def scrape(self, request, response):
if not self.is_html(request, response):
return
content_file = response.body.content_file
encoding = get_heading_encoding(response)
tree = self.parse(content_file, encoding, request.url_info.url)
root = tree.getroot()
if root is None:
return
linked_urls = set()
inline_urls = set()
link_infos = self.iter_links(root)
if 'Refresh' in response.fields:
link = parse_refresh(response.fields['Refresh'])
if link:
link_info = LinkInfo(None, '_refresh', None, to_str(link),
False, True, None, 'refresh')
link_infos = itertools.chain(link_infos, [link_info])
for scraped_link in link_infos:
if self._only_relative:
if scraped_link.base_link or '://' in scraped_link.link:
continue
if not self._is_accepted(scraped_link.tag):
continue
base_url = root.base_url
if scraped_link.base_link:
base_url = wpull.url.urljoin(base_url, scraped_link.base_link)
url = wpull.url.urljoin(base_url,
scraped_link.link,
allow_fragments=False)
# Browsers seems to tolerate URLs with newlines
url = url.replace('\n', '').replace('\r', '')
if scraped_link.inline:
inline_urls.add(url)
if scraped_link.linked:
linked_urls.add(url)
if self._robots and self._robots_cannot_follow(root):
linked_urls.clear()
return {
'inline_urls': inline_urls,
'linked_urls': linked_urls,
'base_url': to_str(root.base_url),
'encoding': to_str(root.getroottree().docinfo.encoding),
}
def iter_links_element(cls, element):
'''Iterate a HTML element.'''
# reference: lxml.html.HtmlMixin.iterlinks()
# NOTE: to_str is needed because on Python 2, only byte strings
# are returned from lxml
attrib = element.attrib
tag = element.tag
if tag == 'link':
iterable = cls.iter_links_link_element(element)
elif tag == 'meta':
iterable = cls.iter_links_meta_element(element)
elif tag in ('object', 'applet'):
iterable = cls.iter_links_object_element(element)
elif tag == 'param':
iterable = cls.iter_links_param_element(element)
elif tag == 'style':
iterable = cls.iter_links_style_element(element)
else:
iterable = cls.iter_links_plain_element(element)
for link_info in iterable:
yield link_info
if 'style' in attrib:
for link in CSSScraper.scrape_urls(attrib['style']):
yield LinkInfo(element, element.tag, 'style', to_str(link),
True, False, None, 'css')
def iter_links_element(cls, element):
'''Iterate a HTML element.'''
# reference: lxml.html.HtmlMixin.iterlinks()
# NOTE: to_str is needed because on Python 2, only byte strings
# are returned from lxml
attrib = element.attrib
tag = element.tag
if tag == 'link':
iterable = cls.iter_links_link_element(element)
elif tag == 'meta':
iterable = cls.iter_links_meta_element(element)
elif tag in ('object', 'applet'):
iterable = cls.iter_links_object_element(element)
elif tag == 'param':
iterable = cls.iter_links_param_element(element)
elif tag == 'style':
iterable = cls.iter_links_style_element(element)
else:
iterable = cls.iter_links_plain_element(element)
for link_info in iterable:
yield link_info
if 'style' in attrib:
for link in CSSScraper.scrape_urls(attrib['style']):
yield LinkInfo(
element, element.tag, 'style',
to_str(link),
True, False,
None,
'css'
)
def iter_links_plain_element(cls, element):
'''Iterate any element for links using generic rules.'''
for attrib_name, link in cls.iter_links_by_attrib(element):
inline = cls.is_link_inline(element.tag, attrib_name)
linked = cls.is_html_link(element.tag, attrib_name)
yield LinkInfo(element, element.tag, attrib_name, to_str(link),
inline, linked, None, 'plain')
def iter_links_param_element(cls, element):
'''Iterate a ``param`` element.'''
valuetype = element.get('valuetype', '')
if valuetype.lower() == 'ref' and 'value' in element.attrib:
yield LinkInfo(element, element.tag, 'value',
to_str(element.get('value')), True, False, None,
'plain')
def iter_links_style_element(self, element):
'''Iterate a ``style`` element.'''
if element.text:
link_iter = itertools.chain(
CSSScraper.scrape_imports(element.text),
CSSScraper.scrape_urls(element.text))
for link in link_iter:
yield LinkInfo(element, element.tag, None, to_str(link), True,
False, None, 'css')
def iter_links_meta_element(cls, element):
'''Iterate the ``meta`` element for links.
This function handles refresh URLs.
'''
if element.get('http-equiv', '').lower() == 'refresh':
content_value = element.get('content')
link = parse_refresh(content_value)
if link:
yield LinkInfo(element, element.tag, 'http-equiv',
to_str(link), False, True, None, 'refresh')
def iter_links_param_element(cls, element):
'''Iterate a ``param`` element.'''
valuetype = element.get('valuetype', '')
if valuetype.lower() == 'ref' and 'value' in element.attrib:
yield LinkInfo(
element, element.tag, 'value',
to_str(element.get('value')),
True, False,
None,
'plain'
)
def iter_links_plain_element(cls, element):
'''Iterate any element for links using generic rules.'''
for attrib_name, link in cls.iter_links_by_attrib(element):
inline = cls.is_link_inline(element.tag, attrib_name)
linked = cls.is_html_link(element.tag, attrib_name)
yield LinkInfo(
element, element.tag, attrib_name,
to_str(link),
inline, linked,
None,
'plain'
)
def iter_links_link_element(cls, element):
'''Iterate a ``link`` for URLs.
This function handles stylesheets and icons in addition to
standard scraping rules.
'''
rel = element.get('rel', '')
inline = 'stylesheet' in rel or 'icon' in rel
for attrib_name, link in cls.iter_links_by_attrib(element):
yield LinkInfo(element, element.tag, attrib_name, to_str(link),
inline, not inline, None, 'plain')
def iter_links_object_element(cls, element):
'''Iterate ``object`` and ``embed`` elements.
This function also looks at ``codebase`` and ``archive`` attributes.
'''
base_link = to_str(element.get('codebase', None))
if base_link:
# lxml returns codebase as inline
yield LinkInfo(element, element.tag, 'codebase', base_link, True,
False, None, 'plain')
for attribute in ('code', 'src', 'classid', 'data'):
if attribute in element.attrib:
yield LinkInfo(element, element.tag, attribute,
to_str(element.get(attribute)), True, False,
base_link, 'plain')
if 'archive' in element.attrib:
for match in re.finditer(r'[^ ]+', element.get('archive')):
value = match.group(0)
yield LinkInfo(element, element.tag, 'archive', to_str(value),
True, False, base_link, 'list')
def iter_links_object_element(cls, element):
'''Iterate ``object`` and ``embed`` elements.
This function also looks at ``codebase`` and ``archive`` attributes.
'''
base_link = to_str(element.get('codebase', None))
if base_link:
# lxml returns codebase as inline
yield LinkInfo(
element, element.tag, 'codebase',
base_link,
True, False,
None,
'plain'
)
for attribute in ('code', 'src', 'classid', 'data'):
if attribute in element.attrib:
yield LinkInfo(
element, element.tag, attribute,
to_str(element.get(attribute)),
True, False,
base_link,
'plain'
)
if 'archive' in element.attrib:
for match in re.finditer(r'[^ ]+', element.get('archive')):
value = match.group(0)
yield LinkInfo(
element, element.tag, 'archive',
to_str(value),
True, False,
base_link,
'list'
)
def iter_links_style_element(self, element):
'''Iterate a ``style`` element.'''
if element.text:
link_iter = itertools.chain(
CSSScraper.scrape_imports(element.text),
CSSScraper.scrape_urls(element.text)
)
for link in link_iter:
yield LinkInfo(
element, element.tag, None,
to_str(link),
True, False,
None,
'css'
)
def iter_links_meta_element(cls, element):
'''Iterate the ``meta`` element for links.
This function handles refresh URLs.
'''
if element.get('http-equiv', '').lower() == 'refresh':
content_value = element.get('content')
link = parse_refresh(content_value)
if link:
yield LinkInfo(
element, element.tag, 'http-equiv',
to_str(link),
False, True,
None,
'refresh'
)
def iter_links_link_element(cls, element):
'''Iterate a ``link`` for URLs.
This function handles stylesheets and icons in addition to
standard scraping rules.
'''
rel = element.get('rel', '')
inline = 'stylesheet' in rel or 'icon' in rel
for attrib_name, link in cls.iter_links_by_attrib(element):
yield LinkInfo(
element, element.tag, attrib_name,
to_str(link),
inline, not inline,
None,
'plain'
)
def test_to_str(self):
self.assertEqual('hi', to_str(b'hi'))
self.assertEqual(['hi'], to_str([b'hi']))
self.assertEqual({'hi': 'hello'}, to_str({b'hi': b'hello'}))
def scrape(self, request, response):
if not self.is_html(request, response):
return
content_file = response.body.content_file
encoding = get_heading_encoding(response)
tree = self.parse(content_file, encoding, request.url_info.url)
root = tree.getroot()
if root is None:
return
linked_urls = set()
inline_urls = set()
link_infos = self.iter_links(root)
if 'Refresh' in response.fields:
link = parse_refresh(response.fields['Refresh'])
if link:
link_info = LinkInfo(
None, '_refresh', None,
to_str(link),
False, True,
None, 'refresh'
)
link_infos = itertools.chain(link_infos, [link_info])
for scraped_link in link_infos:
if self._only_relative:
if scraped_link.base_link or '://' in scraped_link.link:
continue
if not self._is_accepted(scraped_link.tag):
continue
base_url = root.base_url
if scraped_link.base_link:
base_url = wpull.url.urljoin(base_url, scraped_link.base_link)
url = wpull.url.urljoin(base_url, scraped_link.link,
allow_fragments=False)
# Browsers seems to tolerate URLs with newlines
url = url.replace('\n', '').replace('\r', '')
if scraped_link.inline:
inline_urls.add(url)
if scraped_link.linked:
linked_urls.add(url)
if self._robots and self._robots_cannot_follow(root):
linked_urls.clear()
return {
'inline_urls': inline_urls,
'linked_urls': linked_urls,
'base_url': to_str(root.base_url),
'encoding': to_str(root.getroottree().docinfo.encoding),
}
