I'm using AFNetworking and SDURLCache for all my networking operations.
I have SDURLCache set like this:
SDURLCache *urlCache = [[SDURLCache alloc]
initWithMemoryCapacity:1024*1024*2 // 2MB mem cache
diskCapacity:1024*1024*15 // 15MB disk cache
diskPath:[SDURLCache defaultCachePath]];
[urlCache setMinCacheInterval:1];
[NSURLCache setSharedURLCache:urlCache];
All my request are using cachePolicy NSURLRequestUseProtocolCachePolicy, which according to apple docs works like this:
If an NSCachedURLResponse does not exist for the request, then the
data is fetched from the originating source. If there is a cached
response for the request, the URL loading system checks the response
to determine if it specifies that the contents must be revalidated. If
the contents must be revalidated a connection is made to the
originating source to see if it has changed. If it has not changed,
then the response is returned from the local cache. If it has changed,
the data is fetched from the originating source.
If the cached response doesn’t specify that the contents must be revalidated, the maximum age or expiration specified in the response
is examined. If the cached response is recent enough, then the
response is returned from the local cache. If the response is
determined to be stale, the originating source is checked for newer
data. If newer data is available, the data is fetched from the
originating source, otherwise it is returned from the cache.
So everything works perfectly even in airplane mode as long as the cache is not stale. When the cache expires (max-age and others), the failure block gets called.
I've been digging a little inside the SDURLCache and this method returns a response with valid data (I've parsed the data to a string and it contains the cached information)
- (NSCachedURLResponse *)cachedResponseForRequest:(NSURLRequest *)request {
request = [SDURLCache canonicalRequestForRequest:request];
NSCachedURLResponse *memoryResponse =
[super cachedResponseForRequest:request];
if (memoryResponse) {
return memoryResponse;
}
NSString *cacheKey = [SDURLCache cacheKeyForURL:request.URL];
// NOTE: We don't handle expiration here as even staled cache data is
// necessary for NSURLConnection to handle cache revalidation.
// Staled cache data is also needed for cachePolicies which force the
// use of the cache.
__block NSCachedURLResponse *response = nil;
dispatch_sync(get_disk_cache_queue(), ^{
NSMutableDictionary *accesses = [self.diskCacheInfo
objectForKey:kAFURLCacheInfoAccessesKey];
// OPTI: Check for cache-hit in in-memory dictionary before to hit FS
if ([accesses objectForKey:cacheKey]) {
response = [NSKeyedUnarchiver unarchiveObjectWithFile:
[_diskCachePath stringByAppendingPathComponent:cacheKey]];
if (response) {
// OPTI: Log entry last access time for LRU cache eviction
// algorithm but don't save the dictionary
// on disk now in order to save IO and time
[accesses setObject:[NSDate date] forKey:cacheKey];
_diskCacheInfoDirty = YES;
}
}
});
// OPTI: Store the response to memory cache for potential future requests
if (response) {
[super storeCachedResponse:response forRequest:request];
}
return response;
}
So at this point I have no idea what to do, because I believe that the response is handled by the OS and then AFNetworking receives a
- (void)connection:(NSURLConnection *)__unused connection
didFailWithError:(NSError *)error
inside AFURLConnectionOperation.
Well I've finally reached a not so ugly workaround:
First
If you're using IOS5/IOS6 you can drop SDURLCache and use the native one:
//Set Cache
NSURLCache *URLCache = [[NSURLCache alloc] initWithMemoryCapacity:4 * 1024 * 1024
diskCapacity:20 * 1024 * 1024
diskPath:nil];
[NSURLCache setSharedURLCache:URLCache];
But remember that in IOS5 https requests wont be cached in IOS6 they will.
Second
We need to add the following frameworks to our Prefix.pch so AFNetworking can start monitoring our internet connection.
#import <MobileCoreServices/MobileCoreServices.h>
#import <SystemConfiguration/SystemConfiguration.h>
Third
We need and AFHTTPClient instance so we can intercept every outgoing request and change his cachePolicy
-(NSMutableURLRequest *)requestWithMethod:(NSString *)method path:(NSString *)path parameters:(NSDictionary *)parameters {
NSMutableURLRequest * request = [super requestWithMethod:method path:path parameters:parameters];
if (request.cachePolicy == NSURLRequestUseProtocolCachePolicy && self.networkReachabilityStatus == AFNetworkReachabilityStatusNotReachable) {
request.cachePolicy = NSURLRequestReturnCacheDataDontLoad;
}
if (self.networkReachabilityStatus == AFNetworkReachabilityStatusUnknown) {
puts("uknown reachability status");
}
return request;
}
With these peaces of code we can now detect when the wifi/3g is unavailable and the specify the request to use always the cache no matter what. (Offline Mode)
Notes
I still don't know what to do when the networkReachabilityStatus is AFNetworkReachabilityStatusUnknown This can happen is a request is made as soon as the application starts and AF has not obtained the internet status yet.
Remember that in order for this to work the server has to set the correct cache headers in the http response.
UPDATE
Looks like IOS6 has some problems loading cached responses in no-internet situations, so even if the request is cached and the request cache policy is seted to NSURLRequestReturnCacheDataDontLoad the request will fail.
So an ugly workaround is to modify (void)connection:(NSURLConnection __unused *)connection
didFailWithError:(NSError *)error in AFURLConnectionOperation.m to retrieve the cached response if the request fails but only for specific cache policies.
- (void)connection:(NSURLConnection __unused *)connection
didFailWithError:(NSError *)error
{
self.error = error;
[self.outputStream close];
[self finish];
self.connection = nil;
//Ugly hack for making the request succeed if we can find a valid non-empty cached request
//This is because IOS6 is not handling cache responses right when we are in a no-connection sittuation
//Only use this code for cache policies that are supposed to listen to cache regarding it's expiration date
if (self.request.cachePolicy == NSURLRequestUseProtocolCachePolicy ||
self.request.cachePolicy == NSURLRequestReturnCacheDataElseLoad ||
self.request.cachePolicy == NSURLRequestReturnCacheDataDontLoad) {
NSCachedURLResponse * cachedResponse = [[NSURLCache sharedURLCache] cachedResponseForRequest:self.request];
if (cachedResponse.data.length > 0) {
self.responseData = cachedResponse.data;
self.response = cachedResponse.response;
self.error = nil;
}
}
}
Can't tell much without your HTTP headers -- but the most common reason for this is NSURLProtocol forcing revalidation before delivering cached response to WebView.
Please take a look here :
http://robnapier.net/blog/offline-uiwebview-nsurlprotocol-588
It sounds like you want the request to succeed, even though the cache says the data has expired and should be retrieved from the server. You may have some luck setting the cache policy (different policy for online vs. offline) of certain requests where you'd rather use stale data than fail.
NSMutableURLRequest -> setCachePolicy
It looks like NSURLRequestReturnCacheDataDontLoad is the policy you want for offline mode.
Hope that helps!
Related
I am using restkit for sending request, response mapping and all. but sometimes, i am sending same request multiple time which comes from different screens. So it get added to request queue. But i don't want to add request into request queue if the request is already there. How can i check that.
I am calling function like this
//Here I need to check if the following url is in RKRequestQueue. If it not there then call the below method,
[self getDataFromServer];
- (void)getDataFromServer{
RKObjectManager *manager = [[RestKit sharedDataManager] objectManager];
[manager loadObjectsAtResourcePath:#"/getData" usingBlock:^(RKObjectLoader *loader) {
[RKClient sharedClient].requestQueue.showsNetworkActivityIndicatorWhenBusy = YES;
loader.method = RKRequestMethodPOST;
loader.params = inputData;
loader.onDidFailWithError = ^(NSError *error) {
};
loader.onDidLoadObjects = ^(NSArray *objects) {
};
}
Any help is appreciated.
Interesting question, I was pretty sure that you can actually check the URLs of your requests in the RKRequestQueue, but I wasn't able to find anything useful for that in the RKRequestQueue reference. Instead, I've found the containsRequest: method, but this will only compare RKRequest objects, not the actual URLs.
I guess the simplest thing to do is to create some kind of proxy for managing your network activity, implement the requestQueue:didSendRequest: delegate method and monitor which URLs are currently processed.
I am trying to set up a NSURLRequest to download a simple index.html with its externa style.css sheet but I am not quite sure how to do this.. I have only ever just formatted the URL of the request to the file I want.. but this has to be slightly different and I cannot find a good example of what I am trying to do.
this is my code so far:
#pragma mark - NSURLConnection methods
- (void)htmlRequest
{
// Create the request.
NSURLRequest *theRequest=[NSURLRequest requestWithURL:[NSURL URLWithString:#"http://www.mywebsite.com/index.html"]
cachePolicy:NSURLRequestReloadIgnoringCacheData
timeoutInterval:60.0];
// create the connection with the request
// and start loading the data
NSURLConnection *theConnection=[[NSURLConnection alloc] initWithRequest:theRequest delegate:self];
if (theConnection) {
// Create the NSMutableData to hold the received data.
// receivedData is an instance variable declared elsewhere.
receivedData = [NSMutableData data];
} else {
// Inform the user that the connection failed.
NSLog(#"Connection Fail");
}
}
- (void)connection:(NSURLConnection *)connection didReceiveResponse:(NSURLResponse *)response
{
// This method is called when the server has determined that it
// has enough information to create the NSURLResponse.
// It can be called multiple times, for example in the case of a
// redirect, so each time we reset the data.
// receivedData is an instance variable declared elsewhere.
[receivedData setLength:0];
}
- (void)connection:(NSURLConnection *)connection didReceiveData:(NSData *)data
{
// Append the new data to receivedData.
// receivedData is an instance variable declared elsewhere.
[receivedData appendData:data];
}
- (void)connection:(NSURLConnection *)connection didFailWithError:(NSError *)error
{
// inform the developer of error type
}
// This method uses methodName to determin which Initalizer method to send the response data to in EngineResponses.m
- (void)connectionDidFinishLoading:(NSURLConnection *)connection
{
// EngineResponses *engineResponses = [EngineResponses sharedManager];
// [engineResponses GetManufacturers:receivedData];
NSString *myString = [[NSString alloc] initWithData:receivedData encoding:NSUTF8StringEncoding];
NSLog(#"%#", myString);
}
as you can see I am just calling index.html directly.. I would like to know how to format my request so i get the index.html as well as style.css
any help would be greatly appreciated.
I always create a new data structure,which has a -connection property and a -request property,like this
#interface connectionWrapper : NSObject
#property(retain) NSURLRequest *request
#property(retain) NSURLConnection *connection
by retaining this data structure in an mutable array, you can distinguish the connections in callback methods by iterate the array and compare each connectionWrapper instance's -connection property with the connection parameter the of the callback method, if they match(points to a same object), then you can retrieve the -request property of the connectionWrapper instance, then -url property of NSURLRequest instance.
as I'm not an native English speaker, I think code is a better tutor.
-(NSURLRequest*)getRequestByConnection:(NSURLConnection*)connection
{
for(connectionWrapper *w in theArrayContainingAllConnectionWrappers)
{
if(w == connection)
return w.request;
}
}
In callback method:
-(void)connection:(NSURLConnection*)connection didReceiveResponse(NSURLResponse*)response
{
NSURLRequest *request = [self getRequestByConnection:connection];
NSURL *url = [request url];
/*apply different approach to different url/*
}
PS:it's very sad that NSURLConnection don't have a -request property so that we can retrieve the request associated with the connection easily.
One way or another, you will have to make 2 requests. Even if you open a web page directly in a web browser, the browser will make a separate request for the CSS file referenced in the HTML it downloads. If your application needs both the HTML and the CSS file, then you want it to make 2 separate URL requests, first to get the HTML and then to get the CSS file.
Now, just because 2 requests need to be made, that doesn't mean you will always need to write the code that makes those 2 requests. It may be that libraries like the ones recommended by #Slee automatically take the results of a first request, parse them out, and make requests for any referenced CSS files. I have not worked with them so I am not sure what they support, or if any libraries will do this for you.
One thing you may want to consider is loading the HTML and CSS through a UIWebView rather than handling it all manually. UIWebView will attempt to load, parse, and render an HTML file into a UI component. In the process it will load referenced CSS and JavaScript files and apply them to its rendering. If you want to do anything special like intercept the calls it makes to load the CSS file(s), you can implement the UIWebViewDelegate protocol and set the delegate of the the UIWebView. Within that delegate you can implement the -webView:shouldStartLoadWithRequest:navigationType: method to be notified when the web view is loading the CSS file. You can use the call to that method to look at the request that is being issued for the CSS and do something else interesting with the request.
do you know the name of the .css file?
If so I would just make 2 requests otherwise you will have to write a parser to look for the link to the css and make a second request anyways.
I'd also suggest looking into a library to handle the downlading of stuff - lot's of great libraries that can do the heavy lifting for you with advanced features.
Here's 3 I have used:
http://blog.mugunthkumar.com/coding/ios-tutorial-advanced-networking-with-mknetworkkit/
https://github.com/tonymillion/TMHTTPRequest
https://github.com/pokeb/asi-http-request
I've been using NSURLConnection to do a HTTP post to establish the connection. I've also implemented the didReceiveData delegate to process incoming bytes as they become available.
As incoming data comes in via didReceiveData, I add the NSData to a data buffer and try parsing the bytesteam if enough data has come in to complete a message segment. I'm having a hard time managing the data buffer (NSMutableData object) to remove bytes that have been parsed to structs. Was curious if there's an easier way. My didReceiveData delegate is below.
It works, but I don't think I'm managing memory correctly after I copy the message segment (currMsg) out of the responseData buffer and call processMsg. I get double free errors when running under the Simulator -- the program doesn't crash.
NSMutableData/NSData provide methods for appending bytes to the end but I didn't see any methods for removing bytes from the beginning (bytes representing whats already been parsed. I would appreciate some advice on how to best remove the parsed bytes from the responseData buffer. I come from a mostly C background so I'm not sure if there are better ways of manipulating the NSData bytes pointer. I'd like to avoid copying if possible -- just want to process a portion of the responseData buffer and leave the rest in responseData for next time enough bytes are in it for parsing.
Thanks
- (void)connection:(NSURLConnection *)connection didReceiveData:(NSData *)data {
NSData *tmpBuffer = nil;
NSInteger currMsgSize = 10;
[responseData appendData:data];
NSInteger rspDataLen = [responseData length];
while(rspDataLen >= 10) {
currMsg = [[NSData alloc] initWithBytesNoCopy:(void *)[responseData bytes] length:currMsgSize];
[self processMsg:currMsg];
[currMsg release];
[responseData getBytes:tmpBuffer range:NSMakeRange(currMsgSize, rspDataLen - currMsgSize)];
[responseData release];
responseData = [[NSMutableData alloc] initWithBytesNoCopy:(void *)tmpBuffer length:rspDataLen - currMsgSize];
rspDataLen = rspDataLen - currMsgSize;
}
}
Where do you allocate the first responseData?
What is [self processMsg:currMsg] doing with the data? If it is expecting the data to be around after -processMsg: returns, and it isn't explicitly making a copy, then you are in trouble.
Infact, unless you have finished with the received data before didReceiveData: returns, you need to make a copy of it somewhere, which isn't visible in the code shown.
You need to allocate the storage for tempBuffer, not pass in an uninitialised pointer;
You should look probably for a pre-rolled implementation of a simple ring buffer. There are plenty around.
I'm trying to download several images in response to a single http request. On the server side (java) I'm using oreilly multipart response and I'm getting my datas in my iPhone Simulator in didReceiveData (approximately one call for each image) after a call to didReceiveResponse (approximately one call for each image as well) in my delegate.
The problem is this approximately... Has anyone ever managed to handle correctly multipart/x-mixed-re with iPhone SDK ? If yes what is the best strategy here ? Should I play with the expected length ? on server side ? on client side ? should I wait until I've received everything... mmmh that doesn't even seen enough as the calls to didReceiveData happens in a random order (I'm asking picture1,picture2 and I'm sometimes receiving picture2,picture1 even though the order is respected on server side !). Should i temporize between pictures on server side ?
Or should I drop multipart/x-mixed-replace ? what would be the easiest then ?
That's a lot of questions but I'm really stuck here ! Thanks for you help !
I'm not sure what your final use for the images is, but the intended purpose of the multipart/x-midex-replace content type is for each received part to completely replace the previously received responses. Think of it like frames of a video; only one picture is displayed at a time and the previous ones are discarded.
Temporizing is almost never a foolproof solution. Especially on the iPhone you're going to encounter an unimaginable variety of network situations and relying on a magic number delay between frames will probably still fail some of the time.
Since you have control of the server, I'd recommend dropping the multipart. Make sure when you are sending multiple requests to the server that you don't block the main thread of your iPhone app. Use NSOperations or an alternative HTTP library (like ASIHTTPRequest) to make your image fetch operations asynchronous.
I did that successfully using this code. The important thing is to create 2 buffers to receive your data. If you use only one you will have some double access problems (stream access and jpg CODEC access) and corrupted JPG data.
Do not hesitate to ask me for more details.
- (IBAction)startDowload:(id)sender {
NSURL *url = [NSURL URLWithString:#"http://210.236.173.198/axis-cgi/mjpg/video.cgi?resolution=320x240&fps=5"];
NSMutableURLRequest *req = [NSMutableURLRequest requestWithURL:url];
[req setHTTPMethod:#"GET"];
/*
I create 2 NSMutableData buffers. This points is very important.
I swap them each frame.
*/
receivedData = [[NSMutableData data] retain];
receivedData2 = [[NSMutableData data] retain];
currentData = receivedData;
urlCon = [[NSURLConnection alloc] initWithRequest:req delegate:self];
noImg = YES;
frame = 0;
}
- (void)connection:(NSURLConnection *)connection didReceiveResponse:(NSHTTPURLResponse *)response
{
// this method is called when the server has determined that it
// has enough information to create the NSURLResponse
// it can be called multiple times, for example in the case of a
// redirect, so each time we reset the data.
// receivedData is declared as a method instance elsewhere
UIImage *_i;
#try
{
_i = [UIImage imageWithData:currentData];
}
#catch (NSException * e)
{
NSLog(#"%#",[e description]);
}
#finally
{
}
CGSize _s = [_i size];
[imgView setImage:_i];
[imgView setNeedsDisplay];
[[self view] setNeedsDisplay];
}
/*
Buffers swap
*/
if (currentData == receivedData)
{
currentData = receivedData2;
}
else
{
currentData = receivedData;
}
[currendData setLength:0];
}
- (void)connection:(NSURLConnection *)connection didReceiveData:(NSData *)data
{
// append the new data to the currentData (NSData buffer)
[currendData appendData:data];
}
Our iPhone app code currently uses NSURLConnection sendSynchronousRequest and that works fine except we need more visibility into the connection progress and caching so we're moving to an async NSURLConnection.
What's the simplest way to wait for the async code to complete? Wrap it in a NSOperation/NSOperationQueue, performSelector..., or what?
Thanks.
I'm answering this in case anyone else bumps into the issue in the future. Apple's URLCache sample code is a fine example of how this is done. You can find it at:
iOS Developer Library - URLCache
As John points out in the comment above - don't block/wait - notify.
To use NSURLConnection asynchronously you supply a delegate when you init it in initWithRequest:delegate:. The delegate should implement the NSURLConnection delegate methods. NSURLConnection processing takes place on another thread but the delegate methods are called on the thread that started the asynchronous load operation for the associated NSURLConnection object.
Apart from notifications mentioned prior, a common approach is to have the class that needs to know about the URL load finishing set itself as a delegate of the class that's handling the URL callbacks. Then when the URL load is finished the delegate is called and told the load has completed.
Indeed, if you blocked the thread the connection would never go anywhere since it works on the same thread (yes, even if you are using the asynch methods).
I ran into this because our app used NSURLConnection sendSynchronousRequest in quite a few places where it made sense, like having some processing occurring on a background thread occasionally needing extra data to complete the processing. Something like this:
// do some processing
NSData * data = someCachedData;
if (data = nil) {
data = [NSURLConnection sendSynchronousRequest....]
someCachedData = data;
}
// Use data for further processing
If you have something like 3 different places in the same flow that do that, breaking it up into separate functions might not be desirable(or simply not doable if you have a large enough code base).
At some point, we needed to have a delegate for our connections(to do SSL certificate pinning) and I went trolling the internet for solutions and everything was of the form: "just use async and don't fight the framework!". Well, sendSynchronousRequest exists for a reason, this is how to reproduce it with an underlying async connection:
+ (NSData *)sendSynchronousRequest:(NSURLRequest *)request returningResponse:(NSURLResponse *__autoreleasing *)response error:(NSError *__autoreleasing *)error
{
static NSOperationQueue * requestsQueue;
static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^{
requestsQueue = [[NSOperationQueue alloc] init];
requestsQueue.maxConcurrentOperationCount = NSOperationQueueDefaultMaxConcurrentOperationCount;
});
NSCondition * waitLock = [NSCondition new];
[waitLock lock];
__block NSError * returnedError;
__block NSURLResponse * returnedResponse;
__block NSData * returnedData;
__block BOOL done = NO;
[NSURLConnection sendAsynchronousRequest:request
queue:requestsQueue
completionHandler:^(NSURLResponse * response, NSData * data, NSError * connectionError){
returnedError = connectionError;
returnedResponse = response;
returnedData = data;
[waitLock lock];
done = YES;
[waitLock signal];
[waitLock unlock];
}];
if (!done) {
[waitLock wait];
}
[waitLock unlock];
*response = returnedResponse;
*error = returnedError;
return returnedData;
}
Posted here in case anyone comes looking as I did.
Note that NSURLConnection sendAsynchrounousRequest can be replaced by whatever way you use to send an async request, like creating an NSURLConnection object with a delegate or something.