @"台南",
|
|
@"朋友", @"家人", @"工作", @"學習", @"生活",
|
|
@"時間", @"地點", @"方法", @"問題", @"答案",
|
|
@"喜歡", @"愛", @"想念", @"開心", @"快樂",
|
|
@"美麗", @"漂亮", @"帥氣", @"可愛", @"溫柔"
|
|
];
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_loadSimplifiedChineseWords {
|
|
return @[
|
|
@"你好", @"谢谢", @"对不起", @"再见", @"早安",
|
|
@"晚安", @"请问", @"不好意思", @"没关系", @"加油",
|
|
@"中国", @"北京", @"上海", @"广州", @"深圳",
|
|
@"朋友", @"家人", @"工作", @"学习", @"生活",
|
|
@"时间", @"地点", @"方法", @"问题", @"答案",
|
|
@"喜欢", @"爱", @"想念", @"开心", @"快乐",
|
|
@"美丽", @"漂亮", @"帅气", @"可爱", @"温柔"
|
|
];
|
|
}
|
|
|
|
#pragma mark - Pinyin & Bopomofo Map Loading
|
|
|
|
- (NSDictionary<NSString *, NSArray<NSString *> *> *)kb_loadPinyinToTraditionalMap {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"pinyin_to_traditional" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] pinyin_to_traditional.json not found, using empty map");
|
|
return @{};
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read pinyin_to_traditional.json");
|
|
return @{};
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse pinyin_to_traditional.json: %@", error);
|
|
return @{};
|
|
}
|
|
|
|
NSDictionary *mappings = json[@"mappings"];
|
|
if (![mappings isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid mappings in pinyin_to_traditional.json");
|
|
return @{};
|
|
}
|
|
|
|
NSMutableDictionary<NSString *, NSArray<NSString *> *> *result = [NSMutableDictionary dictionary];
|
|
[mappings enumerateKeysAndObjectsUsingBlock:^(NSString *key, id obj, BOOL *stop) {
|
|
if ([obj isKindOfClass:NSArray.class]) {
|
|
NSMutableArray<NSString *> *chars = [NSMutableArray array];
|
|
for (id item in (NSArray *)obj) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[chars addObject:item];
|
|
}
|
|
}
|
|
if (chars.count > 0) {
|
|
result[key] = [chars copy];
|
|
}
|
|
}
|
|
}];
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu pinyin mappings", (unsigned long)result.count);
|
|
return [result copy];
|
|
}
|
|
|
|
- (NSDictionary<NSString *, NSArray<NSString *> *> *)kb_loadBopomofoToChineseMap {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"bopomofo_to_chinese" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] bopomofo_to_chinese.json not found, using empty map");
|
|
return @{};
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read bopomofo_to_chinese.json");
|
|
return @{};
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse bopomofo_to_chinese.json: %@", error);
|
|
return @{};
|
|
}
|
|
|
|
NSDictionary *mappings = json[@"mappings"];
|
|
if (![mappings isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid mappings in bopomofo_to_chinese.json");
|
|
return @{};
|
|
}
|
|
|
|
NSMutableDictionary<NSString *, NSArray<NSString *> *> *result = [NSMutableDictionary dictionary];
|
|
[mappings enumerateKeysAndObjectsUsingBlock:^(NSString *key, id obj, BOOL *stop) {
|
|
if ([obj isKindOfClass:NSArray.class]) {
|
|
NSMutableArray<NSString *> *chars = [NSMutableArray array];
|
|
for (id item in (NSArray *)obj) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[chars addObject:item];
|
|
}
|
|
}
|
|
if (chars.count > 0) {
|
|
result[key] = [chars copy];
|
|
}
|
|
}
|
|
}];
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu bopomofo mappings", (unsigned long)result.count);
|
|
return [result copy];
|
|
}
|
|
|
|
#pragma mark - Spanish Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_spanishSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.spanishWords) {
|
|
self.spanishWords = [self kb_loadSpanishWords];
|
|
}
|
|
NSArray<NSString *> *matches = [self kb_suggestionsFromWordList:self.spanishWords
|
|
prefix:prefix
|
|
limit:limit];
|
|
if (matches.count == 0) {
|
|
return [self kb_latinSuggestionsForPrefix:prefix limit:limit];
|
|
}
|
|
return matches;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_loadSpanishWords {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"spanish_words" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] spanish_words.json not found, using default words");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read spanish_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse spanish_words.json: %@", error);
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSArray *wordsArray = json[@"words"];
|
|
if (![wordsArray isKindOfClass:NSArray.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid words array in spanish_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSMutableArray<NSString *> *result = [NSMutableArray array];
|
|
for (id item in wordsArray) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[result addObject:item];
|
|
}
|
|
}
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu Spanish words", (unsigned long)result.count);
|
|
return result.count > 0 ? [result copy] : [self.class kb_defaultWords];
|
|
}
|
|
|
|
#pragma mark - Portuguese Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_portugueseSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.portugueseWords) {
|
|
self.portugueseWords = [self kb_loadPortugueseWords];
|
|
}
|
|
NSArray<NSString *> *matches = [self kb_suggestionsFromWordList:self.portugueseWords
|
|
prefix:prefix
|
|
limit:limit];
|
|
if (matches.count == 0) {
|
|
return [self kb_latinSuggestionsForPrefix:prefix limit:limit];
|
|
}
|
|
return matches;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_loadPortugueseWords {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"portuguese_words" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] portuguese_words.json not found, using default words");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read portuguese_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse portuguese_words.json: %@", error);
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSArray *wordsArray = json[@"words"];
|
|
if (![wordsArray isKindOfClass:NSArray.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid words array in portuguese_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSMutableArray<NSString *> *result = [NSMutableArray array];
|
|
for (id item in wordsArray) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[result addObject:item];
|
|
}
|
|
}
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu Portuguese words", (unsigned long)result.count);
|
|
return result.count > 0 ? [result copy] : [self.class kb_defaultWords];
|
|
}
|
|
|
|
#pragma mark - Indonesian Suggestions
|
|
|
|
- (NSArray<NSString *> *)kb_indonesianSuggestionsForPrefix:(NSString *)prefix limit:(NSUInteger)limit {
|
|
if (!self.indonesianWords) {
|
|
self.indonesianWords = [self kb_loadIndonesianWords];
|
|
}
|
|
NSArray<NSString *> *matches = [self kb_suggestionsFromWordList:self.indonesianWords
|
|
prefix:prefix
|
|
limit:limit];
|
|
if (matches.count == 0) {
|
|
return [self kb_latinSuggestionsForPrefix:prefix limit:limit];
|
|
}
|
|
return matches;
|
|
}
|
|
|
|
- (NSArray<NSString *> *)kb_loadIndonesianWords {
|
|
NSString *path = [[NSBundle mainBundle] pathForResource:@"indonesian_words" ofType:@"json"];
|
|
if (!path) {
|
|
NSLog(@"[KBSuggestionEngine] indonesian_words.json not found, using default words");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSData *data = [NSData dataWithContentsOfFile:path];
|
|
if (!data) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to read indonesian_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSError *error = nil;
|
|
NSDictionary *json = [NSJSONSerialization JSONObjectWithData:data options:0 error:&error];
|
|
if (error || ![json isKindOfClass:NSDictionary.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Failed to parse indonesian_words.json: %@", error);
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSArray *wordsArray = json[@"words"];
|
|
if (![wordsArray isKindOfClass:NSArray.class]) {
|
|
NSLog(@"[KBSuggestionEngine] Invalid words array in indonesian_words.json");
|
|
return [self.class kb_defaultWords];
|
|
}
|
|
|
|
NSMutableArray<NSString *> *result = [NSMutableArray array];
|
|
for (id item in wordsArray) {
|
|
if ([item isKindOfClass:NSString.class]) {
|
|
[result addObject:item];
|
|
}
|
|
}
|
|
|
|
NSLog(@"[KBSuggestionEngine] Loaded %lu Indonesian words", (unsigned long)result.count);
|
|
return result.count > 0 ? [result copy] : [self.class kb_defaultWords];
|
|
}
|
|
|
|
#pragma mark - Word List Helpers
|
|
|
|
- (NSArray<NSString *> *)kb_suggestionsFromWordList:(NSArray<NSString *> *)words
|
|
prefix:(NSString *)prefix
|
|
limit:(NSUInteger)limit {
|
|
NSString *lower = prefix.lowercaseString;
|
|
NSMutableArray<NSString *> *matches = [NSMutableArray array];
|
|
|
|
for (NSString *word in words) {
|
|
if ([word hasPrefix:lower]) {
|
|
[matches addObject:word];
|
|
if (matches.count >= limit * 2) {
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
if (matches.count == 0) { return @[]; }
|
|
|
|
[matches sortUsingComparator:^NSComparisonResult(NSString *a, NSString *b) {
|
|
NSInteger ca = self.selectionCounts[a].integerValue;
|
|
NSInteger cb = self.selectionCounts[b].integerValue;
|
|
if (ca != cb) {
|
|
return (cb > ca) ? NSOrderedAscending : NSOrderedDescending;
|
|
}
|
|
return [a compare:b];
|
|
}];
|
|
|
|
if (matches.count > limit) {
|
|
return [matches subarrayWithRange:NSMakeRange(0, limit)];
|
|
}
|
|
return matches.copy;
|
|
}
|
|
|
|
- (void)kb_trimCachesForEngineType:(KBSuggestionEngineType)engineType {
|
|
switch (engineType) {
|
|
case KBSuggestionEngineTypeEnglish:
|
|
self.spanishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.words = nil;
|
|
self.traditionalChineseWords = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
case KBSuggestionEngineTypeSpanish:
|
|
self.englishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.words = nil;
|
|
self.traditionalChineseWords = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
case KBSuggestionEngineTypePortuguese:
|
|
self.englishWords = nil;
|
|
self.spanishWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.words = nil;
|
|
self.traditionalChineseWords = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
case KBSuggestionEngineTypeIndonesian:
|
|
self.englishWords = nil;
|
|
self.spanishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.words = nil;
|
|
self.traditionalChineseWords = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
case KBSuggestionEngineTypePinyinTraditional:
|
|
case KBSuggestionEngineTypePinyinSimplified:
|
|
self.words = nil;
|
|
self.englishWords = nil;
|
|
self.spanishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
case KBSuggestionEngineTypeBopomofo:
|
|
self.words = nil;
|
|
self.englishWords = nil;
|
|
self.spanishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
break;
|
|
case KBSuggestionEngineTypeLatin:
|
|
default:
|
|
self.englishWords = nil;
|
|
self.spanishWords = nil;
|
|
self.portugueseWords = nil;
|
|
self.indonesianWords = nil;
|
|
self.traditionalChineseWords = nil;
|
|
self.simplifiedChineseWords = nil;
|
|
self.pinyinToTraditionalMap = nil;
|
|
self.bopomofoToChineseMap = nil;
|
|
break;
|
|
}
|
|
}
|
|
|
|
#pragma mark - Safety Filter
|
|
|
|
- (NSArray<NSString *> *)kb_filterSensitiveSuggestions:(NSArray<NSString *> *)items
|
|
limit:(NSUInteger)limit {
|
|
if (items.count == 0 || limit == 0) { return @[]; }
|
|
NSMutableOrderedSet<NSString *> *result = [NSMutableOrderedSet orderedSet];
|
|
for (id item in items) {
|
|
if (![item isKindOfClass:NSString.class]) { continue; }
|
|
NSString *word = (NSString *)item;
|
|
if (word.length == 0) { continue; }
|
|
if ([self kb_isSensitiveSuggestion:word]) { continue; }
|
|
[result addObject:word];
|
|
if (result.count >= limit) { break; }
|
|
}
|
|
return result.array ?: @[];
|
|
}
|
|
|
|
- (BOOL)kb_isSensitiveSuggestion:(NSString *)word {
|
|
NSString *normalized = [self kb_normalizedSuggestionToken:word];
|
|
if (normalized.length == 0) { return YES; }
|
|
if ([[self.class kb_blockedSuggestionWords] containsObject:normalized]) {
|
|
return YES;
|
|
}
|
|
for (NSString *fragment in [self.class kb_blockedSuggestionFragments]) {
|
|
if ([normalized containsString:fragment]) {
|
|
return YES;
|
|
}
|
|
}
|
|
return NO;
|
|
}
|
|
|
|
- (NSString *)kb_normalizedSuggestionToken:(NSString *)word {
|
|
if (![word isKindOfClass:NSString.class]) { return @""; }
|
|
NSString *value = [[word stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]]
|
|
lowercaseString];
|
|
if (value.length == 0) { return @""; }
|
|
value = [value stringByFoldingWithOptions:NSDiacriticInsensitiveSearch
|
|
locale:[NSLocale currentLocale]];
|
|
NSMutableCharacterSet *trimSet = [[NSCharacterSet punctuationCharacterSet] mutableCopy];
|
|
[trimSet formUnionWithCharacterSet:[NSCharacterSet symbolCharacterSet]];
|
|
return [value stringByTrimmingCharactersInSet:trimSet];
|
|
}
|
|
|
|
+ (NSSet<NSString *> *)kb_blockedSuggestionWords {
|
|
static NSSet<NSString *> *words = nil;
|
|
static dispatch_once_t onceToken;
|
|
dispatch_once(&onceToken, ^{
|
|
// 上架合规优先：过滤常见成人、露骨性行为、毒品、暴力武器等高风险词。
|
|
words = [NSSet setWithArray:@[
|
|
@"sex", @"sexy", @"porn", @"porno", @"xxx", @"nude", @"naked",
|
|
@"fuck", @"fucking", @"shit", @"bitch", @"penis", @"vagina",
|
|
@"boob", @"rape", @"cocaine", @"heroin", @"drug", @"drugs",
|
|
@"kill", @"murder", @"gun", @"weapon",
|
|
@"sexo", @"porno", @"pornografia", @"violacion", @"violacao",
|
|
@"drogas", @"cocaina", @"heroina", @"arma", @"matar", @"muerte",
|
|
@"pene",
|
|
@"色情", @"裸露", @"裸体", @"裸聊", @"裸照",
|
|
@"强奸", @"毒品", @"海洛因", @"可卡因",
|
|
@"枪", @"武器", @"杀人", @"谋杀"
|
|
]];
|
|
});
|
|
return words;
|
|
}
|
|
|
|
+ (NSArray<NSString *> *)kb_blockedSuggestionFragments {
|
|
static NSArray<NSString *> *fragments = nil;
|
|
static dispatch_once_t onceToken;
|
|
dispatch_once(&onceToken, ^{
|
|
fragments = @[
|
|
@"porn", @"fuck", @"rape", @"cocaine", @"heroin",
|
|
@"色情", @"裸聊", @"裸照", @"强奸", @"毒品", @"杀人"
|
|
];
|
|
});
|
|
return fragments;
|
|
}
|
|
|
|
@end
|