def __parse_and_write_file(self, file, link):
"""
Parses the date+time and writes entry to outputfile
@param file: filename
@param link: path
"""
if TIMESTAMP_REGEX.match(file):
# if we found a timestamp too,take hours,min
# and optionally seconds from this timestamp
timestamp = TIMESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdatetimeiso8601(timestamp)
logging.debug("found timestamp: " + orgdate)
else:
datestamp = DATESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdate(datestamp)
orgdate_time_tupel = OrgFormat.datetupeliso8601(datestamp)
file_datetime = time.localtime(os.path.getmtime(link))
# check if the file - time information matches year,month,day,
# then update time
if file_datetime.tm_year == orgdate_time_tupel.tm_year and \
file_datetime.tm_mon == orgdate_time_tupel.tm_mon and \
file_datetime.tm_mday == orgdate_time_tupel.tm_mday:
logging.debug("found a time in file.setting %s-->%s",
orgdate, OrgFormat.date(file_datetime, True))
orgdate = OrgFormat.date(file_datetime, True)
# write entry to org file
output = OrgFormat.link(link=link, description=file)
# we need optional data for hashing due it can be, that more
# than one file have the same timestamp
properties = OrgProperties(data_for_hashing=output)
self._writer.write_org_subitem(timestamp=orgdate,
output=output,
properties=properties)
def __parse_and_write_file(self, file, link):
"""
Parses the date+time and writes entry to outputfile
@param file: filename
@param link: path
"""
if TIMESTAMP_REGEX.match(file):
# if we found a timestamp too,take hours,min
# and optionally seconds from this timestamp
timestamp = TIMESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdatetimeiso8601(timestamp)
logging.debug("found timestamp: " + orgdate)
else:
datestamp = DATESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdate(datestamp)
orgdate_time_tupel = OrgFormat.datetupeliso8601(datestamp)
file_datetime = time.localtime(os.path.getmtime(link))
# check if the file - time information matches year,month,day,
# then update time
if file_datetime.tm_year == orgdate_time_tupel.tm_year and \
file_datetime.tm_mon == orgdate_time_tupel.tm_mon and \
file_datetime.tm_mday == orgdate_time_tupel.tm_mday:
logging.debug("found a time in file.setting %s-->%s", orgdate,
OrgFormat.date(file_datetime, True))
orgdate = OrgFormat.date(file_datetime, True)
# write entry to org file
output = OrgFormat.link(link=link, description=file)
# we need optional data for hashing due it can be, that more
# than one file have the same timestamp
properties = OrgProperties(data_for_hashing=output)
self._writer.write_org_subitem(timestamp=orgdate,
output=output,
properties=properties)
def _handle_row(self, row):
"""
handle a single row
"""
try:
# assume unix timestamp
if not self._args.timestamp_format:
timestamp = datetime.datetime.fromtimestamp(
int(row[self._args.timestamp_field]))
else:
timestamp = time.strptime(row[self._args.timestamp_field],
self._args.timestamp_format)
# show time with the timestamp format, but only
# if it contains at least hours and minutes
if not self._args.timestamp_format or \
any(x in self._args.timestamp_format for x in ['%H', '%M']):
timestamp = OrgFormat.datetime(timestamp)
else:
timestamp = OrgFormat.date(timestamp)
except ValueError, e:
logging.error("timestamp-format does not match: %s", e)
sys.exit(1)
def __get_datetime(self, mydate):
"""
@return string: Datetime - in Org Format
"""
mydate_tupel = OrgFormat.datetupelutctimestamp(mydate)
return OrgFormat.date(mydate_tupel)
def __handle_file(self, file, rootdir):
"""
handles a file
"""
# don't handle emacs tmp files (file~)
if file[-1:] == '~':
return
link = os.path.join(rootdir, file)
logging.debug(link)
if self._args.force_filedate_extraction:
file_datetime = time.localtime(os.path.getmtime(link))
if self._args.skip_filetime_extraction:
orgdate = OrgFormat.date(file_datetime)
else:
orgdate = OrgFormat.datetime(file_datetime)
self.__write_file(file, link, orgdate)
elif DATESTAMP_REGEX.match(file):
try:
# we put this in a try block because:
# if a timestamp is false i.e. 2011-14-19 or false time
# we can handle those not easy with REGEX, therefore we have
# an Exception TimestampParseException, which is thrown,
# wen strptime (parse from string to time tupel) fails
self.__parse_file(file, link)
except TimestampParseException, e:
logging.warning("False date(time) in file: %s", link)
def __get_datetime(self, mydate):
"""
@return string: Datetime - in Org Format
"""
mydate_tupel = OrgFormat.datetupelutctimestamp(mydate)
return OrgFormat.date(mydate_tupel)
def __getTimestamp(self, time, onlyDate=False):
"""
converts xml timestamp into org readable timestamp
Do 6 Nov 21:22:17 2014
"""
time = time.strip().encode('utf-8')
mail_date = datetime.strptime(time,"%c")
if onlyDate is False:
return OrgFormat.datetime(mail_date)
return OrgFormat.date(mail_date)
def __getTimestamp(self, time, onlyDate=False):
"""
converts xml timestamp into org readable timestamp
Do 6 Nov 21:22:17 2014
"""
time = time.strip().encode('utf-8')
mail_date = datetime.strptime(time,"%a %d %b %H:%M:%S %Y")
if onlyDate is False:
return OrgFormat.datetime(mail_date)
return OrgFormat.date(mail_date)
def __parse_file(self, file, link):
"""
Parses the date+time and writes entry to outputfile
@param file: filename
@param link: path
"""
if TIMESTAMP_REGEX.match(file):
# if we found a timestamp too,take hours,min
# and optionally seconds from this timestamp
timestamp = TIMESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdatetimeiso8601(timestamp)
logging.debug("found timestamp: " + orgdate)
else:
datestamp = DATESTAMP_REGEX.match(file).group()
orgdate = OrgFormat.strdate(datestamp)
orgdate_time_tupel = OrgFormat.datetupeliso8601(datestamp)
if self._args.skip_filetime_extraction != True:
if os.path.exists(link):
file_datetime = time.localtime(os.path.getmtime(link))
# check if the file - time information matches year,month,day,
# then update time
if file_datetime.tm_year == orgdate_time_tupel.tm_year and \
file_datetime.tm_mon == orgdate_time_tupel.tm_mon and \
file_datetime.tm_mday == orgdate_time_tupel.tm_mday:
logging.debug("found a time in file.setting %s-->%s",
orgdate,
OrgFormat.date(file_datetime, True))
orgdate = OrgFormat.date(file_datetime, True)
else:
logging.debug(
"item [%s] not found and thus could not determine mtime"
% link)
self.__write_file(file, link, orgdate)
