Backport decode_string from develop

This commit is contained in:
Calum Lind 2014-07-21 11:22:57 +01:00
parent 4bd4c78969
commit 54eb28a097
4 changed files with 42 additions and 24 deletions

View File

@ -626,37 +626,55 @@ def xml_encode(string):
def decode_string(s, encoding="utf8"):
"""
Decodes a string and re-encodes it in utf8. If it cannot decode using
`:param:encoding` then it will try to detect the string encoding and
decode it.
Decodes a string and return unicode. If it cannot decode using
`:param:encoding` then it will try latin1, and if that fails,
try to detect the string encoding. If that fails, decode with
ignore.
:param s: string to decode
:type s: string
:keyword encoding: the encoding to use in the decoding
:type encoding: string
:returns: s converted to unicode
:rtype: unicode
"""
try:
s = s.decode(encoding).encode("utf8", "ignore")
except UnicodeDecodeError:
s = s.decode(chardet.detect(s)["encoding"], "ignore").encode("utf8", "ignore")
if not s:
return u''
elif isinstance(s, unicode):
return s
def utf8_encoded(s):
encodings = [lambda: ("utf8", 'strict'),
lambda: ("iso-8859-1", 'strict'),
lambda: (chardet.detect(s)["encoding"], 'strict'),
lambda: (encoding, 'ignore')]
if not encoding is "utf8":
encodings.insert(0, lambda: (encoding, 'strict'))
for l in encodings:
try:
return s.decode(*l())
except UnicodeDecodeError:
pass
return u''
def utf8_encoded(s, encoding="utf8"):
"""
Returns a utf8 encoded string of s
:param s: (unicode) string to (re-)encode
:type s: basestring
:keyword encoding: the encoding to use in the decoding
:type encoding: string
:returns: a utf8 encoded string of s
:rtype: str
"""
if isinstance(s, str):
s = decode_string(s)
s = decode_string(s, encoding).encode("utf8")
elif isinstance(s, unicode):
s = s.encode("utf8", "ignore")
s = s.encode("utf8")
return s
class VersionSplit(object):

View File

@ -389,7 +389,7 @@ class Torrent(object):
# First we check for an error from libtorrent, and set the state to that
# if any occurred.
status_error = self.handle.status().error
status_error = deluge.common.decode_string(self.handle.status().error)
if status_error or self.error_statusmsg:
# This is an error'd torrent
self.state = "Error"

View File

@ -55,7 +55,7 @@ from deluge.configmanager import ConfigManager, get_config_dir
from deluge.core.torrent import Torrent
from deluge.core.torrent import TorrentOptions
import deluge.core.oldstateupgrader
from deluge.common import utf8_encoded
from deluge.common import utf8_encoded, decode_string
from deluge.log import LOG as log
@ -972,7 +972,7 @@ class TorrentManager(component.Component):
torrent.update_state()
def on_alert_tracker_reply(self, alert):
log.debug("on_alert_tracker_reply: %s", alert.message().decode("utf8"))
log.debug("on_alert_tracker_reply: %s", decode_string(alert.message()))
try:
torrent = self.torrents[str(alert.handle.info_hash())]
except:
@ -1004,7 +1004,7 @@ class TorrentManager(component.Component):
torrent = self.torrents[str(alert.handle.info_hash())]
except:
return
tracker_status = '%s: %s' % (_("Warning"), str(alert.message()))
tracker_status = '%s: %s' % (_("Warning"), decode_string(alert.message()))
# Set the tracker status for the torrent
torrent.set_tracker_status(tracker_status)
@ -1034,7 +1034,7 @@ class TorrentManager(component.Component):
def on_alert_storage_moved_failed(self, alert):
"""Alert handler for libtorrent storage_moved_failed_alert"""
log.debug("on_alert_storage_moved_failed: %s", alert.message())
log.debug("on_alert_storage_moved_failed: %s", decode_string(alert.message()))
try:
torrent_id = str(alert.handle.info_hash())
torrent = self.torrents[torrent_id]
@ -1097,7 +1097,7 @@ class TorrentManager(component.Component):
self.save_resume_data_file()
def on_alert_save_resume_data_failed(self, alert):
log.debug("on_alert_save_resume_data_failed: %s", alert.message())
log.debug("on_alert_save_resume_data_failed: %s", decode_string(alert.message()))
try:
torrent = self.torrents[str(alert.handle.info_hash())]
except:
@ -1111,7 +1111,7 @@ class TorrentManager(component.Component):
def on_alert_file_renamed(self, alert):
log.debug("on_alert_file_renamed")
log.debug("index: %s name: %s", alert.index, alert.name.decode("utf8"))
log.debug("index: %s name: %s", alert.index, decode_string(alert.name))
try:
torrent = self.torrents[str(alert.handle.info_hash())]
torrent_id = str(alert.handle.info_hash())
@ -1149,7 +1149,7 @@ class TorrentManager(component.Component):
torrent.on_metadata_received()
def on_alert_file_error(self, alert):
log.debug("on_alert_file_error: %s", alert.message())
log.debug("on_alert_file_error: %s", decode_string(alert.message()))
try:
torrent = self.torrents[str(alert.handle.info_hash())]
except:
@ -1157,7 +1157,7 @@ class TorrentManager(component.Component):
torrent.update_state()
def on_alert_file_completed(self, alert):
log.debug("file_completed_alert: %s", alert.message())
log.debug("file_completed_alert: %s", decode_string(alert.message()))
try:
torrent_id = str(alert.handle.info_hash())
except:

View File

@ -50,7 +50,7 @@ except ImportError:
from sha import sha
from deluge import bencode
from deluge.common import decode_string, path_join
from deluge.common import utf8_encoded, path_join
from deluge.log import LOG as log
import deluge.configmanager
@ -86,9 +86,9 @@ class TorrentInfo(object):
# Check if 'name.utf-8' is in the torrent and if not try to decode the string
# using the encoding found.
if "name.utf-8" in self.__m_metadata["info"]:
self.__m_name = decode_string(self.__m_metadata["info"]["name.utf-8"])
self.__m_name = utf8_encoded(self.__m_metadata["info"]["name.utf-8"])
else:
self.__m_name = decode_string(self.__m_metadata["info"]["name"], self.encoding)
self.__m_name = utf8_encoded(self.__m_metadata["info"]["name"], self.encoding)
# Get list of files from torrent info
paths = {}
@ -103,7 +103,7 @@ class TorrentInfo(object):
path = os.path.join(prefix, *f["path.utf-8"])
del f["path.utf-8"]
else:
path = decode_string(os.path.join(prefix, decode_string(os.path.join(*f["path"]), self.encoding)), self.encoding)
path = utf8_encoded(os.path.join(prefix, utf8_encoded(os.path.join(*f["path"]), self.encoding)), self.encoding)
f["path"] = path
f["index"] = index
if "sha1" in f and len(f["sha1"]) == 20: