def process_PDFs(site):
cat = site.portal_catalog
filebrains = cat(portal_type='File', contentType='application/pdf')
print('Processing {} PDFs'.format(len(filebrains)))
count = 0
for brain in filebrains:
obj = brain.getObject()
blob = obj.file.open('r')
tmp_dir = tempfile.mkdtemp()
filepath = os.path.join(tmp_dir,
obj.file.filename).encode('utf-8').strip()
file = open(filepath, 'wb')
file.write(blob.read())
file.close()
blob.close()
exception = False
try:
gs_pdf(filepath)
except Exception:
logger.warn(
u'Could not strip additional metadata with gs {}'.format(
filepath),
exc_info=True) # noqa
exception = True
if not exception:
try:
exiftool(filepath)
except Exception:
logger.warn(
u'Could not strip metadata with exiftool {}'.format(
filepath),
exc_info=True) # noqa
exception = True
if not exception:
try:
qpdf(filepath)
except Exception:
logger.warn(
u'Could not strip additional metadata with qpdf {}'.format(
filepath),
exc_info=True) # noqa
exception = True
if not exception:
file = open(filepath, 'rb')
obj.file = NamedBlobFile(file, filename=obj.file.filename)
file.close()
count += 1
if count % 50 == 0:
print('Processed {} PDFs'.format(count))
transaction.commit()
else:
logger.warn(
'{sitepath} will not be modified because an exception occured.'
.format(sitepath=brain.getURL())) # noqa
shutil.rmtree(tmp_dir)
transaction.commit()
print('Done.')
def create_object(self, folder, type_, info):
filename = info['name']
name = filename.decode("utf8")
chooser = INameChooser(folder)
chooser_name = name.lower().replace('aq_', '')
newid = chooser.chooseName(chooser_name, folder.aq_parent)
# strip metadata from file
if (type_ in ('Image', 'File', 'Video', 'Audio')
and exiftool is not None and 'tmp_file' in info):
try:
exiftool(info['tmp_file'])
except Exception:
logger.warn('Could not strip metadata from file: %s' %
info['tmp_file'])
fi = open(info['tmp_file'], 'r')
try:
# Try to determine which kind of NamedBlob we need
# This will suffice for standard p.a.contenttypes File/Image
# and any other custom type that would have 'File' or 'Image' in
# its type name
filename = ploneutils.safe_unicode(filename)
create_opts = dict(type=type_, id=newid, container=folder)
if 'Image' in type_:
image = NamedBlobImage(data=fi, filename=filename)
try:
image.focal_point = [
float(self.request.form.get('focalX')),
float(self.request.form.get('focalY'))
]
except Exception:
pass
create_opts['image'] = image
else:
create_opts['file'] = NamedBlobFile(data=fi, filename=filename)
for field in get_upload_fields():
if not field.get('name'):
continue
name = field['name']
if not self.request.form.get(name):
continue
if name in ('tags', 'subject'):
# tags needs to be converted
create_opts['subject'] = self.request.form.get(name).split(
';')
else:
create_opts[name] = self.request.form.get(name, '')
return api.content.create(**create_opts)
finally:
fi.close()