from functools import partial
import os
from lxml import html, cssselect
from sync_media_function import sync_media
from metafuncs import branch, combine, maybe, tryit, getitem, cache, Pipe
# Support functions
is_not_none = lambda obj: obj is not None
unique = lambda sequence: list(set(sequence))
get_html = html.parse # :: str -> ElementTree
img_tags = cssselect.CSSSelector('img') # :: ElementTree -> List[Element]
get_src = maybe(getitem('src'), # normal
maybe(getitem('data-src'), # photo pages
maybe(getitem('data-srcset')))) # home page
BACKGROUND_IMAGE_REGEX = re.compile(r'background-image: url\((.*?\))')
read_page = lambda url: urllib2.urlopen(url).read()
path_to_url = lambda path: "file://"+os.path.abspath(path)
PATH_SPLIT_REGEX = re.compile(r'^(/media)/(.*)/(.*?\..*?)$')
# PATH_SPLIT_REGEX = re.compile(r'/media/(.*)/(')
split_path_parts = lambda src: PATH_SPLIT_REGEX.match(src).groups() # :: str -> PathPats
MEDIA_PATH_REGEX = re.compile(r'/media/')
is_media_path = lambda path: MEDIA_PATH_REGEX.search(path) is not None
fixup_cdn = lambda url: "http:"+url if url.startswith('//cdn') else url
def crop(prefix):
def wrapped(word):
from sync_media_function import sync_media
from metafuncs import branch, combine, maybe, tryit, getitem, cache
# Composable = Pipe
from monad.composable import Composable
import functools
F = functools.partial
# Support functions
is_not_none = lambda obj: obj is not None
unique = lambda sequence: list(set(sequence))
get_html = html.parse # :: str -> ElementTree
img_tags = cssselect.CSSSelector('img') # :: ElementTree -> List[Element]
get_src = maybe(getitem('src'), # normal
maybe(getitem('data-src'), # photo pages
maybe(getitem('data-srcset')))) # home page
BACKGROUND_IMAGE_REGEX = re.compile(r'background-image: url\((.*?\))')
read_page = lambda url: urllib2.urlopen(url).read()
PATH_SPLIT_REGEX = re.compile(r'^(/media)/(.*)/(.*?\..*?)$')
# PATH_SPLIT_REGEX = re.compile(r'/media/(.*)/(')
split_path_parts = lambda src: PATH_SPLIT_REGEX.match(src).groups() # :: str -> PathPats
MEDIA_PATH_REGEX = re.compile(r'/media/')
is_media_path = lambda path: MEDIA_PATH_REGEX.search(path) is not None
# Composite work-horse functions
# Retreive src-like properties from <img> tags
get_img_srcs = (
Composable() # :: Location
>> cache(get_html) # :: ElementTree
>> img_tags # :: List[Element]