CC-430: Audio normalization (Replaygain Support)

-Added support on media-monitor's side
This commit is contained in:
Martin Konecny 2012-07-05 21:54:44 -04:00
parent 1cbb0345b3
commit a687e48d80
7 changed files with 195 additions and 116 deletions

View File

@ -5,11 +5,8 @@ import time
import pyinotify
import shutil
from subprocess import Popen, PIPE
from api_clients import api_client
class AirtimeMediaMonitorBootstrap():
"""AirtimeMediaMonitorBootstrap constructor
Keyword Arguments:
@ -29,11 +26,11 @@ class AirtimeMediaMonitorBootstrap():
self.curr_mtab_file = "/var/tmp/airtime/media-monitor/currMtab"
self.logger.info("Adding %s on watch list...", self.mount_file)
self.wm.add_watch(self.mount_file, pyinotify.ALL_EVENTS, rec=False, auto_add=False)
tmp_dir = os.path.dirname(self.curr_mtab_file)
if not os.path.exists(tmp_dir):
os.makedirs(tmp_dir)
# create currMtab file if it's the first time
if not os.path.exists(self.curr_mtab_file):
shutil.copy('/etc/mtab', self.curr_mtab_file)
@ -77,10 +74,10 @@ class AirtimeMediaMonitorBootstrap():
dir -- pathname of the directory
"""
def sync_database_to_filesystem(self, dir_id, dir):
dir = os.path.normpath(dir)+"/"
dir = os.path.normpath(dir) + "/"
"""
set to hold new and/or modified files. We use a set to make it ok if files are added
twice. This is because some of the tests for new files return result sets that are not
@ -109,13 +106,13 @@ class AirtimeMediaMonitorBootstrap():
if os.path.exists(self.mmc.timestamp_file):
"""find files that have been modified since the last time media-monitor process started."""
time_diff_sec = time.time() - os.path.getmtime(self.mmc.timestamp_file)
command = "find '%s' -iname '*.ogg' -o -iname '*.mp3' -type f -readable -mmin -%d" % (dir, time_diff_sec/60+1)
command = "find '%s' -iname '*.ogg' -o -iname '*.mp3' -type f -readable -mmin -%d" % (dir, time_diff_sec / 60 + 1)
else:
command = "find '%s' -iname '*.ogg' -o -iname '*.mp3' -type f -readable" % dir
self.logger.debug(command)
stdout = self.mmc.exec_command(command)
if stdout is None:
self.logger.error("Unrecoverable error when syncing db to filesystem.")
return

View File

@ -1,13 +1,14 @@
# -*- coding: utf-8 -*-
import replay_gain
import os
import hashlib
import mutagen
import logging
import math
import re
import traceback
from api_clients import api_client
"""
list of supported easy tags in mutagen version 1.20
@ -18,39 +19,39 @@ class AirtimeMetadata:
def __init__(self):
self.airtime2mutagen = {\
"MDATA_KEY_TITLE": "title",\
"MDATA_KEY_CREATOR": "artist",\
"MDATA_KEY_SOURCE": "album",\
"MDATA_KEY_GENRE": "genre",\
"MDATA_KEY_MOOD": "mood",\
"MDATA_KEY_TRACKNUMBER": "tracknumber",\
"MDATA_KEY_BPM": "bpm",\
"MDATA_KEY_LABEL": "organization",\
"MDATA_KEY_COMPOSER": "composer",\
"MDATA_KEY_ENCODER": "encodedby",\
"MDATA_KEY_CONDUCTOR": "conductor",\
"MDATA_KEY_YEAR": "date",\
"MDATA_KEY_URL": "website",\
"MDATA_KEY_ISRC": "isrc",\
"MDATA_KEY_COPYRIGHT": "copyright",\
"MDATA_KEY_TITLE": "title", \
"MDATA_KEY_CREATOR": "artist", \
"MDATA_KEY_SOURCE": "album", \
"MDATA_KEY_GENRE": "genre", \
"MDATA_KEY_MOOD": "mood", \
"MDATA_KEY_TRACKNUMBER": "tracknumber", \
"MDATA_KEY_BPM": "bpm", \
"MDATA_KEY_LABEL": "organization", \
"MDATA_KEY_COMPOSER": "composer", \
"MDATA_KEY_ENCODER": "encodedby", \
"MDATA_KEY_CONDUCTOR": "conductor", \
"MDATA_KEY_YEAR": "date", \
"MDATA_KEY_URL": "website", \
"MDATA_KEY_ISRC": "isrc", \
"MDATA_KEY_COPYRIGHT": "copyright", \
}
self.mutagen2airtime = {\
"title": "MDATA_KEY_TITLE",\
"artist": "MDATA_KEY_CREATOR",\
"album": "MDATA_KEY_SOURCE",\
"genre": "MDATA_KEY_GENRE",\
"mood": "MDATA_KEY_MOOD",\
"tracknumber": "MDATA_KEY_TRACKNUMBER",\
"bpm": "MDATA_KEY_BPM",\
"organization": "MDATA_KEY_LABEL",\
"composer": "MDATA_KEY_COMPOSER",\
"encodedby": "MDATA_KEY_ENCODER",\
"conductor": "MDATA_KEY_CONDUCTOR",\
"date": "MDATA_KEY_YEAR",\
"website": "MDATA_KEY_URL",\
"isrc": "MDATA_KEY_ISRC",\
"copyright": "MDATA_KEY_COPYRIGHT",\
"title": "MDATA_KEY_TITLE", \
"artist": "MDATA_KEY_CREATOR", \
"album": "MDATA_KEY_SOURCE", \
"genre": "MDATA_KEY_GENRE", \
"mood": "MDATA_KEY_MOOD", \
"tracknumber": "MDATA_KEY_TRACKNUMBER", \
"bpm": "MDATA_KEY_BPM", \
"organization": "MDATA_KEY_LABEL", \
"composer": "MDATA_KEY_COMPOSER", \
"encodedby": "MDATA_KEY_ENCODER", \
"conductor": "MDATA_KEY_CONDUCTOR", \
"date": "MDATA_KEY_YEAR", \
"website": "MDATA_KEY_URL", \
"isrc": "MDATA_KEY_ISRC", \
"copyright": "MDATA_KEY_COPYRIGHT", \
}
self.logger = logging.getLogger()
@ -67,9 +68,9 @@ class AirtimeMetadata:
## return format hh:mm:ss.uuu
def format_length(self, mutagen_length):
t = float(mutagen_length)
h = int(math.floor(t/3600))
h = int(math.floor(t / 3600))
t = t % 3600
m = int(math.floor(t/60))
m = int(math.floor(t / 60))
s = t % 60
# will be ss.uuu
@ -94,12 +95,12 @@ class AirtimeMetadata:
for key in m:
if key in self.airtime2mutagen:
value = m[key]
if value is not None:
value = unicode(value)
else:
value = unicode('');
#if len(value) > 0:
self.logger.debug("Saving key '%s' with value '%s' to file", key, value)
airtime_file[self.airtime2mutagen[key]] = value
@ -123,8 +124,14 @@ class AirtimeMetadata:
self.logger.info("getting info from filepath %s", filepath)
md = {}
replay_gain_val = replay_gain.calculate_replay_gain(filepath)
self.logger.info('ReplayGain calculated as %s for %s' % (replay_gain_val, filepath))
md['MDATA_KEY_REPLAYGAIN'] = replay_gain_val
try:
md = {}
md5 = self.get_md5(filepath)
md['MDATA_KEY_MD5'] = md5
@ -135,6 +142,7 @@ class AirtimeMetadata:
self.logger.error("Exception %s", e)
return None
#check if file has any metadata
if file_info is None:
return None
@ -149,6 +157,7 @@ class AirtimeMetadata:
except Exception, e:
self.logger.error('Exception: %s', e)
self.logger.error("traceback: %s", traceback.format_exc())
if 'MDATA_KEY_TITLE' not in md:
#get rid of file extension from original name, name might have more than 1 '.' in it.
original_name = os.path.basename(filepath)
@ -165,8 +174,6 @@ class AirtimeMetadata:
pass
if isinstance(md['MDATA_KEY_TRACKNUMBER'], basestring):
match = re.search('^(\d*/\d*)?', md['MDATA_KEY_TRACKNUMBER'])
try:
md['MDATA_KEY_TRACKNUMBER'] = int(md['MDATA_KEY_TRACKNUMBER'].split("/")[0], 10)
except Exception, e:
@ -223,7 +230,7 @@ class AirtimeMetadata:
md['MDATA_KEY_BITRATE'] = getattr(file_info.info, "bitrate", None)
md['MDATA_KEY_SAMPLERATE'] = getattr(file_info.info, "sample_rate", None)
self.logger.info( "Bitrate: %s , Samplerate: %s", md['MDATA_KEY_BITRATE'], md['MDATA_KEY_SAMPLERATE'] )
self.logger.info("Bitrate: %s , Samplerate: %s", md['MDATA_KEY_BITRATE'], md['MDATA_KEY_SAMPLERATE'])
try: md['MDATA_KEY_DURATION'] = self.format_length(file_info.info.length)
except Exception as e: self.logger.warn("File: '%s' raises: %s", filepath, str(e))

View File

@ -8,13 +8,11 @@ import traceback
# For RabbitMQ
from kombu.connection import BrokerConnection
from kombu.messaging import Exchange, Queue, Consumer, Producer
from kombu.messaging import Exchange, Queue, Consumer
import pyinotify
from pyinotify import Notifier
#from api_clients import api_client
from api_clients import api_client
from airtimemetadata import AirtimeMetadata
class AirtimeNotifier(Notifier):
@ -153,7 +151,6 @@ class AirtimeNotifier(Notifier):
md.update(file_md)
else:
file_md = None
data = None
if (os.path.exists(filepath) and (mode == self.config.MODE_CREATE)):
if file_md is None:
@ -184,7 +181,7 @@ class AirtimeNotifier(Notifier):
self.api_client.update_media_metadata(md, mode)
elif (mode == self.config.MODE_DELETE):
self.api_client.update_media_metadata(md, mode)
elif (mode == self.config.MODE_DELETE_DIR):
self.api_client.update_media_metadata(md, mode)

View File

@ -9,7 +9,6 @@ import difflib
import traceback
from subprocess import Popen, PIPE
import pyinotify
from pyinotify import ProcessEvent
from airtimemetadata import AirtimeMetadata
@ -59,8 +58,8 @@ class AirtimeProcessEvent(ProcessEvent):
if "-unknown-path" in path:
unknown_path = path
pos = path.find("-unknown-path")
path = path[0:pos]+"/"
path = path[0:pos] + "/"
list = self.api_client.list_all_watched_dirs()
# case where the dir that is being watched is moved to somewhere
if path in list[u'dirs'].values():
@ -81,14 +80,14 @@ class AirtimeProcessEvent(ProcessEvent):
self.logger.info("Removing watch on: %s wd %s", unknown_path, wd)
self.wm.rm_watch(wd, rec=True)
self.file_events.append({'mode': self.config.MODE_DELETE_DIR, 'filepath': path})
def process_IN_DELETE_SELF(self, event):
#we only care about files that have been moved away from imported/ or organize/ dir
if event.path in self.config.problem_directory or event.path in self.config.organize_directory:
return
self.logger.info("event: %s", event)
path = event.path + '/'
if event.dir:
@ -103,7 +102,7 @@ class AirtimeProcessEvent(ProcessEvent):
self.logger.info("%s removed from watch folder list successfully.", path)
else:
self.logger.info("Removing the watch folder failed: %s", res['msg']['error'])
def process_IN_CREATE(self, event):
if event.path in self.mount_file_dir:
return
@ -111,7 +110,7 @@ class AirtimeProcessEvent(ProcessEvent):
if not event.dir:
# record the timestamp of the time on IN_CREATE event
self.create_dict[event.pathname] = time.time()
#event.dir: True if the event was raised against a directory.
#event.name: filename
#event.pathname: pathname (str): Concatenation of 'path' and 'name'.
@ -122,7 +121,7 @@ class AirtimeProcessEvent(ProcessEvent):
return
self.logger.info("event: %s", event)
self.logger.info("create_dict: %s", self.create_dict)
try:
del self.create_dict[event.pathname]
self.handle_created_file(event.dir, event.pathname, event.name)
@ -130,8 +129,8 @@ class AirtimeProcessEvent(ProcessEvent):
pass
#self.logger.warn("%s does not exist in create_dict", event.pathname)
#Uncomment the above warning when we fix CC-3830 for 2.1.1
def handle_created_file(self, dir, pathname, name):
if not dir:
self.logger.debug("PROCESS_IN_CLOSE_WRITE: %s, name: %s, pathname: %s ", dir, name, pathname)
@ -145,12 +144,12 @@ class AirtimeProcessEvent(ProcessEvent):
self.temp_files[pathname] = None
elif self.mmc.is_audio_file(name):
if self.mmc.is_parent_directory(pathname, self.config.organize_directory):
#file was created in /srv/airtime/stor/organize. Need to process and move
#to /srv/airtime/stor/imported
file_md = self.md_manager.get_md_from_file(pathname)
playable = self.mmc.test_file_playability(pathname)
if file_md and playable:
self.mmc.organize_new_file(pathname, file_md)
else:
@ -182,7 +181,7 @@ class AirtimeProcessEvent(ProcessEvent):
if self.mmc.is_audio_file(name):
is_recorded = self.mmc.is_parent_directory(pathname, self.config.recorded_directory)
self.file_events.append({'filepath': pathname, 'mode': self.config.MODE_MODIFY, 'is_recorded_show': is_recorded})
# if change is detected on /etc/mtab, we check what mount(file system) was added/removed
# and act accordingly
def handle_mount_change(self):
@ -192,41 +191,41 @@ class AirtimeProcessEvent(ProcessEvent):
shutil.move(self.curr_mtab_file, self.prev_mtab_file)
# create the file
shutil.copy(self.mount_file, self.curr_mtab_file)
d = difflib.Differ()
curr_fh = open(self.curr_mtab_file, 'r')
prev_fh = open(self.prev_mtab_file, 'r')
diff = list(d.compare(prev_fh.readlines(), curr_fh.readlines()))
added_mount_points = []
removed_mount_points = []
for dir in diff:
info = dir.split(' ')
if info[0] == '+':
added_mount_points.append(info[2])
elif info[0] == '-':
removed_mount_points.append(info[2])
self.logger.info("added: %s", added_mount_points)
self.logger.info("removed: %s", removed_mount_points)
# send current mount information to Airtime
self.api_client.update_file_system_mount(added_mount_points, removed_mount_points);
def handle_watched_dir_missing(self, dir):
self.api_client.handle_watched_dir_missing(dir);
#if a file is moved somewhere, this callback is run. With details about
#where the file is being moved from. The corresponding process_IN_MOVED_TO
#callback is only called if the destination of the file is also in a watched
#directory.
def process_IN_MOVED_FROM(self, event):
#we don't care about files that have been moved from problem_directory
if event.path in self.config.problem_directory:
return
self.logger.info("process_IN_MOVED_FROM: %s", event)
if not event.dir:
if event.pathname in self.temp_files:
@ -241,10 +240,10 @@ class AirtimeProcessEvent(ProcessEvent):
def process_IN_MOVED_TO(self, event):
self.logger.info("process_IN_MOVED_TO: %s", event)
# if /etc/mtab is modified
filename = self.mount_file_dir +"/mtab"
filename = self.mount_file_dir + "/mtab"
if event.pathname in filename:
self.handle_mount_change()
if event.path in self.config.problem_directory:
return
@ -265,15 +264,15 @@ class AirtimeProcessEvent(ProcessEvent):
#to /srv/airtime/stor/imported
file_md = self.md_manager.get_md_from_file(pathname)
playable = self.mmc.test_file_playability(pathname)
if file_md and playable:
filepath = self.mmc.organize_new_file(pathname, file_md)
else:
#move to problem_files
self.mmc.move_to_problem_dir(pathname)
else:
filepath = event.pathname
@ -283,23 +282,23 @@ class AirtimeProcessEvent(ProcessEvent):
#file's original location is from outside an inotify watched dir.
pathname = event.pathname
if self.mmc.is_parent_directory(pathname, self.config.organize_directory):
#file was created in /srv/airtime/stor/organize. Need to process and move
#to /srv/airtime/stor/imported
file_md = self.md_manager.get_md_from_file(pathname)
playable = self.mmc.test_file_playability(pathname)
if file_md and playable:
self.mmc.organize_new_file(pathname, file_md)
else:
#move to problem_files
self.mmc.move_to_problem_dir(pathname)
else:
#show moved from unwatched folder into a watched folder. Do not "organize".
@ -309,33 +308,33 @@ class AirtimeProcessEvent(ProcessEvent):
#When we move a directory into a watched_dir, we only get a notification that the dir was created,
#and no additional information about files that came along with that directory.
#need to scan the entire directory for files.
if event.cookie in self.cookies_IN_MOVED_FROM:
del self.cookies_IN_MOVED_FROM[event.cookie]
mode = self.config.MODE_MOVED
else:
mode = self.config.MODE_CREATE
files = self.mmc.scan_dir_for_new_files(event.pathname)
if self.mmc.is_parent_directory(event.pathname, self.config.organize_directory):
for pathname in files:
#file was created in /srv/airtime/stor/organize. Need to process and move
#to /srv/airtime/stor/imported
file_md = self.md_manager.get_md_from_file(pathname)
playable = self.mmc.test_file_playability(pathname)
if file_md and playable:
self.mmc.organize_new_file(pathname, file_md)
#self.file_events.append({'mode': mode, 'filepath': filepath, 'is_recorded_show': False})
else:
#move to problem_files
self.mmc.move_to_problem_dir(pathname)
else:
for file in files:
self.file_events.append({'mode': mode, 'filepath': file, 'is_recorded_show': False})
@ -368,12 +367,12 @@ class AirtimeProcessEvent(ProcessEvent):
for event in self.file_events:
self.multi_queue.put(event)
self.mmc.touch_index_file()
self.file_events = []
#yield to worker thread
time.sleep(0)
#use items() because we are going to be modifying this
#dictionary while iterating over it.
for k, pair in self.cookies_IN_MOVED_FROM.items():
@ -390,7 +389,7 @@ class AirtimeProcessEvent(ProcessEvent):
#it from the Airtime directory.
del self.cookies_IN_MOVED_FROM[k]
self.handle_removed_file(False, event.pathname)
# we don't want create_dict grow infinitely
# this part is like a garbage collector
for k, t in self.create_dict.items():
@ -404,14 +403,14 @@ class AirtimeProcessEvent(ProcessEvent):
if os.path.exists(k):
# check if file is open
try:
command = "lsof "+k
command = "lsof " + k
#f = os.popen(command)
f = Popen(command, shell=True, stdout=PIPE).stdout
except Exception, e:
self.logger.error('Exception: %s', e)
self.logger.error("traceback: %s", traceback.format_exc())
continue
if not f.readlines():
self.logger.info("Handling file: %s", k)
self.handle_created_file(False, k, os.path.basename(k))

View File

@ -10,7 +10,6 @@ import traceback
from subprocess import Popen, PIPE
from airtimemetadata import AirtimeMetadata
from api_clients import api_client
import pyinotify
class MediaMonitorCommon:
@ -75,7 +74,7 @@ class MediaMonitorCommon:
try:
return self.is_user_readable(item, 'www-data', 'www-data') \
and self.is_user_readable(item, 'pypo', 'pypo')
except Exception, e:
except Exception:
self.logger.warn(u"Failed to check owner/group/permissions for %s", item)
return False
@ -110,7 +109,7 @@ class MediaMonitorCommon:
else:
pathname = dirname
is_dir = True
except Exception, e:
except Exception:
#something went wrong while we were trying to make world readable.
return False
@ -194,7 +193,7 @@ class MediaMonitorCommon:
break
except Exception, e:
self.logger.error("Exception %s", e)
self.logger.error("Exception %s", e)
return filepath
@ -273,7 +272,7 @@ class MediaMonitorCommon:
File name charset encoding is UTF-8.
"""
stdout = stdout.decode("UTF-8")
except Exception, e:
except Exception:
stdout = None
self.logger.error("Could not decode %s using UTF-8" % stdout)

View File

@ -0,0 +1,81 @@
from subprocess import Popen, PIPE
import re
import os
import sys
def get_process_output(command):
"""
Run subprocess and return stdout
"""
p = Popen(command, shell=True, stdout=PIPE)
return p.communicate()[0].strip()
def run_process(command):
"""
Run subprocess and return "return code"
"""
p = Popen(command, shell=True)
return os.waitpid(p.pid, 0)[1]
def get_mime_type(file_path):
"""
Attempts to get the mime type but will return prematurely if the process
takes longer than 5 seconds. Note that this function should only be called
for files which do not have a mp3/ogg/flac extension.
"""
return get_process_output("timeout 5 file -b --mime-type %s" % file_path)
def calculate_replay_gain(file_path):
"""
This function accepts files of type mp3/ogg/flac and returns a calculated ReplayGain value.
If the value cannot be calculated for some reason, then we default to 1.
TODO:
Currently some of the subprocesses called will actually insert metadata into the file itself,
which we do *not* want as this changes the file's hash. Need to make a copy of the file before
we run this function.
http://wiki.hydrogenaudio.org/index.php?title=ReplayGain_1.0_specification
"""
search = None
if re.search(r'mp3$', file_path, re.IGNORECASE) or get_mime_type(file_path) == "audio/mpeg":
if run_process("which mp3gain > /dev/null") == 0:
out = get_process_output('mp3gain -q "%s" 2> /dev/null' % file_path)
search = re.search(r'Recommended "Track" dB change: (.*)', out)
else:
print "mp3gain not found"
#Log warning
elif re.search(r'ogg$', file_path, re.IGNORECASE) or get_mime_type(file_path) == "application/ogg":
if run_process("which vorbisgain > /dev/null && which ogginfo > dev/null") == 0:
run_process('vorbisgain -q -f "%s" 2>/dev/null >/dev/null' % file_path)
out = get_process_output('ogginfo "%s"' % file_path)
search = re.search(r'REPLAYGAIN_TRACK_GAIN=(.*) dB', out)
else:
print "vorbisgain/ogginfo not found"
#Log warning
elif re.search(r'flac$', file_path, re.IGNORECASE) or get_mime_type(file_path) == "audio/x-flac":
if run_process("which metaflac > /dev/null") == 0:
out = get_process_output('metaflac --show-tag=REPLAYGAIN_TRACK_GAIN "%s"' % file_path)
search = re.search(r'REPLAYGAIN_TRACK_GAIN=(.*) dB', out)
else:
print "metaflac not found"
#Log warning
else:
pass
#Log unknown file type.
replay_gain = 1
if search:
matches = search.groups()
if len(matches) == 1:
replay_gain = matches[0]
return replay_gain
# Example of running from command line:
# python replay_gain.py /path/to/filename.mp3
if __name__ == "__main__":
print calculate_replay_gain(sys.argv[1])

View File

@ -1,11 +1,10 @@
# -*- coding: utf-8 -*-
from mediaconfig import AirtimeMediaConfig
import traceback
import os
class MediaMonitorWorkerProcess:
def __init__(self, config, mmc):
self.config = config
self.mmc = mmc