AFNetworking 中 GCD 的使用(3)

GCD in AFNetworking 的使用

AFHTTPSessionManager 处理HTTP的各种请求

只有异步到主线程操作

if (serializationError) {
        if (failure) {
            dispatch_async(self.completionQueue ?: dispatch_get_main_queue(), ^{
                failure(nil, serializationError);
            });
        }
        
        return nil;
    }

AFSecurityPolicy 管理加密证书 没有dispatch相关操作

AFURLResponseSerialization 响应序列化

​static NSLock* imageLock = nil;

@implementation UIImage (AFNetworkingSafeImageLoading)

+ (UIImage *)af_safeImageWithData:(NSData *)data {
    UIImage* image = nil;
    static dispatch_once_t onceToken;
    dispatch_once(&onceToken, ^{
        imageLock = [[NSLock alloc] init];
    });
    
    [imageLock lock];
    image = [UIImage imageWithData:data];
    [imageLock unlock];
    return image;
}

@end

AFURLRequestSerialization 请求序列化

1. 只初始化一次 dispatch_once

static NSArray * AFHTTPRequestSerializerObservedKeyPaths() {
    static NSArray *_AFHTTPRequestSerializerObservedKeyPaths = nil;
    static dispatch_once_t onceToken;
    dispatch_once(&onceToken, ^{
        _AFHTTPRequestSerializerObservedKeyPaths = @[NSStringFromSelector(@selector(allowsCellularAccess)), NSStringFromSelector(@selector(cachePolicy)), NSStringFromSelector(@selector(HTTPShouldHandleCookies)), NSStringFromSelector(@selector(HTTPShouldUsePipelining)), NSStringFromSelector(@selector(networkServiceType)), NSStringFromSelector(@selector(timeoutInterval))];
    });

    return _AFHTTPRequestSerializerObservedKeyPaths;
}

2. 创建并行队列

    self.requestHeaderModificationQueue = dispatch_queue_create("requestHeaderModificationQueue", DISPATCH_QUEUE_CONCURRENT);

3. 同步并发读 dispatch_sync 会在调用所在队列进行读操作，系统的一个优化

- (NSDictionary *)HTTPRequestHeaders {
    NSDictionary __block *value;
    dispatch_sync(self.requestHeaderModificationQueue, ^{
        value = [NSDictionary dictionaryWithDictionary:self.mutableHTTPRequestHeaders];
    });
    return value;
}


- (NSString *)valueForHTTPHeaderField:(NSString *)field {
    NSString __block *value;
    dispatch_sync(self.requestHeaderModificationQueue, ^{
        value = [self.mutableHTTPRequestHeaders valueForKey:field];
    });
    return value;
}

// 并发同步的编码
- (void)encodeWithCoder:(NSCoder *)coder {
    dispatch_sync(self.requestHeaderModificationQueue, ^{
        [coder encodeObject:self.mutableHTTPRequestHeaders forKey:NSStringFromSelector(@selector(mutableHTTPRequestHeaders))];
    });
    [coder encodeObject:@(self.queryStringSerializationStyle) forKey:NSStringFromSelector(@selector(queryStringSerializationStyle))];
}

// 并发同步设置属性，经过优化，还是在当前队列执行的
- (instancetype)copyWithZone:(NSZone *)zone {
    AFHTTPRequestSerializer *serializer = [[[self class] allocWithZone:zone] init];
    dispatch_sync(self.requestHeaderModificationQueue, ^{
        serializer.mutableHTTPRequestHeaders = [self.mutableHTTPRequestHeaders mutableCopyWithZone:zone];
    });
    serializer.queryStringSerializationStyle = self.queryStringSerializationStyle;
    serializer.queryStringSerialization = self.queryStringSerialization;

    return serializer;
}

4. 同步并发写 dispatch_barrier_sync 保证之前提交到操作队列中的都完毕，之后提交的，等自己执行完毕， 并发屏障

- (void)setValue:(NSString *)value
forHTTPHeaderField:(NSString *)field
{
    dispatch_barrier_sync(self.requestHeaderModificationQueue, ^{
        [self.mutableHTTPRequestHeaders setValue:value forKey:field];
    });
}

- (void)clearAuthorizationHeader {
    dispatch_barrier_sync(self.requestHeaderModificationQueue, ^{
        [self.mutableHTTPRequestHeaders removeObjectForKey:@"Authorization"];
    });
}

5. 后台线程，开启runloop， 读取 stream，完毕之后，异步到主线程

- (NSMutableURLRequest *)requestWithMultipartFormRequest:(NSURLRequest *)request
                             writingStreamContentsToFile:(NSURL *)fileURL
                                       completionHandler:(void (^)(NSError *error))handler
{
    NSParameterAssert(request.HTTPBodyStream);
    NSParameterAssert([fileURL isFileURL]);

    NSInputStream *inputStream = request.HTTPBodyStream;
    NSOutputStream *outputStream = [[NSOutputStream alloc] initWithURL:fileURL append:NO];
    __block NSError *error = nil;

    dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
        [inputStream scheduleInRunLoop:[NSRunLoop currentRunLoop] forMode:NSDefaultRunLoopMode];
        [outputStream scheduleInRunLoop:[NSRunLoop currentRunLoop] forMode:NSDefaultRunLoopMode];

        [inputStream open];
        [outputStream open];

        while ([inputStream hasBytesAvailable] && [outputStream hasSpaceAvailable]) {
            uint8_t buffer[1024];

            NSInteger bytesRead = [inputStream read:buffer maxLength:1024];
            if (inputStream.streamError || bytesRead < 0) {
                error = inputStream.streamError;
                break;
            }

            NSInteger bytesWritten = [outputStream write:buffer maxLength:(NSUInteger)bytesRead];
            if (outputStream.streamError || bytesWritten < 0) {
                error = outputStream.streamError;
                break;
            }

            if (bytesRead == 0 && bytesWritten == 0) {
                break;
            }
        }

        [outputStream close];
        [inputStream close];

        if (handler) {
            dispatch_async(dispatch_get_main_queue(), ^{
                handler(error);
            });
        }
    });

    NSMutableURLRequest *mutableRequest = [request mutableCopy];
    mutableRequest.HTTPBodyStream = nil;

    return mutableRequest;
}

6. 保证在主线程执行如下代码, dispatch_sync 跟 main queue 会保证在主线程执行

- (BOOL)transitionToNextPhase {
    if (![[NSThread currentThread] isMainThread]) {
        dispatch_sync(dispatch_get_main_queue(), ^{
            [self transitionToNextPhase];
        });
        return YES;
    }
  
        ...

    return YES;
}

总结：

1. 判定主线程，保证主线程执行, 主线程判定，sync to main queue
2. 并发队列的同步， 保证多路径执行。 dispatch_sync 会优化在当前线程；dispatch_barrier_sync 的同步，会保证之前的已完成，多线程中，只有一个在执行，之后在它之后执行。
3. 对于流的读写，配合 runloop 使用。
4. 并发队列的读同步，可以直接 dispatch_sync 读。
5. 并发队列的写同步，dispatch_barrier_sync， 也可以写异步dispatch_barrier_async