added file conversion

process_data.py

@@ -57,7 +57,7 @@ def process_raw_video_file(args, s_config, r_config):
 
             else:
                 logger.info('Enter correct video(*.mp4) file path.')
-                
+
     except Exception as e:
         logger.error('Failed to process mp4 file.')
         pf.remove_file(video_file[0])
@@ -76,15 +76,15 @@ def process_raw_audio_file(args, s_config, r_config):
 
             if len(audio_file)>0:
                 logger.info('Calculating raw variables...')
-                
+
                 out_path = os.path.join(args.output_path, 'raw_variables')
                 pf.process_acoustic(audio_file[0], out_path, args.dbm_group, r_config)
-                
+
             else:
                 logger.info('Enter correct audio(*.wav) file path.')
     except Exception as e:
         logger.error('Failed to process wav file.')
-    
+
 def process_raw_video_dir(args, s_config, r_config):
     """
     Processing video file
@@ -94,17 +94,19 @@ def process_raw_video_dir(args, s_config, r_config):
         r_config: raw feature config object
     """
     if args.output_path != None:
-        vid_loc = glob.glob(args.input_path + '/*.mp4')
+        vid_loc = glob.glob(args.input_path + '/*.mp4') + glob.glob(args.input_path + '/*.mov')
-        
+
         if len(vid_loc) == 0:
             logger.info('Directory does not have any MP4 files.')
             return
-        
+
         logger.info('Calculating raw variables...')
         for vid_file in vid_loc:
             try:
-                
+                fname, file_ext = os.path.splitext(vid_file)
-                common_video(vid_file, args, r_config)  
+                if file_ext == '.mov':
+                    convert_file(vid_file)
+                common_video(fname+'.mp4', args, r_config)
             except Exception as e:
                 logger.error('Failed to process mp4 file.')
                 pf.remove_file(vid_file)
@@ -118,21 +120,39 @@ def process_raw_audio_dir(args, s_config, r_config):
         r_config: raw feature config object
     """
     if args.output_path != None:
-        audio_loc = glob.glob(args.input_path + '/*.wav')
+        audio_loc = glob.glob(args.input_path + '/*.wav') + glob.glob(args.input_path + '/*.mp3')
-        
+
         if len(audio_loc) == 0:
             logger.info('Directory does not have any WAV files.')
             return
-        
+
         logger.info('Calculating raw variables...')
         for audio in audio_loc:
             try:
-                
+                fname, file_ext = os.path.splitext(audio)
+                if file_ext == '.mp3':
+                    convert_file(audio)
                 out_path = os.path.join(args.output_path, 'raw_variables')
-                pf.process_acoustic(audio, out_path, args.dbm_group, r_config)
+                pf.process_acoustic(fname+'.wav', out_path, args.dbm_group, r_config)
             except Exception as e:
                 logger.error('Failed to process wav file.')
-        
+
+def convert_file(input_filepath):
+    _, file_ext = os.path.splitext(os.path.basename(input_filepath))
+    fname, _ = splitext(input_filepath)
+
+    if file_ext == '.mp3':
+        output_filepath = fname + '.wav'
+        logger.info('Converting audio from {} to wav'.format(input_filepath))
+        call = ['ffmpeg', '-i', input_filepath, output_filepath]
+
+    if file_ext == '.mov':
+        output_filepath = fname + '.mp4'
+        logger.info('Converting video from {} to mp4'.format(input_filepath))
+        call = ['ffmpeg', '-i', input_filepath, output_filepath]
+
+    subprocess.check_output(call)
+
 def process_derive(args, r_config, d_config, input_type):
     """
     Processing dbm derived variables
@@ -141,43 +161,49 @@ def process_derive(args, r_config, d_config, input_type):
         input_file = glob.glob(args.input_path)
     else:
         input_file = glob.glob(args.input_path + '/*')
-        
+
     out_raw_path = os.path.join(args.output_path, 'raw_variables')
     out_derive_path = os.path.join(args.output_path, 'derived_variables')
-    
+
     logger.info('Calculating derived variables...')
     feature_df = der.run_derive(input_file, out_raw_path, out_derive_path, r_config, d_config)
-    
+
 if __name__=="__main__":
     start_time = time.time()
     parser = argparse.ArgumentParser(description="Process video/audio......")
-    
+
     parser.add_argument("--input_path", help="path to the input files", required=True)
     parser.add_argument("--output_path", help="path to the raw and derived variable output", required=True)
     parser.add_argument("--dbm_group", help="list of feature groups", nargs='+')
-    
+
     args = parser.parse_args()
     s_config = config_reader.ConfigReader()
     r_config = config_raw_feature.ConfigRawReader()
     d_config = config_derive_feature.ConfigDeriveReader()
-    
+
     _, file_ext = os.path.splitext(os.path.basename(args.input_path))
-    
+
     if file_ext:
+        #add check for mov and mp3 here
         input_type = 'file'
-        if file_ext.lower() == '.mp4':
+
+        if file_ext.lower() in ['.mp4','mov']:
+            if file_ext.lower() == '.mov':
+                convert_file(args.input_path)
             process_raw_video_file(args, s_config, r_config)
 
-        elif file_ext.lower() == '.wav':
+        elif file_ext.lower() in ['.wav','.mp3']:
+            if file_ext.lower() == '.mp3':
+                convert_file(args.input_path)
             process_raw_audio_file(args, s_config, r_config)
-            
+
         else:
-            logger.error('No WAV or MP4 files detected in input path')
+            logger.error('No WAV/MP3 or MOV/MP4 files detected in input path')
     else:
         input_type = 'dir'
         process_raw_video_dir(args, s_config, r_config)
         process_raw_audio_dir(args, s_config, r_config)
-        
+
     process_derive(args, r_config, d_config, input_type)
     exec_time = time.time() - start_time
-    logger.info('Done! Processing time: {} seconds'.format(exec_time))
+    logger.info('Done! Processing time: {} seconds'.format(exec_time))