def transmogrify(self, item):
self.seen_count += 1
original_path_key = self.originalPathKey(*item.keys())[0]
if not original_path_key:
# not enough info
raise NothingToDoHere
path = self.get_path(item)
original_paths = item[original_path_key]
if isinstance(original_paths, basestring):
original_paths = [
original_paths,
]
original_paths = [self._prepare_path(p) for p in original_paths]
if self.assure_target_exists:
# Bails out if can't retrieve object at item's current path:
if self.get_object(item, raise_=False) is None:
logger.warn("Ignoring item at %s - object not created" % path)
raise NothingToDoHere
if not path.startswith(self.portal_path):
path = self.portal_path + "/" + path.lstrip("/")
for original_path in original_paths:
if original_path != path:
self.redirector.add(original_path, path)
self.changed_count += 1
return item
def transmogrify(self, item):
self.seen_count += 1
original_path_key = self.originalPathKey(*item.keys())[0]
if not original_path_key:
# not enough info
raise NothingToDoHere
path = self.get_path(item)
original_paths = item[original_path_key]
if isinstance(original_paths, basestring):
original_paths = [original_paths, ]
original_paths = [self._prepare_path(p) for p in original_paths]
if self.assure_target_exists:
# Bails out if can't retrieve object at item's current path:
if self.get_object(item, raise_=False) is None:
logger.warn("Ignoring item at %s - object not created" % path)
raise NothingToDoHere
if not path.startswith(self.portal_path):
path = self.portal_path + "/" + path.lstrip("/")
for original_path in original_paths:
if original_path != path:
self.redirector.add(original_path, path)
self.changed_count += 1
return item
def get_remote_image(url, item, img_title="", pathkey="_path",
jsonmigrator = False):
# FIXME: not making an extra call to get the real pathkey
logger.info("""Fetching image %s for article %s """ % (url,
item.get(pathkey, "")))
# Strip plone view from URL:
url, it_worked = _strip_view(url)
# it won't work if the image url does not have a proper image
if not it_worked:
jsonmigrator = False
if jsonmigrator:
url += "/get_item"
try:
http = urllib.urlopen(url)
image_data = http.read()
if http.code > 399:
# we can't get the image here
raise IOError
real_url = http.url
except Exception as error:
logger.error("Could not retrieve image at %s: %s - skipping" %
(url, error))
return None, None, []
if jsonmigrator and real_url.endswith("/get_item"):
real_url = real_url[:len("/get_item")]
image_filename, post_parts = _get_filename(real_url)
if jsonmigrator:
try:
image = json.loads(image_data)
except ValueError:
logger.warn("Could not retrieve image json contents at %s " % url)
raise None, None, []
else: # build object item for the pipeline
image = {}
image["_type"] = "Image"
image["image"] = image_data
image["creation_date"] = item.get("creation_date", None)
image["modification_date"] = item.get("modification_date", None)
image["_transitions"] = item.get("_transitions", "published")
if not img_title:
img_title = image_filename.split(".")[0].encode("utf-8")
image["title"] = img_title
image["_filename"] = image_filename
return real_url, image, post_parts
def transmogrify(self, item):
if (not "image" in item):
raise NothingToDoHere
if isinstance(item["image"] , dict):
data = item["image"]["data"]
else:
data = item["image"]
try:
data= StringIO(data)
img = Image.open(data)
except (TypeError, IOError) as error:
# We caught it: a fake image! :-)
path = self.get_path(item)
if item.get("_type", "") == "Image":
logger.warn("Image at %s contains no image data - discarding"
% path)
raise ThouShallNotPass
logger.warn("Item at %s contains no image data - scrubbing"
" incorrecet data" % path)
del item["image"]
return item
