pictuga/morss
1
1
Fork 0
Code Issues Pull Requests Actions Packages Releases Activity

readabilite: change scoring algorithm

Browse Source 
Use 3 groups of keywords instead
This commit is contained in:
pictuga
2017-07-17 00:01:44 +02:00
parent df22396838
commit 843dc97fbf
1 changed files with 38 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

64
morss/readabilite.py
View File

@@ -35,19 +35,29 @@ def count_words(string):
return count return count
regex_bad = re.compile('|'.join(['robots-nocontent', 'combx', 'comment', regex_bad = re.compile('|'.join(['comment', 'community', 'extra', 'foot',
'community', 'disqus', 'extra', 'foot', 'header', 'menu', 'remark', 'rss', 'sponsor', 'pagination', 'pager', 'tweet', 'twitter', 'com-', 'masthead',
'shoutbox', 'sidebar', 'sponsor', 'ad-', 'agegate', 'pagination', 'media', 'meta', 'related', 'shopping', 'tags', 'tool', 'author', 'about']),
'pager', 'popup', 'tweet', 'twitter', 'com-', 'sharing', 'share', 'social', re.I)
'contact', 'footnote', 'masthead', 'media', 'meta', 'outbrain', 'promo',
'related', 'scroll', 'shoutbox', 'shopping', 'tags',
'tool', 'widget', 'hide', 'author', 'about']), re.I)
regex_good = re.compile('|'.join(['and', 'article', 'body', 'column', regex_junk = re.compile('|'.join(['robots-nocontent', 'combx', 'disqus',
'main', 'shadow', 'content', 'entry', 'hentry', 'main', 'page', 'header', 'menu', 'remark', 'rss', 'shoutbox', 'sidebar', 'ad-', 'agegate',
'pagination', 'post', 'text', 'blog', 'story', 'par', 'editorial']), re.I) 'popup', 'sharing', 'share', 'social', 'contact', 'footnote', 'outbrain',
'promo', 'scroll', 'hidden', 'widget', 'hide']), re.I)
regex_good = re.compile('|'.join(['and', 'article', 'body', 'column', 'main',
'shadow', 'content', 'entry', 'hentry', 'main', 'page', 'pagination',
'post', 'text', 'blog', 'story', 'par', 'editorial']), re.I)
tags_bad = ['a']
tags_junk = ['script', 'head', 'iframe', 'object', 'noscript', 'param', 'embed',
'layer', 'applet', 'style', 'form', 'input', 'textarea', 'button', 'footer']
tags_good = ['h1', 'h2', 'h3', 'article', 'p', 'cite', 'section', 'img',
'figcaption', 'figure']
tags_junk = ['script', 'head', 'iframe', 'object', 'noscript', 'param', 'embed', 'layer', 'applet', 'style', 'form', 'input', 'textarea', 'button']
attributes_fine = ['title', 'src', 'href', 'type', 'name', 'for', 'value'] attributes_fine = ['title', 'src', 'href', 'type', 'name', 'for', 'value']
@@ -58,28 +68,30 @@ def score_node(node):
if isinstance(node, lxml.html.HtmlComment): if isinstance(node, lxml.html.HtmlComment):
return 0 return 0
class_id = node.get('class', '') + node.get('id', '')
score -= len(regex_bad.findall(class_id))
score -= len(regex_junk.findall(class_id))
score += len(regex_good.findall(class_id))
wc = count_words(''.join([node.text or ''] + [x.tail or '' for x in node])) wc = count_words(''.join([node.text or ''] + [x.tail or '' for x in node]))
# the .tail part is to include *everything* in that node # the .tail part is to include *everything* in that node
if wc < 5: if wc > 10:
return 0 score += 1
if node.tag in ['h1', 'h2', 'h3', 'article']: if wc > 20:
score += 8 score += 1
if node.tag in ['p', 'cite', 'section']: if wc > 30:
score += 1
if node.tag in tags_bad or node.tag in tags_junk:
score = -1 * abs(score)
if node.tag in tags_good:
score += 3 score += 3
class_id = node.get('class', '') + node.get('id', '')
score += len(regex_good.findall(class_id) * 5)
score -= len(regex_bad.findall(class_id) * 3)
score += wc / 5.
if node.tag in tags_junk or node.tag in ['a']:
score *= -1
return score return score