add pysox, 2x faster

lib/render.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python2
 # -*- coding: UTF-8 -*-
 # File: render.py
-# Date: Thu Dec 25 10:18:20 2014 +0800
+# Date: Thu Dec 25 22:05:41 2014 +0800
 # Author: Yuxin Wu <[email protected]>
 
 import os
@@ -20,7 +20,7 @@ except:
     css_compress = lambda x: x
 
 from .msg import *
-from .utils import ensure_unicode, ProgressReporter, pmap
+from .utils import ensure_unicode, ProgressReporter, pmap, timing
 from .smiley import SmileyProvider
 from .msgslice import MessageSlicerByTime, MessageSlicerBySize
 
@@ -97,7 +97,6 @@ class HTMLRender(object):
                 return fallback()
             bigimgpath = self.parser.imginfo.get(msg.msgSvrId)
             fnames = [k for k in [imgpath, bigimgpath] if k]
-            assert len(fnames) > 0, msg.msg_str()
             bigimg, smallimg = self.res.get_img(fnames)
             if not smallimg:
                 logger.warn("No image thumbnail found for {}".format(imgpath))
@@ -130,8 +129,6 @@ class HTMLRender(object):
 
     def _render_partial_msgs(self, msgs):
         """ return single html"""
-        talker_name = msgs[0].talker
-        avatars = self.get_avatar_pair(talker_name)
         slicer = MessageSlicerByTime()
         slices = slicer.slice(msgs)
 
@@ -151,12 +148,12 @@ class HTMLRender(object):
                                 extra_js=self.js_string,
                                 talker=msgs[0].talker_name,
                                 messages=u''.join(blocks),
-                                avatars=avatars)
-
+                                avatars=self.avatars)
 
     def render_msgs(self, msgs):
         """ render msgs of one friend, return a list of html"""
         talker_name = msgs[0].talker
+        self.avatars = self.get_avatar_pair(talker_name)
         logger.info(u"Rendering {} messages of {}({})".format(
             len(msgs), self.parser.contacts[talker_name], talker_name))
 

lib/res.py

@@ -1,7 +1,7 @@
 #!/usr/bin/env python2
 # -*- coding: UTF-8 -*-
 # File: res.py
-# Date: Thu Dec 25 10:04:29 2014 +0800
+# Date: Thu Dec 25 23:19:32 2014 +0800
 # Author: Yuxin Wu <[email protected]>
 
 import glob
@@ -14,9 +14,10 @@ import logging
 import imghdr
 logger = logging.getLogger(__name__)
 
-import eyed3
+import pysox
 
-from lib.avatar import AvatarReader
+from .avatar import AvatarReader
+from .utils import timing
 
 LIB_PATH = os.path.dirname(os.path.abspath(__file__))
 INTERNAL_EMOJI_DIR = os.path.join(LIB_PATH, 'static', 'internal_emoji')
@@ -53,21 +54,25 @@ class Resource(object):
                     "Error interpreting the protocol, this is a bug!"
                 self.speak_data[key] = full_path
 
+    @timing(total=True)
     def get_voice_mp3(self, imgpath):
         """ return base64 string, and voice duration"""
         amr_fpath = self.speak_data[imgpath]
         assert amr_fpath.endswith('.amr')
         mp3_file = os.path.join('/tmp',
                                 os.path.basename(amr_fpath)[:-4] + '.mp3')
-        # TODO is there a library to use?
-        ret = os.system('sox {} {}'.format(amr_fpath, mp3_file))
-        if ret != 0:
-            logger.warn("Sox Failed!")
-            return ""
-        mp3_string = open(mp3_file, 'rb').read()
-        duration = eyed3.load(mp3_file).info.time_secs
+
+        infile = pysox.CSoxStream(amr_fpath)
+        outfile = pysox.CSoxStream(mp3_file, 'w', infile.get_signal())
+        chain = pysox.CEffectsChain(infile, outfile)
+        chain.flow_effects()
+        outfile.close()
+
+        signal = infile.get_signal().get_signalinfo()
+        duration = signal['length'] * 1.0 / signal['rate']
+        mp3_string = Resource.get_file_b64(mp3_file)
         os.unlink(mp3_file)
-        return base64.b64encode(mp3_string), duration
+        return mp3_string, duration
 
     @staticmethod
     def get_file_b64(fname):

lib/utils.py

@@ -1,11 +1,16 @@
 #!/usr/bin/env python2
 # -*- coding: UTF-8 -*-
 # File: utils.py
-# Date: Thu Dec 25 10:11:21 2014 +0800
+# Date: Thu Dec 25 16:27:39 2014 +0800
 # Author: Yuxin Wu <[email protected]>
 
 import sys
 import time
+import logging
+logger = logging.getLogger(__name__)
+
+import time, functools
+from collections import defaultdict
 
 def ensure_bin_str(s):
     if type(s) == str:
@@ -111,3 +116,43 @@ def pimap(map_func, iterator, nr_proc=None, nr_precompute=None):
 def pmap(map_func, iterator, nr_proc=None, nr_precompute=None):
     '''parallel map'''
     return list(pimap(map_func, iterator, nr_proc, nr_precompute))
+
+
+class TotalTimer(object):
+    def __init__(self):
+        self.times = defaultdict(float)
+
+    def add(self, name, t):
+        self.times[name] += t
+
+    def reset(self):
+        self.times = defaultdict(float)
+
+    def __del__(self):
+        for k, v in self.times.iteritems():
+            logger.info("{} took {} seconds in total.".format(k, v))
+
+_total_timer = TotalTimer()
+class timing(object):
+    def __init__(self, total=False):
+        self.total = total
+
+    def __call__(self, func):
+        @functools.wraps(func)
+        def wrapper(*args, **kwargs):
+            start_time = time.time()
+            ret = func(*args, **kwargs)
+            duration = time.time() - start_time
+
+            if hasattr(func, '__name__'):
+                func_name = func.__name__
+            else:
+                func_name = 'function in module {}'.format(func.__module__)
+            if self.total:
+                _total_timer.add(func_name, duration)
+            else:
+                logger.info('Duration for `{}\': {}'.format(
+                    func_name, duration))
+            return ret
+        return wrapper
+