def configure_general_logging():
""" Configure the general server logger to output to screen if a TTY is
attached, and server.log always.
Logs are output to a directory configurable via LOGDIR.
"""
# import here to avoid import loop from config.py
from .config import config
assert 'LOGDIR' in config, "LOGDIR not specified in config file"
assert 'LOGLEVEL' in config, "LOGLEVEL not specified in config file"
mkdirp(config['LOGDIR'])
# configure properties that should cascade
inbox_root_logger = logging.getLogger('inbox.server')
inbox_root_logger.setLevel(int(config['LOGLEVEL']))
# don't pass messages up to the root root logger
inbox_root_logger.propagate = False
# log everything to screen
if sys.stdout.isatty():
inbox_root_logger.addHandler(get_tty_handler())
logger = get_logger()
for handler in logger.handlers:
logger.removeHandler(handler)
logfile = os.path.join(config['LOGDIR'], 'server.log')
file_handler = logging.FileHandler(logfile, encoding='utf-8')
file_handler.setFormatter(file_formatter)
logger.addHandler(file_handler)
return logger
def configure_general_logging():
"""
Configure the general server logger to output to screen and server.log.
Logs are output to a directory configurable via LOGDIR.
"""
logdir = config.get_required('LOGDIR')
loglevel = config.get_required('LOGLEVEL')
mkdirp(logdir)
# configure properties that should cascade
inbox_root_logger = logging.getLogger('inbox')
inbox_root_logger.setLevel(int(loglevel))
# don't pass messages up to the root root logger
inbox_root_logger.propagate = False
# log everything to screen (or main logfile if redirecting)
inbox_root_logger.addHandler(get_tty_handler())
logger = get_logger()
for handler in logger.handlers:
logger.removeHandler(handler)
logfile = os.path.join(logdir, 'server.log')
file_handler = logging.FileHandler(logfile, encoding='utf-8')
file_handler.setFormatter(file_formatter)
logger.addHandler(file_handler)
return logger
def log(request, config):
"""
Returns root server logger. For others loggers, use this fixture
for setup but then call inbox.log.get_logger().
Testing log file is removed at the end of the test run!
"""
import logging
from inbox.util.file import mkdirp
root_logger = logging.getLogger()
for handler in root_logger.handlers:
root_logger.removeHandler(handler)
logdir = config.get_required('LOGDIR')
mkdirp(logdir)
logfile = config.get_required('TEST_LOGFILE')
fileHandler = logging.FileHandler(logfile, encoding='utf-8')
root_logger.addHandler(fileHandler)
root_logger.setLevel(logging.DEBUG)
def remove_logs():
try:
os.remove(logfile)
except OSError:
pass
request.addfinalizer(remove_logs)
def log(request, config):
"""
Returns root server logger. For others loggers, use this fixture
for setup but then call inbox.log.get_logger().
Testing log file is removed at the end of the test run!
"""
import logging
from inbox.util.file import mkdirp
root_logger = logging.getLogger()
for handler in root_logger.handlers:
root_logger.removeHandler(handler)
logdir = config.get_required('LOGDIR')
mkdirp(logdir)
logfile = config.get_required('TEST_LOGFILE')
fileHandler = logging.FileHandler(logfile, encoding='utf-8')
root_logger.addHandler(fileHandler)
root_logger.setLevel(logging.DEBUG)
def remove_logs():
try:
os.remove(logfile)
except OSError:
pass
request.addfinalizer(remove_logs)
def set_cache(key, val):
path = _path_from_key(key)
dirname = os.path.dirname(path)
mkdirp(dirname)
log.info("Saving cache to {0}".format(dirname))
with open(path, 'w') as f:
msgpack.pack(val, f)
def set_cache(key, val):
path = _path_from_key(key)
dirname = os.path.dirname(path)
mkdirp(dirname)
log.info("Saving cache to {0}".format(dirname))
with open(path, 'w') as f:
msgpack.pack(val, f)
def configure_general_logging():
""" Configure the general server logger to output to screen if a TTY is
attached, and server.log always.
Logs are output to a directory configurable via LOGDIR.
"""
# import here to avoid import loop from config.py
from .config import config
assert 'LOGDIR' in config, "LOGDIR not specified in config file"
assert 'LOGLEVEL' in config, "LOGLEVEL not specified in config file"
mkdirp(config['LOGDIR'])
# configure properties that should cascade
inbox_root_logger = logging.getLogger('inbox.server')
inbox_root_logger.setLevel(int(config['LOGLEVEL']))
# don't pass messages up to the root root logger
inbox_root_logger.propagate = False
# log everything to screen
if sys.stdout.isatty():
inbox_root_logger.addHandler(get_tty_handler())
logger = get_logger()
for handler in logger.handlers:
logger.removeHandler(handler)
logfile = os.path.join(config['LOGDIR'], 'server.log')
file_handler = logging.FileHandler(logfile, encoding='utf-8')
file_handler.setFormatter(file_formatter)
logger.addHandler(file_handler)
return logger
def _keyfile(self, create_dir=True):
assert self.key
assert KEY_DIR
if create_dir:
mkdirp(KEY_DIR)
key_filename = '{0}'.format(sha256(self.key).hexdigest())
return os.path.join(KEY_DIR, key_filename)
def _keyfile(self, create_dir=True):
assert self.key
assert KEY_DIR
if create_dir:
mkdirp(KEY_DIR)
key_filename = "{0}".format(sha256(self.key).hexdigest())
return os.path.join(KEY_DIR, key_filename)
def _keyfile(self, create_dir=True):
assert self.key
key_dir = config.get('KEY_DIR', None)
assert key_dir
if create_dir:
mkdirp(key_dir)
key_filename = '{0}'.format(sha256(self.key).hexdigest())
return os.path.join(key_dir, key_filename)
def _get_errfilename(account_id, folder_name, uid):
try:
errdir = os.path.join(config['LOGDIR'], str(account_id), 'errors',
folder_name)
errfile = os.path.join(errdir, str(uid))
mkdirp(errdir)
except UnicodeEncodeError:
# Rather than wrangling character encodings, just base64-encode the
# folder name to construct a directory.
b64_folder_name = base64.b64encode(folder_name.encode('utf-8'))
return _get_errfilename(account_id, b64_folder_name, uid)
return errfile
def _get_errfilename(account_id, folder_name, uid):
try:
errdir = os.path.join(config['LOGDIR'], str(account_id), 'errors',
folder_name)
errfile = os.path.join(errdir, str(uid))
mkdirp(errdir)
except UnicodeEncodeError:
# Rather than wrangling character encodings, just base64-encode the
# folder name to construct a directory.
b64_folder_name = base64.b64encode(folder_name.encode('utf-8'))
return _get_errfilename(account_id, b64_folder_name, uid)
return errfile
def configure_logging(account_id, purpose):
logger = get_logger(account_id, purpose)
logger.propagate = True
logdir = os.path.join(config['LOGDIR'], str(account_id))
mkdirp(logdir)
logfile = os.path.join(logdir, '{0}.log'.format(purpose))
handler = logging.FileHandler(logfile, encoding='utf-8')
handler.setFormatter(file_formatter)
logger.addHandler(handler)
return logger
def configure_logging(account_id, purpose):
# avoid import loop from config.py
from .config import config
logger = get_logger(account_id, purpose)
logger.propagate = True
logdir = os.path.join(config['LOGDIR'], str(account_id))
mkdirp(logdir)
logfile = os.path.join(logdir, '{0}.log'.format(purpose))
handler = logging.FileHandler(logfile, encoding='utf-8')
handler.setFormatter(file_formatter)
logger.addHandler(handler)
return logger
def save_to_blockstore(data_sha256, data):
assert data is not None
assert type(data) is not unicode
if len(data) == 0:
log.warning('Not saving 0-length data blob')
return
if STORE_MSG_ON_S3:
_save_to_s3(data_sha256, data)
else:
directory = _data_file_directory(data_sha256)
mkdirp(directory)
with open(_data_file_path(data_sha256), 'wb') as f:
f.write(data)
def save_to_blockstore(data_sha256, data):
assert data is not None
assert type(data) is not unicode
if len(data) == 0:
log.warning('Not saving 0-length data blob')
return
if STORE_MSG_ON_S3:
_save_to_s3(data_sha256, data)
else:
directory = _data_file_directory(data_sha256)
mkdirp(directory)
with open(_data_file_path(data_sha256), 'wb') as f:
f.write(data)
def save_to_blockstore(data_sha256, data):
# type: (str, bytes) -> None
assert data is not None
assert isinstance(data, bytes)
if len(data) == 0:
log.warning("Not saving 0-length data blob")
return
if STORE_MSG_ON_S3:
_save_to_s3(data_sha256, data)
else:
directory = _data_file_directory(data_sha256)
mkdirp(directory)
with open(_data_file_path(data_sha256), "wb") as f:
f.write(data)
def save_to_blockstore(data_sha256, data):
# this code is save attachments unnecessary to the disk so disabled
return
assert data is not None
assert type(data) is not unicode
if len(data) == 0:
log.warning('Not saving 0-length data blob')
return
if STORE_MSG_ON_S3:
_save_to_s3(data_sha256, data)
else:
directory = _data_file_directory(data_sha256)
mkdirp(directory)
with open(_data_file_path(data_sha256), 'wb') as f:
f.write(data)
def _get_errfilename(account_id, folder_name, uid):
errdir = os.path.join(config['LOGDIR'], str(account_id), 'errors',
folder_name)
errfile = os.path.join(errdir, str(uid))
mkdirp(errdir)
return errfile
def _save_to_disk(self, data):
mkdirp(self._data_file_directory)
with open(self._data_file_path, 'wb') as f:
f.write(data)
def _save_to_disk(self, data):
mkdirp(self._data_file_directory)
with open(self._data_file_path, 'wb') as f:
f.write(data)
def _save_to_disk(self, data):
directory = _data_file_directory(self.data_sha256)
mkdirp(directory)
with open(_data_file_path(self.data_sha256), 'wb') as f:
f.write(data)
def calculate_sanitized_body(self):
plain_part, html_part = self.body
snippet_length = 191
if html_part:
assert '\r' not in html_part, "newlines not normalized"
# Rudimentary stripping out quoted text in 'gmail_quote' div
# Wrap this in a try/catch because sometimes BeautifulSoup goes
# down a dark spiral of recursion death
try:
soup = BeautifulSoup(html_part.strip(), "lxml")
for div in soup.findAll('div', 'gmail_quote'):
div.extract()
for container in soup.findAll('div', 'gmail_extra'):
if container.contents is not None:
for tag in reversed(container.contents):
if not hasattr(tag, 'name') or tag.name != 'br':
break
else:
tag.extract()
if container.contents is None:
# we emptied it!
container.extract()
# Paragraphs don't need trailing line-breaks.
for container in soup.findAll('p'):
if container.contents is not None:
for tag in reversed(container.contents):
if not hasattr(tag, 'name') or tag.name != 'br':
break
else:
tag.extract()
# Misc other crap.
dtd = [item for item in soup.contents if isinstance(
item, Doctype)]
comments = soup.findAll(text=lambda text: isinstance(
text, Comment))
for tag in chain(dtd, comments):
tag.extract()
self.sanitized_body = unicode(soup)
# trim for snippet
for tag in soup.findAll(['style', 'head', 'title']):
tag.extract()
self.snippet = soup.get_text(' ')[:191]
except RuntimeError as exc:
err_prefix = 'maximum recursion depth exceeded'
# e.message is deprecated in Python 3
if exc.args[0].startswith(err_prefix):
full_traceback = 'Ignoring error: {}\nOuter stack:\n{}{}'\
.format(exc, ''.join(traceback.format_stack()[:-2]),
traceback.format_exc(exc))
# Note that python doesn't support tail call recursion
# optimizations
# http://neopythonic.blogspot.com/2009/04/tail-recursion-elimination.html
full_traceback = 'Error in BeautifulSoup.' + \
'System recursion limit: {0}'.format(
sys.getrecursionlimit()) + \
'\n\n\n' + \
full_traceback
# TODO have a better logging service for storing these
errdir = os.path.join(config['LOGDIR'],
'bs_parsing_errors', )
errfile = os.path.join(errdir, str(self.data_sha256))
mkdirp(errdir)
with open("{0}_traceback".format(errfile), 'w') as fh:
fh.write(full_traceback)
# Write the file in binary mode, since it might also have
# decoding errors.
with open("{0}_data".format(errfile), 'wb') as fh:
fh.write(html_part.encode("utf-8"))
log.error("BeautifulSoup parsing error. Data logged to\
{0}_data and {0}_traceback".format(errfile))
self.decode_error = True
# Not sanitized, but will still work
self.sanitized_body = html_part
self.snippet = soup.get_text(' ')[:191]
else:
log.error("Unknown BeautifulSoup exception: {0}".format(
exc))
raise exc
elif plain_part is None:
self.sanitized_body = u''
self.snippet = u''
else:
stripped = strip_plaintext_quote(plain_part.strip())
self.sanitized_body = plaintext2html(stripped)
self.snippet = stripped[:snippet_length]
def _get_errfilename(account_id, folder_name, uid):
errdir = os.path.join(config['LOGDIR'], str(account_id), 'errors',
folder_name)
errfile = os.path.join(errdir, str(uid))
mkdirp(errdir)
return errfile
def _save_to_disk(data_sha256, data):
directory = _data_file_directory(data_sha256)
mkdirp(directory)
with open(_data_file_path(data_sha256), 'wb') as f:
f.write(data)
