Cyclone-Team/gramps
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Gedcom fixes for mime and finding media (#795)

Browse Source 
* Fix Gedcom import/export for more comprehensive mime types

Fixes #11041

* Fix Gedcom import to better find media files

Issue #11041

* Fix Gedcom test files for media mime and error location message
This commit is contained in:
Paul Culley 2019-03-16 22:44:42 -05:00 committed by Sam Manzi
parent d96b6d19b8
commit 100d6d0672
6 changed files with 124 additions and 98 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
data/tests/exp_sample_ged.ged
View File

@ -1,12 +1,12 @@
0 HEAD
1 SOUR Gramps
2 VERS 5.0.1
2 VERS 5.0.2
2 NAME Gramps
1 DATE 7 NOV 2018
2 TIME 16:03:33
1 DATE 5 MAR 2019
2 TIME 09:11:15
1 SUBM @SUBM@
1 FILE C:\Users\prc\AppData\Roaming\gramps\temp\exp_sample_ged.ged
1 COPR Copyright (c) 2018 Alex Roitman,,,.
1 COPR Copyright (c) 2019 Alex Roitman,,,.
1 GEDC
2 VERS 5.5.1
2 FORM LINEAGE-LINKED
@ -1420,8 +1420,8 @@
0 @N0018@ NOTE Another Citation Note
0 @N0019@ NOTE A bad photo for sure
0 @O0000@ OBJE
1 FILE c:\users\prc\workspace\grampsm\main\data\tests\O0.jpg
2 FORM jpeg
1 FILE c:\msys64\mingw64\share\gramps\tests\O0.jpg
2 FORM jpg
2 TITL Michael O'Toole 2015-11
1 NOTE @N0019@
1 CHAN

65
data/tests/imp_FTM_16dec2015a-mod1.gramps
View File

@ -3,41 +3,41 @@
"http://gramps-project.org/xml/1.7.1/grampsxml.dtd">
<database xmlns="http://gramps-project.org/xml/1.7.1/">
<header>
<created date="2016-08-29" version="5.0.0-alpha1"/>
<created date="2019-03-13" version="5.0.2"/>
<researcher>
</researcher>
</header>
<events>
<event handle="_0000000500000005" change="1472500305" id="E0000">
<event handle="_0000000500000005" change="1" id="E0000">
<type>Birth</type>
<dateval val="1816"/>
<place hlink="_0000000700000007"/>
<citationref hlink="_0000000600000006"/>
</event>
<event handle="_0000000800000008" change="1472500305" id="E0001">
<event handle="_0000000800000008" change="1" id="E0001">
<type>Residence</type>
<dateval val="1850"/>
<place hlink="_0000000a0000000a"/>
<citationref hlink="_0000000900000009"/>
</event>
<event handle="_0000000b0000000b" change="1472500305" id="E0002">
<event handle="_0000000b0000000b" change="1" id="E0002">
<type>Death</type>
<datestr val="1850/1860"/>
<place hlink="_0000000c0000000c"/>
</event>
<event handle="_0000000f0000000f" change="1472500305" id="E0003">
<event handle="_0000000f0000000f" change="1" id="E0003">
<type>Marriage</type>
<dateval val="1841" type="about"/>
<place hlink="_0000001000000010"/>
</event>
<event handle="_0000001100000011" change="1472500305" id="E0004">
<event handle="_0000001100000011" change="1" id="E0004">
<type>Marriage</type>
<dateval val="1847-08"/>
<place hlink="_0000001200000012"/>
</event>
</events>
<people>
<person handle="_0000000100000001" change="1472500305" id="I0278">
<person handle="_0000000100000001" change="1" id="I0278">
<gender>M</gender>
<name type="Birth Name">
<first>Andrew</first>
@ -54,13 +54,13 @@
</person>
</people>
<families>
<family handle="_0000000d0000000d" change="1472500305" id="F0073">
<family handle="_0000000d0000000d" change="1" id="F0073">
<rel type="Married"/>
<father hlink="_0000000100000001"/>
<eventref hlink="_0000000f0000000f" role="Family"/>
<childref hlink="_0000000100000001"/>
</family>
<family handle="_0000000e0000000e" change="1472500305" id="F0074">
<family handle="_0000000e0000000e" change="1" id="F0074">
<rel type="Married"/>
<father hlink="_0000000100000001"/>
<eventref hlink="_0000001100000011" role="Family"/>
@ -68,19 +68,19 @@
</family>
</families>
<citations>
<citation handle="_0000000400000004" change="1472500305" id="C0000">
<citation handle="_0000000400000004" change="1" id="C0000">
<page>Year: 1850; Census Place: District 14, Cape Girardeau, Missouri; Roll: M432_394; Page: 435B; Image: 248</page>
<confidence>2</confidence>
<objref hlink="_0000000300000003"/>
<sourceref hlink="_0000000200000002"/>
</citation>
<citation handle="_0000000600000006" change="1472500305" id="C0001">
<citation handle="_0000000600000006" change="1" id="C0001">
<page>Year: 1850; Census Place: District 14, Cape Girardeau, Missouri; Roll: M432_394; Page: 435B; Image: 248</page>
<confidence>2</confidence>
<objref hlink="_0000000300000003"/>
<sourceref hlink="_0000000200000002"/>
</citation>
<citation handle="_0000000900000009" change="1472500305" id="C0002">
<citation handle="_0000000900000009" change="1" id="C0002">
<page>Year: 1850; Census Place: District 14, Cape Girardeau, Missouri; Roll: M432_394; Page: 435B; Image: 248</page>
<confidence>2</confidence>
<objref hlink="_0000000300000003"/>
@ -88,7 +88,7 @@
</citation>
</citations>
<sources>
<source handle="_0000000200000002" change="1472500305" id="S0029">
<source handle="_0000000200000002" change="1" id="S0029">
<stitle>1850 United States Federal Census</stitle>
<sauthor>Ancestry.com</sauthor>
<spubinfo>Name: Ancestry.com Operations, Inc.; Location: Provo, UT, USA; Date: 2009;</spubinfo>
@ -96,42 +96,43 @@
</source>
</sources>
<places>
<placeobj handle="_0000000700000007" change="1472500305" id="P0000" type="Unknown">
<placeobj handle="_0000000700000007" change="1" id="P0000" type="Unknown">
<ptitle>Tennessee, USA</ptitle>
<pname value="Tennessee, USA"/>
</placeobj>
<placeobj handle="_0000000a0000000a" change="1472500305" id="P0001" type="Unknown">
<placeobj handle="_0000000a0000000a" change="1" id="P0001" type="Unknown">
<ptitle>District 14, Cape Girardeau, Missouri, USA</ptitle>
<pname value="District 14, Cape Girardeau, Missouri, USA"/>
</placeobj>
<placeobj handle="_0000000c0000000c" change="1472500305" id="P0002" type="Unknown">
<placeobj handle="_0000000c0000000c" change="1" id="P0002" type="Unknown">
<ptitle>Bollinger Co. MO</ptitle>
<pname value="Bollinger Co. MO"/>
</placeobj>
<placeobj handle="_0000001000000010" change="1472500305" id="P0003" type="Unknown">
<placeobj handle="_0000001000000010" change="1" id="P0003" type="Unknown">
<ptitle>Union Co.?, IL</ptitle>
<pname value="Union Co.?, IL"/>
</placeobj>
<placeobj handle="_0000001200000012" change="1472500305" id="P0004" type="Unknown">
<placeobj handle="_0000001200000012" change="1" id="P0004" type="Unknown">
<ptitle>Wayne, Missouri, United States</ptitle>
<pname value="Wayne, Missouri, United States"/>
</placeobj>
</places>
<objects>
<object handle="_0000000300000003" change="1472500305" id="M159">
<object handle="_0000000300000003" change="1" id="M159">
<file src="1850 United States Federal Census(11)-1.jpg" mime="image/jpeg" description="1850 United States Federal Census"/>
<noteref hlink="_0000001400000014"/>
<noteref hlink="_0000001500000015"/>
</object>
<object handle="_0000001600000016" change="1472500305" id="M158">
<file src="D:/Users/PRC/Downloads/1850 United States Federal Census(11)-1.jpg" mime="image/jpeg" description="D:\Users\PRC\Downloads\1850 United States Federal Census(11)-1.jpg"/>
<object handle="_0000001600000016" change="1" id="M158">
<file src="/Users/PRC/Downloads/1850 United States Federal Census(11)-1.jpg" mime="image/jpeg" description="/Users/PRC/Downloads/1850 United States Federal Census(11)-1.jpg"/>
<noteref hlink="_0000001700000017"/>
</object>
<object handle="_0000001700000017" change="1472500305" id="M157">
<object handle="_0000001800000018" change="1" id="M157">
<file src="http://1.gravatar.com/avatar/77e02a3c8c665155ad1acaac8c2742e0?s=120&amp;d=mm&amp;r=pg" mime="unknown" description="http://1.gravatar.com/avatar/77e02a3c8c665155ad1acaac8c2742e0?s=120&amp;d=mm&amp;r=pg"/>
</object>
</objects>
<repositories>
<repository handle="_0000001300000013" change="1472500305" id="R0001">
<repository handle="_0000001300000013" change="1" id="R0001">
<rname>Ancestry.com</rname>
<type>Library</type>
<address>
@ -140,16 +141,26 @@
</repository>
</repositories>
<notes>
<note handle="_0000001400000014" change="1472500305" id="N0000" type="Media Note">
<text>Year: 1850; Census Place: District 14, Cape Girardeau, Missouri; Roll: M432_394; Page: 435B; Image: 248</text>
<note handle="_0000001400000014" change="1" id="N0000" type="Media Note">
<text>Year: 1850; Census Place: District 14, Cape Girardeau, Missouri; Roll: M432_394; Page: 435B; Image: 248 </text>
</note>
<note handle="_0000001500000015" change="1472500305" id="N0001" type="GEDCOM import">
<note handle="_0000001500000015" change="1" id="N0001" type="GEDCOM import">
<text>Records not imported into OBJE (multi-media object) Gramps ID M159:
Could not import 1850 United States Federal Census(11)-1.jpg Line 70: 1 FILE 1850 United States Federal Census(11)-1.jpg</text>
Could not import 1850 United States Federal Census(11)-1.jpg Line 70: 1 FILE 1850 United States Federal Census(11)-1.jpg
</text>
<style name="fontface" value="Monospace">
<range start="0" end="200"/>
</style>
</note>
<note handle="_0000001700000017" change="1" id="N0002" type="GEDCOM import">
<text>Records not imported into OBJE (multi-media object) Gramps ID M158:
Could not import D:\Users\PRC\Downloads\1850 United States Federa Line 75: 1 FILE D:\Users\PRC\Downloads\1850 United States Federal Census(11)-1.jpg
</text>
<style name="fontface" value="Monospace">
<range start="0" end="223"/>
</style>
</note>
</notes>
</database>

8
data/tests/imp_MediaTest.gramps
View File

@ -3,7 +3,7 @@
"http://gramps-project.org/xml/1.7.1/grampsxml.dtd">
<database xmlns="http://gramps-project.org/xml/1.7.1/">
<header>
<created date="2019-02-01" version="5.0.2"/>
<created date="2019-03-05" version="5.0.2"/>
<researcher>
</researcher>
</header>
@ -69,7 +69,7 @@
<noteref hlink="_0000000b0000000b"/>
</object>
<object handle="_0000000400000004" change="548708291" id="M1">
<file src="" mime="" description="Multimedia link to linked form v5.5 blob"/>
<file src="" mime="image/jpeg" description="Multimedia link to linked form v5.5 blob"/>
<attribute type="REFN" value="Ref12345M1">
<noteref hlink="_0000001200000012"/>
</attribute>
@ -118,7 +118,7 @@
<citationref hlink="_0000002700000027"/>
</object>
<object handle="_0000000c0000000c" change="1" id="O0002">
<file src="http://www.geni.com/photo/view?photo_id=6000000001341319061" mime="unknown" description="Multimedia link embedded form v5.5 URL"/>
<file src="http://www.geni.com/photo/view?photo_id=6000000001341319061" mime="text/html" description="Multimedia link embedded form v5.5 URL"/>
<noteref hlink="_0000000b0000000b"/>
</object>
<object handle="_0000000d0000000d" change="1" id="M7">
@ -130,7 +130,7 @@
<noteref hlink="_0000002900000029"/>
</object>
<object handle="_0000001500000015" change="1" id="M2">
<file src="" mime="" description="2nd blob Multimedia link to linked form v5.5 blob"/>
<file src="" mime="image/jpeg" description="2nd blob Multimedia link to linked form v5.5 blob"/>
<noteref hlink="_0000001600000016"/>
</object>
</objects>

49
data/tests/imp_sample.gramps
View File

@ -3,7 +3,7 @@
"http://gramps-project.org/xml/1.7.1/grampsxml.dtd">
<database xmlns="http://gramps-project.org/xml/1.7.1/">
<header>
<created date="2016-10-24" version="5.0.0-alpha1"/>
<created date="2019-03-05" version="5.0.2"/>
<researcher>
<resname>Alex Roitman,,,</resname>
<resaddr>Not Provided</resaddr>
@ -1456,13 +1456,13 @@
<noteref hlink="_000000f9000000f9"/>
</object>
<object handle="_000000d1000000d1" change="1" id="O0001">
<file src="Magnes&amp;Anna_smiths_marr_cert.jpg" mime="unknown" description="Magnes&amp;Anna_smiths_marr_cert.jpg"/>
<file src="Magnes&amp;Anna_smiths_marr_cert.jpg" mime="image/jpeg" description="Magnes&amp;Anna_smiths_marr_cert.jpg"/>
</object>
<object handle="_000000d8000000d8" change="1" id="O0002">
<file src="John&amp;Alice_smiths_marr_cert.jpg" mime="unknown" description="John&amp;Alice_smiths_marr_cert.jpg"/>
<file src="John&amp;Alice_smiths_marr_cert.jpg" mime="image/jpeg" description="John&amp;Alice_smiths_marr_cert.jpg"/>
</object>
<object handle="_000000f1000000f1" change="1" id="O0003">
<file src="Attic_photo.jpg" mime="unknown" description="Attic_photo.jpg"/>
<file src="Attic_photo.jpg" mime="image/jpeg" description="Attic_photo.jpg"/>
</object>
</objects>
<repositories>
@ -1506,7 +1506,8 @@
<note handle="_0000000100000001" change="1" id="N0000" type="GEDCOM import">
<text>Records not imported into HEAD (header):
GEDCOM FORM not supported Line 14: 2 FORM NOT LINEAGE-LINKED</text>
GEDCOM FORM not supported Line 14: 2 FORM NOT LINEAGE-LINKED
</text>
<style name="fontface" value="Monospace">
<range start="0" end="148"/>
</style>
@ -1514,7 +1515,8 @@ GEDCOM FORM not supported Line 14:
<note handle="_0000000200000002" change="1" id="N0001" type="GEDCOM import">
<text>Records not imported into SUBM (Submitter): (@SUBM@) Alex Roitman,,,:
Line ignored as not understood Line 23: 2 NOTE No address provided (note not supported)</text>
Line ignored as not understood Line 23: 2 NOTE No address provided (note not supported)
</text>
<style name="fontface" value="Monospace">
<range start="0" end="199"/>
</style>
@ -1523,7 +1525,8 @@ Line ignored as not understood Line 23:
<text>Records not imported into FAM (family) Gramps ID F0003:
Line ignored as not understood Line 46: 2 SOUR Not really allowed here
Filename omitted Line 48: 1 OBJE</text>
Filename omitted Line 48: 1 OBJE
</text>
<style name="fontface" value="Monospace">
<range start="0" end="256"/>
</style>
@ -1562,7 +1565,8 @@ Filename omitted Line 48:
<text>Records not imported into INDI (individual) Gramps ID I0016:
Warn: ADDR overwritten Line 204: 3 ADR1 456 Main St again
ADDR element ignored '459 Main St.' Line 202: 2 ADDR 459 Main St., The Village, San Francisco, CA, USA</text>
ADDR element ignored '459 Main St.' Line 202: 2 ADDR 459 Main St., The Village, San Francisco, CA, USA
</text>
<style name="fontface" value="Monospace">
<range start="0" end="304"/>
</style>
@ -1573,7 +1577,8 @@ ADDR element ignored '459 Main St.' Line 202:
<note handle="_0000004700000047" change="1" id="N0014" type="GEDCOM import">
<text>Records not imported into INDI (individual) Gramps ID I0018:
Tag recognized but not supported Line 245: 2 TYPE first generaton</text>
Tag recognized but not supported Line 245: 2 TYPE first generaton
</text>
<style name="fontface" value="Monospace">
<range start="0" end="165"/>
</style>
@ -1604,7 +1609,8 @@ Company. He enlisted in the army at Sparks 7 December 1917 and served as a Corpo
<note handle="_000000cf000000cf" change="1" id="N0021" type="GEDCOM import">
<text>Records not imported into FAM (family) Gramps ID F0010:
Tag recognized but not supported Line 863: 2 _STAT</text>
Tag recognized but not supported Line 863: 2 _STAT
</text>
<style name="fontface" value="Monospace">
<range start="0" end="146"/>
</style>
@ -1613,7 +1619,8 @@ Tag recognized but not supported Line 863:
<text>Records not imported into FAM (family) Gramps ID F0011:
Could not import Magnes&amp;Anna_smiths_marr_cert.jpg Line 878: 3 OBJE
Could not import Magnes&amp;Anna_smiths_marr_cert.jpg Line 881: 2 OBJE</text>
Could not import Magnes&amp;Anna_smiths_marr_cert.jpg Line 881: 2 OBJE
</text>
<style name="fontface" value="Monospace">
<range start="0" end="233"/>
</style>
@ -1621,7 +1628,8 @@ Could not import Magnes&amp;Anna_smiths_marr_cert.jpg Line 8
<note handle="_000000d9000000d9" change="1" id="N0023" type="GEDCOM import">
<text>Records not imported into FAM (family) Gramps ID F0012:
Could not import John&amp;Alice_smiths_marr_cert.jpg Line 905: 1 OBJE</text>
Could not import John&amp;Alice_smiths_marr_cert.jpg Line 905: 1 OBJE
</text>
<style name="fontface" value="Monospace">
<range start="0" end="145"/>
</style>
@ -1629,7 +1637,8 @@ Could not import John&amp;Alice_smiths_marr_cert.jpg Line 9
<note handle="_000000e4000000e4" change="1" id="N0024" type="GEDCOM import">
<text>Records not imported into FAM (family) Gramps ID F0008:
Tag recognized but not supported Line 1005: 1 ADDR 123 Main st, Grantville, Virginia, USA</text>
Tag recognized but not supported Line 1005: 1 ADDR 123 Main st, Grantville, Virginia, USA
</text>
<style name="fontface" value="Monospace">
<range start="0" end="183"/>
</style>
@ -1653,7 +1662,8 @@ Tag recognized but not supported Line 1005:
<text>Records not imported into SOUR (source) Gramps ID S0003:
Tag recognized but not supported Line 1045: 1 DATA
Skipped subordinate line Line 1046: 2 AGNC NYC Public Library</text>
Skipped subordinate line Line 1046: 2 AGNC NYC Public Library
</text>
<style name="fontface" value="Monospace">
<range start="0" end="252"/>
</style>
@ -1669,7 +1679,8 @@ Skipped subordinate line Line 1046:
REFN ignored Line 1075: 3 REFN blah blah
Skipped subordinate line Line 1076: 4 TYPE who knows
Could not import Attic_photo.jpg Line 1079: 3 OBJE</text>
Could not import Attic_photo.jpg Line 1079: 3 OBJE
</text>
<style name="fontface" value="Monospace">
<range start="0" end="344"/>
</style>
@ -1677,7 +1688,8 @@ Could not import Attic_photo.jpg Line 1079:
<note handle="_000000f6000000f6" change="1" id="N0034" type="GEDCOM import">
<text>Records not imported into Top Level:
Unknown tag Line 1106: 0 XXX an unknown token at level 0</text>
Unknown tag Line 1106: 0 XXX an unknown token at level 0
</text>
<style name="fontface" value="Monospace">
<range start="0" end="152"/>
</style>
@ -1685,12 +1697,13 @@ Unknown tag Line 1106:
<note handle="_000000f8000000f8" change="1" id="N0035" type="GEDCOM import">
<text>Records not imported into Top Level:
Unknown tag Line 1109: 1 @X1@ XXX and unknown token xref definition</text>
Unknown tag Line 1109: 1 @X1@ XXX and unknown token xref definition
</text>
<style name="fontface" value="Monospace">
<range start="0" end="163"/>
</style>
</note>
<note handle="_000000f9000000f9" change="1477325896" id="N0036" type="General">
<note handle="_000000f9000000f9" change="1551799781" id="N0036" type="General">
<text>Objects referenced by this note were missing in a file imported on 12/25/1999 12:00:00 AM.</text>
</note>
</notes>

12
gramps/plugins/export/exportgedcom.py
View File

@ -104,15 +104,6 @@ LANGUAGES = {
#
#-------------------------------------------------------------------------
MIME2GED = {
"image/bmp" : "bmp",
"image/gif" : "gif",
"image/jpeg" : "jpeg",
"image/x-pcx" : "pcx",
"image/tiff" : "tiff",
"audio/x-wav" : "wav"
}
QUALITY_MAP = {
Citation.CONF_VERY_HIGH : "3",
Citation.CONF_HIGH : "2",
@ -1467,8 +1458,7 @@ class GedcomWriter(UpdateCallback):
gramps_id = media.get_gramps_id()
self._writeln(0, '@%s@' % gramps_id, 'OBJE')
mime = media.get_mime_type()
form = MIME2GED.get(mime, mime)
form = os.path.splitext(media.get_path())[1][1:]
path = media_path_full(self.dbase, media.get_path())
self._writeln(1, 'FILE', path, limit=255)
if form:

72
gramps/plugins/lib/libgedcom.py
View File

@ -94,6 +94,7 @@ import time
# from xml.parsers.expat import ParserCreate
from collections import defaultdict, OrderedDict
import string
import mimetypes
from io import StringIO, TextIOWrapper
from urllib.parse import urlparse
@ -123,7 +124,7 @@ from gramps.gen.lib import (
Surname, Tag, Url, UrlType, PlaceType, PlaceRef, PlaceName)
from gramps.gen.db import DbTxn
from gramps.gen.updatecallback import UpdateCallback
from gramps.gen.mime import get_type
from gramps.gen.utils.file import media_path
from gramps.gen.utils.id import create_id
from gramps.gen.utils.lds import TEMPLES
from gramps.gen.utils.unknown import make_unknown, create_explanation_note
@ -531,15 +532,6 @@ PEDIGREE_TYPES = {
'adopted': TYPE_ADOPT,
'foster' : TYPE_FOSTER, }
MIME_MAP = {
'jpeg' : 'image/jpeg', 'rtf' : 'text/rtf',
'jpg' : 'image/jpeg', 'pdf' : 'application/pdf',
'mpeg' : 'video/mpeg', 'gif' : 'image/gif',
'mpg' : 'video/mpeg', 'bmp' : 'image/x-ms-bmp',
'tiff' : 'image/tiff', 'aif' : 'audio/x-aiff',
'text' : 'text/plain', 'w8bn' : 'application/msword',
'wav' : 'audio/x-wav', 'mov' : 'video/quicktime', }
FTW_BAD_PLACE = [
EventType.OCCUPATION,
EventType.RELIGION,
@ -2728,7 +2720,6 @@ class GedcomParser(UpdateCallback):
self.attrs = list(amap.values())
self.gedattr = dict([key, val] for val, key in amap.items())
self.search_paths = []
def parse_gedcom_file(self, use_trans=False):
"""
@ -3021,6 +3012,9 @@ class GedcomParser(UpdateCallback):
# FIXME: problem possibly caused by umlaut/accented character
# in filename
return (0, fullname)
# strip off Windows drive letter, if present
if len(fullname) > 3 and fullname[1] == ':':
fullname = fullname[2:]
# look where we found the '.ged', using the full path in fullname
other = os.path.join(altpath, fullname)
if os.path.isfile(other):
@ -3029,15 +3023,15 @@ class GedcomParser(UpdateCallback):
other = os.path.join(altpath, os.path.basename(fullname))
if os.path.isfile(other):
return (1, other)
# I don't think the following code does anything because search_paths
# is never initialized...
if len(fullname) > 3:
if fullname[1] == ':':
fullname = fullname[2:]
for path in self.search_paths:
other = os.path.normpath("%s/%s" % (path, fullname))
# lets try using the base path for relative media paths
other = os.path.join(media_path(self.dbase), fullname)
if os.path.isfile(other):
return (1, other)
return (1, fullname)
# lets try using the base path for relative media paths with base name
other = os.path.join(media_path(self.dbase),
os.path.basename(fullname))
if os.path.isfile(other):
return (1, os.path.basename(fullname))
return (0, fullname)
def __get_next_line(self):
@ -5366,8 +5360,9 @@ class GedcomParser(UpdateCallback):
# to allow import of references to URLs (especially for import from
# geni.com), do not try to find the file if it is blatently a URL
res = urlparse(sub_state.filename)
if sub_state.filename != '' and \
(res.scheme == '' or res.scheme == 'file'):
if sub_state.filename != '' and (res.scheme == '' or
len(res.scheme) == 1 or
res.scheme == 'file'):
(valid, path) = self.__find_file(sub_state.filename,
self.dir_path)
if not valid:
@ -5384,13 +5379,19 @@ class GedcomParser(UpdateCallback):
if sub_state.title:
photo.set_description(sub_state.title)
else:
photo.set_description(path)
photo.set_description(path.replace('\\', '/'))
full_path = os.path.abspath(path)
if os.path.isfile(full_path):
photo.set_mime_type(get_type(full_path))
# deal with mime types
value = mimetypes.guess_type(full_path)
if value and value[0]: # found from filename
photo.set_mime_type(value[0])
else: # get from OBJE.FILE.FORM
if '/' in sub_state.form: # already has expanded mime type
photo.set_mime_type(sub_state.form)
else:
photo.set_mime_type(MIME_MAP.get(sub_state.form,
'unknown'))
value = mimetypes.types_map.get('.' + sub_state.form,
_('unknown'))
photo.set_mime_type(value)
if sub_state.attr:
photo.attribute_list.append(sub_state.attr)
self.dbase.add_media(photo, self.trans)
@ -6691,6 +6692,17 @@ class GedcomParser(UpdateCallback):
if state.media.get_path() == "":
self.__add_msg(_("Filename omitted"), line, state)
# deal with mime types
value = mimetypes.guess_type(state.media.get_path())
if value and value[0]: # found from filename
state.media.set_mime_type(value[0])
else: # get from OBJE.FILE.FORM
if '/' in state.form: # already has expanded mime type
state.media.set_mime_type(state.form)
else:
value = mimetypes.types_map.get('.' + state.form,
_('unknown'))
state.media.set_mime_type(value)
# Add the default reference if no source has found
self.__add_default_source(media)
@ -6726,21 +6738,21 @@ class GedcomParser(UpdateCallback):
self.__skip_subordinate_levels(state.level + 1, state)
return
res = urlparse(line.data)
if line.data != '' and (res.scheme == '' or res.scheme == 'file'):
if line.data != '' and (res.scheme == '' or
len(res.scheme) == 1 or res.scheme == 'file'):
(file_ok, filename) = self.__find_file(line.data, self.dir_path)
if state.form != "url":
# Might not work if FORM doesn't precede FILE
if not file_ok:
self.__add_msg(_("Could not import %s") % filename, line,
self.__add_msg(_("Could not import %s") % line.data, line,
state)
path = filename
else:
path = line.data
state.media.set_path(path)
state.media.set_mime_type(get_type(path))
if not state.media.get_description():
state.media.set_description(path)
state.media.set_description(path.replace('\\', '/'))
def __obje_title(self, line, state):
"""