New git url and updated date in provided index.html

Without it, some websites were accidentally matching it (false positives)

README.md

@@ -24,15 +24,13 @@ hand-written rules (ie. there's no automatic detection of links to build feeds).
 Please mind that feeds based on html files may stop working unexpectedly, due to
 html structure changes on the target website.
 
-Additionally morss can grab the source xml feed of iTunes podcast, and detect
+Additionally morss can detect rss feeds in html pages' `<meta>`.
-rss feeds in html pages' `<meta>`.
 
 You can use this program online for free at **[morss.it](https://morss.it/)**.
 
 Some features of morss:
 - Read RSS/Atom feeds
 - Create RSS feeds from json/html pages
-- Convert iTunes podcast links into xml links
 - Export feeds as RSS/JSON/CSV/HTML
 - Fetch full-text content of feed items
 - Follow 301/meta redirects
@@ -48,6 +46,7 @@ You do need:
 
 - [python](http://www.python.org/) >= 2.6 (python 3 is supported)
 - [lxml](http://lxml.de/) for xml parsing
+- [bs4](https://pypi.org/project/bs4/) for badly-formatted html pages
 - [dateutil](http://labix.org/python-dateutil) to parse feed dates
 - [chardet](https://pypi.python.org/pypi/chardet)
 - [six](https://pypi.python.org/pypi/six), a dependency of chardet
@@ -74,9 +73,10 @@ The arguments are:
 
 - Change what morss does
 	- `json`: output as JSON
+	- `html`: outpout as HTML
+	- `csv`: outpout as CSV
 	- `proxy`: doesn't fill the articles
 	- `clip`: stick the full article content under the original feed content (useful for twitter)
-	- `keep`: by default, morss does drop feed description whenever the full-content is found (so as not to mislead users who use Firefox, since the latter only shows the description in the feed preview, so they might believe morss doens't work), but with this argument, the description is kept
 	- `search=STRING`: does a basic case-sensitive search in the feed
 - Advanced
 	- `csv`: export to csv
@@ -88,11 +88,9 @@ The arguments are:
 	- `mono`: disable multithreading while fetching, makes debugging easier
 	- `theforce`: force download the rss feed and ignore cached http errros
 	- `silent`: don't output the final RSS (useless on its own, but can be nice when debugging)
-	- `encoding=ENCODING`: overrides the encoding auto-detection of the crawler. Some web developers did not quite understand the importance of setting charset/encoding tags correctly...
 - http server only
 	- `callback=NAME`: for JSONP calls
 	- `cors`: allow Cross-origin resource sharing (allows XHR calls from other servers)
-	- `html`: changes the http content-type to html, so that python cgi erros (written in html) are readable in a web browser
 	- `txt`: changes the http content-type to txt (for faster "`view-source:`")
 - Custom feeds: you can turn any HTML page into a RSS feed using morss, using xpath rules. The article content will be fetched as usual (with readabilite). Please note that you will have to **replace** any `/` in your rule with a `|` when using morss as a webserver
 	- `items`: (**mandatory** to activate the custom feeds function) xpath rule to match all the RSS entries

main.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python
 
-from morss import main, cgi_wrapper as application
+from morss import main, cgi_standalone_app as application
 
 if __name__ == '__main__':
     main()

morss/crawler.py

@@ -27,13 +27,33 @@ except NameError:
 
 MIMETYPE = {
     'xml': ['text/xml', 'application/xml', 'application/rss+xml', 'application/rdf+xml', 'application/atom+xml', 'application/xhtml+xml'],
+    'rss': ['application/rss+xml', 'application/rdf+xml', 'application/atom+xml'],
     'html': ['text/html', 'application/xhtml+xml', 'application/xml']}
 
 
 DEFAULT_UA = 'Mozilla/5.0 (X11; Linux x86_64; rv:25.0) Gecko/20100101 Firefox/25.0'
 
 
-def custom_handler(accept=None, strict=False, delay=None, encoding=None, basic=False):
+def get(*args, **kwargs):
+    return adv_get(*args, **kwargs)[0]
+
+
+def adv_get(url, timeout=None, *args, **kwargs):
+    if timeout is None:
+        con = custom_handler(*args, **kwargs).open(url)
+
+    else:
+        con = custom_handler(*args, **kwargs).open(url, timeout=timeout)
+
+    data = con.read()
+
+    contenttype = con.info().get('Content-Type', '').split(';')[0]
+    encoding= detect_encoding(data, con)
+
+    return data, con, contenttype, encoding
+
+
+def custom_handler(follow=None, delay=None, encoding=None):
     handlers = []
 
     # as per urllib2 source code, these Handelers are added first
@@ -51,14 +71,12 @@ def custom_handler(accept=None, strict=False, delay=None, encoding=None, basic=F
     handlers.append(HTTPEquivHandler())
     handlers.append(HTTPRefreshHandler())
     handlers.append(UAHandler(DEFAULT_UA))
-
+    handlers.append(BrowserlyHeaderHandler())
-    if not basic:
-        handlers.append(AutoRefererHandler())
 
     handlers.append(EncodingFixHandler(encoding))
 
-    if accept:
+    if follow:
-        handlers.append(ContentNegociationHandler(MIMETYPE[accept], strict))
+        handlers.append(AlternateHandler(MIMETYPE[follow]))
 
     handlers.append(CacheHandler(force_min=delay))
 
@@ -196,45 +214,33 @@ class UAHandler(BaseHandler):
     https_request = http_request
 
 
-class AutoRefererHandler(BaseHandler):
+class BrowserlyHeaderHandler(BaseHandler):
+    """ Add more headers to look less suspicious """
+
     def http_request(self, req):
-        req.add_unredirected_header('Referer', 'http://%s' % req.host)
+        req.add_unredirected_header('Accept', 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8')
+        req.add_unredirected_header('Accept-Language', 'en-US,en;q=0.5')
         return req
 
     https_request = http_request
 
 
-class ContentNegociationHandler(BaseHandler):
+class AlternateHandler(BaseHandler):
-    " Handler for content negociation. Also parses <link rel='alternate' type='application/rss+xml' href='...' /> "
+    " Follow <link rel='alternate' type='application/rss+xml' href='...' /> "
 
-    def __init__(self, accept=None, strict=False):
+    def __init__(self, follow=None):
-        self.accept = accept
+        self.follow = follow or []
-        self.strict = strict
-
-    def http_request(self, req):
-        if self.accept is not None:
-            if isinstance(self.accept, basestring):
-                self.accept = (self.accept,)
-
-            string = ','.join(self.accept)
-
-            if self.strict:
-                string += ',*/*;q=0.9'
-
-            req.add_unredirected_header('Accept', string)
-
-        return req
 
     def http_response(self, req, resp):
         contenttype = resp.info().get('Content-Type', '').split(';')[0]
-        if 200 <= resp.code < 300 and self.accept is not None and self.strict and contenttype in MIMETYPE['html'] and contenttype not in self.accept:
+        if 200 <= resp.code < 300 and len(self.follow) and contenttype in MIMETYPE['html'] and contenttype not in self.follow:
             # opps, not what we were looking for, let's see if the html page suggests an alternative page of the right types
 
             data = resp.read()
             links = lxml.html.fromstring(data[:10000]).findall('.//link[@rel="alternate"]')
 
             for link in links:
-                if link.get('type', '') in self.accept:
+                if link.get('type', '') in self.follow:
                     resp.code = 302
                     resp.msg = 'Moved Temporarily'
                     resp.headers['location'] = link.get('href')
@@ -246,7 +252,6 @@ class ContentNegociationHandler(BaseHandler):
 
         return resp
 
-    https_request = http_request
     https_response = http_response
 
 
@@ -384,7 +389,7 @@ class CacheHandler(BaseHandler):
 
         elif  self.force_min is None and ('no-cache' in cc_list
                                         or 'no-store' in cc_list
-                                        or ('private' in cc_list and not self.private)):
+                                        or ('private' in cc_list and not self.private_cache)):
             # kindly follow web servers indications, refresh
             return None
 
@@ -419,7 +424,7 @@ class CacheHandler(BaseHandler):
 
             cc_list = [x for x in cache_control if '=' not in x]
 
-            if 'no-cache' in cc_list or 'no-store' in cc_list or ('private' in cc_list and not self.private):
+            if 'no-cache' in cc_list or 'no-store' in cc_list or ('private' in cc_list and not self.private_cache):
                 # kindly follow web servers indications
                 return resp
 
@@ -513,18 +518,20 @@ import pymysql.cursors
 
 
 class MySQLCacheHandler(BaseCache):
-    " NB. Requires mono-threading, as pymysql isn't thread-safe "
     def __init__(self, user, password, database, host='localhost'):
-        self.con = pymysql.connect(host=host, user=user, password=password, database=database, charset='utf8', autocommit=True)
+        self.user = user
+        self.password = password
+        self.database = database
+        self.host = host
 
-        with self.con.cursor() as cursor:
+        with self.cursor() as cursor:
             cursor.execute('CREATE TABLE IF NOT EXISTS data (url VARCHAR(255) NOT NULL PRIMARY KEY, code INT, msg TEXT, headers TEXT, data BLOB, timestamp INT)')
 
-    def __del__(self):
+    def cursor(self):
-        self.con.close()
+        return pymysql.connect(host=self.host, user=self.user, password=self.password, database=self.database, charset='utf8', autocommit=True).cursor()
 
     def __getitem__(self, url):
-        cursor = self.con.cursor()
+        cursor = self.cursor()
         cursor.execute('SELECT * FROM data WHERE url=%s', (url,))
         row = cursor.fetchone()
 
@@ -535,10 +542,10 @@ class MySQLCacheHandler(BaseCache):
 
     def __setitem__(self, url, value): # (code, msg, headers, data, timestamp)
         if url in self:
-            with self.con.cursor() as cursor:
+            with self.cursor() as cursor:
                 cursor.execute('UPDATE data SET code=%s, msg=%s, headers=%s, data=%s, timestamp=%s WHERE url=%s',
                     value + (url,))
 
         else:
-            with self.con.cursor() as cursor:
+            with self.cursor() as cursor:
                 cursor.execute('INSERT INTO data VALUES (%s,%s,%s,%s,%s,%s)', (url,) + value)

morss/feedify.ini

@@ -90,6 +90,9 @@ item_updated = updated
 [html]
 mode = html
 
+path =
+  http://localhost/
+
 title = //div[@id='header']/h1
 desc = //div[@id='header']/h2
 items = //div[@id='content']/div
@@ -99,7 +102,7 @@ item_link = ./a/@href
 item_desc = ./div[class=desc]
 item_content = ./div[class=content]
 
-base = <!DOCTYPE html> <html> <head> <title>Feed reader by morss</title> <meta name="viewport" content="width=device-width; initial-scale=1.0; maximum-scale=1.0;" /> </head> <body> <div id="header"> <h1>@feed.title</h1> <h2>@feed.desc</h2> <p>- via morss</p> </div> <div id="content"> <div class="item"> <a class="title link" href="@item.link" target="_blank">@item.title</a> <div class="desc">@item.desc</div> <div class="content">@item.content</div> </div> </div> <script> var items = document.getElementsByClassName('item') for (var i in items) items[i].onclick = function() { this.classList.toggle('active') document.body.classList.toggle('noscroll') } </script> </body> </html>
+base = file:reader.html.template
 
 [twitter]
 mode = html

morss/feedify.py

@@ -1,28 +0,0 @@
-import re
-import json
-
-from . import crawler
-
-try:
-    basestring
-except NameError:
-    basestring = str
-
-
-def pre_worker(url):
-    if url.startswith('http://itunes.apple.com/') or url.startswith('https://itunes.apple.com/'):
-        match = re.search('/id([0-9]+)(\?.*)?$', url)
-        if match:
-            iid = match.groups()[0]
-            redirect = 'https://itunes.apple.com/lookup?id=%s' % iid
-
-            try:
-                con = crawler.custom_handler(basic=True).open(redirect, timeout=4)
-                data = con.read()
-
-            except (IOError, HTTPException):
-                raise
-
-            return json.loads(data.decode('utf-8', 'replace'))['results'][0]['feedUrl']
-
-    return None

morss/feeds.py

@@ -15,6 +15,7 @@ import dateutil.parser
 from copy import deepcopy
 
 import lxml.html
+from .readabilite import parse as html_parse
 
 json.encoder.c_make_encoder = None
 
@@ -46,13 +47,17 @@ def parse_rules(filename=None):
 
     for section in rules.keys():
         for arg in rules[section].keys():
-            if '\n' in rules[section][arg]:
+            if rules[section][arg].startswith('file:'):
+                import_file = os.path.join(os.path.dirname(__file__), rules[section][arg][5:])
+                rules[section][arg] = open(import_file).read()
+
+            elif '\n' in rules[section][arg]:
                 rules[section][arg] = rules[section][arg].split('\n')[1:]
 
     return rules
 
 
-def parse(data, url=None, mimetype=None):
+def parse(data, url=None, mimetype=None, encoding=None):
     " Determine which ruleset to use "
 
     rulesets = parse_rules()
@@ -66,26 +71,20 @@ def parse(data, url=None, mimetype=None):
                 for path in ruleset['path']:
                     if fnmatch(url, path):
                         parser = [x for x in parsers if x.mode == ruleset['mode']][0]
-                        return parser(data, ruleset) 
+                        return parser(data, ruleset, encoding=encoding) 
 
-    # 2) Look for a parser based on mimetype
+    # 2) Try each and every parser
-
-    if mimetype is not None:
-        parser_candidates = [x for x in parsers if mimetype in x.mimetype]
-
-    if mimetype is None or parser_candidates is None:
-        parser_candidates = parsers
 
     # 3) Look for working ruleset for given parser
         # 3a) See if parsing works
         # 3b) See if .items matches anything
 
-    for parser in parser_candidates:
+    for parser in parsers:
         ruleset_candidates = [x for x in rulesets.values() if x['mode'] == parser.mode and 'path' not in x]
             # 'path' as they should have been caught beforehands
 
         try:
-            feed = parser(data)
+            feed = parser(data, encoding=encoding)
 
         except (ValueError):
             # parsing did not work
@@ -112,7 +111,7 @@ def parse(data, url=None, mimetype=None):
 
 
 class ParserBase(object):
-    def __init__(self, data=None, rules=None, parent=None):
+    def __init__(self, data=None, rules=None, parent=None, encoding=None):
         if rules is None:
             rules = parse_rules()[self.default_ruleset]
 
@@ -121,9 +120,10 @@ class ParserBase(object):
         if data is None:
             data = rules['base']
 
-        self.root = self.parse(data)
         self.parent = parent
+        self.encoding = encoding
 
+        self.root = self.parse(data)
 
     def parse(self, raw):
         pass
@@ -148,15 +148,15 @@ class ParserBase(object):
         c = csv.writer(out, dialect=csv.excel)
 
         for item in self.items:
-            row = [getattr(item, x) for x in item.dic]
+            c.writerow([getattr(item, x) for x in item.dic])
-
-            if encoding != 'unicode':
-                row = [x.encode(encoding) if isinstance(x, unicode) else x for x in row]
-
-            c.writerow(row)
 
         out.seek(0)
-        return out.read()
+        out = out.read()
+
+        if encoding != 'unicode':
+            out = out.encode(encoding)
+
+        return out
 
     def tohtml(self, **k):
         return self.convert(FeedHTML).tostring(**k)
@@ -267,8 +267,15 @@ class ParserBase(object):
 
         except AttributeError:
             # does not exist, have to create it
-            self.rule_create(self.rules[rule_name])
+            try:
-            self.rule_set(self.rules[rule_name], value)
+                self.rule_create(self.rules[rule_name])
+
+            except AttributeError:
+                # no way to create it, give up
+                pass
+
+            else:
+                self.rule_set(self.rules[rule_name], value)
 
     def rmv(self, rule_name):
         # easy deleter
@@ -401,13 +408,14 @@ class ParserXML(ParserBase):
 
         else:
             if html_rich:
-                # atom stuff
-                if 'atom' in rule:
-                    match.attrib['type'] = 'xhtml'
-
                 self._clean_node(match)
                 match.append(lxml.html.fragment_fromstring(value, create_parent='div'))
-                match.find('div').drop_tag()
+
+                if self.rules['mode'] == 'html':
+                    match.find('div').drop_tag() # not supported by lxml.etree
+
+                else: # i.e. if atom
+                    match.attrib['type'] = 'xhtml'
 
             else:
                 if match is not None and len(match):
@@ -440,8 +448,7 @@ class ParserHTML(ParserXML):
     mimetype = ['text/html', 'application/xhtml+xml']
 
     def parse(self, raw):
-        parser = etree.HTMLParser(remove_blank_text=True) # remove_blank_text needed for pretty_print
+        return html_parse(raw, encoding=self.encoding)
-        return etree.fromstring(raw, parser)
 
     def tostring(self, encoding='unicode', **k):
         return lxml.html.tostring(self.root, encoding=encoding, **k)
@@ -467,6 +474,9 @@ class ParserHTML(ParserXML):
             element = deepcopy(match)
             match.getparent().append(element)
 
+        else:
+            raise AttributeError('no way to create item')
+
 
 def parse_time(value):
     if value is None or value == 0:
@@ -474,13 +484,13 @@ def parse_time(value):
 
     elif isinstance(value, basestring):
         if re.match(r'^[0-9]+$', value):
-            return datetime.fromtimestamp(int(value), tz.UTC)
+            return datetime.fromtimestamp(int(value), tz.tzutc())
 
         else:
-            return dateutil.parser.parse(value)
+            return dateutil.parser.parse(value).replace(tzinfo=tz.tzutc())
 
     elif isinstance(value, int):
-        return datetime.fromtimestamp(value, tz.UTC)
+        return datetime.fromtimestamp(value, tz.tzutc())
 
     elif isinstance(value, datetime):
         return value

morss/morss.py

@@ -1,7 +1,10 @@
 import sys
 import os
 import os.path
+
 import time
+from datetime import datetime
+from dateutil import tz
 
 import threading
 
@@ -12,25 +15,25 @@ import lxml.etree
 import lxml.html
 
 from . import feeds
-from . import feedify
 from . import crawler
 from . import readabilite
 
 import wsgiref.simple_server
 import wsgiref.handlers
+import cgitb
 
 
 try:
     # python 2
     from Queue import Queue
     from httplib import HTTPException
-    from urllib import quote_plus
+    from urllib import unquote
     from urlparse import urlparse, urljoin, parse_qs
 except ImportError:
     # python 3
     from queue import Queue
     from http.client import HTTPException
-    from urllib.parse import quote_plus
+    from urllib.parse import unquote
     from urllib.parse import urlparse, urljoin, parse_qs
 
 LIM_ITEM = 100  # deletes what's beyond
@@ -44,7 +47,7 @@ THREADS = 10  # number of threads (1 for single-threaded)
 DEBUG = False
 PORT = 8080
 
-PROTOCOL = ['http', 'https', 'ftp']
+PROTOCOL = ['http', 'https']
 
 
 def filterOptions(options):
@@ -52,7 +55,7 @@ def filterOptions(options):
 
     # example of filtering code below
 
-    #allowed = ['proxy', 'clip', 'keep', 'cache', 'force', 'silent', 'pro', 'debug']
+    #allowed = ['proxy', 'clip', 'cache', 'force', 'silent', 'pro', 'debug']
     #filtered = dict([(key,value) for (key,value) in options.items() if key in allowed])
 
     #return filtered
@@ -66,6 +69,7 @@ def log(txt, force=False):
     if DEBUG or force:
         if 'REQUEST_URI' in os.environ:
             open('morss.log', 'a').write("%s\n" % repr(txt))
+
         else:
             print(repr(txt))
 
@@ -73,6 +77,7 @@ def log(txt, force=False):
 def len_html(txt):
     if len(txt):
         return len(lxml.html.fromstring(txt).text_content())
+
     else:
         return 0
 
@@ -80,6 +85,7 @@ def len_html(txt):
 def count_words(txt):
     if len(txt):
         return len(lxml.html.fromstring(txt).text_content().split())
+
     return 0
 
 
@@ -88,12 +94,14 @@ class Options:
         if len(args):
             self.options = args
             self.options.update(options or {})
+
         else:
             self.options = options or {}
 
     def __getattr__(self, key):
         if key in self.options:
             return self.options[key]
+
         else:
             return False
 
@@ -107,17 +115,23 @@ class Options:
 def parseOptions(options):
     """ Turns ['md=True'] into {'md':True} """
     out = {}
+
     for option in options:
         split = option.split('=', 1)
+
         if len(split) > 1:
             if split[0].lower() == 'true':
                 out[split[0]] = True
+
             elif split[0].lower() == 'false':
                 out[split[0]] = False
+
             else:
                 out[split[0]] = split[1]
+
         else:
             out[split[0]] = True
+
     return out
 
 
@@ -158,6 +172,11 @@ def ItemFix(item, feedurl='/'):
         item.link = parse_qs(urlparse(item.link).query)['url'][0]
         log(item.link)
 
+    # pocket
+    if fnmatch(item.link, 'https://getpocket.com/redirect?url=*'):
+        item.link = parse_qs(urlparse(item.link).query)['url'][0]
+        log(item.link)
+
     # facebook
     if fnmatch(item.link, 'https://www.facebook.com/l.php?u=*'):
         item.link = parse_qs(urlparse(item.link).query)['u'][0]
@@ -208,6 +227,7 @@ def ItemFill(item, options, feedurl='/', fast=False):
         if len(match):
             link = match[0]
             log(link)
+
         else:
             link = None
 
@@ -217,6 +237,7 @@ def ItemFill(item, options, feedurl='/', fast=False):
         if len(match) and urlparse(match[0]).netloc != 'www.facebook.com':
             link = match[0]
             log(link)
+
         else:
             link = None
 
@@ -232,19 +253,17 @@ def ItemFill(item, options, feedurl='/', fast=False):
         delay = -2
 
     try:
-        con = crawler.custom_handler('html', False, delay, options.encoding).open(link, timeout=TIMEOUT)
+        data, con, contenttype, encoding = crawler.adv_get(url=link, delay=delay, timeout=TIMEOUT)
-        data = con.read()
 
     except (IOError, HTTPException) as e:
         log('http error')
         return False # let's just delete errors stuff when in cache mode
 
-    contenttype = con.info().get('Content-Type', '').split(';')[0]
     if contenttype not in crawler.MIMETYPE['html'] and contenttype != 'text/plain':
         log('non-text page')
         return True
 
-    out = readabilite.get_article(data, link, options.encoding or crawler.detect_encoding(data, con))
+    out = readabilite.get_article(data, url=con.geturl(), encoding=encoding)
 
     if out is not None:
         item.content = out
@@ -268,9 +287,6 @@ def ItemAfter(item, options):
         item.content = item.desc + "<br/><br/><center>* * *</center><br/><br/>" + item.content
         del item.desc
 
-    if not options.keep and not options.proxy:
-        del item.desc
-
     if options.nolink and item.content:
         content = lxml.html.fromstring(item.content)
         for link in content.xpath('//a'):
@@ -284,27 +300,23 @@ def ItemAfter(item, options):
     return item
 
 
-def FeedFetch(url, options):
+def UrlFix(url):
-    # basic url clean-up
     if url is None:
         raise MorssException('No url provided')
 
+    if isinstance(url, bytes):
+        url = url.decode()
+
     if urlparse(url).scheme not in PROTOCOL:
         url = 'http://' + url
         log(url)
 
     url = url.replace(' ', '%20')
 
-    if isinstance(url, bytes):
+    return url
-        url = url.decode()
 
-    # allow for code execution for feedify
-    pre = feedify.pre_worker(url)
-    if pre:
-        url = pre
-        log('url redirect')
-        log(url)
 
+def FeedFetch(url, options):
     # fetch feed
     delay = DELAY
 
@@ -312,35 +324,34 @@ def FeedFetch(url, options):
         delay = 0
 
     try:
-        con = crawler.custom_handler(accept='xml', strict=True, delay=delay,
+        xml, con, contenttype, encoding = crawler.adv_get(url=url, follow='rss', delay=delay, timeout=TIMEOUT * 2)
-            encoding=options.encoding, basic=not options.items) \
-            .open(url, timeout=TIMEOUT * 2)
-        xml = con.read()
 
     except (IOError, HTTPException):
         raise MorssException('Error downloading feed')
 
-    contenttype = con.info().get('Content-Type', '').split(';')[0]
-
     if options.items:
         # using custom rules
-        rss = feeds.FeedHTML(xml, url, contenttype)
+        rss = feeds.FeedHTML(xml, encoding=encoding)
-        feed.rule
+
+        rss.rules['title'] = options.title              if options.title        else '//head/title'
+        rss.rules['desc'] = options.desc                if options.desc         else '//head/meta[@name="description"]/@content'
 
         rss.rules['items'] = options.items
 
-        if options.item_title:
+        rss.rules['item_title'] = options.item_title    if options.item_title   else './/a|.'
-            rss.rules['item_title'] = options.item_title
+        rss.rules['item_link'] = options.item_link      if options.item_link    else './@href|.//a/@href'
-        if options.item_link:
+
-            rss.rules['item_link'] = options.item_link
         if options.item_content:
             rss.rules['item_content'] = options.item_content
+
         if options.item_time:
             rss.rules['item_time'] = options.item_time
 
+        rss = rss.convert(feeds.FeedXML)
+
     else:
         try:
-            rss = feeds.parse(xml, url, contenttype)
+            rss = feeds.parse(xml, url, contenttype, encoding=encoding)
             rss = rss.convert(feeds.FeedXML)
                 # contains all fields, otherwise much-needed data can be lost
 
@@ -375,6 +386,7 @@ def FeedGather(rss, url, options):
             value = queue.get()
             try:
                 worker(*value)
+
             except Exception as e:
                 log('Thread Error: %s' % e.message)
             queue.task_done()
@@ -411,9 +423,12 @@ def FeedGather(rss, url, options):
         t.daemon = True
         t.start()
 
-    for i, item in enumerate(list(rss.items)):
+    now = datetime.now(tz.tzutc())
+    sorted_items = sorted(rss.items, key=lambda x:x.updated or x.time or now, reverse=True)
+    for i, item in enumerate(sorted_items):
         if threads == 1:
             worker(*[i, item])
+
         else:
             queue.put([i, item])
 
@@ -433,37 +448,38 @@ def FeedGather(rss, url, options):
     return rss
 
 
-def FeedFormat(rss, options):
+def FeedFormat(rss, options, encoding='utf-8'):
     if options.callback:
         if re.match(r'^[a-zA-Z0-9\.]+$', options.callback) is not None:
-            return '%s(%s)' % (options.callback, rss.tojson())
+            out = '%s(%s)' % (options.callback, rss.tojson(encoding='unicode'))
+            return out if encoding == 'unicode' else out.encode(encoding)
 
         else:
             raise MorssException('Invalid callback var name')
 
     elif options.json:
         if options.indent:
-            return rss.tojson(encoding='UTF-8', indent=4)
+            return rss.tojson(encoding=encoding, indent=4)
 
         else:
-            return rss.tojson(encoding='UTF-8')
+            return rss.tojson(encoding=encoding)
 
     elif options.csv:
-        return rss.tocsv(encoding='UTF-8')
+        return rss.tocsv(encoding=encoding)
 
-    elif options.reader:
+    elif options.html:
         if options.indent:
-            return rss.tohtml(encoding='UTF-8', pretty_print=True)
+            return rss.tohtml(encoding=encoding, pretty_print=True)
 
         else:
-            return rss.tohtml(encoding='UTF-8')
+            return rss.tohtml(encoding=encoding)
 
     else:
         if options.indent:
-            return rss.torss(xml_declaration=True, encoding='UTF-8', pretty_print=True)
+            return rss.torss(xml_declaration=(not encoding == 'unicode'), encoding=encoding, pretty_print=True)
 
         else:
-            return rss.torss(xml_declaration=True, encoding='UTF-8')
+            return rss.torss(xml_declaration=(not encoding == 'unicode'), encoding=encoding)
 
 
 def process(url, cache=None, options=None):
@@ -475,14 +491,16 @@ def process(url, cache=None, options=None):
     if cache:
         crawler.default_cache = crawler.SQLiteCache(cache)
 
+    url = UrlFix(url)
     rss = FeedFetch(url, options)
     rss = FeedGather(rss, url, options)
 
     return FeedFormat(rss, options)
 
 
-def cgi_app(environ, start_response):
+def cgi_parse_environ(environ):
     # get options
+
     if 'REQUEST_URI' in environ:
         url = environ['REQUEST_URI'][1:]
     else:
@@ -496,7 +514,7 @@ def cgi_app(environ, start_response):
     if url.startswith(':'):
         split = url.split('/', 1)
 
-        options = split[0].replace('|', '/').replace('\\\'', '\'').split(':')[1:]
+        raw_options = unquote(split[0]).replace('|', '/').replace('\\\'', '\'').split(':')[1:]
 
         if len(split) > 1:
             url = split[1]
@@ -504,15 +522,22 @@ def cgi_app(environ, start_response):
             url = ''
 
     else:
-        options = []
+        raw_options = []
 
     # init
-    options = Options(filterOptions(parseOptions(options)))
+    options = Options(filterOptions(parseOptions(raw_options)))
-    headers = {}
 
     global DEBUG
     DEBUG = options.debug
 
+    return (url, options)
+
+
+def cgi_app(environ, start_response):
+    url, options = cgi_parse_environ(environ)
+
+    headers = {}
+
     # headers
     headers['status'] = '200 OK'
     headers['cache-control'] = 'max-age=%s' % DELAY
@@ -520,7 +545,7 @@ def cgi_app(environ, start_response):
     if options.cors:
         headers['access-control-allow-origin'] = '*'
 
-    if options.html or options.reader:
+    if options.html:
         headers['content-type'] = 'text/html'
     elif options.txt or options.silent:
         headers['content-type'] = 'text/plain'
@@ -534,9 +559,12 @@ def cgi_app(environ, start_response):
     else:
         headers['content-type'] = 'text/xml'
 
+    headers['content-type'] += '; charset=utf-8'
+
     crawler.default_cache = crawler.SQLiteCache(os.path.join(os.getcwd(), 'morss-cache.db'))
 
     # get the work done
+    url = UrlFix(url)
     rss = FeedFetch(url, options)
 
     if headers['content-type'] == 'text/xml':
@@ -547,18 +575,42 @@ def cgi_app(environ, start_response):
     rss = FeedGather(rss, url, options)
     out = FeedFormat(rss, options)
 
-    if not options.silent:
+    if options.silent:
-        return out
+        return ['']
+
+    else:
+        return [out]
 
 
-def cgi_wrapper(environ, start_response):
+def middleware(func):
-    # simple http server for html and css
+    " Decorator to turn a function into a wsgi middleware "
+    # This is called when parsing the code
+
+    def app_builder(app):
+        # This is called when doing app = cgi_wrapper(app)
+
+        def app_wrap(environ, start_response):
+            # This is called when a http request is being processed
+
+            return func(environ, start_response, app)
+
+        return app_wrap
+
+    return app_builder
+
+
+@middleware
+def cgi_file_handler(environ, start_response, app):
+    " Simple HTTP server to serve static files (.html, .css, etc.) "
+
     files = {
         '': 'text/html',
-        'index.html': 'text/html'}
+        'index.html': 'text/html',
+        'sheet.xsl': 'text/xsl'}
 
     if 'REQUEST_URI' in environ:
         url = environ['REQUEST_URI'][1:]
+
     else:
         url = environ['PATH_INFO'][1:]
 
@@ -587,16 +639,87 @@ def cgi_wrapper(environ, start_response):
             start_response(headers['status'], list(headers.items()))
             return ['Error %s' % headers['status']]
 
-    # actual morss use
+    else:
+        return app(environ, start_response)
+
+
+def cgi_get(environ, start_response):
+    url, options = cgi_parse_environ(environ)
+
+    # get page
+    PROTOCOL = ['http', 'https']
+
+    if urlparse(url).scheme not in ['http', 'https']:
+        url = 'http://' + url
+
+    data, con, contenttype, encoding = crawler.adv_get(url=url)
+
+    if contenttype in ['text/html', 'application/xhtml+xml', 'application/xml']:
+        if options.get == 'page':
+            html = readabilite.parse(data, encoding=encoding)
+            html.make_links_absolute(con.geturl())
+
+            kill_tags = ['script', 'iframe', 'noscript']
+
+            for tag in kill_tags:
+                for elem in html.xpath('//'+tag):
+                    elem.getparent().remove(elem)
+
+            output = lxml.etree.tostring(html.getroottree(), encoding='utf-8')
+
+        elif options.get == 'article':
+            output = readabilite.get_article(data, url=con.geturl(), encoding=encoding, debug=options.debug)
+
+        else:
+            raise MorssException('no :get option passed')
+
+    else:
+        output = data
+
+    # return html page
+    headers = {'status': '200 OK', 'content-type': 'text/html; charset=utf-8'}
+    start_response(headers['status'], list(headers.items()))
+    return [output]
+
+
+dispatch_table = {
+    'get': cgi_get,
+    }
+
+
+@middleware
+def cgi_dispatcher(environ, start_response, app):
+    url, options = cgi_parse_environ(environ)
+
+    for key in dispatch_table.keys():
+        if key in options:
+            return dispatch_table[key](environ, start_response)
+
+    return app(environ, start_response)
+
+
+@middleware
+def cgi_error_handler(environ, start_response, app):
     try:
-        return [cgi_app(environ, start_response) or '(empty)']
+        return app(environ, start_response)
+
     except (KeyboardInterrupt, SystemExit):
         raise
+
     except Exception as e:
-        headers = {'status': '500 Oops', 'content-type': 'text/plain'}
+        headers = {'status': '500 Oops', 'content-type': 'text/html'}
         start_response(headers['status'], list(headers.items()), sys.exc_info())
-        log('ERROR <%s>: %s' % (url, e.message), force=True)
+        log('ERROR: %s' % repr(e), force=True)
-        return ['An error happened:\n%s' % e.message]
+        return [cgitb.html(sys.exc_info())]
+
+
+@middleware
+def cgi_encode(environ, start_response, app):
+    out = app(environ, start_response)
+    return [x if isinstance(x, bytes) else str(x).encode('utf-8') for x in out]
+
+
+cgi_standalone_app = cgi_encode(cgi_error_handler(cgi_dispatcher(cgi_file_handler(cgi_app))))
 
 
 def cli_app():
@@ -608,12 +731,13 @@ def cli_app():
 
     crawler.default_cache = crawler.SQLiteCache(os.path.expanduser('~/.cache/morss-cache.db'))
 
+    url = UrlFix(url)
     rss = FeedFetch(url, options)
     rss = FeedGather(rss, url, options)
-    out = FeedFormat(rss, options)
+    out = FeedFormat(rss, options, 'unicode')
 
     if not options.silent:
-        print(out.decode('utf-8', 'replace') if isinstance(out, bytes) else out)
+        print(out)
 
     log('done')
 
@@ -622,6 +746,7 @@ def isInt(string):
     try:
         int(string)
         return True
+
     except ValueError:
         return False
 
@@ -629,7 +754,13 @@ def isInt(string):
 def main():
     if 'REQUEST_URI' in os.environ:
         # mod_cgi
-        wsgiref.handlers.CGIHandler().run(cgi_wrapper)
+
+        app = cgi_app
+        app = cgi_dispatcher(app)
+        app = cgi_error_handler(app)
+        app = cgi_encode(app)
+
+        wsgiref.handlers.CGIHandler().run(app)
 
     elif len(sys.argv) <= 1 or isInt(sys.argv[1]) or '--root' in sys.argv[1:]:
         # start internal (basic) http server
@@ -638,22 +769,31 @@ def main():
             argPort = int(sys.argv[1])
             if argPort > 0:
                 port = argPort
+
             else:
                 raise MorssException('Port must be positive integer')
 
         else:
             port = PORT
 
-        print('Serving http://localhost:%s/'%port)
+        app = cgi_app
-        httpd = wsgiref.simple_server.make_server('', port, cgi_wrapper)
+        app = cgi_file_handler(app)
+        app = cgi_dispatcher(app)
+        app = cgi_error_handler(app)
+        app = cgi_encode(app)
+
+        print('Serving http://localhost:%s/' % port)
+        httpd = wsgiref.simple_server.make_server('', port, app)
         httpd.serve_forever()
 
     else:
         # as a CLI app
         try:
             cli_app()
+
         except (KeyboardInterrupt, SystemExit):
             raise
+
         except Exception as e:
             print('ERROR: %s' % e.message)
 

morss/readabilite.py

@@ -1,13 +1,17 @@
 import lxml.etree
 import lxml.html
+from bs4 import BeautifulSoup
 import re
 
 
 def parse(data, encoding=None):
     if encoding:
-        parser = lxml.html.HTMLParser(remove_blank_text=True, remove_comments=True, encoding=encoding)
+        data = BeautifulSoup(data, 'lxml', from_encoding=encoding).prettify('utf-8')
+
     else:
-        parser = lxml.html.HTMLParser(remove_blank_text=True, remove_comments=True)
+        data = BeautifulSoup(data, 'lxml').prettify('utf-8')
+
+    parser = lxml.html.HTMLParser(remove_blank_text=True, remove_comments=True, encoding='utf-8')
 
     return lxml.html.fromstring(data, parser=parser)
 
@@ -43,6 +47,12 @@ def count_content(node):
     return count_words(node.text_content()) + len(node.findall('.//img'))
 
 
+def percentile(N, P):
+    # https://stackoverflow.com/a/7464107
+    n = max(int(round(P * len(N) + 0.5)), 2)
+    return N[n-2]
+
+
 class_bad = ['comment', 'community', 'extra', 'foot',
     'sponsor', 'pagination', 'pager', 'tweet', 'twitter', 'com-', 'masthead',
     'media', 'meta', 'related', 'shopping', 'tags', 'tool', 'author', 'about',
@@ -62,7 +72,7 @@ regex_good = re.compile('|'.join(class_good), re.I)
 
 tags_junk = ['script', 'head', 'iframe', 'object', 'noscript',
     'param', 'embed', 'layer', 'applet', 'style', 'form', 'input', 'textarea',
-    'button', 'footer']
+    'button', 'footer', 'link', 'meta']
 
 tags_bad = tags_junk + ['a', 'aside']
 
@@ -93,10 +103,18 @@ def score_node(node):
     class_id = node.get('class', '') + node.get('id', '')
 
     if (isinstance(node, lxml.html.HtmlComment)
-            or node.tag in tags_bad
+            or isinstance(node, lxml.html.HtmlProcessingInstruction)):
-            or regex_bad.search(class_id)):
         return 0
 
+    if node.tag in tags_junk:
+        score += -1 # actuall -2 as tags_junk is included tags_bad
+
+    if node.tag in tags_bad:
+        score += -1
+
+    if regex_bad.search(class_id):
+        score += -1
+
     if node.tag in tags_good:
         score += 4
 
@@ -114,33 +132,42 @@ def score_node(node):
     return score
 
 
-def score_all(node, grades=None):
+def score_all(node):
     " Fairly dumb loop to score all worthwhile nodes. Tries to be fast "
 
-    if grades is None:
-        grades = {}
-
     for child in node:
         score = score_node(child)
         child.attrib['seen'] = 'yes, ' + str(int(score))
 
-        if score > 0:
+        if score > 0 or len(list(child.iterancestors())) <= 2:
-            spread_score(child, score, grades)
+            spread_score(child, score)
-            score_all(child, grades)
+            score_all(child)
-
-    return grades
 
 
-def spread_score(node, score, grades):
+def set_score(node, value):
+    node.attrib['morss_score'] = str(float(value))
+
+
+def get_score(node):
+    return float(node.attrib.get('morss_score', 0))
+
+
+def incr_score(node, delta):
+    set_score(node, get_score(node) + delta)
+
+
+def get_all_scores(node):
+    return {x:get_score(x) for x in list(node.iter()) if get_score(x) != 0}
+
+
+def spread_score(node, score):
     " Spread the node's score to its parents, on a linear way "
 
     delta = score / 2
+
     for ancestor in [node,] + list(node.iterancestors()):
         if score >= 1 or ancestor is node:
-            try:
+            incr_score(ancestor, score)
-                grades[ancestor] += score
-            except KeyError:
-                grades[ancestor] = score
 
             score -= delta
 
@@ -148,26 +175,24 @@ def spread_score(node, score, grades):
             break
 
 
-def write_score_all(root, grades):
+def clean_root(root, keep_threshold=None):
-    " Useful for debugging "
-
-    for node in root.iter():
-        node.attrib['score'] = str(int(grades.get(node, 0)))
-
-
-def clean_root(root):
     for node in list(root):
-        clean_root(node)
+        # bottom-up approach, i.e. starting with children before cleaning current node
-        clean_node(node)
+        clean_root(node, keep_threshold)
+        clean_node(node, keep_threshold)
 
 
-def clean_node(node):
+def clean_node(node, keep_threshold=None):
     parent = node.getparent()
 
     if parent is None:
         # this is <html/> (or a removed element waiting for GC)
         return
 
+    if keep_threshold is not None and get_score(node) >= keep_threshold:
+        # high score, so keep
+        return
+
     gdparent = parent.getparent()
 
     # remove shitty tags
@@ -266,41 +291,45 @@ def lowest_common_ancestor(nodeA, nodeB, max_depth=None):
     return nodeA # should always find one tho, at least <html/>, but needed for max_depth
 
 
-def rank_nodes(grades):
+def rank_grades(grades):
+    # largest score to smallest
     return sorted(grades.items(), key=lambda x: x[1], reverse=True)
 
 
-def get_best_node(grades):
+def get_best_node(ranked_grades):
     " To pick the best (raw) node. Another function will clean it "
 
-    if len(grades) == 1:
+    if len(ranked_grades) == 1:
-        return grades[0]
+        return ranked_grades[0]
 
-    top = rank_nodes(grades)
+    lowest = lowest_common_ancestor(ranked_grades[0][0], ranked_grades[1][0], 3)
-    lowest = lowest_common_ancestor(top[0][0], top[1][0], 3)
 
     return lowest
 
 
-def get_article(data, url=None, encoding=None):
+def get_article(data, url=None, encoding=None, debug=False):
     " Input a raw html string, returns a raw html string of the article "
 
     html = parse(data, encoding)
-    scores = score_all(html)
+    score_all(html)
+    scores = rank_grades(get_all_scores(html))
 
     if not len(scores):
         return None
 
     best = get_best_node(scores)
+
+    if not debug:
+        keep_threshold = percentile([x[1] for x in scores], 0.1)
+        clean_root(best, keep_threshold)
+
     wc = count_words(best.text_content())
     wca = count_words(' '.join([x.text_content() for x in best.findall('.//a')]))
 
-    if wc - wca < 50 or float(wca) / wc > 0.3:
+    if not debug and (wc - wca < 50 or float(wca) / wc > 0.3):
         return None
 
     if url:
         best.make_links_absolute(url)
 
-    clean_root(best)
+    return lxml.etree.tostring(best if not debug else html, pretty_print=True)
-
-    return lxml.etree.tostring(best, pretty_print=True)

morss/reader.html.template

@@ -1,11 +1,9 @@
-@require(feed)
 <!DOCTYPE html>
 <html>
 	<head>
-		<title>@feed.title &#8211; via morss</title>
+		<title>Feed reader by morss</title>
-		<meta charset="UTF-8" />
-		<meta name="description" content="@feed.desc (via morss)" />
 		<meta name="viewport" content="width=device-width; initial-scale=1.0; maximum-scale=1.0;" />
+		<meta name="robots" content="noindex" />
 
 		<style type="text/css">
 			/* columns - from https://thisisdallas.github.io/Simple-Grid/simpleGrid.css */
@@ -32,7 +30,7 @@
 				padding-right: 20px; /* column-space */
 			}
 
-			@@media handheld, only screen and (max-width: 767px) { /* @@ to escape from the template engine */
+			@media handheld, only screen and (max-width: 767px) {
 				#content {
 					width: 100%;
 					min-width: 0;
@@ -82,6 +80,7 @@
 
 			#content {
 				text-align: justify;
+				line-height: 1.5em;
 			}
 
 				.item .title {
@@ -171,30 +170,17 @@
 
 	<body>
 		<div id="header">
-			<h1>@feed.title</h1>
+			<h1>RSS feed</h1>
-			@if feed.desc:
+			<h2>with full text articles</h2>
-				<h2>@feed.desc</h2>
-			@end
 			<p>- via morss</p>
 		</div>
 
 		<div id="content">
-			@for item in feed.items:
 				<div class="item">
-					@if item.link:
+					<a class="title link" href="@item.link" target="_blank"></a>
-						<a class="title link" href="@item.link" target="_blank">@item.title</a>
+					<div class="desc"></div>
-					@else:
+					<div class="content"></div>
-						<span class="title">@item.title</span>
-					@end
-					<div class="article">
-						@if item.content:
-							@item.content
-						@else:
-							@item.desc
-						@end
-					</div>
 				</div>
-			@end
 		</div>
 
 	<script>

requirements.txt

@@ -1,4 +1,5 @@
 lxml
+bs4
 python-dateutil <= 1.5
 chardet
 pymysql

www/index.html

@@ -35,8 +35,8 @@
 			<input type="text" id="url" name="url" placeholder="Feed url (http://example.com/feed.xml)" />
 		</form>
 
-		<code>Copyright: pictuga 2013-2014<br/>
+		<code>Copyright: pictuga 2013-2020<br/>
-		Source code: https://github.com/pictuga/morss</code>
+		Source code: https://git.pictuga.com/pictuga/morss</code>
 
 		<script>
 			form = document.forms[0]

www/sheet.xsl

@@ -13,6 +13,7 @@
 				body {
 					overflow-wrap: anywhere;
 					word-wrap: anywhere;
+					font-family: sans;
 				}
 
 				#url {