Camera - interruped video recording

Man, it took me forever to find a way to repair a video from when recording crashed on my Jolla.

There's a tool called "untrunc" that is supposed to be able to rebuild the missing "moov" index. You can find it at http://vcg.isti.cnr.it/~ponchio/untrunc.php

Unfortunately, when I tried it, it did not yet support straight mp4v video. But eventually, I managed to beat it into working with the following changes:

diff --git a/track.cpp b/track.cpp
index 2afd0ad..e00414e 100644
--- a/track.cpp
+++ b/track.cpp
@@ -206,9 +206,25 @@ int Codec::getLength(unsigned char *start, int maxlength) {
         return consumed;

     } else if(name == "mp4v") {
-        //found no way to guess size, probably the only way is to use some functions in ffmpeg
-        //to decode the stream
-        return -1;
+
+        AVFrame *frame = avcodec_alloc_frame();
+        if(!frame)
+            throw string("Could not create AVFrame");
+        AVPacket avp;
+        av_init_packet(&avp);
+
+        int got_frame;
+        avp.data=(uint8_t *)(start);
+        avp.size = maxlength;
+        int consumed = avcodec_decode_video2(context, frame, &got_frame, &avp);
+        av_freep(&frame);
+
+        if (consumed <= 0 || consumed >= maxlength) return consumed;
+
+        if (start[consumed] == 0x7f)
+            consumed++; // hack
+
+        return consumed;

     } else if(name == "avc1") {

@@ -359,6 +375,8 @@ void Track::parse(Atom *t, Atom *mdat) {
     codec.codec = avcodec_find_decoder(codec.context->codec_id);
     //if audio use next?

+    codec.context->thread_type = 0; // if you're using ffmpeg's libavcodec
+
     if(!codec.codec) throw string("No codec found!");
     if(avcodec_open2(codec.context, codec.codec, NULL)<0)
         throw string("Could not open codec: ") + codec.context->codec_name;

Extending the previous (working) answer with all-in-one solution in Docker :)

Dockerfile

FROM ubuntu:xenial
ENV TERM xterm
ENV UNTRUNC /opt/video-recovery
WORKDIR $UNTRUNC

RUN apt-get update && \
  apt-get -y install build-essential curl unzip libavformat-dev libavcodec-dev libavutil-dev && \
  apt-get -y autoremove && apt-get clean
RUN curl -SLO https://github.com/ponchio/untrunc/archive/master.zip && \
  unzip *.zip -d $UNTRUNC && \
  rm *.zip && \
  mv untrunc-master/* ./ && \
  rm -R untrunc-master
COPY track.cpp.patch ./
RUN patch track.cpp track.cpp.patch
RUN g++ -o untrunc file.cpp main.cpp track.cpp atom.cpp mp4.cpp -L/usr/local/lib -lavformat -lavcodec -lavutil
CMD ["./untrunc"]

track.cpp.patch

219,236c219,236
<       /*     THIS DOES NOT SEEM TO WORK FOR SOME UNKNOWN REASON. IT JUST CONSUMES ALL BYTES.
<   *     AVFrame *frame = avcodec_alloc_frame();
<       if(!frame)
<           throw string("Could not create AVFrame");
<       AVPacket avp;
<       av_init_packet(&avp);
< 
<       int got_frame;
<       avp.data=(uint8_t *)(start);
<       avp.size = maxlength;
<       int consumed = avcodec_decode_video2(context, frame, &got_frame, &avp);
<       av_freep(&frame);
<       return consumed; */
< 
<       //found no way to guess size, probably the only way is to use some functions in ffmpeg
<       //to decode the stream
<       cout << "Unfortunately I found no way to guess size of mp4v packets. Sorry\n";
<       return -1;
---
>          AVFrame *frame = avcodec_alloc_frame();
>          if(!frame)
>              throw string("Could not create AVFrame");
>          AVPacket avp;
>          av_init_packet(&avp);
> 
>          int got_frame;
>          avp.data=(uint8_t *)(start);
>          avp.size = maxlength;
>          int consumed = avcodec_decode_video2(context, frame, &got_frame, &avp);
>          av_freep(&frame);
> 
>          if (consumed <= 0 || consumed >= maxlength) return consumed;
> 
>          if (start[consumed] == 0x7f)
>              consumed++; // hack
> 
>          return consumed;
387a388
>   codec.context->thread_type = 0; // if you're using ffmpeg's libavcodec

Usage:

• paste the code above to the respective files

• cd to the folder with Dockerfile and track.cpp.patch

docker build -t video-recovery .

docker run --rm -it -v /your/path/to/broken_and_template/videos:/untrunc --name video-recovery video-recovery bash

(or adjust that command to your needs) however, you are in the container now, you may run

 ./untrunc /untrunc/functionalSample.mp4 /untrunc/brokenVideo.mp4

you should see the result at the following path:

/your/path/to/broken_and_template/videos/brokenVideo.mp4_fixed.mp4

the following error (with different numbers) is somehow ok:

[aac @ 0x1a07da0] decode_band_types: Input buffer exhausted before END element found

Invalid length. -1094995529. Wrong match in track: 1
Found 2814 packets

Please for your safety: The described symptoms correspond to a lack of space on your device. Verify that and apply the btrfs balance proposed to solve it. Then go ahead with your recording problem. There is some priorities to respect if you do not want to increase the difficulties. Hope to help you! Best regards Thoric