Archives
/
python-readability
2
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Improved title shortener method, and added it to the Document class.

Browse Source
pull/1/head
Yuri Baburov 13 years ago
parent f925e3ef05
commit 96f476181c
2 changed files with 135 additions and 106 deletions
Show Stats Download Patch File Download Diff File

208
readability/htmls.py
Unescape Escape View File

@ -1,96 +1,112 @@
from cleaners import normalize_spaces, clean_attributes
from encodings import get_encoding
from lxml.html import tostring
import logging
import lxml.html
import re
logging.getLogger().setLevel(logging.DEBUG)
utf8_parser = lxml.html.HTMLParser(encoding='utf-8')
def build_doc(page):
enc = get_encoding(page)
page_enc = page.decode(enc, 'replace').encode('utf-8')
doc = lxml.html.document_fromstring(page_enc, parser=utf8_parser)
return doc
def js_re(src, pattern, flags, repl):
return re.compile(pattern, flags).sub(src, repl.replace('$', '\\'))
def normalize_entities(cur_title):
entities = {
u'\u2014':'-',
u'\u2013':'-',
u'—': '-',
u'–': '-',
u'\u00A0': ' ',
u'\u00AB': '"',
u'\u00BB': '"',
u'"': '"',
}
for c, r in entities.iteritems():
if c in cur_title:
cur_title = cur_title.replace(c, r)
return cur_title
def norm_title(title):
return normalize_entities(normalize_spaces(title))
def get_title(doc):
title = doc.find('.//title').text
if not title:
return '[no-title]'
return norm_title(title)
def shortify_title(doc):
title = doc.find('.//title').text
if not title:
return '[no-title]'
title = orig = norm_title(title)
for delimiter in [' | ', ' - ', ' :: ', ' / ']:
if delimiter in title:
parts = orig.split(delimiter)
if len(parts[0].split()) >= 4:
title = parts[0]
break
elif len(parts[-1].split()) >= 4:
title = parts[-1]
break
else:
if ': ' in title:
parts = orig.split(': ')
if len(parts[-1].split()) >= 4:
title = parts[-1]
else:
title = orig.split(': ', 1)[1]
if len(title.split()) <= 4:
h1 = list(doc.iterfind('.//h1'))
if len(h1) == 1:
title = norm_title(h1[0].text)
elif len(h1) == 0:
h2 = list(doc.iterfind('.//h2'))
if len(h1) == 1:
title = norm_title(h2[1].text)
if not 15 < len(title) < 150:
return orig
return title
def get_body(doc):
[ elem.drop_tree() for elem in doc.xpath('.//script | .//link | .//style') ]
raw_html = unicode(tostring(doc.body or doc))
cleaned = clean_attributes(raw_html)
try:
#BeautifulSoup(cleaned) #FIXME do we really need to try loading it?
return cleaned
except Exception: #FIXME find the equivalent lxml error
logging.error("cleansing broke html content: %s\n---------\n%s" % (raw_html, cleaned))
return raw_html
from cleaners import normalize_spaces, clean_attributes
from encodings import get_encoding
from lxml.html import tostring
import logging
import lxml.html
import re
logging.getLogger().setLevel(logging.DEBUG)
utf8_parser = lxml.html.HTMLParser(encoding='utf-8')
def build_doc(page):
enc = get_encoding(page)
page_enc = page.decode(enc, 'replace').encode('utf-8')
doc = lxml.html.document_fromstring(page_enc, parser=utf8_parser)
return doc
def js_re(src, pattern, flags, repl):
return re.compile(pattern, flags).sub(src, repl.replace('$', '\\'))
def normalize_entities(cur_title):
entities = {
u'\u2014':'-',
u'\u2013':'-',
u'&mdash;': '-',
u'&ndash;': '-',
u'\u00A0': ' ',
u'\u00AB': '"',
u'\u00BB': '"',
u'&quot;': '"',
}
for c, r in entities.iteritems():
if c in cur_title:
cur_title = cur_title.replace(c, r)
return cur_title
def norm_title(title):
return normalize_entities(normalize_spaces(title))
def get_title(doc):
title = doc.find('.//title').text
if not title:
return '[no-title]'
return norm_title(title)
def add_match(collection, text, orig):
text = norm_title(text)
if len(text.split()) >= 2 and len(text) >= 15:
if text.replace('"', '') in orig.replace('"', ''):
collection.add(text)
def shorten_title(doc):
title = doc.find('.//title').text
if not title:
return ''
title = orig = norm_title(title)
candidates = set()
for item in ['.//h1', './/h2', './/h3']:
for e in list(doc.iterfind(item)):
if e.text:
add_match(candidates, e.text, orig)
if e.text_content():
add_match(candidates, e.text_content(), orig)
for item in ['#title', '#head', '#heading', '.pageTitle', '.news_title', '.title', '.head', '.heading', '.contentheading', '.small_header_red']:
for e in doc.cssselect(item):
if e.text:
add_match(candidates, e.text, orig)
if e.text_content():
add_match(candidates, e.text_content(), orig)
if candidates:
title = sorted(candidates, key=len)[-1]
else:
for delimiter in [' | ', ' - ', ' :: ', ' / ']:
if delimiter in title:
parts = orig.split(delimiter)
if len(parts[0].split()) >= 4:
title = parts[0]
break
elif len(parts[-1].split()) >= 4:
title = parts[-1]
break
else:
if ': ' in title:
parts = orig.split(': ')
if len(parts[-1].split()) >= 4:
title = parts[-1]
else:
title = orig.split(': ', 1)[1]
if not 15 < len(title) < 150:
return orig
return title
def get_body(doc):
[ elem.drop_tree() for elem in doc.xpath('.//script | .//link | .//style') ]
raw_html = unicode(tostring(doc.body or doc))
cleaned = clean_attributes(raw_html)
try:
#BeautifulSoup(cleaned) #FIXME do we really need to try loading it?
return cleaned
except Exception: #FIXME find the equivalent lxml error
logging.error("cleansing broke html content: %s\n---------\n%s" % (raw_html, cleaned))
return raw_html

33
readability/readability.py
Unescape Escape View File

@ -1,7 +1,7 @@
#!/usr/bin/env python
from collections import defaultdict
from cleaners import html_cleaner, clean_attributes
from htmls import build_doc, get_body, get_title
from htmls import build_doc, get_body, get_title, shorten_title
from lxml.etree import tostring, tounicode
import logging
import re
@ -15,12 +15,12 @@ REGEXES = {
'positiveRe': re.compile('caption|article|body|content|entry|hentry|page|pagination|post|text',re.I),
'negativeRe': re.compile('adwrapper|ad_wrapper|share|bookmark|nav|combx|comment|contact|foot|footer|footnote|link|media|meta|promo|related|scroll|shoutbox|sponsor|tags|widget',re.I),
'divToPElementsRe': re.compile('<(a|blockquote|dl|div|img|ol|p|pre|table|ul)',re.I),
'replaceBrsRe': re.compile('(<br[^>]*>[ \n\r\t]*){2,}',re.I),
'replaceFontsRe': re.compile('<(\/?)font[^>]*>',re.I),
'trimRe': re.compile('^\s+|\s+$/'),
'normalizeRe': re.compile('\s{2,}/'),
'killBreaksRe': re.compile('(<br\s*\/?>(\s|&nbsp;?)*){1,}/'),
'videoRe': re.compile('http:\/\/(www\.)?(youtube|vimeo)\.com', re.I),
#'replaceBrsRe': re.compile('(<br[^>]*>[ \n\r\t]*){2,}',re.I),
#'replaceFontsRe': re.compile('<(\/?)font[^>]*>',re.I),
#'trimRe': re.compile('^\s+|\s+$/'),
#'normalizeRe': re.compile('\s{2,}/'),
#'killBreaksRe': re.compile('(<br\s*\/?>(\s|&nbsp;?)*){1,}/'),
#'videoRe': re.compile('http:\/\/(www\.)?(youtube|vimeo)\.com', re.I),
}
def describe(node):
@ -37,6 +37,15 @@ def log_candidates(candidates, print_format=""):
#def _text(node):
# return " ".join(node.findall(text=True))
def to_int(x):
if not x: return None
x = x.strip()
if x.endswith('px'):
return int(x[:-2])
if x.endswith('em'):
return int(x[:-2]) * 12
return int(x)
class Unparseable(ValueError):
pass
@ -72,6 +81,9 @@ class Document:
def title(self):
return get_title(self._html(True))
def short_title(self):
return shorten_title(self._html(True))
def summary(self):
try:
ruthless = True
@ -263,9 +275,10 @@ class Document:
def sanitize(self, node, candidates):
for header in self.tags(node, "h1", "h2", "h3", "h4", "h5", "h6"):
if self.class_weight(header) < 0 or self.get_link_density(header) > 0.33: header.drop_tree()
if self.class_weight(header) < 0 or self.get_link_density(header) > 0.33:
header.drop_tree()
for elem in self.tags(node, "form", "iframe"):
for elem in self.tags(node, "form", "iframe", "textarea"):
elem.drop_tree()
allowed = {}
# Conditionally clean <table>s, <ul>s, and <div>s
@ -338,7 +351,7 @@ class Document:
height = img.get('height')
width = img.get('width')
self.debug ("height %s width %s" %(repr(height), repr(width)))
if (height and int(height) >= 50) or (width and int(width) >= 50):
if to_int(height) >= 100 or to_int(width) >= 100:
valid_img = True
self.debug("valid image" + tounicode(img))
break

Write Preview
Loading…
Cancel
Save