def cleaner_parameters():
reject_list = [
'script', 'noscript', 'style', 'meta', 'semantics', 'img', 'label',
'table', 'li', 'ul', 'ol', 'nav', 'dl', 'dd', 'sub', 'sup', 'math'
]
accept_list = [
'div', 'p', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6'
'span', 'b', 'a', 'u', 'i', 'body'
]
html_cleaner = Cleaner()
html_cleaner.remove_unknown_tags = True
html_cleaner.processing_instructions = True
html_cleaner.style = True
html_cleaner.comments = True
html_cleaner.scripts = True
html_cleaner.javascript = True
html_cleaner.meta = True
html_cleaner.links = True
html_cleaner.embedded = True
html_cleaner.annoying_tags = True
html_cleaner.frames = True
html_cleaner.forms = True
html_cleaner.remove_tags = accept_list
html_cleaner.kill_tags = reject_list
return html_cleaner
def lxml_extractor(html, url):
'''LXML PARSER'''
cleaner = Cleaner()
cleaner.javascript = True # This is True because we want to activate the javascript filter
cleaner.style = True # This is True because we want to activate the styles & stylesheet filter
cleaner.comments = True
cleaner.embedded = True
cleaner.forms= True
cleaner.frames = True
cleaner.annoying_tags = True
cleaner.kill_tags = NEGATIVE_K
cleaner.allow_tag = POSITIVE_K
cleaner.safe_attrs_only = True
#~ oc = document_fromstring(html, parser=parser, base_url=base_url, **kw)
#~ File "/usr/local/lib/python2.7/dist-packages/lxml/html/__init__.py", line 752, in document_fromstring
#~ value = etree.fromstring(html, parser, **kw)
try:
html = lxml.html.fromstring(html, base_url="url")
tree = cleaner.clean_html(html)
#tree.make_links_absolute(url)
doc = lxml.html.tostring(tree)
doc = soup_extractor(doc, url)
except ValueError:
doc = soup_extractor(html, url)
#~ (title, doc, article, text) = read_extractor(html, url)
#~ print title
#~ doc = (self.doc).replace(unichr(160), " ")
#~ doc = re.sub(spaces,"",self.doc)
return doc
def clean_html(html_text,
javascript=True,
scripts=True,
style=True,
embedded=True,
links=True,
forms=True,
frames=True,
comments=True,
annoying_tags=True,
meta=True,
safe_attrs_only=True,
remove_unknown_tags=True,
processing_instructions=True):
"""Clean all the javascript and styles from the HTML returning the string with only the html content"""
# True = Remove | False = Keep
cleaner = Cleaner()
cleaner.javascript = javascript # This is True because we want to activate the javascript filter
cleaner.scripts = scripts # This is True because we want to activate the scripts filter
cleaner.style = style
cleaner.embedded = embedded
cleaner.links = links
cleaner.forms = forms
cleaner.frames = frames
cleaner.comments = comments
cleaner.page_structure = False # Keep page structure
cleaner.annoying_tags = annoying_tags
cleaner.meta = meta
cleaner.safe_attrs_only = safe_attrs_only
cleaner.remove_unknown_tags = remove_unknown_tags
cleaner.processing_instructions = processing_instructions
clean_content = cleaner.clean_html(lxml.html.fromstring(html_text))
return lxml.html.tostring(clean_content)
def html2text(html):
cleaner = Cleaner()
cleaner.javascript = True # This is True because we want to activate the javascript filter
cleaner.style = True
cleaner.scripts = True
cleaner.comments = True
cleaner.links = True
cleaner.meta = True
cleaner.page_structure = True
cleaner.processing_instructions = True
cleaner.forms = True
cleaner.add_nofollow = True
#html = unicodedata.normalize('NFKD', html).encode('ascii','ignore')
try:
document = lxml.html.document_fromstring(html)
c = cleaner.clean_html(document)
html = lxml.html.tostring(c)
soup = BeautifulSoup(html, 'lxml')
parsed_text = soup.get_text()
if (len(parsed_text) > MINSIZE_CHARSDOC):
return parsed_text.lower()
else:
return None
except:
return None
def cleanpage(html):
# cleaner setup
cleaner = Cleaner()
cleaner.html = True
cleaner.page_structure = False
cleaner.meta = False
cleaner.safe_attrs_only = False
cleaner.links = False
cleaner.javascript = True # activate the javascript filter
cleaner.style = True # activate the styles & stylesheet filter
cleaner.links = False
cleaner.frames = True
cleaner.embedded = True
cleaner.comments = True
cleaner.annoying_tags = True
cleaner.inline_style = True
cleaner.page_structure = False
# cleaner.remove_tags = ['b','img','h']
cleaner.kill_tags = ['img','script']
#invoke cleaner
try:
content=cleaner.clean_html(html)
except:
#error: ValueError: Unicode strings with encoding declaration are not supported. Please use bytes input or XML fr
content = u""
return content
def get_data(self):
html_body = self._by_pass_get_html(self.url)
cleaner = Cleaner()
cleaner.comments = True
doc = cleaner.clean_html(html_body)
return doc
def cleaned_html(self):
# Try to parse the provided HTML string using lxml
# strip all unnecessary information to save space
cleaner = Cleaner()
cleaner.scripts = True
cleaner.javascript = True
cleaner.comments = True
cleaner.style = True
self.dom = cleaner.clean_html(self.dom)
assert len(self.dom), 'The html needs to be parsed to get the cleaned html'
return lxml.html.tostring(self.dom)
def cleaned_html(self):
# Try to parse the provided HTML string using lxml
# strip all unnecessary information to save space
cleaner = Cleaner()
cleaner.scripts = True
cleaner.javascript = True
cleaner.comments = True
cleaner.style = True
self.dom = cleaner.clean_html(self.dom)
assert len(self.dom), 'The html needs to be parsed to get the cleaned html'
return lxml.html.tostring(self.dom)
def cleaned_html(htmlString):
# Try to parse the provided HTML string using lxml
# strip all unnecessary information to save space
cleaner = Cleaner()
cleaner.comments = True
cleaner.javascript = True
cleaner.scripts = True
cleaner.style = True
htmlString = cleaner.clean_html(htmlString)
return htmlString
def clean_text(data):
cleaner = Cleaner()
cleaner.javascript = True
cleaner.style = True
cleaner.scripts = True
cleaner.comments = True
cleaner.meta = True
cleaner.annoying_tags = True
stuff = lxml.html.tostring(cleaner.clean_html(data))
soup = BeautifulSoup(stuff.decode('utf-8', 'ignore'))
all_text = ' '.join(filter(lambda val: val, \
map(lambda x: x.strip(), soup.findAll(text=True))))
return all_text
def crawNews(self, url):
cleaner = Cleaner()
cleaner.javascript = True
cleaner.style = True
cleaner.comments = True
tech_content = lxml.html.parse(url)
tech_content = (lxml.html.tostring(tech_content))
re_title = re.compile(r'<h1.*>(.*)</h1', re.S)
re_content = re.compile(
r'<!-- Begin: Wordpress Article Content -->(.*)<!-- End: Wordpress Article Content -->',
re.S)
re_published = re.compile(r'name="sailthru.date"\scontent="(.*?)"')
re_author = re.compile(
r'<a\shref="(.*?)"\stitle.*?rel="author">(.*?)<\/a>.*?rel="external">(.*?)<\/a>'
)
match_title = re.search(re_title, tech_content)
match_content = re.search(re_content, tech_content)
match_date = re.search(re_published, tech_content)
match_author = re.search(re_author, tech_content)
author_url = "http://techcrunch.com" + match_author.group(1)
author_name = match_author.group(2)
author_twitter = match_author.group(3)
title = re.sub(r'<[^>]*?>', '',
cleaner.clean_html(match_title.group(1)))
title = re.sub(r'\s+', ' ', title)
title = title.decode('utf-8').strip()
content = re.sub(r'<[^>]*?>', '',
cleaner.clean_html(match_content.group(1)))
content = re.sub(r'\s+', ' ', content)
content = content.decode('utf-8').strip()
content = content.strip('\n')
published_on = datetime.datetime.strptime(match_date.group(1),
'%Y-%m-%d %H:%M:%S')
news = self.save_news(url, title, content, published_on)
author = self.findAuthorByUrl(author_url)
if (isinstance(author, Author) == False):
author = self.save_author(author_url, author_name, author_twitter,
'')
self.newsAuthor(news, author)
def clean_html(self):
"""
Cleaner removes HTML tags prior to processing. Note: cleaning removes
the Title tags from HTML. Do not clean before grabbing titles!
"""
if len(self.response.content):
cleaner = Cleaner()
cleaner.javascript = True
cleaner.scripts = True
cleaner.style = True
cleaner.comments = True
try:
return html.fromstring(cleaner.clean_html(self.response.content))
except Exception as e:
logging.error(e)
return None
def clean_text(data):
cleaner = Cleaner()
cleaner.javascript = True
cleaner.style = True
cleaner.scripts = True
cleaner.comments = True
cleaner.meta = True
cleaner.annoying_tags = True
doc = UnicodeDammit(data, is_html=True)
parser = html.HTMLParser(encoding=doc.original_encoding)
root = html.document_fromstring(data, parser=parser)
stuff = lxml.html.tostring(cleaner.clean_html(root))
soup = BeautifulSoup(stuff.decode('utf-8', 'ignore'))
all_text = ' '.join(filter(lambda val: val, \
map(lambda x: x.strip(), soup.findAll(text=True))))
return all_text.encode('ascii', 'ignore')
def extract_content(bytehtml, doc):
"""
extracts blog post content from html
"""
lxmldoc = lxml.html.document_fromstring(bytehtml)
cleaner = Cleaner()
cleaner.scripts = True
cleaner.comments = True
cleaner.style = True
#cleaner.page_structure = True
cleaner.kill_tags = ['head', 'noscript']
cleaner.remove_tags = ['p', 'i', 'b', 'strong', 'em', 'blockquote']
cleaner(lxmldoc)
content_el = find_content_element(lxmldoc)
if content_el:
debug(3, 'content quality {}'.format(content_el._quality))
text = tidy_content(content_el.text_content())
return text
else:
debug(2, 'no content found!')
raise Exception('no content')
def crawNews(self, url):
cleaner = Cleaner()
cleaner.javascript = True
cleaner.style = True
cleaner.comments = True
tech_content = lxml.html.parse(url)
tech_content = (lxml.html.tostring(tech_content))
re_title = re.compile(r'<h1.*>(.*)</h1', re.S)
re_content = re.compile(r'<!-- Begin: Wordpress Article Content -->(.*)<!-- End: Wordpress Article Content -->', re.S)
re_published = re.compile(r'name="sailthru.date"\scontent="(.*?)"')
re_author = re.compile(r'<a\shref="(.*?)"\stitle.*?rel="author">(.*?)<\/a>.*?rel="external">(.*?)<\/a>')
match_title = re.search(re_title, tech_content)
match_content = re.search(re_content, tech_content)
match_date = re.search(re_published, tech_content)
match_author = re.search(re_author, tech_content)
author_url = "http://techcrunch.com" + match_author.group(1)
author_name = match_author.group(2)
author_twitter = match_author.group(3)
title = re.sub(r'<[^>]*?>', '', cleaner.clean_html(match_title.group(1)))
title = re.sub(r'\s+', ' ', title)
title = title.decode('utf-8').strip()
content = re.sub(r'<[^>]*?>', '', cleaner.clean_html(match_content.group(1)))
content = re.sub(r'\s+', ' ', content)
content = content.decode('utf-8').strip()
content = content.strip('\n')
published_on = datetime.datetime.strptime(match_date.group(1), '%Y-%m-%d %H:%M:%S')
news = self.save_news(url, title, content, published_on)
author = self.findAuthorByUrl(author_url)
if (isinstance(author, Author) == False):
author = self.save_author(author_url, author_name, author_twitter, '')
self.newsAuthor(news, author)
def scrape(lineHashDB, html, encoding):
# cleaner setup
cleaner = Cleaner(allow_tags=['div', 'p'], remove_unknown_tags=False)
cleaner.javascript = True # activate the javascript filter
cleaner.style = True # activate the styles & stylesheet filter
cleaner.comments = True
cleaner.annoying_tags = True
cleaner.inline_style = True
cleaner.page_structure = False
cleaner.remove_tags = ['b', 'a', 'h']
cleaner.kill_tags = ['script']
#invoke cleaner
try:
page = cleaner.clean_html(html)
except:
#error: ValueError: Unicode strings with encoding declaration are not supported. Please use bytes input or XML fr
content = u""
return content
page8 = page
page8 = re.sub(u'\n', ' ', page8) # remove NL
# page8 = re.sub(u'\s','',page8,re.UNICODE) # blanks -> space
page8 = re.sub(u' ', ' ', page8) # remove CR
page8 = re.sub(u'<!--.*?-->', ' ', page8) # remove comments
page8 = re.sub(u' class=".*?"', ' ', page8) # remove attributes
page8 = re.sub(u' id=".*?"', ' ', page8)
page8 = re.sub(u' rel=".*?"', ' ', page8)
page8 = re.sub(u'\[an error occurred while processing this directive\]',
' ', page8)
page8 = re.sub(u'>\s*?<', '><', page8) # remove blanks between tags
# cycle to remove spurious divs
for count in range(1, 20):
page8 = re.sub(u'>.{0,10}<', '><',
page8) # remove words under 10 chars between tags
page8 = re.sub(u'<div></div>', ' ', page8)
page8 = re.sub(u'<p></p>', ' ', page8)
page8 = re.sub(u'<span></span>', ' ', page8)
page8 = re.sub(u'\s+', ' ', page8) # remove repeated blanks
#XPATHs
xpath = '//*[((p) or (a) or (b) or (div) or (span)) ]/node()[(string-length() > 300)]/text()'
xpath = '//*[((p) or (div))]/node()[(string-length() > 100)]/text()'
sel = Selector(text=page8, type="html")
text = sel.xpath(xpath).extract()
content = u""
if text:
for s in text:
# squash duplicate whitespaces
' '.join(s.split())
# remove short lines
# on empirical analysis, no unfrequent sentence under 40 chars is a relevant part of the article text, excluding repetition of title, authors, dates, etc.
if len(s) < 40:
next
# remove leading whitespace
#if s.endswith(" "): s = s[:-1]
if s.startswith(" "): s = s[1:]
content += s
content += "\n"
return content
def f_parse(args):
def isAlphabet(word):
alphabet = [
'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'j', 'k', 'i', 'l', 'm',
'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'x', 'y', 'w', 'z',
'à', 'è', 'é', 'ì', 'í', 'ò', 'ó', 'ù', 'ú'
]
guard = True
for t in word:
if t not in alphabet:
guard = False
return guard
loc = args[0]
corpuses = args[1]
MINSIZE_WORD = 4
MAXSIZE_WORD = 15
MINSIZE_CHARSDOC = 100
MINSIZE_WORDSDOC = 50
cleaner = Cleaner()
cleaner.javascript = True # This is True because we want to activate the javascript filter
cleaner.style = True
cleaner.scripts = True
cleaner.comments = True
cleaner.links = True
cleaner.meta = True
cleaner.page_structure = True
cleaner.processing_instructions = True
cleaner.forms = True
cleaner.add_nofollow = True
ret = []
for document in corpuses:
#html = unicodedata.normalize('NFKD', html).encode('ascii','ignore')
if len(document) > 0:
try:
document = lxml.html.document_fromstring(document)
c = cleaner.clean_html(document)
html = lxml.html.tostring(c)
soup = BeautifulSoup(html, 'lxml')
parsed_text = soup.get_text()
if (len(parsed_text) > MINSIZE_CHARSDOC):
parsed_text = parsed_text.lower()
tokenizer = RegexpTokenizer(r'\w+')
# create English stop words list
en_stop = get_stop_words('en')
it_stop = get_stop_words('it')
sp_stop = get_stop_words('es')
ge_stop = get_stop_words('de')
fr_stop = get_stop_words('fr')
# Create p_stemmer of class PorterStemmer
#p_stemmer = PorterStemmer()
# clean and tokenize document string
tokens = tokenizer.tokenize(parsed_text)
# remove stop words from tokens
stopped_tokens1 = [i for i in tokens if not i in en_stop]
stopped_tokens2 = [
i for i in stopped_tokens1 if not i in it_stop
]
stopped_tokens3 = [
i for i in stopped_tokens2 if not i in sp_stop
]
stopped_tokens4 = [
i for i in stopped_tokens3 if not i in ge_stop
]
stopped_tokens5 = [
i for i in stopped_tokens4 if not i in fr_stop
]
for word in stopped_tokens5:
if not any(char.isdigit() for char in word):
if len(word) > 1:
#check if the word has the alphabet character
if isAlphabet(word):
ret.append(word)
except:
print('Exception : Document empty')
return [loc, ret]
import lxml
from lxml import etree
from lxml.html.clean import autolink_html
from lxml.html.clean import Cleaner
#LXML PARSER
cleaner = Cleaner()
cleaner.javascript = True # This is True because we want to activate the javascript filter
cleaner.style = True # This is True because we want to activate the styles & stylesheet filter
cleaner.comments = True
cleaner.embedded = True
cleaner.forms= True
cleaner.frames = True
#cleaner.safe_attrs_only = True
import re
notalpha = re.compile('[^a-zA-Z]')
#BS PARSER
from bs4 import BeautifulSoup as bs
from bs4 import Comment
def make_links_absolute(soup, url):
return [urlparse.urljoin(url, tag['href']) for tag in soup.findAll('a', href=True)]
def clean_html(soup):
soup = bs(" ".join([s.extract() for s in soup('script')]))
soup = bs(" ".join([s.extract() for s in soup('iframe')]))
soup = bs(" ".join([s.extract() for s in soup('form')]))
soup = bs(" ".join([s.extract() for s in soup('embed')]))
soup = bs(" ".join([s.extract() for s in soup('style')]))
from lxml import etree from lxml.html.clean import Cleaner from .filters import duplicate_test, textfilter from .settings import CUT_EMPTY_ELEMS, DEFAULT_CONFIG, MANUALLY_CLEANED, MANUALLY_STRIPPED from .utils import trim LOGGER = logging.getLogger(__name__) # HTML_CLEANER config # http://lxml.de/api/lxml.html.clean.Cleaner-class.html # https://lxml.de/apidoc/lxml.html.clean.html HTML_CLEANER = Cleaner() HTML_CLEANER.annoying_tags = False # True HTML_CLEANER.comments = True HTML_CLEANER.embedded = False # True HTML_CLEANER.forms = False # True HTML_CLEANER.frames = False # True HTML_CLEANER.javascript = False HTML_CLEANER.links = False HTML_CLEANER.meta = False HTML_CLEANER.page_structure = False HTML_CLEANER.processing_instructions = True HTML_CLEANER.remove_unknown_tags = False HTML_CLEANER.safe_attrs_only = False HTML_CLEANER.scripts = False HTML_CLEANER.style = False #HTML_CLEANER.remove_tags = MANUALLY_STRIPPED #HTML_CLEANER.kill_tags = MANUALLY_CLEANED
DATE_EXPRESSIONS = [
"//*[contains(@id, 'date') or contains(@id, 'Date') or contains(@id, 'datum') or contains(@id, 'Datum') or contains(@id, 'time') or contains(@class, 'post-meta-time')]",
"//*[contains(@class, 'date') or contains(@class, 'Date') or contains(@class, 'datum') or contains(@class, 'Datum')]",
"//*[contains(@class, 'postmeta') or contains(@class, 'post-meta') or contains(@class, 'entry-meta') or contains(@class, 'postMeta') or contains(@class, 'post_meta') or contains(@class, 'post__meta')]",
"//*[@class='meta' or @class='meta-before' or @class='asset-meta' or contains(@id, 'article-metadata') or contains(@class, 'article-metadata') or contains(@class, 'byline') or contains(@class, 'subline')]",
"//*[contains(@class, 'published') or contains(@class, 'posted') or contains(@class, 'submitted') or contains(@class, 'created-post')]",
"//*[contains(@id, 'lastmod') or contains(@itemprop, 'date') or contains(@class, 'time')]",
"//footer",
"//*[@class='post-footer' or @class='footer' or @id='footer']",
"//small",
"//*[contains(@class, 'author') or contains(@class, 'autor') or contains(@class, 'field-content') or @class='meta' or contains(@class, 'info') or contains(@class, 'fa-clock-o')]",
]
CLEANER = Cleaner()
CLEANER.comments = False
CLEANER.embedded = True
CLEANER.forms = False
CLEANER.frames = True
CLEANER.javascript = True
CLEANER.links = False
CLEANER.meta = False
CLEANER.page_structure = True
CLEANER.processing_instructions = True
CLEANER.remove_unknown_tags = False
CLEANER.safe_attrs_only = False
CLEANER.scripts = False
CLEANER.style = True
CLEANER.kill_tags = [
'audio', 'canvas', 'label', 'map', 'math', 'object', 'picture', 'rdf',
'svg', 'video'
E.BODY(
E.H1(E.CLASS("heading"), title),
lxml.html.fromstring(html)
)
)
html_out.getroottree().write(file="summarized-roanoke.html", method="html")
if __name__ == "__main__":
cleaner = Cleaner()
cleaner.javascript = True
cleaner.scripts = True
cleaner.frame = True
cleaner.meta = True
cleaner.comments = True
cleaner.links = True
cleaner.style = True
cleaner.kill_tags = ["cite", "sup", "img", "noscript", "label", "video"]
url = "https://en.wikipedia.org/wiki/Roanoke_Colony"
doc = urllib2.urlopen(url)
tree = lxml.html.parse(doc)
title = tree.find(".//title").text
tree = cleaner.clean_html(tree)
netloc = urlparse(url).netloc
if netloc == "en.wikipedia.org":
parse_wiki(tree, title)
