mirror of
https://github.com/taroved/pol
synced 2025-05-16 22:30:09 -07:00
102 lines
3.6 KiB
Python
102 lines
3.6 KiB
Python
import w3lib.url
|
|
import w3lib.html
|
|
|
|
from lxml import etree
|
|
import re
|
|
from hashlib import md5
|
|
|
|
from feedgenerator import Rss201rev2Feed, Enclosure
|
|
import datetime
|
|
|
|
import MySQLdb
|
|
from settings import DATABASES, DOWNLOADER_USER_AGENT
|
|
|
|
url_hash_regexp = re.compile('(#.*)?$')
|
|
|
|
def element_to_string(element):
|
|
s = [element.text] if element.text else []
|
|
for sub_element in element:
|
|
s.append(etree.tostring(sub_element))
|
|
if sub_element.tail:
|
|
s.append(element.tail)
|
|
return ''.join(s)
|
|
|
|
def _build_link(html, doc_url, url):
|
|
base_url = w3lib.html.get_base_url(html, doc_url)
|
|
return w3lib.url.urljoin_rfc(base_url, url)
|
|
|
|
def buildFeed(response, feed_config):
|
|
response.selector.remove_namespaces()
|
|
|
|
tree = response.selector._root.getroottree()
|
|
# get data from html
|
|
items = []
|
|
for node in tree.xpath(feed_config['xpath']):
|
|
item = {}
|
|
title_link = None
|
|
for field_name in ['title', 'description']:
|
|
if field_name in feed_config['fields']:
|
|
element = node.xpath(feed_config['fields'][field_name])
|
|
if element:
|
|
item[field_name] = element_to_string(element[0])
|
|
# get item link
|
|
if field_name == 'title':
|
|
anchor = element[0].xpath('ancestor-or-self::node()[name()="a"]')
|
|
if anchor and anchor[0].get('href'):
|
|
title_link = _build_link(response.body_as_unicode(), feed_config['uri'], anchor[0].get('href'))
|
|
|
|
if len(item) == len(feed_config['fields']): # all fields are required
|
|
item['title_link'] = title_link
|
|
items.append(item)
|
|
|
|
title = response.selector.xpath('//title/text()').extract()
|
|
|
|
#build feed
|
|
feed = Rss201rev2Feed(
|
|
title = title[0] if title else 'Polite Pol: ' + feed_config['uri'],
|
|
link=feed_config['uri'],
|
|
description="Generated by PolitePol.com.\n"+\
|
|
"Source page url: " + feed_config['uri'],
|
|
language="en",
|
|
)
|
|
for item in items:
|
|
title = item['title'] if 'title' in item else ''
|
|
desc = item['description'] if 'description' in item else ''
|
|
if item['title_link']:
|
|
link = item['title_link']
|
|
else:
|
|
link = url_hash_regexp.sub('#' + md5((title+desc).encode('utf-8')).hexdigest(), feed_config['uri'])
|
|
feed.add_item(
|
|
title = title,
|
|
link = link,
|
|
description = desc,
|
|
#enclosure=Enclosure(fields[4], "32000", "image/jpeg") if 4 in fields else None, #"Image"
|
|
pubdate=datetime.datetime.now()
|
|
)
|
|
return feed.writeString('utf-8')
|
|
|
|
def getFeedData(request, feed_id):
|
|
# get url, xpathes
|
|
creds = DATABASES['default']
|
|
db = MySQLdb.connect(host=creds['HOST'], port=int(creds['PORT']), user=creds['USER'], passwd=creds['PASSWORD'], db=creds['NAME'])
|
|
feed = {}
|
|
with db:
|
|
cur = db.cursor()
|
|
cur.execute("""select f.uri, f.xpath, fi.name, ff.xpath from frontend_feed f
|
|
right join frontend_feedfield ff on ff.feed_id=f.id
|
|
left join frontend_field fi on fi.id=ff.field_id
|
|
where f.id=%s""", (feed_id,))
|
|
rows = cur.fetchall()
|
|
|
|
for row in rows:
|
|
if not feed:
|
|
feed['uri'] = row[0]
|
|
feed['xpath'] = row[1]
|
|
feed['fields'] = {}
|
|
feed['fields'][row[2]] = row[3]
|
|
|
|
if feed:
|
|
return [feed['uri'], feed]
|
|
else:
|
|
return 'Feed generator error: config of feed is empty'
|