From 27bd9b54752d412d9d869dc02a9ed2b4444b9206 Mon Sep 17 00:00:00 2001
From: j <0x006A@0x2620.org>
Date: Wed, 12 Dec 2007 13:58:19 +0000
Subject: [PATCH] extract ogg clips

---
 oxdbarchive/cache.py                      |  19 ++-
 oxdbarchive/controllers.py                |   6 +-
 oxdbarchive/extract.py                    | 166 ++++++++++++++++++++++
 oxdbarchive/model.py                      |  36 ++++-
 oxdbarchive/subtitles.py                  | 114 ---------------
 oxdbarchive/tools/extract_poster_still.py |   6 +-
 6 files changed, 223 insertions(+), 124 deletions(-)
 create mode 100644 oxdbarchive/extract.py

diff --git a/oxdbarchive/cache.py b/oxdbarchive/cache.py
index 953f5ef..d771644 100644
--- a/oxdbarchive/cache.py
+++ b/oxdbarchive/cache.py
@@ -40,7 +40,7 @@ def loadFrame(afile, position):
     return loadFile(frame)
   return ''
 
-def loadClip(afile, position):
+def loadFlvClip(afile, position):
   position = basename(position).replace(':', '.')
   positions = position.split(';')
   if len(positions) > 1:
@@ -50,11 +50,26 @@ def loadClip(afile, position):
     outpoint = -1
   flash = join(afile.frameFolder, '%s.%s' % (position, 'flv'))
   if not exists(flash):
-    afile.extractClip(position, outpoint)
+    afile.extractFlvClip(position, outpoint)
   if exists(flash):
     return loadFile(flash)
   return ''
 
+def loadOggClip(afile, position):
+  position = basename(position).replace(':', '.')
+  positions = position.split(';')
+  if len(positions) > 1:
+    position = positions[0]
+    outpoint = positions[1]
+  else:
+    outpoint = -1
+  clip = join(afile.frameFolder, '%s.%s' % (position, 'ogg'))
+  if not exists(clip):
+    afile.extractOggClip(position, outpoint)
+  if exists(clip):
+    return loadFile(clip)
+  return ''
+
 def loadTimeline(afile):
   timeline = afile.timelineFile
   if not exists(timeline):
diff --git a/oxdbarchive/controllers.py b/oxdbarchive/controllers.py
index 6504e2d..b8056c7 100644
--- a/oxdbarchive/controllers.py
+++ b/oxdbarchive/controllers.py
@@ -43,7 +43,11 @@ class Root(controllers.RootController):
       if action == 'clip':
         cherrypy.response.headerMap['Content-Type'] = "video/x-flv"
         cherrypy.response.headerMap["Expires"] = httpExpires(60*60*24*15)
-        return f.clip(position)
+        return f.flvClip(position)
+      if action == 'ogg':
+        cherrypy.response.headerMap['Content-Type'] = "video/x-ogg"
+        cherrypy.response.headerMap["Expires"] = httpExpires(60*60*24*15)
+        return f.oggClip(position)
       elif action == 'frame':
         cherrypy.response.headerMap['Content-Type'] = "image/jpeg"
         cherrypy.response.headerMap["Expires"] = httpExpires(60*60*24*15)
diff --git a/oxdbarchive/extract.py b/oxdbarchive/extract.py
new file mode 100644
index 0000000..88288f0
--- /dev/null
+++ b/oxdbarchive/extract.py
@@ -0,0 +1,166 @@
+# -*- coding: utf-8 -*-
+# -*- Mode: Python; -*-
+# vi:si:et:sw=2:sts=2:ts=2
+
+import re
+import os
+from os.path import abspath, join, dirname
+import shutil
+import time
+
+from subtitles import *
+
+
+def extract_flash_ng(movie_file, flash_file, inpoint, outpoint, width=128, height=96, offset = 0):
+  ext = movie_file.split('.')[-1]
+  if ext in ('sub', 'srt'):
+    print "this is not a movie file, will not try to extract frames"
+    return
+  if offset:
+    print "Inpoint ", inpoint,
+    inpoint = shift_time(-offset, inpoint)
+    outpoint = shift_time(-offset, outpoint)
+    print " becomes ", inpoint
+  
+  print "extracting %s -> %s" % (inpoint, outpoint)
+  duration = time_str2msec(outpoint) - time_str2msec(inpoint)
+  inpoint = time_str2msec(inpoint)
+  extractClipScript = abspath(join(dirname(__file__), "tools/extract_clip.py"))
+
+  cmd = '''%s "%s" %s %s %s''' % (extractClipScript, movie_file, flash_file, inpoint, duration)
+  os.system(cmd.encode('utf-8'))
+
+def extract_flash(movie_file, flash_file, inpoint, outpoint, width=128, height=96, offset = 0):
+  import warnings
+  warnings.filterwarnings("ignore", "tempnam")
+  ext = movie_file.split('.')[-1]
+  if ext in ('sub', 'srt', 'mkv'):
+    print "this is not a movie file, will not try to extract frames"
+    return
+  framedir = os.tempnam()
+  os.mkdir(framedir)
+  os.chdir(framedir)
+  if offset:
+    print "Inpoint ", inpoint,
+    inpoint = shift_time(-offset, inpoint)
+    outpoint = shift_time(-offset, outpoint)
+    print " becomes ", inpoint
+  print "extracting %s -> %s" % (inpoint, outpoint)
+  outpoint = float(time_str2msec(outpoint) - time_str2msec(inpoint)) / 1000 + 1
+
+  audiorate = "44100"
+  if os.path.exists(movie_file):
+    mencoder_options = ''
+    mencoder_options += " '%s'" % movie_file
+    mencoder_options += " -ss '%s' -endpos %0.2f" % (inpoint, outpoint)
+    mencoder_options += ' -ovc copy -oac copy -o tempfile.avi '
+    mencoder = "mencoder %s >/dev/null 2>&1" % mencoder_options
+    #print mencoder.encode('utf-8')
+    os.system(mencoder.encode('utf-8'))
+    
+    #create flash file
+    ffmpeg_options = ''
+    #ffmpeg_options += " -ss '%s' -t %0.2f" % (inpoint, outpoint)
+    ffmpeg_options += " -y -i 'tempfile.avi'"
+    ffmpeg_options += " -ar %s -b 128000 '%s'" % (audiorate, flash_file)
+    ffmpeg = "ffmpeg %s >/dev/null 2>&1" % ffmpeg_options
+    #print ffmpeg.encode('utf-8')
+    os.system(ffmpeg.encode('utf-8'))
+  else:
+    print "update the cache %s missing" % movie_file.encode('utf-8')
+  shutil.rmtree(framedir)
+
+
+def extract_ogg(movie_file, clip_file, inpoint, outpoint, width=128, height=96, offset = 0):
+  import warnings
+  warnings.filterwarnings("ignore", "tempnam")
+  ext = movie_file.split('.')[-1]
+  if ext in ('sub', 'srt', 'mkv'):
+    print "this is not a movie file, will not try to extract frames"
+    return
+  framedir = os.tempnam()
+  os.mkdir(framedir)
+  os.chdir(framedir)
+  if offset:
+    print "Inpoint ", inpoint,
+    inpoint = shift_time(-offset, inpoint)
+    outpoint = shift_time(-offset, outpoint)
+    print " becomes ", inpoint
+  print "extracting %s -> %s" % (inpoint, outpoint)
+  outpoint = float(time_str2msec(outpoint) - time_str2msec(inpoint)) / 1000 + 1
+
+  audiorate = "44100"
+  if os.path.exists(movie_file):
+    mencoder_options = ''
+    mencoder_options += " '%s'" % movie_file
+    mencoder_options += " -ss '%s' -endpos %0.2f" % (inpoint, outpoint)
+    mencoder_options += ' -ovc copy -oac copy -o tempfile.avi '
+    mencoder = "mencoder %s >/dev/null 2>&1" % mencoder_options
+    #print mencoder.encode('utf-8')
+    os.system(mencoder.encode('utf-8'))
+    
+    #create ogg file
+    ffmpeg2theora_options = ''
+    ffmpeg2theora_options += " 'tempfile.avi' -H %s -o '%s'" % (audiorate, flash_file)
+    ffmpeg2theora = "ffmpeg2theora %s >/dev/null 2>&1" % ffmpeg2theora_options
+    print ffmpeg2theora.encode('utf-8')
+    os.system(ffmpeg2theora.encode('utf-8'))
+  else:
+    print "update the cache %s missing" % movie_file.encode('utf-8')
+  shutil.rmtree(framedir)
+
+def extract_frame(movie_file, timestamp, img_folder, width=128, offset = 0, redo = False):
+  import warnings
+  warnings.filterwarnings("ignore", "tempnam")
+  ext = movie_file.split('.')[-1]
+  if ext in ('sub', 'srt'):
+    print "this is not a movie file, will not try to extract frames"
+    return
+  framedir = os.tempnam()
+  
+  os.mkdir(framedir)
+  os.chdir(framedir)
+  if offset:
+    timestamp_in_file = shift_time(-offset, timestamp)
+  else:
+    timestamp_in_file = timestamp
+  if os.path.exists(movie_file):
+    mplayer_options = ''
+    mplayer_options += " '%s'" % movie_file
+    mplayer_options += " -ss '%s' -frames 2" % (timestamp_in_file)
+    mplayer_options += " -vo jpeg:quality=90 -vf scale -zoom -xy %d " % width
+    mplayer_options += " -ao null"
+    mplayer = "mplayer %s >/dev/null 2>&1" % mplayer_options
+    frame = os.path.join(img_folder, "%s.%s" % (timestamp.replace(':', '.'), img_extension))
+    if redo or not os.path.exists(frame):
+      print mplayer.encode('utf-8')
+      os.system (mplayer.encode('utf-8'))
+      files = os.listdir(framedir)
+      if files:
+        print "creating frame ", frame
+        shutil.move(os.path.join(framedir,files[-1]), frame)
+        if len(files)>1:
+          for f in files[:-2]:
+            print "unlink", f
+            os.unlink(f)
+    time.sleep(0.1)
+  else:
+    print "update the cache %s missing" % movie_file
+  shutil.rmtree(framedir)
+
+def extract_poster_still(movie_file, png_file, inpoint):
+  ext = movie_file.split('.')[-1]
+  if ext in ('sub', 'srt'):
+    print "this is not a movie file, will not try to extract frames"
+    return
+  inpoint = time_str2msec(inpoint)
+  extractClipScript = abspath(join(dirname(__file__), "tools/extract_poster_still.py"))
+
+  cmd = '''%s "%s" "%s" %s''' % (extractClipScript, movie_file, png_file, inpoint)
+  os.system(cmd.encode('utf-8'))
+
+def extract_subtitles(movie_file, srt, img_folder, width=128, offset = 0, redo = False):
+  subtitles = srt2dict(srt)  
+  for k in sorted([int(k) for k in subtitles]):
+    timestamp = subtitles["%s" % k]['start']
+    extract_frame(movie_file, timestamp, img_folder, width, offset, redo)
diff --git a/oxdbarchive/model.py b/oxdbarchive/model.py
index fcddc2c..1de177d 100644
--- a/oxdbarchive/model.py
+++ b/oxdbarchive/model.py
@@ -24,6 +24,7 @@ import cache
 import oxdb_import
 from oxdb_utils import oxdb_title, oxdb_director, oxdb_id, oxdb_makedir
 from subtitles import *
+from extract import *
 import midentify
 
 
@@ -447,8 +448,8 @@ class ArchiveFile(SQLObject):
       #enable this later
       #self.extractFrames()
       #self.extractClips()
-    
-  def extractClip(self, inpoint, outpoint=-1, flash_folder=-1):
+  
+  def extractFlvClip(self, inpoint, outpoint=-1, flash_folder=-1):
     if flash_folder == -1:
       flash_folder = self.frameFolder
     movie_file = self.mini_movie_file
@@ -469,9 +470,30 @@ class ArchiveFile(SQLObject):
     extract_flash(movie_file, flash_movie, inpoint, outpoint, width, height, offset = 0)
     #extract_flash_ng(self.absolutePath, flash_movie, inpoint, outpoint, width, height, offset)
   
+  def extractOggClip(self, inpoint, outpoint=-1, flash_folder=-1):
+    if clip_folder == -1:
+      clip_folder = self.frameFolder
+    movie_file = self.mini_movie_file
+    position = inpoint.replace(':', '.')
+    clip_movie = join(self.frameFolder, '%s.%s' % (position, 'ogg'))
+    width = 128
+    height = int(width / self.frameAspect)
+    height = height - height % 2
+    inpoint = inpoint.replace('.', ':')
+    if outpoint == -1:
+      s = self._findSubtitleByStart(inpoint)
+      if s:
+        outpoint = s['stop']
+      else:
+        outpoint = shift_time(5000, inpoint)
+    else:
+      outpoint = outpoint.replace('.', ':')
+    extract_ogg(movie_file, clip_movie, inpoint, outpoint, width, height, offset = 0)
+
+  
   def extractClips(self, img_folder=cache.frame_cache_root):
     for p in self._startPoints():
-      self.extractClip(p)
+      self.extractFlvClip(p)
   
   def extractFrame(self, position, img_folder=-1): 
     if img_folder == -1:
@@ -508,6 +530,7 @@ class ArchiveFile(SQLObject):
     if exists(mini_movie_file) and not force:
       print "clip exists, skipping extraction", mini_movie_file
       return
+    self.extracted = False
     oxdb_makedir(dirname(mini_movie_file))
     
     options = ''
@@ -609,8 +632,11 @@ class ArchiveFile(SQLObject):
       pixels = self.updatePixels()
     return pixels
   
-  def clip(self, position):
-    return cache.loadClip(self, position)
+  def flvClip(self, position):
+    return cache.loadFlvClip(self, position)
+  
+  def oggClip(self, position):
+    return cache.loadOggClip(self, position)
   
   def frame(self, position):
     return cache.loadFrame(self, position)
diff --git a/oxdbarchive/subtitles.py b/oxdbarchive/subtitles.py
index 320e997..9a156ed 100644
--- a/oxdbarchive/subtitles.py
+++ b/oxdbarchive/subtitles.py
@@ -128,120 +128,6 @@ def split_subtitle(subtitles, offset):
       two[k] = subtitle[k]
   two = shift_subtitles(-offset, -len(two), two)
 
-def extract_flash_ng(movie_file, flash_file, inpoint, outpoint, width=128, height=96, offset = 0):
-  ext = movie_file.split('.')[-1]
-  if ext in ('sub', 'srt'):
-    print "this is not a movie file, will not try to extract frames"
-    return
-  if offset:
-    print "Inpoint ", inpoint,
-    inpoint = shift_time(-offset, inpoint)
-    outpoint = shift_time(-offset, outpoint)
-    print " becomes ", inpoint
-  
-  print "extracting %s -> %s" % (inpoint, outpoint)
-  duration = time_str2msec(outpoint) - time_str2msec(inpoint)
-  inpoint = time_str2msec(inpoint)
-  extractClipScript = abspath(join(dirname(__file__), "tools/extract_clip.py"))
-
-  cmd = '''%s "%s" %s %s %s''' % (extractClipScript, movie_file, flash_file, inpoint, duration)
-  os.system(cmd.encode('utf-8'))
-  
-def extract_flash(movie_file, flash_file, inpoint, outpoint, width=128, height=96, offset = 0):
-  import warnings
-  warnings.filterwarnings("ignore", "tempnam")
-  ext = movie_file.split('.')[-1]
-  if ext in ('sub', 'srt', 'mkv'):
-    print "this is not a movie file, will not try to extract frames"
-    return
-  framedir = os.tempnam()
-  os.mkdir(framedir)
-  os.chdir(framedir)
-  if offset:
-    print "Inpoint ", inpoint,
-    inpoint = shift_time(-offset, inpoint)
-    outpoint = shift_time(-offset, outpoint)
-    print " becomes ", inpoint
-  print "extracting %s -> %s" % (inpoint, outpoint)
-  outpoint = float(time_str2msec(outpoint) - time_str2msec(inpoint)) / 1000 + 1
-
-  audiorate = "44100"
-  if os.path.exists(movie_file):
-    mencoder_options = ''
-    mencoder_options += " '%s'" % movie_file
-    mencoder_options += " -ss '%s' -endpos %0.2f" % (inpoint, outpoint)
-    mencoder_options += ' -ovc copy -oac copy -o tempfile.avi '
-    mencoder = "mencoder %s >/dev/null 2>&1" % mencoder_options
-    #print mencoder.encode('utf-8')
-    os.system(mencoder.encode('utf-8'))
-    
-    ffmpeg_options = ''
-    #ffmpeg_options += " -ss '%s' -t %0.2f" % (inpoint, outpoint)
-    ffmpeg_options += " -y -i 'tempfile.avi'"
-    ffmpeg_options += " -ar %s -b 128000 '%s'" % (audiorate, flash_file)
-    ffmpeg = "ffmpeg %s >/dev/null 2>&1" % ffmpeg_options
-    #print ffmpeg.encode('utf-8')
-    os.system(ffmpeg.encode('utf-8'))
-  else:
-    print "update the cache %s missing" % movie_file.encode('utf-8')
-  shutil.rmtree(framedir)
-  
-def extract_frame(movie_file, timestamp, img_folder, width=128, offset = 0, redo = False):
-  import warnings
-  warnings.filterwarnings("ignore", "tempnam")
-  ext = movie_file.split('.')[-1]
-  if ext in ('sub', 'srt'):
-    print "this is not a movie file, will not try to extract frames"
-    return
-  framedir = os.tempnam()
-  
-  os.mkdir(framedir)
-  os.chdir(framedir)
-  if offset:
-    timestamp_in_file = shift_time(-offset, timestamp)
-  else:
-    timestamp_in_file = timestamp
-  if os.path.exists(movie_file):
-    mplayer_options = ''
-    mplayer_options += " '%s'" % movie_file
-    mplayer_options += " -ss '%s' -frames 2" % (timestamp_in_file)
-    mplayer_options += " -vo jpeg:quality=90 -vf scale -zoom -xy %d " % width
-    mplayer_options += " -ao null"
-    mplayer = "mplayer %s >/dev/null 2>&1" % mplayer_options
-    frame = os.path.join(img_folder, "%s.%s" % (timestamp.replace(':', '.'), img_extension))
-    if redo or not os.path.exists(frame):
-      print mplayer.encode('utf-8')
-      os.system (mplayer.encode('utf-8'))
-      files = os.listdir(framedir)
-      if files:
-        print "creating frame ", frame
-        shutil.move(os.path.join(framedir,files[-1]), frame)
-        if len(files)>1:
-          for f in files[:-2]:
-            print "unlink", f
-            os.unlink(f)
-    time.sleep(0.1)
-  else:
-    print "update the cache %s missing" % movie_file
-  shutil.rmtree(framedir)
-
-def extract_poster_still(movie_file, png_file, inpoint):
-  ext = movie_file.split('.')[-1]
-  if ext in ('sub', 'srt'):
-    print "this is not a movie file, will not try to extract frames"
-    return
-  inpoint = time_str2msec(inpoint)
-  extractClipScript = abspath(join(dirname(__file__), "tools/extract_poster_still.py"))
-
-  cmd = '''%s "%s" "%s" %s''' % (extractClipScript, movie_file, png_file, inpoint)
-  os.system(cmd.encode('utf-8'))
-
-def extract_subtitles(movie_file, srt, img_folder, width=128, offset = 0, redo = False):
-  subtitles = srt2dict(srt)  
-  for k in sorted([int(k) for k in subtitles]):
-    timestamp = subtitles["%s" % k]['start']
-    extract_frame(movie_file, timestamp, img_folder, width, offset, redo)
-
 def detectEncoding(fp):
     bomDict={ # bytepattern : name              
              (0x00, 0x00, 0xFE, 0xFF) : "utf_32_be",        
diff --git a/oxdbarchive/tools/extract_poster_still.py b/oxdbarchive/tools/extract_poster_still.py
index 79d6c6a..e446f3a 100755
--- a/oxdbarchive/tools/extract_poster_still.py
+++ b/oxdbarchive/tools/extract_poster_still.py
@@ -139,7 +139,7 @@ class PosterStillMplayer:
   
   def scaleImage(self, tmpImage):
     img = Image.open(tmpImage)
-    output_d = self.scaleto(i.size[0], i.size[1])
+    output_d = self.scaleto(img.size[0], img.size[1])
     img = img.resize(output_d, Image.ANTIALIAS)
     img.save(self.png_frame)
   
@@ -158,4 +158,6 @@ if __name__ == "__main__":
     sys.exit()
   outputFile = sys.argv[2]
   offset = int(float(sys.argv[3]) * gst.MSECOND)
-  f = PosterStill(inputFile, outputFile, offset, height)
+  #f = PosterStill(inputFile, outputFile, offset, height)
+  f = PosterStillMplayer(inputFile, outputFile, offset, height)
+