I had some code to post to a friend's facebook wall that upon the deprecation of this via Open Graph I changed to use the deprecated headers and Facebook dialog to include the user in the posting. That has been working for weeks, but seems to have stopped working (worked fine Tuesday, failed Wednesday) in production application (i.e. no code change or re-build.)
I've spent hours trying all sort of things (running on main thread, asserting sessions are live, re-writting to SDK 3.2, uninstalling app/re-installing, completely changing the dialog type and post contents) to restore the ability but whatever I do the dialog immediately dismisses. [Note: There are no developer alerts on our app, and I believe that none of the app settings have changed.]
Here is the code. The facebookManager "performConnectedAction" is from the Facebook sample to ensure an active session (and I assert it w/o problem.) :
NSMutableDictionary *wallPost = [NSMutableDictionary dictionary];
// Content populating not show (and I've tried various simplifications)
// Who to...
[wallPost setObject:_selectedUser.id forKey:kFB_FIELD_TO];
[facebookManager performConnectedAction:^{
FBSession *facebookSession = facebookManager.facebookSession;
Assert( facebookSession.isOpen, #"Need an open Facebook session.");
[FBWebDialogs presentFeedDialogModallyWithSession:facebookSession parameters:wallPost handler:^(FBWebDialogResult result, NSURL *resultURL,NSError *error) {
if ( !!error ) {
[self logEventWithFormat:#"Failed to invite on Facebook [%#]", error];
}
}];
}];
Basically ... not only does the dialog flash up/down, but no callback is given (no error or result provided.) All other parts of the Facebook application continue to operate (our graph calls, our Facebook Friends dialog, and so on.)
I'd appreciate any thoughts on ways to investigate this further. I have filed a bug report:
http://developers.facebook.com/bugs/165456656937602?browse=search_5130e2d7663dd6543665811
I had the same issue and fixed it by delaying the method call like this. My method is fbWallPost
[self performSelector:#selector(fbWallPost) withObject:nil afterDelay:0.5];
Related
Background story: I need to obtain the handles of the tagged Twitter users from an attached Twitter media. There's no current API method to do that unfortunately (see https://twittercommunity.com/t/how-to-get-tags-of-a-media-in-a-tweet/185614 and https://github.com/twitterdev/open-evolution/issues/34).
I have no other choice but to scrape, this is an example URL: https://twitter.com/justinwood_/status/1626275168157851650/media_tags. This is the page which pops up when you click on the tags link under the media of the parent Tweet: https://twitter.com/justinwood_/status/1626275168157851650/
The React generated DOM is deep and ugly, but would be scrapeable, however I do not want to log in with any account to get banned. Unfortunately when you visit https://twitter.com/justinwood_/status/1626275168157851650/media_tags in an Incognito window the popup shows up dead empty. However when I dig into the network requests the /TweetDetail GraphQL endpoint is full of messages about the anonymous page visit, fortunately it still contains the list of handles I need despite of all of this.
So what I need to have is a scraper which is able to process JavaScript, and capture the response for that specific GraphQL call. Selenium uses a headless Chrome under the hood, so it is able to process JavaScript, and Selenium-Wire offers the ability to capture the response.
Unfortunately my crafted Selenium-Wire script only has the TweetResultByRestId and UsersByRestId GraphQL requests but is missing the TweetDetail. I don't know what to tweak to make all the requests to happen. I iterated over a ton of Chrome options. Here is a variation of my script:
from seleniumwire import webdriver
from selenium.webdriver.chrome.service import Service
chrome_options = webdriver.ChromeOptions()
chrome_options.add_argument("--disable-extensions")
chrome_options.add_argument("--disable-gpu")
chrome_options.add_argument("--no-sandbox")
chrome_options.add_argument("--headless") # for Jenkins
chrome_options.add_argument("--disable-dev-shm-usage") # Jenkins
chrome_options.add_argument('--start-maximized')
chrome_options.add_argument('--window-size=1900,1080')
chrome_options.add_argument('--ignore-certificate-errors-spki-list')
chrome_options.add_argument('--ignore-ssl-errors')
selenium_options = {
'request_storage_base_dir': '/tmp', # Use /tmp to store captured data
'exclude_hosts': ''
}
ser = Service('/usr/bin/chromedriver')
ser.service_args=["--verbose", "--log-path=test.log"]
driver = webdriver.Chrome(service=ser, options=chrome_options, seleniumwire_options=selenium_options)
tweet_id = "1626275168157851650"
twitter_media_url = f"https://twitter.com/justinwood_/status/{tweet_id}/media_tags"
driver.get(twitter_media_url)
driver.wait_for_request("/TweetDetail", timeout=10)
Any ideas?
Apparently it looks like I'd rather need to scrape the parent Tweet URL https://twitter.com/justinwood_/status/1626275168157851650/ and right now it seems my craved GraphQL call happens. Probably I got confused while trying 100 combinations.
I created a webview (BrowserComponent) and added a listener:
BComp.addWebEventListener(BrowserComponent.onMessage, e->{
Display.getInstance().callSerially(()->{
Dialog.show("Message", (String)e.getSource(), "OK", null);
});
});
Then, in the JavaScript of the embedded website, I called cn1PostMessage and postMesage. This works well in the simulator! But when building the application, on Android, it does nothing (cn1PostMessage is undefined and postMessage is not received by the main program).
var msg = "test";
if(window.cn1PostMessage) {
window.cn1PostMessage(msg);
} else {
window.parent.postMessage(msg, '*');
// Tried: window.postMessage(msg, '*');
// Tried: window.top.postMessage(msg, '*');
}
What can I do?
Thanks!
I just tried building the BrowserComponentPostMessageSample, and it seemed to work fine on my Galaxy S8, Android 8.
If you're finding that window.cn1PostMessage is undefined, then either there was a javascript error in page load that prevented that callback from being installed, or your code is running before the CodenameOne initialization code has run.
I have just added support for the "android.webContentsDebuggingEnabled" display property, which will make the app's web views debuggable using Chrome's remote development tools. This will make it easier for your to track down such issues. See usage example:
Display.getInstance().setProperty("android.webContentsDebuggingEnabled", "true");
This will be available in Friday's update (Dec. 6/19).
In the mean time, I recommend starting with the BrowserComponentPostMessageSample and modifying it to suit your needs from there.
Alternatively, if you can post a test case to demonstrate the issue, I can look at it.
I'm implementing Offline Playback with HLS Fairplay following the demo in the FairPlay Streaming Server SDK v4.0.1 that uses AVContentSessionKey.
I download three contents, each content is downloaded and persisted correctly, both the .movpkg and its content key on the documents directory, when I turn off the WIFI these three contents downloaded plays correctly without any problems, before playing Im using this code:
let urlAsset = element.urlAsset!
ContentKeyManager.shared.contentKeySession.addContentKeyRecipient(urlAsset)
if !urlAsset.resourceLoader.preloadsEligibleContentKeys {
urlAsset.resourceLoader.preloadsEligibleContentKeys = true
}
self.present(playerViewController, animated: true, completion: {
AssetPlaybackManager.sharedManager.setAssetForPlayback(urlAsset)
})
So far so good. But the problem is when I close the application (Home button to close applications) and then play the downloaded contents, only the last content downloaded plays correctly, the other ones (first and second) send these error on the console.
Error Domain=AVFoundationErrorDomain Code=-11800 "The operation could not be completed"
UserInfo={NSUnderlyingError=0x1c065d760 {Error Domain=NSOSStatusErrorDomain Code=-16227 "(null)"},
NSLocalizedFailureReason=An unknown error occurred (-16227),
NSURL=file:///private/var/mobile/Containers/Data/Application/A950D8DB-B711-47E3-AAF5-C95CC9682430/Library/com.apple.UserManagedAssets.kkG8Ih/644986_7798B8476A473F68.movpkg/, NSLocalizedDescription=The operation could not be completed}
I double check the .movpkg with the keys in the documents directory and appears correctly
/Documents/.keys/one-key
/Documents/.keys/two-key
/Documents/.keys/three-key
Before the error occurs the ContentKeyDelegate is called and the key is loaded and passed to the request correctly.
if persistableContentKeyExistsOnDisk(withContentKeyIdentifier: assetIDString) {
let urlToPersistableKey = urlForPersistableContentKey(withContentKeyIdentifier: assetIDString)
guard let contentKey = FileManager.default.contents(atPath: urlToPersistableKey.path) else {
/
pendingPersistableContentKeyIdentifiers.remove(assetIDString)
return
}
/
Create an AVContentKeyResponse from the persistent key data to use for requesting a key for
decrypting content.
*/
let keyResponse = AVContentKeyResponse(fairPlayStreamingKeyResponseData: contentKey)
/
keyRequest.processContentKeyResponse(keyResponse)
return
}
If I print the contentKeyRecipients the three contents appears correctly
- (lldb) po
ContentKeyManager.shared.contentKeySession.contentKeyRecipients ▿ 3
elements
- 0 : AVURLAsset: 0x1c0234d40, URL = file:///private/var/mobile/Containers/Data/Application/E791A4DE-4261-46B7-A84D-D10B27035FAE/Library/com.apple.UserManagedAssets.kkG8Ih/539628_20469336224AA388.movpkg
- 1 : AVURLAsset: 0x1c0234fa0, URL = file:///private/var/mobile/Containers/Data/Application/E791A4DE-4261-46B7-A84D-D10B27035FAE/Library/com.apple.UserManagedAssets.kkG8Ih/644986_7798B8476A473F68.movpkg
- 2 : AVURLAsset: 0x1c42391c0, URL = file:///private/var/mobile/Containers/Data/Application/E791A4DE-4261-46B7-A84D-D10B27035FAE/Library/com.apple.UserManagedAssets.kkG8Ih/573744_62377F9549C45B93.movpkg
My tests are in iOS 11.1.2 and iOS 11.2 beta 2
I'm not sure what is happening, but seems to be a problem with the persisted key, I don't how if each content needs to be associated with one AVContentKeySession at time.
If someone faced a similar problem, any help would be appreciated.
Thanks in advance
I'm having similar issue.
however, since I need to support iOS 10, I'm not using the new AVContentKeyResponse class. Instead, I'm loading the persistent content key myself, and pass it to the loading request.
Anyway, I'm getting exact the same error as you and same behavior. One thing to note is that if I remove the code that loads persistent content key from disk, and always fetch the key from server, then everything works. But this defeats the purpose of "offline" playback...
So it seems like the system thinks the persistent content key is invalid...
Which TLLV you used on the server side to specify Rental Duration of the downloaded content? Did you use Content key duration TLLV or Offline Key TLLV? If you used Offline Key TLLV you need to double check that "Content ID" field is different for every downloaded movie.
We had encounter this error message, too.
It will happen when content exceed over expiration date which set in server side.
For example :
We give 10 minutes of expiration date for Video A
Download this Video A, and verify CKC delivery correctly (print log)
Play Video A without connection
Take a break (after 11 minutes), close App, and launch App again, select Video A to Play
Show below error message from AVPlayerItem.error.description:
Error Domain=AVFoundationErrorDomain Code=-11800
"The operation could not be completed" UserInfo={NSLocalizedFailureReason=An unknown error occurred (-16227),
NSLocalizedDescription=The operation could not be completed, NSUnderlyingError=0x1d4257310
{Error Domain=NSOSStatusErrorDomain Code=-16227 "(null)"}}
You can refresh encrypted data again by
AVAssetResourceLoaderDelegate
or use AVContentSessionKey
Reference : https://developer.apple.com/videos/play/wwdc2018/507/
Make sure you set correct offline content identifier on a serverside. The identifier you set should be associated with the specific rendition/stream allowed by the license. This helped me.
I am trying to use FBPlacePickerViewController and it doesn't seem to load any data.
Here is my code:
FBPlacePickerViewController *picker = (FBPlacePickerViewController*)segue.destinationViewController;
picker.navigationController.navigationBarHidden = YES;
picker.delegate = self;
picker.radiusInMeters = 1000;
picker.resultsLimit = 30;
if([TonerAppDelegate instance].lastLocation != nil){
picker.locationCoordinate = [TonerAppDelegate instance].lastLocation.coordinate;
[picker loadData];
}
[TonerAppDelegate instance].lastLocationUpdateFunction = ^{
picker.locationCoordinate = [TonerAppDelegate instance].lastLocation.coordinate;
[picker loadData];
};
It is an embed segue (iOS 6). I verify that the picker is a valid object. The [picker loadData] method does get called, and the coordinate data is perfectly valid. I am not getting any exceptions or warnings. I've allowed my app to access to my location in iOS and I double-verified that in Settings. My iPod is connected to the Internet and the connection works perfectly. All the other apps can use location services without any problem. So, there probably is a problem with my implementation of the place picker. I've also implemented the -(void)placePickerViewControllerDataDidChange:(FBPlacePickerViewController *)placePicker and -(BOOL)placePickerViewController:(FBPlacePickerViewController *)placePicker shouldIncludePlace:(id<FBGraphPlace>) methods of the delegate, and they aren't getting called either. What am I doing wrong?
Thanks,
Can.
Found the answer: I wasn't creating the Facebook session before displaying the place picker. I totally forgot about the session. It'd be nice to see Facebook add an assertion check in loadData method of the picker for an existing Facebook session. I've created the session, and THEN tried my code, and it works perfectly now.
I am trying to set a maintenance page so that when the site is disabled, it should appear no matter what page was requested.
I currently tried doing this with $this->cakeError():
in app_controller.php:
function beforeFilter(){
....
if($this->__get_config('maintenance_status') == 1){
$this->cakeError('maintenance', array('message' => $this->__get_config('maintenance_message')));
}
....
}
and in app_error.php:
function maintenance($message){
$this->controller->set('message', $message['message']);
($this->controller->RequestHandler->isAjax()) ? $this->_outputMessage('ajax_maintenance') : $this->_outputMessage('maintenance');
}
The problem is that a Fatal Error occurs, which says: Call to a member function isAjax() on a non-object. But I have obviously set the RequestHandler Component in app_controller.php. Moreover, I have tried calling this error from within another controller and it doesn't give me any Fatal Error.
What could be the problem? Why doesn't it recognize that I have initalized the Component?
From the CakePHP book:
Calling this method will show an error page to the user and halt any further processing in your application
I am assuming that you're calling the error in some callback in AppController.
If that is the case you may very likely be halting execution of your script before your components are instantiated. This would certainly cause your error.
Now, I think this error is a good chance to reevaluate how you're dealing with the problem. Is this really an error? You know the maintenance status is set so it's expected that the user be shown this page. It isn't an error. Furthermore, you certainly wouldn't want 10,000 messages in your log telling you that you turned maintenance on!
I think this could be better solved by utilizing some controller callbacks and a little bit of code.
I don't know what _get_config() is so I assume it is a custom user function that you can call in this callback.
We'll be using the beforeFilter() controller callback.
class AppController extends Controller {
public function beforeFilter() {
if ($this->_get_config('maintenance_status') === 1) {
$this->redirect('/maintenance');
}
}
}
Now, you can just setup a maintenance controller, attached to its own view, that will properly show your maintenance message, and won't log all those connection attempts during maintenance in your error log.
Slightly better would also be to use the Configure::read( "System.maintenance" ) or similar. (I tend to namespace my config data, System being the namespace for stuff like maintenance flags etc.)
Also, as Charles said - don't use an error page for an expected event. Errors are to show the user, and for the application to handle notifications etc, about unexpected failures. The maintenance page could simply be a view file in the /app/views/pages/ folder. Redirect to that if the config key is set to true/1.
Your approach seems to be intelligent, but you might be overdoing it a little.
I have a similar setup in a site I am currently developing and I simply use the auth component to take care of it for me.
To help out, I setup a new offline layout that I force the application to use if status of the site is 0 (offline). If the status is 0 app_controller denies access to the entire site.
$this->Auth->deny('*');
$this->layout = "offline";
Also, in this layout I have a hidden login form that appears if the user clicks the message.If user is able to authenticate (all users for now - development) access is granted to the entire site using the default template.
Check it out, it might help you out...
Click Here
Some of the code, but you can read more about it in the link above
function beforeFilter(){
// Site Offline = 0 , Site Online = 1
if($this->Configuration->get_site_status() == 1){
// Allow access to the site to all users and perform all required
// beforeFilter code
}else{
...
// If site is OFFLINE but User is logged in allow access.
// Later I will need to change it to only allow admin access if logged in as I am still developing
// Everyone else will be denied access even if they are able to authenticate
if(!$this->Auth->user() == null){
$this->layout = 'default';
$this->Auth->allow('*');
}else{
$this->layout = 'offline';
$this->Auth->deny('*');
}
...
}
}