Files
keyboard/CustomKeyboard/Network/KBStreamFetcher.m

381 lines
17 KiB
Mathematica
Raw Normal View History

2025-11-12 14:18:56 +08:00
//
// KBStreamFetcher.m
//
#import "KBStreamFetcher.h"
@interface KBStreamFetcher () <NSURLSessionDataDelegate>
@property (nonatomic, strong) NSURLSession *session;
@property (nonatomic, strong) NSURLSessionDataTask *task;
@property (nonatomic, strong) NSMutableData *buffer; //
@property (nonatomic, assign) NSStringEncoding textEncoding; // UTF-8
@property (nonatomic, assign) BOOL isSSE; // SSE
@property (nonatomic, strong) NSMutableString *sseTextBuffer; // SSE
@property (nonatomic, assign) NSInteger decodedPrefixBytes; // sseTextBuffer SSE
@property (nonatomic, assign) NSInteger deliveredCharCount; // SSE
@property (nonatomic, assign) BOOL hasEmitted; // 1 \t
2025-11-12 15:31:22 +08:00
@property (nonatomic, assign) BOOL lastChunkEndedWithTab; // "\t" \t
2025-11-12 14:36:15 +08:00
@property (nonatomic, strong) NSMutableArray<NSString *> *pendingQueue; //
@property (nonatomic, strong) NSTimer *flushTimer; //
@property (nonatomic, strong, nullable) NSError *finishError; //
2025-11-12 16:49:19 +08:00
// Metrics
@property (nonatomic, assign) CFAbsoluteTime tStart; // start()
@property (nonatomic, assign) CFAbsoluteTime tFirstByte; //
@property (nonatomic, assign) CFAbsoluteTime tFinish; // /
@property (nonatomic, assign) NSInteger emittedChunkCount; //
2025-11-12 14:18:56 +08:00
@end
// UTF-8
static NSUInteger kb_validUTF8PrefixLen(NSData *data) {
const unsigned char *bytes = (const unsigned char *)data.bytes;
NSUInteger n = data.length;
if (n == 0) return 0;
NSInteger i = (NSInteger)n - 1;
while (i >= 0 && (bytes[i] & 0xC0) == 0x80) { i--; } // 10xxxxxx
if (i < 0) return 0; //
unsigned char b = bytes[i];
NSUInteger expected = 1;
if ((b & 0x80) == 0x00) expected = 1; // 0xxxxxxx
else if ((b & 0xE0) == 0xC0) expected = 2; // 110xxxxx
else if ((b & 0xF0) == 0xE0) expected = 3; // 1110xxxx
else if ((b & 0xF8) == 0xF0) expected = 4; // 11110xxx
else return (NSUInteger)i; // i
NSUInteger remain = n - (NSUInteger)i;
return (remain >= expected) ? n : (NSUInteger)i;
}
@implementation KBStreamFetcher
+ (instancetype)fetcherWithURL:(NSURL *)url {
KBStreamFetcher *f = [[self alloc] init];
f.url = url;
return f;
}
- (instancetype)init {
if (self = [super init]) {
_acceptEventStream = NO;
_disableCompression = YES;
_treatSlashTAsTab = YES;
_trimLeadingTabOnce = YES;
_requestTimeout = 30.0;
_textEncoding = NSUTF8StringEncoding;
_buffer = [NSMutableData data];
_sseTextBuffer = [NSMutableString string];
2025-11-12 14:36:15 +08:00
_pendingQueue = [NSMutableArray array];
2025-11-12 15:31:22 +08:00
_flushInterval = 0.1;
2025-11-12 14:36:15 +08:00
_splitLargeDeltasOnWhitespace = YES;
2025-11-12 16:49:19 +08:00
_loggingEnabled = YES;
2025-11-12 14:18:56 +08:00
}
return self;
}
- (void)start {
if (!self.url) return;
[self cancel];
NSURLSessionConfiguration *cfg = [NSURLSessionConfiguration defaultSessionConfiguration];
cfg.requestCachePolicy = NSURLRequestReloadIgnoringLocalCacheData;
cfg.timeoutIntervalForRequest = self.requestTimeout;
cfg.timeoutIntervalForResource = MAX(self.requestTimeout, 60.0);
self.session = [NSURLSession sessionWithConfiguration:cfg delegate:self delegateQueue:[NSOperationQueue mainQueue]];
NSMutableURLRequest *req = [NSMutableURLRequest requestWithURL:self.url];
req.HTTPMethod = @"GET";
if (self.disableCompression) { [req setValue:@"identity" forHTTPHeaderField:@"Accept-Encoding"]; }
if (self.acceptEventStream) { [req setValue:@"text/event-stream" forHTTPHeaderField:@"Accept"]; }
[req setValue:@"no-cache" forHTTPHeaderField:@"Cache-Control"];
[req setValue:@"keep-alive" forHTTPHeaderField:@"Connection"];
[self.extraHeaders enumerateKeysAndObjectsUsingBlock:^(NSString *k, NSString *v, BOOL *stop){ [req setValue:v forHTTPHeaderField:k]; }];
//
[self.buffer setLength:0];
[self.sseTextBuffer setString:@""];
self.isSSE = NO;
self.textEncoding = NSUTF8StringEncoding;
self.decodedPrefixBytes = 0;
self.deliveredCharCount = 0;
self.hasEmitted = NO;
2025-11-12 15:31:22 +08:00
self.lastChunkEndedWithTab = NO;
2025-11-12 14:36:15 +08:00
[self.pendingQueue removeAllObjects];
[self.flushTimer invalidate]; self.flushTimer = nil;
self.finishError = nil;
2025-11-12 14:18:56 +08:00
2025-11-12 16:49:19 +08:00
self.tStart = CFAbsoluteTimeGetCurrent();
self.tFirstByte = 0;
self.tFinish = 0;
self.emittedChunkCount = 0;
if (self.loggingEnabled) {
NSLog(@"[KBStream] start url=%@ acceptSSE=%@ disableCompression=%@ flush=%.0fms splitWords=%@",
self.url.absoluteString,
self.acceptEventStream?@"YES":@"NO",
self.disableCompression?@"YES":@"NO",
self.flushInterval*1000.0,
self.splitLargeDeltasOnWhitespace?@"YES":@"NO");
}
2025-11-12 14:18:56 +08:00
self.task = [self.session dataTaskWithRequest:req];
[self.task resume];
}
- (void)cancel {
[self.task cancel];
self.task = nil;
[self.session invalidateAndCancel];
self.session = nil;
[self.buffer setLength:0];
[self.sseTextBuffer setString:@""];
self.decodedPrefixBytes = 0;
self.deliveredCharCount = 0;
self.hasEmitted = NO;
2025-11-12 15:31:22 +08:00
self.lastChunkEndedWithTab = NO;
2025-11-12 14:36:15 +08:00
[self.pendingQueue removeAllObjects];
[self.flushTimer invalidate]; self.flushTimer = nil;
self.finishError = nil;
2025-11-12 14:18:56 +08:00
}
#pragma mark - NSURLSessionDataDelegate
- (void)URLSession:(NSURLSession *)session dataTask:(NSURLSessionDataTask *)dataTask didReceiveResponse:(NSURLResponse *)response completionHandler:(void (^)(NSURLSessionResponseDisposition))completionHandler {
self.isSSE = NO;
self.textEncoding = NSUTF8StringEncoding;
if ([response isKindOfClass:[NSHTTPURLResponse class]]) {
NSHTTPURLResponse *r = (NSHTTPURLResponse *)response;
NSString *ct = r.allHeaderFields[@"Content-Type"] ?: r.allHeaderFields[@"content-type"];
if ([ct isKindOfClass:[NSString class]]) {
NSString *lower = [ct lowercaseString];
if ([lower containsString:@"text/event-stream"]) self.isSSE = YES;
NSRange pos = [lower rangeOfString:@"charset="];
if (pos.location != NSNotFound) {
NSString *charset = [[lower substringFromIndex:pos.location + pos.length] componentsSeparatedByString:@";"][0];
if ([charset containsString:@"utf-8"] || [charset containsString:@"utf8"]) {
self.textEncoding = NSUTF8StringEncoding;
} else if ([charset containsString:@"iso-8859-1"] || [charset containsString:@"latin1"]) {
self.textEncoding = NSISOLatin1StringEncoding;
}
}
}
}
[self.sseTextBuffer setString:@""];
self.decodedPrefixBytes = 0;
if (completionHandler) completionHandler(NSURLSessionResponseAllow);
}
- (void)URLSession:(NSURLSession *)session dataTask:(NSURLSessionDataTask *)dataTask didReceiveData:(NSData *)data {
if (data.length == 0) return;
[self.buffer appendData:data];
NSUInteger validLen = (self.textEncoding == NSUTF8StringEncoding)
? kb_validUTF8PrefixLen(self.buffer)
: self.buffer.length;
2025-11-12 16:49:19 +08:00
if (validLen > 0 && self.tFirstByte == 0) {
self.tFirstByte = CFAbsoluteTimeGetCurrent();
if (self.loggingEnabled) {
NSLog(@"[KBStream] first-bytes after %.0fms (encoding=%@, SSE=%@)",
(self.tFirstByte - self.tStart)*1000.0,
(self.textEncoding==NSUTF8StringEncoding?@"UTF-8":@"Other"),
self.isSSE?@"YES":@"NO");
}
}
2025-11-12 14:18:56 +08:00
if (validLen == 0) return; //
if (self.isSSE) {
if ((NSUInteger)self.decodedPrefixBytes < validLen) {
NSRange rng = NSMakeRange((NSUInteger)self.decodedPrefixBytes, validLen - (NSUInteger)self.decodedPrefixBytes);
NSString *piece = [[NSString alloc] initWithBytes:(const char *)self.buffer.bytes + rng.location
length:rng.length
encoding:self.textEncoding];
if (piece.length > 0) {
[self.sseTextBuffer appendString:piece];
self.decodedPrefixBytes = (NSInteger)validLen;
}
}
// SSE \n\n
if (self.sseTextBuffer.length > 0) {
NSString *normalized = [self.sseTextBuffer stringByReplacingOccurrencesOfString:@"\r\n" withString:@"\n"];
[self.sseTextBuffer setString:normalized];
while (1) {
NSRange sep = [self.sseTextBuffer rangeOfString:@"\n\n"]; //
if (sep.location == NSNotFound) break;
NSString *event = [self.sseTextBuffer substringToIndex:sep.location];
[self.sseTextBuffer deleteCharactersInRange:NSMakeRange(0, sep.location + sep.length)];
// data:
NSArray<NSString *> *lines = [event componentsSeparatedByString:@"\n"];
NSMutableString *payload = [NSMutableString string];
for (NSString *ln in lines) {
if ([ln hasPrefix:@"data:"]) {
NSString *v = [ln substringFromIndex:5];
if (v.length > 0 && [v hasPrefix:@" "]) v = [v substringFromIndex:1];
[payload appendString:v ?: @""];
}
}
2025-11-12 14:36:15 +08:00
if (payload.length > 0) { [self enqueueChunk:payload]; }
2025-11-12 14:18:56 +08:00
}
}
return;
}
// SSE
NSString *prefix = [[NSString alloc] initWithBytes:self.buffer.bytes length:validLen encoding:self.textEncoding];
if (!prefix) return;
if (self.deliveredCharCount < (NSInteger)prefix.length) {
NSString *delta = [prefix substringFromIndex:self.deliveredCharCount];
self.deliveredCharCount = prefix.length;
2025-11-12 14:36:15 +08:00
if (self.splitLargeDeltasOnWhitespace && delta.length > 16) {
// 使
NSArray<NSString *> *parts = [delta componentsSeparatedByString:@" "];
for (NSUInteger i = 0; i < parts.count; i++) {
NSString *w = parts[i];
if (w.length == 0) { [self enqueueChunk:@" "]; continue; }
if (i + 1 < parts.count) {
[self enqueueChunk:[w stringByAppendingString:@" "]];
} else {
[self enqueueChunk:w];
}
}
} else {
[self enqueueChunk:delta];
}
2025-11-12 14:18:56 +08:00
}
}
- (void)URLSession:(NSURLSession *)session task:(NSURLSessionTask *)task didCompleteWithError:(NSError *)error {
if (!error && self.isSSE && self.sseTextBuffer.length > 0) {
// \n\n
NSString *normalized = [self.sseTextBuffer stringByReplacingOccurrencesOfString:@"\r\n" withString:@"\n"];
NSArray<NSString *> *lines = [normalized componentsSeparatedByString:@"\n"];
NSMutableString *payload = [NSMutableString string];
for (NSString *ln in lines) {
if ([ln hasPrefix:@"data:"]) {
NSString *v = [ln substringFromIndex:5];
if (v.length > 0 && [v hasPrefix:@" "]) v = [v substringFromIndex:1];
[payload appendString:v ?: @""];
}
}
if (payload.length > 0) {
2025-11-12 16:49:19 +08:00
NSString *delta = nil;
if ((NSInteger)payload.length >= self.deliveredCharCount) {
delta = [payload substringFromIndex:self.deliveredCharCount];
} else {
delta = payload;
}
self.deliveredCharCount = payload.length;
if (delta.length > 0) { [self emitChunk:delta]; }
2025-11-12 14:18:56 +08:00
}
}
2025-11-12 16:49:19 +08:00
self.tFinish = CFAbsoluteTimeGetCurrent();
if (self.loggingEnabled) {
double t0 = (self.tFirstByte>0? (self.tFirstByte - self.tStart)*1000.0 : -1);
double t1 = (self.tFirstByte>0? (self.tFinish - self.tFirstByte)*1000.0 : -1);
double tt = (self.tFinish - self.tStart)*1000.0;
NSLog(@"[KBStream] finish chunks=%ld firstByte=%.0fms after start, tail=%.0fms, total=%.0fms error=%@",
(long)self.emittedChunkCount, t0, t1, tt, error);
}
2025-11-12 14:36:15 +08:00
// finish
if (self.pendingQueue.count > 0) {
self.finishError = error;
[self startFlushTimerIfNeeded];
} else {
if (self.onFinish) dispatch_async(dispatch_get_main_queue(), ^{ self.onFinish(error); });
[self cancel];
}
2025-11-12 14:18:56 +08:00
}
#pragma mark - Helpers
- (void)emitChunk:(NSString *)rawText {
if (rawText.length == 0) return;
NSString *text = rawText;
2025-11-12 17:55:59 +08:00
// 0) \r/\n "\n\t""\r\n\t""\r\t" "\t"
text = [text stringByReplacingOccurrencesOfString:@"\r\n\t" withString:@"\t"];
text = [text stringByReplacingOccurrencesOfString:@"\n\t" withString:@"\t"];
text = [text stringByReplacingOccurrencesOfString:@"\r\t" withString:@"\t"];
while (text.length > 0) {
unichar c0 = [text characterAtIndex:0];
if (c0 == '\n' || c0 == '\r') { text = [text substringFromIndex:1]; continue; }
break;
}
2025-11-12 15:31:22 +08:00
// 1) /t -> \t
2025-11-12 14:18:56 +08:00
if (self.treatSlashTAsTab) {
text = [text stringByReplacingOccurrencesOfString:@"/t" withString:@"\t"];
}
2025-11-12 15:31:22 +08:00
// 2) "\t"
2025-11-12 14:18:56 +08:00
if (!self.hasEmitted && self.trimLeadingTabOnce) {
2025-11-12 15:31:22 +08:00
if (text.length > 0 && [text characterAtIndex:0] == '\t') {
NSUInteger start = 1;
if (start < text.length && [text characterAtIndex:start] == ' ') start++;
text = [text substringFromIndex:start];
2025-11-12 14:18:56 +08:00
}
2025-11-12 15:31:22 +08:00
}
// 3) \t -> \t
if (text.length > 0) {
// \t
if (self.lastChunkEndedWithTab) {
NSUInteger j = 0;
while (j < text.length && [text characterAtIndex:j] == ' ') { j++; }
if (j > 0) {
text = [text substringFromIndex:1]; //
}
2025-11-12 14:18:56 +08:00
}
2025-11-12 15:31:22 +08:00
// \t \t
text = [text stringByReplacingOccurrencesOfString:@"\t " withString:@"\t"];
2025-11-12 14:18:56 +08:00
}
2025-11-12 15:31:22 +08:00
if (text.length == 0) { self.lastChunkEndedWithTab = NO; return; }
2025-11-12 16:49:19 +08:00
self.emittedChunkCount += 1;
if (self.loggingEnabled) {
NSLog(@"[KBStream] chunk#%ld len=%lu text=\"%@\"",
(long)self.emittedChunkCount, (unsigned long)text.length, KBPrintableSnippet(text, 160));
}
2025-11-12 14:18:56 +08:00
if (self.onChunk) dispatch_async(dispatch_get_main_queue(), ^{ self.onChunk(text); });
self.hasEmitted = YES;
2025-11-12 15:31:22 +08:00
// \t
unichar lastc = [text characterAtIndex:text.length - 1];
self.lastChunkEndedWithTab = (lastc == '\t');
2025-11-12 14:18:56 +08:00
}
2025-11-12 14:36:15 +08:00
#pragma mark - Queue/Flush
2025-11-12 14:18:56 +08:00
2025-11-12 14:36:15 +08:00
- (void)enqueueChunk:(NSString *)s {
if (s.length == 0) return;
[self.pendingQueue addObject:s];
[self startFlushTimerIfNeeded];
}
- (void)startFlushTimerIfNeeded {
if (self.flushTimer) return;
__weak typeof(self) weakSelf = self;
self.flushTimer = [NSTimer scheduledTimerWithTimeInterval:MAX(0.01, self.flushInterval)
repeats:YES
block:^(NSTimer * _Nonnull t) {
__strong typeof(weakSelf) self = weakSelf; if (!self) { [t invalidate]; return; }
if (self.pendingQueue.count == 0) {
[t invalidate]; self.flushTimer = nil;
if (self.finishError || self.finishError == nil) {
NSError *err = self.finishError; self.finishError = nil;
if (self.onFinish) dispatch_async(dispatch_get_main_queue(), ^{ self.onFinish(err); });
[self cancel];
}
return;
}
NSString *first = self.pendingQueue.firstObject;
[self.pendingQueue removeObjectAtIndex:0];
[self emitChunk:first];
}];
}
2025-11-12 16:49:19 +08:00
#pragma mark - Logging helpers
static NSString *KBPrintableSnippet(NSString *s, NSUInteger maxLen) {
if (!s) return @"";
NSString *x = [s stringByReplacingOccurrencesOfString:@"\n" withString:@"\\n"];
if (x.length > maxLen) {
x = [[x substringToIndex:maxLen] stringByAppendingString:@"…"];
}
return x;
}
2025-11-12 14:36:15 +08:00
@end