def text(self):
text = None
if self.data.show_text:
item = self.existing_content_item
if IRichText.providedBy(item) and IRichTextBehavior(item).text:
text = IRichTextBehavior(item).text.output
return text
def SearchableText(obj):
text = u''
richtext = IRichText(obj, None)
if richtext:
textvalue = richtext.text
if IRichTextValue.providedBy(textvalue):
transforms = getToolByName(obj, 'portal_transforms')
# Before you think about switching raw/output
# or mimeType/outputMimeType, first read
# https://github.com/plone/Products.CMFPlone/issues/2066
raw = safe_unicode(textvalue.raw)
if six.PY2:
raw = raw.encode('utf-8', 'replace')
text = transforms.convertTo(
'text/plain',
raw,
mimetype=textvalue.mimeType,
).getData().strip()
subject = u' '.join(
[safe_unicode(s) for s in obj.Subject()]
)
return u' '.join((
safe_unicode(obj.id),
safe_unicode(obj.title) or u'',
safe_unicode(obj.description) or u'',
safe_unicode(text),
safe_unicode(subject),
))
def post_creation(self, obj):
field_data = self.field_data
bdata = ILayoutAware(obj, None)
if bdata:
bdata.contentLayout = '++contentlayout++default/document.html'
bdata = IRichText(obj, None)
if bdata:
bdata.text = RichTextValue(field_data['text'], 'text/html',
'text/html')
bdata = IBasic(obj, None)
if bdata:
bdata.title = field_data['title']
bdata.description = field_data['description']
else:
obj.title = field_data['title']
obj.description = field_data['description']
bdata = ICategorization(obj, None)
if bdata:
bdata.subjects = field_data['subject']
bdata = IPublication(obj)
if field_data['effectiveDate']:
bdata.effective = field_data['effectiveDate'].asdatetime()
ldata = ILocation(obj, None)
if ldata:
if field_data.get('location'):
ldata.locations = [field_data['location']]
if field_data.get('newsLocation'):
if ldata.locations:
ldata.locations.append(field_data['newsLocation'])
else:
ldata.locations = [field_data['newsLocation']]
obj.modification_date = field_data['modification_date']
obj.creation_date = field_data['creation_date']
bdata = ILayoutAware(obj, None)
if bdata:
if self.data[
'portal_type'] == 'Folder' and 'text' in self.field_data:
bdata.content = FOLDER_DEFAULT_PAGE_LAYOUT % self.field_data[
'text']
elif self.layout:
bdata.contentLayout = self.layout
def SearchableText(obj):
text = u''
richtext = IRichText(obj, None)
if richtext:
textvalue = richtext.text
if IRichTextValue.providedBy(textvalue):
transforms = getToolByName(obj, 'portal_transforms')
text = transforms.convertTo(
'text/plain',
safe_unicode(textvalue.output).encode('utf8'),
mimetype=textvalue.mimeType,
).getData().strip()
subject = u' '.join([safe_unicode(s) for s in obj.Subject()])
return u' '.join((
safe_unicode(obj.id),
safe_unicode(obj.title) or u'',
safe_unicode(obj.description) or u'',
safe_unicode(text),
safe_unicode(subject),
))
title = unidecode(dom.cssselect('h1')[0].text_content())
except IndexError:
print('bad url: ' + found_url)
continue
_id = normalizer.normalize(found_url.split('/')[-1])
if _id in container.objectIds():
continue
kw = {}
imgs = dom.cssselect('#mw-content-text img')
if len(imgs) > 0:
im_url = imgs[0].attrib['src']
if im_url.startswith('//'):
im_url = 'https:' + im_url
resp = requests.get(im_url)
kw['image'] = NamedBlobImage(
data=resp.content,
filename=unidecode(im_url.split('/')[-1]).decode('utf8'))
obj = api.content.create(type='Document', id=_id, title=title,
exclude_from_nav=True, container=container, **kw)
text = ''
for p in dom.cssselect('#mw-content-text p'):
text += tostring(p)
bdata = IRichText(obj, None)
bdata.text = RichTextValue(text, 'text/html', 'text/html')
obj.reindexObject()
parsed.append(found_url)
transaction.commit()
def test_richtext_behavior(self):
IRichText.providedBy(self.portal.doc1)
def post_creation(self,
obj,
pdb_if_exception=False,
post_creation_data=None):
if obj is None:
return
field_data = self.field_data
bdata = ILayoutAware(obj, None)
if bdata:
try:
bdata.contentLayout = self.layout
except Exception:
bdata.contentLayout = '++contentlayout++default/document.html'
bdata = IRichText(obj, None)
if bdata:
try:
bdata.text = RichTextValue(field_data['text'], 'text/html',
'text/html')
except Exception:
try:
bdata.text = RichTextValue(
field_data[
'plone.app.contenttypes.behaviors.richtext.IRichText']
['text'], # noqa
'text/html',
'text/html').raw
except Exception:
bdata.text = ''
bdata = IBasic(obj, None)
if bdata:
try:
bdata.title = field_data['title']
except Exception:
try:
bdata.title = field_data[
'plone.app.content.interfaces.INameFromTitle']['title']
except Exception:
bdata.description = field_data[dublin]['title']
try:
bdata.description = field_data['description']
except Exception:
try:
bdata.description = field_data[dublin]['description']
except Exception:
bdata.description = field_data[basic]['description']
else:
try:
obj.title = field_data['title']
obj.description = field_data['description']
except Exception:
obj.title = field_data[dublin]['title']
obj.description = field_data[dublin]['description']
bdata = ICategorization(obj, None)
if bdata:
try:
bdata.subjects = field_data['subject']
except Exception:
try:
bdata.subjects = self.field_data[dublin]['subjects']
except Exception:
try:
bdata.subjects = self.field_data[categorization][
'subjects']
except Exception:
pass # no keywords found
bdata = IPublication(obj)
try:
if field_data['effectiveDate']:
bdata.effective = pydt(field_data['effectiveDate'])
except Exception:
try:
if field_data[dublin]['effective']:
bdata.effective = pydt(field_data[dublin]['effective'])
except Exception:
try:
if field_data[publication]['effective']:
bdata.effective = pydt(
field_data[publication]['effective'])
except Exception:
bdata.effective = None
ldata = ILocation(obj, None)
if ldata:
if field_data.get('location'):
ldata.locations = [field_data['location']]
if field_data.get('newsLocation'):
if ldata.locations:
ldata.locations.append(field_data['newsLocation'])
else:
ldata.locations = [field_data['newsLocation']]
try:
obj.modification_date = field_data['modification_date']
except Exception:
try:
obj.modification_date = obj.modified()
except Exception:
obj.modification_date = None
try:
obj.creation_date = field_data['creation_date']
except Exception:
try:
obj.creation_date = obj.created()
except Exception:
obj.creation_date = None
bdata = IDublinCore(obj, None)
if bdata:
if IDublinCore.__identifier__ in field_data:
dublin_core = field_data[IDublinCore.__identifier__]
bdata.expires = dublin_core['expires']
bdata.rights = dublin_core['rights']
bdata.creators = tuple(dublin_core['creators'])
bdata.language = dublin_core['language']
bdata.effective = pydt(dublin_core['effective'])
bdata.subjects = dublin_core['subjects']
bdata.contributors = tuple(dublin_core['contributors'])
else:
bdata.expires = pydt(field_data.get('expirationDate'))
bdata.rights = field_data.get('rights')
creators = field_data.get('creators')
bdata.creators = tuple(creators) if creators else ()
language = field_data.get('language')
bdata.language = language if language is not None else ""
bdata.effective = pydt(field_data.get('effectiveDate'))
bdata.subjects = field_data.get('subject')
contributors = field_data.get('contributors')
bdata.contributors = tuple(contributors) if contributors else (
)
bdata = ILayoutAware(obj, None)
if bdata:
if self.data['portal_type'] == 'Folder' and (
self.field_data.get('text') or '').strip():
bdata.content = FOLDER_DEFAULT_PAGE_LAYOUT % self.field_data[
'text']
# need to explicitly reset contentLayout value because this data
# could be overwritten
bdata.contentLayout = None
elif self.layout:
if layoutaware in field_data and 'contentLayout' in field_data[
layoutaware]:
bdata.contentLayout = field_data[layoutaware][
'contentLayout']
if layoutaware in field_data and 'content' in field_data[
layoutaware]:
bdata.content = field_data[
'plone.app.blocks.layoutbehavior.ILayoutAware'][
'content']
if 'rendered_layout' in self.data['data']:
bdata.rendered_layout = self.data['data'][
'rendered_layout']
inv_field_mapping = {v: k for k, v in self.fields_mapping.iteritems()}
for IBehavior, field_name in self.behavior_data_mappers:
original_field_name = inv_field_mapping.get(field_name, field_name)
if original_field_name not in self.field_data:
# data not here...
continue
behavior = IBehavior(obj, None)
if behavior is None:
# behavior not valid for obj type
continue
val = self.field_data[original_field_name]
if field_name in self.data_converters:
val = self.data_converters[field_name](val)
setattr(behavior, field_name, val)
# handle lead images
for field_name in self.lead_image_field_names:
if self.field_data.get(field_name):
if field_name == 'plone.app.contenttypes.behaviors.leadimage.ILeadImage':
im_obj = self.field_data.get(field_name)['image']
else:
im_obj = self.field_data.get(field_name)
if hasattr(im_obj, 'read'):
im_data = im_obj.read()
else:
im_data = im_obj
if not im_data:
continue
filename = self.field_data.get('image_filename')
if not filename:
if hasattr(im_obj, 'filename'):
filename = im_obj.filename
else:
filename = self.field_data['id']
obj.image = im_data
if not isinstance(obj.image, NamedBlobImage):
is_stringio = isinstance(im_obj, StringIO)
if is_stringio:
namedblobimage_data = im_data
elif isinstance(im_obj, Image):
namedblobimage_data = im_data.data
else:
if pdb_if_exception:
pdb.set_trace()
logger.info(" lead image is type %s" % type(im_obj))
obj.image = NamedBlobImage(data=namedblobimage_data,
contentType='',
filename=filename)
if hasattr(obj.image, 'contentType') and isinstance(
obj.image.contentType, unicode):
obj.image.contentType = obj.image.contentType.encode(
'ascii')
else:
if isinstance(im_obj, Image):
data = im_obj.data
elif hasattr(im_obj, 'buf'):
data = im_obj.buf
elif hasattr(im_obj, '_blob'):
if hasattr(im_obj._blob, '_p_blob_uncommitted'):
f = open(im_obj._blob._p_blob_uncommitted, 'r')
data = f.read()
f.close()
else:
raise Exception(
"no _p_blob_uncommitted attr in im_obj._blob")
else:
raise Exception("no _blob attr in im_obj")
if data == '' or data is None:
data = base64.b64decode(
'R0lGODlhAQABAAAAACH5BAEKAAEALAAAAAABAAEAAAICTAEAOw=='
)
image_type = what('', h=data)
if image_type in [
'png', 'bmp', 'jpeg', 'xbm', 'tiff', 'gif'
]:
obj.image.contentType = 'image/%s' % image_type
elif image_type == 'rast':
obj.image.contentType = 'image/cmu-raster'
elif image_type == 'ppm':
obj.image.contentType = 'image/x-portable-pixmap'
elif image_type == 'pgm':
obj.image.contentType = 'image/x-portable-greymap'
elif image_type == 'pbm':
obj.image.contentType = 'image/x-portable-bitmap'
elif image_type == 'rgb':
obj.image.contentType = 'image/x-rgb'
else:
# look at filename extension
contentType, encoding = guess_type(obj.image.filename,
strict=False)
if contentType:
obj.image.contentType = contentType
else:
logger.info(
"Unknown image type {};"
" defaulting to jpeg".format(image_type))
pdb.set_trace()
obj.image.contentType = 'image/jpeg' # default
for caption_field_name in self.lead_image_caption_field_names:
if caption_field_name in self.field_data:
obj.imageCaption = self.field_data.get(
caption_field_name)
def test_richtext_behavior(self):
IRichText.providedBy(self.portal.doc1)
def post_creation(self, obj):
field_data = self.field_data
bdata = ILayoutAware(obj, None)
if bdata:
bdata.contentLayout = '++contentlayout++default/document.html'
bdata = IRichText(obj, None)
if bdata:
bdata.text = RichTextValue(field_data['text'], 'text/html',
'text/html')
bdata = IBasic(obj, None)
if bdata:
bdata.title = field_data['title']
bdata.description = field_data['description']
else:
obj.title = field_data['title']
obj.description = field_data['description']
bdata = ICategorization(obj, None)
if bdata:
bdata.subjects = field_data['subject']
bdata = IPublication(obj)
if field_data['effectiveDate']:
bdata.effective = pydt(field_data['effectiveDate'])
ldata = ILocation(obj, None)
if ldata:
if field_data.get('location'):
ldata.locations = [field_data['location']]
if field_data.get('newsLocation'):
if ldata.locations:
ldata.locations.append(field_data['newsLocation'])
else:
ldata.locations = [field_data['newsLocation']]
obj.modification_date = field_data['modification_date']
obj.creation_date = field_data['creation_date']
bdata = ILayoutAware(obj, None)
if bdata:
if self.data[
'portal_type'] == 'Folder' and 'text' in self.field_data:
bdata.content = FOLDER_DEFAULT_PAGE_LAYOUT % self.field_data[
'text']
# need to explicitly reset contentLayout value because this data
# could be overwritten
bdata.contentLayout = None
elif self.layout:
bdata.contentLayout = self.layout
inv_field_mapping = {v: k for k, v in self.fields_mapping.iteritems()}
for IBehavior, field_name in self.behavior_data_mappers:
original_field_name = inv_field_mapping.get(field_name, field_name)
if original_field_name not in self.field_data:
# data not here...
continue
behavior = IBehavior(obj, None)
if behavior is None:
# behavior not valid for obj type
continue
val = self.field_data[original_field_name]
if field_name in self.data_converters:
val = self.data_converters[field_name](val)
setattr(behavior, field_name, val)
# handle lead images
for field_name in self.lead_image_field_names:
if self.field_data.get(field_name):
im_obj = self.field_data.get(field_name)
if hasattr(im_obj, 'read'):
im_data = im_obj.read()
else:
im_data = im_obj
if not im_data:
continue
filename = self.field_data.get('image_filename')
if not filename:
if hasattr(im_obj, 'filename'):
filename = im_obj.filename
else:
filename = self.field_data['id']
obj.image = NamedBlobImage(data=decode_file_data(im_data),
filename=to_unicode(filename))
if not obj.image.contentType:
obj.image.contentType = 'image/jpeg'
for caption_field_name in self.lead_image_caption_field_names:
if caption_field_name in self.field_data:
obj.imageCaption = self.field_data.get(
caption_field_name)