def truncate_text(text):
settings = get_feeds_settings()
truncate_words = settings.getint("FEEDS_CONFIG_TRUNCATE_WORDS")
if truncate_words and truncate_words > 0:
tree = build_tree(text)[0]
# Assume that a word has 5 characters on average.
truncate_tree(tree, truncate_words * 5)
text = serialize_tree(tree)
return text
def truncate_text(text):
settings = get_feeds_settings()
truncate_words = settings.getint("FEEDS_CONFIG_TRUNCATE_WORDS")
if truncate_words and truncate_words > 0:
tree = build_tree(text)[0]
# Assume that a word has 5 characters on average.
truncate_tree(tree, truncate_words * 5)
text = serialize_tree(tree)
return text
def remove_elems(tree, loader_context):
remove_elems = []
settings = get_feeds_settings()
remove_images = settings.getbool("FEEDS_CONFIG_REMOVE_IMAGES")
if remove_images:
remove_elems += ["img"]
# Remove tags.
for elem_sel in loader_context.get("remove_elems", []) + remove_elems:
selector = CSSSelector(elem_sel)
for elem in selector(tree):
elem.drop_tree()
for elem_sel in loader_context.get("remove_elems_xpath", []):
for elem in tree.xpath(elem_sel):
elem.drop_tree()
return [tree]
def remove_elems(tree, loader_context):
remove_elems = []
settings = get_feeds_settings()
remove_images = settings.getbool("FEEDS_CONFIG_REMOVE_IMAGES")
if remove_images:
remove_elems += ["img"]
# Remove tags.
for elem_sel in loader_context.get("remove_elems", []) + remove_elems:
selector = CSSSelector(elem_sel)
for elem in selector(tree):
elem.drop_tree()
for elem_sel in loader_context.get("remove_elems_xpath", []):
for elem in tree.xpath(elem_sel):
elem.drop_tree()
return [tree]