add ffprobe wrapper

ox/file.py

@@ -13,6 +13,12 @@ from ox.utils import json
 
 __all__ = ['sha1sum', 'oshash', 'avinfo', 'makedirs']
 
+def cmd(program):
+    local = os.path.expanduser('~/.ox/bin/%s' % program)
+    if os.path.exists(local):
+        program = local
+    return program
+
 def sha1sum(filename):
     sha1 = hashlib.sha1()
     file=open(filename)
@@ -62,10 +68,7 @@ def oshash(filename):
 
 def avinfo(filename):
     if os.path.getsize(filename):
-        ffmpeg2theora = 'ffmpeg2theora'
-        local = os.path.expanduser('~/.ox/bin/ffmpeg2theora')
-        if os.path.exists(local):
-            ffmpeg2theora = local
+        ffmpeg2theora = cmd('ffmpeg2theora')
         p = subprocess.Popen([ffmpeg2theora], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
         info, error = p.communicate()
         version = info.split('\n')[0].split(' - ')[0].split(' ')[-1]
@@ -90,6 +93,83 @@ def avinfo(filename):
 
     return {'path': filename, 'size': 0}
 
+def ffprobe(filename):
+    p = subprocess.Popen([
+        cmd('ffprobe'),
+        '-show_format',
+        '-show_streams',
+        '-print_format',
+        'json',
+        '-i', filename
+
+    ], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    info, error = p.communicate()
+    ffinfo = json.loads(info)
+
+    def fix_value(key, value):
+        if key == 'r_frame_rate':
+            value = value.replace('/', ':')
+        elif key == 'bit_rate':
+            value = float(value) / 1000
+        elif key == 'duration':
+            value = float(value)
+        elif key == 'size':
+            value = int(value)
+        return value
+
+    info = {}
+    for key in ('duration', 'size', 'bit_rate'):
+        info[{
+            'bit_rate': 'bitrate'
+        }.get(key, key)] = fix_value(key, ffinfo['format'][key])
+    info['audio'] = []
+    info['video'] = []
+    info['metadata'] = ffinfo['format'].get('tags', {})
+    for s in ffinfo['streams']:
+        tags =  s.pop('tags', {})
+        for t in tags:
+            info['metadata'][t] = tags[t]
+        if s.get('codec_type') in ('audio', 'video'):
+            stream = {}
+            keys = [ 
+                'codec_name',
+                'width',
+                'height',
+                'bit_rate',
+                'index',
+                'display_aspect_ratio',
+                'sample_rate',
+                'channels',
+            ]
+            if s['codec_type'] == 'video':
+                keys += [
+                    'sample_aspect_ratio',
+                    'r_frame_rate',
+                    'pix_fmt',
+                ]
+
+            for key in keys:
+                if key in s:
+                    stream[{
+                        'codec_name': 'codec',
+                        'bit_rate': 'bitrate',
+                        'index': 'id',
+                        'r_frame_rate': 'framerate',
+                        'sample_rate': 'samplerate',
+                        'pix_fmt': 'pixel_format',
+                    }.get(key, key)] = fix_value(key, s[key])
+            info[s['codec_type']].append(stream)
+        else:
+            pass
+            #print s
+    for v in info['video']:
+        if not 'display_aspect_ratio' in v and 'width' in v:
+            v['display_aspect_ratio'] = '%d:%d' % (v['width'], v['height'])
+            v['pixel_aspect_ratio'] = '1:1'
+    info['oshash'] = ox.oshash(filename)
+    info['path'] = os.path.basename(filename)
+    return info
+
 def makedirs(path):
     if not os.path.exists(path):
         try: