7212: convert invalid dates to text on import

If the date is invalid, it will be converted to text mode,
and the text field will be set to the problematic XML.
A warning will be printed on the console as well.
The log message will use the invalid date attached as
DateError.date, see [8e1659].
This commit is contained in:
Vassilii Khachaturov 2013-11-17 16:12:43 +02:00
parent befcd21d0b
commit cca72aa579

View File

@ -29,10 +29,12 @@
# Standard Python Modules
#
#-------------------------------------------------------------------------
from __future__ import unicode_literals
import os
import sys
import time
from xml.parsers.expat import ExpatError, ParserCreate
from xml.sax.saxutils import escape
from gramps.gen.const import GRAMPS_LOCALE as glocale
_ = glocale.translation.gettext
import re
@ -47,7 +49,7 @@ LOG = logging.getLogger(".ImportXML")
#-------------------------------------------------------------------------
from gramps.gen.mime import get_type
from gramps.gen.lib import (Address, Attribute, AttributeType, ChildRef,
ChildRefType, Citation, Date, Event, EventRef,
ChildRefType, Citation, Date, DateError, Event, EventRef,
EventRoleType, EventType, Family, LdsOrd, Location,
MediaObject, MediaRef, Name, NameOriginType,
NameType, Note, NoteType, Person, PersonRef,
@ -2294,10 +2296,16 @@ class GrampsParser(UpdateCallback):
else:
newyear = Date.newyear_to_code(newyear)
try:
date_value.set(qual, mode, cal,
(day, month, year, dualdated,
rng_day, rng_month, rng_year, dualdated),
newyear=newyear)
except DateError as e:
self._set_date_to_xml_text(date_value, e,
xml_element_name = ("datespan" if mode == Date.MOD_SPAN
else "daterange"),
xml_attrs = attrs)
def start_dateval(self, attrs):
if self.citation:
@ -2375,8 +2383,27 @@ class GrampsParser(UpdateCallback):
else:
newyear = Date.newyear_to_code(newyear)
try:
date_value.set(qual, mod, cal, (day, month, year, dualdated),
newyear=newyear)
except DateError as e:
self._set_date_to_xml_text(date_value, e, 'dateval', attrs)
def _set_date_to_xml_text(self, date_value, date_error, xml_element_name, xml_attrs):
"""
Common handling of invalid dates for the date... element handlers.
Prints warning on console and sets date_value to a text-only date
with the problematic XML inside.
"""
xml = "<{element_name} {attrs}/>".format(
element_name = xml_element_name,
attrs = " ".join(
['{}="{}"'.format(k,escape(v, entities={'"' : "&quot;"}))
for k,v in xml_attrs.iteritems()]))
LOG.warning(_("Invalid date {} in XML {}, preserving XML as text"
).format(date_error.date.to_struct(), xml))
date_value.set(modifier=Date.MOD_TEXTONLY, text=xml)
def start_datestr(self, attrs):
if self.citation: