Update the mode of the barcode detector only if it is a real detector.
Otherwise, the app will crash due to '-[BarcodeDetectorManagerMlkit setMode:queue:]: unrecognized selector sent to instance'.
Co-authored-by: René Fischer <rene.fischer@zollsoft.de>
* add TouchEvent on android
* add GestureHandler to detect touches on android
* add property to enable touchDetector on android
* add onTouch event to ios
* add GestureRecognizer to ios
* add onTouch property and js setup
* add missing semicolons
* fix literal notation
* add missing ":"
* fix copy-paste error (wrong var-name)
* pass the native event to the onTouch callback
* replace : with ;
* add onTouch type defs
* add documentation for onTouch property
* scale postion before emitting since the event coordinates are raw pixels
* migrate advanced example to native pinch zoom and onTouch
* split onTouch property into onTap and onDoubleTap
* add native zoom on android
* add native zoom on ios
* add useNativeZoom property
* extract compution of maxZoomFactor to a method
* use optional maxZoomFactor for native zoom
* add useNativeZoom Property to typescript declaration
* add doc entry for useNativeZoom
* This update tries to improve audio recording interruptions on iOS due to phone calls or background music.
- Use a more generic event to handle session interruptions. This removes the need to listen to foreground/background events, and stopping the session this way was actually redundant/wrong (see https://forums.developer.apple.com/thread/61406). This also makes session stopping detection more reliable (calls, suspension due to a call or notification, etc., which would previously not set the recording interrupted flag on every case)
From the above docs: "No, incorrect. You _never_ need to stop your capture session. The capture session automatically stops itself when your app goes to the background and resumes itself when you come back to the foreground."
- Allow for `captureAudio` updates to also update the audio connections internally so the prop can be correctly updated on the fly without remounting.
- add onAudioInterrupted and onAudioConnected events so the UI can handle scenarios where audio is wanted but not available. This should also help in keeping the preview active even if audio is interrupted and we have captureAudio={true}. Lastly, it can be used to detect if we can record audio or not due to the dummy implementation of the audio permission on iOS always returning true.
- check, activate, and release audio sessions (if captureAudio) so we can detect early if audio is available before attempting to connect the input. This will also allow us to detect if we can record even if there was already a call before opening the camera.
- use proper observer for session error instead of of the strong self block. No benefit, but makes code more readable and allows access to instance variables
- getDeviceOrientationWithBlock might fire more than once under some circumstances, ending up taking a picture or video twice. Add a lock and additional check to prevent this.
* no need for change check,
* do not resume audio if we were hinted not to (e.g., music playback happening)
* Move heavy work to a dedicated background thread. Improves camera initial loading and resumes from background.
Details:
- Use a HandlerThread to delegate heavy tasks to background. The thread is managed by the view, and passed down to the implementation in case it also needs to use it. The view will fire start calls and other possibly heavy operations in this thread to avoid ANRs. Some code sent to this thread:
- start calls: start is extremely heavy and will cause ANRs on some devices, especially when coming back from background
- Camera1: some preset changes fire a stop/start sequence. These will now happen in the background thread
- take picture and start recording (from view class) will also start in this thread
- Add some extra null checks
- View was not properly cleaning up itself on destroy (host destroy event was never fired)
* Fix for a possible crash when changing devices and changing focus. If the new device resets the focus, "defocusing" might not be possible if the new device does not support auto focus. For this reason, we need to do a different cleanup on the focus and exposed flags and events.
* start session here also on session queue.
* check for session running before trying to record or capture.
This should fix a possible race condition where both the session start call happens at the same time as the record call
* set preview orientation also in session queue
* no need to set orientation on constructor, and set it on session queue to prevent race conditions
* move device init and checks also to session queue. This prevents possible double initializations.
* catch possible errors when starting camera preview. This might still randomly fail on some devices for some reason.
* delay capture in progress until we have resumed/paused preview.
* do not crash the app if set texture setup failed
* more synchronized checks to prevent crashes due to concurrent camera updates
* remove unused imports
* Handle audio interruption in session queue. This prevents the session isRunning flag from getting corrupted due to concurrent updates to the session.
* Fix possible crash when attempting to retrieve camera parameters.
* Preserve exif/metadata on photo capture. Add a few comments to each step of the capture process
* orientation must be fixed before mirroring
* x/y dimensions are redundantly updated (storing the image sets them automatically). However, orientation must be reset on any image change since the final stored image is automatically rotated when it is modified in place.
* revert mirrorImage order, has to be first since forceUpOrientation adjusts the image afterwards.
* Minor change: also implement `onPictureTaken` for iOS in case anyone needs the early event.
* - Improve Android code so skipProcessing is not needed, the code is more in line with iOS, and is "fast" by default. This means that skipProcessing is no longer needed (nor used), and adding additional options will "slow down" the capture as expected, rather than having always a lot of processing. This shouldn't be a breaking change.
- document the writeExif option, and implement it for iOS as well.
* This is a fix for events possibly being fired not on the main thread, which could cause the whole app to freeze.
* Fire events on the right thread. No need to use UI thread, and fix the still works
* Release CF object which could cause a memleak
* Fix Objects.equals that is only available after API 19
* allow for audio session to be kept even after unmounts
* readme typo
* readme typo
* Make camera ready events to fire also on camera/device change to be consistent with Android. Fire unmount error when session or device fails to start.
* update example app to properly use camera ready event
* Make camera ready events to fire also on camera/device change to be consistent with Android. Fire unmount error when session or device fails to start. Update advanced example app to use camera ready event instead.
* Android crash fixes. Make sure no unsupported aspect ratio is used, and do not crash when there are no cameras available.
* stop/release camera in non-UI thread so we prevent ANRs and UI freezing.
Some phones may take up to a second to release the camera and preview.
* move codec, max duration, and max file size settings to the session queue.
This might prevent a race condition when changing presets/quality.
* android crash fix
* Add onRecordingStart and onRecordingEnd events
* fix for surface destroy and resume events.
* add missing types
* fix for surface destroy and resume events.
* add missing types
Co-authored-by: Cristiano Coelho <cristianocca@hotmail.com>
* - Improve Android code so skipProcessing is not needed, the code is more in line with iOS, and is "fast" by default. This means that skipProcessing is no longer needed (nor used), and adding additional options will "slow down" the capture as expected, rather than having always a lot of processing. This shouldn't be a breaking change.
- document the writeExif option, and implement it for iOS as well.
* Release CF object which could cause a memleak
* Preserve exif/metadata on photo capture. Add a few comments to each step of the capture process
* orientation must be fixed before mirroring
* x/y dimensions are redundantly updated (storing the image sets them automatically). However, orientation must be reset on any image change since the final stored image is automatically rotated when it is modified in place.
* revert mirrorImage order, has to be first since forceUpOrientation adjusts the image afterwards.
* Minor change: also implement `onPictureTaken` for iOS in case anyone needs the early event.
* This update tries to improve audio recording interruptions on iOS due to phone calls or background music.
- Use a more generic event to handle session interruptions. This removes the need to listen to foreground/background events, and stopping the session this way was actually redundant/wrong (see https://forums.developer.apple.com/thread/61406). This also makes session stopping detection more reliable (calls, suspension due to a call or notification, etc., which would previously not set the recording interrupted flag on every case)
From the above docs: "No, incorrect. You _never_ need to stop your capture session. The capture session automatically stops itself when your app goes to the background and resumes itself when you come back to the foreground."
- Allow for `captureAudio` updates to also update the audio connections internally so the prop can be correctly updated on the fly without remounting.
- add onAudioInterrupted and onAudioConnected events so the UI can handle scenarios where audio is wanted but not available. This should also help in keeping the preview active even if audio is interrupted and we have captureAudio={true}. Lastly, it can be used to detect if we can record audio or not due to the dummy implementation of the audio permission on iOS always returning true.
- check, activate, and release audio sessions (if captureAudio) so we can detect early if audio is available before attempting to connect the input. This will also allow us to detect if we can record even if there was already a call before opening the camera.
- use proper observer for session error instead of of the strong self block. No benefit, but makes code more readable and allows access to instance variables
- getDeviceOrientationWithBlock might fire more than once under some circumstances, ending up taking a picture or video twice. Add a lock and additional check to prevent this.
* no need for change check,
* do not resume audio if we were hinted not to (e.g., music playback happening)
* start session here also on session queue.
* check for session running before trying to record or capture.
This should fix a possible race condition where both the session start call happens at the same time as the record call
* no need to set orientation on constructor, and set it on session queue to prevent race conditions
* move device init and checks also to session queue. This prevents possible double initializations.
* Torch fixes for iOS and a few nil checks.
- Torch turned on by default should work as expected (turned on after mounting and camera type switch)
- Torch maintained when video recording starts (it will flicked if preset has to change)
- Torch resumed when app comes back from background (it is disabled on session pause)
* Fix Torch mode while recording for Android as well.
* Use connection for muting instead of adding/removing outputs to reduce/remove flickering while recording.
* Add a tiny delay before recording starts to prevent the first frame of the video from being underexposed/black.
* feat: autoexposure on point of interest and defocus on subject change
* Impelement cameraIds for iOS, and various iOS bug fixes.
Summary:
- Implement getCameraIds and cameraId property to manually select a camera device.
- Fix the internal preset being used to properly use a photo and video preset, and check for invalid presets on camera switch.
- This fixes a bug that would happen after recording. After a recording is complete, the camera preset was set to 4k permanently, and attempting to select any camera that does not support that preset would result in an totally unusable camera. Now, the "Photo" preset is the default which should be the highest quality option for photos, and the previously used "High" preset will be used as video default if no value is provided.
- This also adds proper support to setting the undocumented "pictureSize" property. Setting this property will change the default photo quality to be that one.
- Minor cleanup to the camera initialization so the same code is not called as many times redundantly (start session / device happening multiple times)
- The reduced redundant calls to start session and session updates should make the camera startup slightly faster.
- Fix for warnings issued due to misuse of the "isRecording" property. This property was defined as both as a property and a method and was both getting set manually and retrieved with a function. It should now consistently only be retrieved with a function based on the real recording status. Should have no behaivour change, but removes a pesky warning.
- Fix an issue related to the camera getting stuck after a background resume by removing the use of the queue for those two events.
- Fixes a wrong event unsubscription (UIDeviceOrientationDidChangeNotification to UIApplicationDidChangeStatusBarOrientationNotification) from a previous change. Also move event subscriptions to superview changes for consistency and so they are not used/consumed if not needed. After testing, the View of RNCamera is instantiated twice (for some reason I couldn't find) but used only once. RN keeps a reference to the view in memory even after the camera is destroyed, resulting in these events getting fired and handled all the time during the app's lifetime.
* add maxZoom property to iOS.
Android: Not needed as of now. Also, zoom multiplier behaves different on Android and can't be used with the same value as iOS.
* add auto exposure option.
This branch is temporary from the other changes until the proper changes from https://github.com/react-native-community/react-native-camera/pull/2423/files are merged.
* allow for expose POI to also be cleared
* Add ultra-wide lense support. Requires XCode with SDK for iOS 13
* check that video capture device is ready before capturing, fix for string cameraId assignment
* add missing ios 13 camera types and exposure check.
* remove duplicates from camera id list, return its type on IOS, handle WB crash on unsupported devices.
* Do not return virtual devices (we can't really use them with the library), update zoom and flash on preset change so they are maintained while recording/restoring.
* remove redundant duplicates check, add SDK macro check for ultra wide lenses, remove virtual cameras from query
* Add more info about deviceType and iOS camera IDs behaviour.
* Impelement cameraIds for iOS, and various iOS bug fixes.
Summary:
- Implement getCameraIds and cameraId property to manually select a camera device.
- Fix the internal preset being used to properly use a photo and video preset, and check for invalid presets on camera switch.
- This fixes a bug that would happen after recording. After a recording is complete, the camera preset was set to 4k permanently, and attempting to select any camera that does not support that preset would result in an totally unusable camera. Now, the "Photo" preset is the default which should be the highest quality option for photos, and the previously used "High" preset will be used as video default if no value is provided.
- This also adds proper support to setting the undocumented "pictureSize" property. Setting this property will change the default photo quality to be that one.
- Minor cleanup to the camera initialization so the same code is not called as many times redundantly (start session / device happening multiple times)
- The reduced redundant calls to start session and session updates should make the camera startup slightly faster.
- Fix for warnings issued due to misuse of the "isRecording" property. This property was defined as both as a property and a method and was both getting set manually and retrieved with a function. It should now consistently only be retrieved with a function based on the real recording status. Should have no behaivour change, but removes a pesky warning.
- Fix an issue related to the camera getting stuck after a background resume by removing the use of the queue for those two events.
- Fixes a wrong event unsubscription (UIDeviceOrientationDidChangeNotification to UIApplicationDidChangeStatusBarOrientationNotification) from a previous change. Also move event subscriptions to superview changes for consistency and so they are not used/consumed if not needed. After testing, the View of RNCamera is instantiated twice (for some reason I couldn't find) but used only once. RN keeps a reference to the view in memory even after the camera is destroyed, resulting in these events getting fired and handled all the time during the app's lifetime.
* add maxZoom property to iOS.
Android: Not needed as of now. Also, zoom multiplier behaves different on Android and can't be used with the same value as iOS.
* Update RNCamera.m
Add videoBitrate option for ios, setting video bitrate requires a codec to be set
* Update API documentation
* Update documentation
* Update typings
Since the code relies on the status bar orientation, it makes sense to use UIApplicationDidChangeStatusBarOrientationNotification instead. This should fix some issues where the orientation value of the status bar has not been updated when the event is received.
* Use standardized exposure value between 0 and 1
* Add iOS native exposure control
* Delete logging
* Add documentation + spell fixes
* Update RNCamera.m
* Remove reset of exposure after tap-to-focus
When VCARD doesnt contain ALL name fields, app crashes because it is trying to insert nil values into dictionary. This provides blank strings for nil values in vcard
* wrap captureStillImageAsynchronouslyFromConnection with try/catch
the diff may look pretty big but the only change made here is
try/catch block added for captureStillImageAsynchronouslyFromConnection
call.
For some reason, this method may throw an exception saying about
inconsistent state.
* make sure null in cameraHandle isn't passed to Native code
