tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

karaoke files: try to decode non-ascii text

Browse Source
This commit is contained in:
Jean-Francois Dockes 2011-01-29 12:15:51 +01:00
parent e8fcd35fef
commit b528289a09
1 changed files with 107 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

130
src/filters/rclkar
View File

@ -5,6 +5,9 @@
import rclexecm
import sys
import os.path
import string
import re
import codecs
try:
import midi
@ -12,7 +15,7 @@ except:
print "RECFILTERROR HELPERNOTFOUND python:midi"
sys.exit(1);
# prototype for the html document we're returning
# Prototype for the html document we're returning
htmltemplate = '''
<html>
<head>
@ -28,25 +31,103 @@ htmltemplate = '''
'''
class KarTextExtractor:
# Afaik, the only charset encodings with null bytes are variations on
# utf-16 and utf-32 and iso relatives. A hopefully comprehensive
# list follows, compiled from iconv and python values. This is used for
# stripping garbage from some files.
acceptnullencodings = \
set(('csucs4', 'csunicode', 'csunicode11', 'iso-10646-ucs-2',
'iso-10646-ucs-4', 'u16', 'u32', 'ucs-2', 'ucs-2-internal',
'ucs-2-swapped', 'ucs-2be', 'ucs-2le', 'ucs-4', 'ucs-4-internal',
'ucs-4-swapped', 'ucs-4be', 'ucs-4le', 'unicode-1-1', 'unicodebig',
'unicodelittle', 'utf-16', 'utf-16be', 'utf-16le', 'utf-32',
'utf-32be', 'utf-32le', 'utf16', 'utf32', 'utf_16', 'utf_16_be',
'utf_16_le', 'utf_32', 'utf_32_be', 'utf_32_le'))
def __init__(self, em):
self.em = em
self.currentindex = 0
self.encoding = ""
self.defaultencoding = ""
self.acceptnulls = False
try:
self.defaultencoding = sys.getfilesystemencoding()
except:
pass
if self.defaultencoding is None:
self.defaultencoding = sys.getdefaultencoding()
if not self.defaultencoding or \
self.defaultencoding.lower().find('ascii') != -1:
self.defaultencoding = 'latin_1'
try:
codecs.lookup(self.defaultencoding)
except:
self.defaultencoding = 'latin_1'
# Try to decode input binary string then encode to utf-8 for output
def reencode(self, data):
text = ""
if not data:
return text
# Some files have garbage data after a null byte.
if not self.acceptnulls:
firstnull = data.find(chr(0))
if firstnull != -1:
data = data[0 : firstnull]
try:
text = data.decode(self.encoding, 'ignore')
except Exception, err:
self.em.rclog("Decode failed: " + str(err))
return ""
try:
text = text.encode('utf-8')
except Exception, err:
self.em.rclog("Encode failed: " + str(err))
return ""
text = self.em.htmlescape(text).replace("\n", "<br>\n")
return text
# Some karaoke files have the encoding as part of the file name
# as 'some title (encoding).xxx' Not sure the whitespace before
# the '(' has to be there, so not relying on this
def encodingfromfilename(self, fn):
rexp = r'\(([^\)]+)\)\.[a-zA-Z]+$'
m = re.search(rexp, fn)
if m:
return m.group(1)
else:
return ""
def extractone(self, params):
docdata = ""
ok = False
if not params.has_key("filename:"):
self.em.rclog("extractone: no mime or file name")
return (ok, docdata, "", rclexecm.RclExecM.eofnow)
filename = params["filename:"]
self.encoding = self.encodingfromfilename(filename)
try:
codecs.lookup(self.encoding)
except:
self.em.rclog("Encoding [%s] not found, defaulting to [%s]" % \
(self.encoding, self.defaultencoding))
self.encoding = self.defaultencoding
self.acceptnulls = self.encoding.lower() in self.acceptnullencodings
# Mimetype not used for now
if not params.has_key("mimetype:"):
mimetype = 'audio/x-midi'
else:
mimetype = params["mimetype:"]
if not params.has_key("filename:"):
self.em.rclog("extractone: no mime or file name")
return (ok, docdata, "", rclexecm.RclExecM.eofnow)
filename = params["filename:"]
try:
stream = midi.read_midifile(filename)
except Exception, err:
@ -57,36 +138,46 @@ class KarTextExtractor:
author = None
language = None
lyrics = ""
for event in stream.iterevents():
edata = ""
if isinstance(event, midi.TextMetaEvent):
if not event.data:
continue
elif event.data[0] == '/' or event.data[0] == '\\':
lyrics += "\n" + event.data[1:]
edata += "\n" + event.data[1:]
elif event.data[0] == '[' or event.data[0] == ']':
lyrics += event.data[1:]
edata += event.data[1:]
elif event.data[0] == '@':
if len(event.data) == 1:
continue
else:
if event.data[1] == 'I':
lyrics += event.data[2:] + '\n'
edata += event.data[2:] + '\n'
elif event.data[1] == 'L':
language = event.data[2:]
language = self.reencode(event.data[2:])
elif event.data[1] == 'T':
if title is None:
title = event.data[2:]
title = self.reencode(event.data[2:])
elif author is None:
author = event.data[2:]
author = self.reencode(event.data[2:])
else:
lyrics += event.data
elif isinstance(event, midi.LryricsEvent):
edata += event.data
elif isinstance(event, midi.LryricsEvent) or \
isinstance(event, midi.TrackNameEvent):
space = ""
if isinstance(event, midi.TrackNameEvent):
nl = "\n"
if not event.data:
continue
elif event.data[0] == '/' or event.data[0] == '\\':
lyrics += "\n" + event.data[1:]
edata += "\n" + event.data[1:] + nl
else:
lyrics += event.data
edata += event.data + nl
lyrics += self.reencode(edata)
if title is None:
title = ""
if author is None:
@ -94,13 +185,6 @@ class KarTextExtractor:
if language is None:
language = ""
if lyrics != "":
try:
lyrics = self.em.htmlescape(lyrics.encode("utf-8"))
lyrics = lyrics.replace("\n", "<br>")
except Exception, err:
print "ENCODE FAILED", err
lyrics = ""
self.em.setmimetype("text/html")
docdata = htmltemplate % (title, author, language, lyrics)