如何从java读取ffmpeg响应并使用它来创建进度条？

Question

I am working on creating a progress bar for ffmpeg in java. So for that I need to execute a command, then read all of the progress:

String[] command = {"gnome-terminal", "-x", "/bin/sh", "-c","ffmpeg -i /home/tmp/F.webm /home/tmp/converted1.mp4"};

Process process = Runtime.getRuntime().exec(command);

This runs perfectly. However, I need to capture the all of the progress to make a progress bar. So how can I read that data from java?

Answer 1

Here's a complete example for you which should get you started

import java.io.*;
import java.util.Scanner;
import java.util.regex.Pattern;

class Test {
  public static void main(String[] args) throws IOException {
    ProcessBuilder pb = new ProcessBuilder("ffmpeg","-i","in.webm","out.mp4");
    final Process p = pb.start();

    new Thread() {
      public void run() {

        Scanner sc = new Scanner(p.getErrorStream());

        // Find duration
        Pattern durPattern = Pattern.compile("(?<=Duration: )[^,]*");
        String dur = sc.findWithinHorizon(durPattern, 0);
        if (dur == null)
          throw new RuntimeException("Could not parse duration.");
        String[] hms = dur.split(":");
        double totalSecs = Integer.parseInt(hms[0]) * 3600
                         + Integer.parseInt(hms[1]) *   60
                         + Double.parseDouble(hms[2]);
        System.out.println("Total duration: " + totalSecs + " seconds.");

        // Find time as long as possible.
        Pattern timePattern = Pattern.compile("(?<=time=)[\\d.]*");
        String match;
        while (null != (match = sc.findWithinHorizon(timePattern, 0))) {
          double progress = Double.parseDouble(match) / totalSecs;
          System.out.printf("Progress: %.2f%%%n", progress * 100);
        }
      }
    }.start();

  }
}

Output:

Total duration: 117.7 seconds.
Progress: 7.71%
Progress: 16.40%
Progress: 25.00%
Progress: 33.16%
Progress: 42.67%
Progress: 51.35%
Progress: 60.57%
Progress: 69.07%
Progress: 78.02%
Progress: 86.49%
Progress: 95.94%
Progress: 99.97%

You may also consider using some kind of Java bindings for ffmpeg such as jjmpeg which may provide what you need in a more robust way.

EDIT

With ffmpeg 2.0, time output is HH:mm:ss.S so the timePattern needs a to incorporate a :

Pattern timePattern = Pattern.compile("(?<=time=)[\\d:.]*");

In addition, the dur will need to be split on : and summed together

String[] matchSplit;
while (null != (match = sc.findWithinHorizon(timePattern, 0))) {
    matchSplit = match.split(":")
    double progress = Integer.parseInt(matchSplit[0]) * 3600 +
        Integer.parseInt(matchSplit[1]) * 60 +
        Double.parseDouble(matchSplit[2]) / totalSecs;
//...

Answer 2

You can try to parse ffmpeg output and somehow understand what work is already done. But this is hard and not stable anyway. Neither we (the ffmpeg users) nor ffmpeg itself does not know and cannot know in terms of time how long the processing will take.

According to my experience the easiest way is to implement a kind of heuristics. Assume that the time of processing linearly depends on the file size. This approach is "wrong" but good enough and very simple. Now run your processing with exactly the same options you are using in real life with several files of different size. Create mapping of size-to-time. Do statistical analysis and create formula like time = something + coef * size .

Now you can create you process bar. As most process bars it should arrive to ~95% and then wait for real termination of the process.

It is very simple and works not worse than any other more sophisticated solution.

Answer 3

*I have Successfully Display ProgressBar for ffmpeg command using Following Code.

  try {
                Scanner sc = new Scanner(process.getErrorStream());

                // Find duration
                Pattern durPattern = Pattern.compile("(?<=Duration: )[^,]*");
                String dur = sc.findWithinHorizon(durPattern, 0);
                Log.e("duration"+dur);
                String givenDateString = dur;
                SimpleDateFormat sdf = new SimpleDateFormat("HH:mm:ss.S");
                sdf.setTimeZone(TimeZone.getTimeZone("GMT"));
                try {
                    Date mDate = sdf.parse(givenDateString);
                    totalDuration = mDate.getTime();
                    System.out.println("Duration in milli :: " + totalDuration);
                } catch (ParseException e) {
                    e.printStackTrace();
                }

                // Find time as long as possible.
                Pattern timePattern = Pattern.compile("(?<=time=)[\\d:.]*");
                String match;
                String[] matchSplit;
                while (null != (match = sc.findWithinHorizon(timePattern, 0))) {
                    if (isCancelled()) {
                        return;
                    }
                    Log.e("match"+match);
                    String givenDateString1 = match;
                    SimpleDateFormat sdf1 = new SimpleDateFormat("HH:mm:ss.S");
                    sdf1.setTimeZone(TimeZone.getTimeZone("GMT"));
                    try {
                        Date mDate = sdf1.parse(givenDateString1);
                        currentDuration = mDate.getTime();
                        System.out.println("Time in milli :: " + currentDuration);
                    } catch (ParseException e) {
                        e.printStackTrace();
                    }
                    Double percentage = (double) 0;

                    long currentSeconds = (int) (currentDuration);
                    long totalSeconds = (int) (totalDuration);

                    // calculating percentage
                     percentage =(((double)currentSeconds)/totalSeconds)*100;


                    Log.e("Progress"+percentage);
                    publishProgress(""+percentage);
                }
            }catch (Exception e){
                e.printStackTrace();
            }