comt: src/cm/views/export.py@6b3d72136e85 (annotated)

0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	1	from django import forms
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	2	from django.core.urlresolvers import reverse
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	3	from django.http import HttpResponse, HttpResponseRedirect, Http404
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	4	from django.shortcuts import render_to_response
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	5	from django.template.loader import render_to_string
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	6	from django.template import RequestContext
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	7	from django.utils.translation import ugettext as _, ugettext_lazy
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	8	from django.contrib.auth.models import User
455 33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	9	from django.conf import settings
364 41dd28557b5d tidyfy html before conversion with abiword gibus parents: 363 diff changeset	10	from cm.converters.pandoc_converters import pandoc_convert, do_tidy
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	11	from cm.models import Text, TextVersion, Attachment, Comment
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	12	from cm.security import get_viewable_comments
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	13	import mimetypes
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	14	import simplejson
455 33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	15	import imghdr
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	16	import base64
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	17	import re
360 bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	18	from cm.cm_settings import USE_ABI
454 b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	19
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	20	EXPORT2_INFOS = {
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	21	# key -> { mimetype, extension}
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	22	's5' : {},
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	23	'pdf' : {'mimetype': 'application/pdf', 'extension':'pdf'},
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	24	'markdown' : {'mimetype': 'text/plain', 'extension':'mkd'},
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	25	'odt' : {'mimetype': 'application/vnd.oasis.opendocument.text', 'extension':'odt'},
433 056d92bffb23 Added export in .doc and .docx formats. gibus parents: 367 diff changeset	26	'doc' : {'mimetype': 'application/msword', 'extension':'doc'},
056d92bffb23 Added export in .doc and .docx formats. gibus parents: 367 diff changeset	27	'docx' : {'mimetype': 'application/vnd.openxmlformats-officedocument.wordprocessingml.document', 'extension':'docx'},
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	28	'latex' :{'mimetype': 'text/x-tex', 'extension':'tex'},
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	29	'html' :{'mimetype': 'text/html', 'extension':'html'},
443 cacd524f5279 Adds export to epub. gibus parents: 441 diff changeset	30	'epub' :{'mimetype': 'application/epub+zip', 'extension':'epub'},
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	31	'raw' : {'mimetype': 'text/plain', 'extension':'txt'},
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	32	'xml' : {'mimetype': 'text/xml', 'extension':'xml'},
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	33	}
454 b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	34
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	35	HTML_HEADER = u"""
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	36	<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	37	<html><head>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	38	<STYLE TYPE='text/css'>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	39	div.pagebreakhere {
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	40	page-break-before: always ;
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	41	}
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	42	</STYLE>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	43	<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/></head>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	44	<body>%s</body>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	45	</html>
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	46	"""
b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	47
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	48	def content_export2(request, content, title, content_format, format, use_pandoc, download_response):
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	49	# TODO : formats must be imported from converters
56 bd8a4ffc7dad BUG FIX : pdf export PhA tests, TODO fix special html caracters in comments reno parents: 0 diff changeset	50	# import pdb;pdb.set_trace()
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	51	if format == 'raw' :
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	52	export_content = content
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	53	elif content_format == 'html' and format == 'html':
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	54	export_content = HTML_HEADER % content
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	55	elif content_format == 'markdown' and format == 'markdown':
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	56	export_content = content
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	57	else:
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	58	if use_pandoc :
367 e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	59	# markdown2pdf is buggy => convert to HTML and use abiword to export in PDF
454 b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	60	if format in ('pdf', 'odt', 'docx', 'doc') and USE_ABI:
367 e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	61	html_content = pandoc_convert(content, content_format, 'html', full=True)
e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	62	from cm.converters.abi_converters import AbiFileConverter
e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	63	converter = AbiFileConverter()
454 b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	64	fix_content = converter.add_html_header(html_content)
367 e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	65	export_content = converter.convert_from_html(fix_content, format)
e4a0c2fe8df2 markdown2pdf is buggy => convert to HTML and use abiword to export in PDF gibus parents: 364 diff changeset	66	else:
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	67	export_content = pandoc_convert(content, content_format, format, full=True)
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	68	else :
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	69	fix_content = content
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	70	if content_format == 'html':
360 bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	71	if USE_ABI:
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	72	from cm.converters.abi_converters import AbiFileConverter
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	73	converter = AbiFileConverter()
454 b7a092a52eae Cleaned export. gibus parents: 453 diff changeset	74	fix_content = converter.add_html_header(content)
360 bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	75	else:
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	76	from cm.converters.oo_converters import combine_css_body
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	77	fix_content = combine_css_body(content, '')
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	78	if USE_ABI:
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	79	from cm.converters.abi_converters import AbiFileConverter
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	80	converter = AbiFileConverter()
494 6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	81	try:
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	82	export_content = converter.convert_from_html(fix_content, format)
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	83	except:
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	84	from cm.converters.oo_converters import combine_css_body
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	85	fix_content = combine_css_body(content, '')
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	86	from cm.converters.oo_converters import convert_html as oo_convert
6b3d72136e85 Use libreoffice to convert html to pdf when abiword crashes. gibus parents: 460 diff changeset	87	export_content = oo_convert(fix_content, format)
360 bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	88	else:
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	89	from cm.converters.oo_converters import convert_html as oo_convert
bfaab8740995 Add abiword as an alternative to open office for conversions gibus parents: 77 diff changeset	90	export_content = oo_convert(fix_content, format)
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	91
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	92	export_infos = EXPORT2_INFOS[format]
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	93
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	94	if download_response:
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	95	return _response_download(export_content, title, export_infos['mimetype'], export_infos['extension']) ;
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	96	else:
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	97	return _response_write(export_content)
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	98
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	99
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	100	def _response_download(content, title, mimetype, extension):
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	101	response = HttpResponse(mimetype=mimetype)
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	102	file_title = title + '.' + extension
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	103	from email.header import Header
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	104	encoded_name = str(Header(file_title.encode('utf8'), charset='utf8', maxlinelen=500))
52 9fa013909d9a prevent header error if newline in header (pb not fully solved) raph parents: 0 diff changeset	105	# TODO: find a way to include long (more than 76 chars) into header
9fa013909d9a prevent header error if newline in header (pb not fully solved) raph parents: 0 diff changeset	106	encoded_name = encoded_name.replace('\n','')
9fa013909d9a prevent header error if newline in header (pb not fully solved) raph parents: 0 diff changeset	107	response['Content-Disposition'] = 'attachment; filename=%s' % encoded_name
0 40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	108	response.write(content)
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	109	return response
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	110
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	111	def _response_write(content):
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	112	response = HttpResponse()
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	113	response.write(content)
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	114	return response
40c8f766c9b8 import from internal svn r 4007 raph parents: diff changeset	115
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	116	def xml_export(request, text_version, whichcomments):
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	117	# Text version infos
460 2fdb7d095d5c Added import from XML file, including text, comments and attachments. gibus parents: 455 diff changeset	118	template_dict = { 'title': text_version.title, 'created': text_version.created, 'modified': text_version.modified, 'format': text_version.format, 'content': text_version.content, 'tags': text_version.tags, }
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	119
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	120	# Comments
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	121	comments = [] # whichcomments=="none"
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	122	if whichcomments == "filtered" or whichcomments == "all":
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	123	_comments = text_version.comment_set.all()
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	124	if whichcomments == "filtered" :
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	125	filteredIds = []
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	126	if request.method == 'POST' :
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	127	ll = request.POST.get('filteredIds',[]).split(",")
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	128	filteredIds = [ int(l) for l in ll if l]
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	129	_comments = text_version.comment_set.filter(id__in=filteredIds)
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	130	comments = get_viewable_comments(request, _comments, text_version, order_by=('start_wrapper','start_offset','end_wrapper','end_offset'))
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	131	# Add user name/email if missing comment name/email
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	132	for comment in comments:
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	133	users = User.objects.filter(id=comment.user_id)
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	134	if not(comment.name):
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	135	comment.name = users[0].username
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	136	if not(comment.email):
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	137	comment.email = users[0].email
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	138
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	139	template_dict['comments'] = comments
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	140
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	141	# Author
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	142	users = User.objects.filter(id=text_version.user_id)
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	143	if text_version.name:
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	144	template_dict['name'] = text_version.name
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	145	else:
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	146	template_dict['name'] = users[0].username
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	147	if text_version.email:
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	148	template_dict['email'] = text_version.email
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	149	else:
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	150	template_dict['email'] = users[0].email
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	151
455 33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	152	# Attachments
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	153	attachments = []
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	154	template_dict['content'] = re.sub("%s" %settings.SITE_URL, '', template_dict['content']) # replaces absolute urls by relative urls
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	155	attach_re = r'(?:/text/(?P<key>\w))?/attach/(?P<attach_key>\w)/'
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	156	attach_str_textversion = r'/text/%s/attach/%s/'
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	157	attach_str = r'/attach/%s/'
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	158	for match in re.findall(attach_re, template_dict['content']):
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	159	if match[0]: # removes text_version, attachements do not need it
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	160	template_dict['content'] = template_dict['content'].replace(attach_str_textversion %match, attach_str %match[1])
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	161
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	162	attach = Attachment.objects.get(key=match[1])
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	163	img_fmt = imghdr.what(attach.data.path)
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	164	img = open(attach.data.path, 'rb')
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	165	attachments.append({'key': match[1], 'data': base64.b64encode(img.read())})
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	166	img.close()
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	167	template_dict['attachments'] = attachments
33c7e20efcb7 Added export of attachements as inline b64 images for appropriate formats. gibus parents: 454 diff changeset	168
453 1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	169	# Renders template
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	170	export_content = render_to_string('site/export.xml', template_dict, context_instance=RequestContext(request))
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	171
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	172	# Returns HTTP response
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	173	export_infos = EXPORT2_INFOS['xml']
1d314f629611 Added export to XML for re-import (nb. without attachements). gibus parents: 443 diff changeset	174	return _response_download(export_content, text_version.title, export_infos['mimetype'], export_infos['extension']) ;

author	gibus
	Mon, 28 Jan 2013 12:19:11 +0100
changeset 494	6b3d72136e85
parent 460	2fdb7d095d5c
child 495	f86b611f1876
permissions	-rw-r--r--