dos2unix for url.py, argh
This commit is contained in:
Florian Bruhin
parent
b50eba38d3
commit
dadfc952d1
@ -1,110 +1,110 @@
|
|||||||
"""Utils regarding URL handling."""
|
"""Utils regarding URL handling."""
|
||||||
|
|
||||||
import re
|
import re
|
||||||
import socket
|
import socket
|
||||||
import logging
|
import logging
|
||||||
import urllib.parse
|
import urllib.parse
|
||||||
|
|
||||||
from PyQt5.QtCore import QUrl
|
from PyQt5.QtCore import QUrl
|
||||||
|
|
||||||
import qutebrowser.utils.config as config
|
import qutebrowser.utils.config as config
|
||||||
|
|
||||||
|
|
||||||
def qurl(url):
|
def qurl(url):
|
||||||
"""Get a QUrl from an url string."""
|
"""Get a QUrl from an url string."""
|
||||||
return url if isinstance(url, QUrl) else QUrl(url)
|
return url if isinstance(url, QUrl) else QUrl(url)
|
||||||
|
|
||||||
|
|
||||||
def urlstring(url):
|
def urlstring(url):
|
||||||
"""Return an QUrl as string.
|
"""Return an QUrl as string.
|
||||||
|
|
||||||
qurl -- URL as string or QUrl.
|
qurl -- URL as string or QUrl.
|
||||||
"""
|
"""
|
||||||
return url.url() if isinstance(url, QUrl) else url
|
return url.url() if isinstance(url, QUrl) else url
|
||||||
|
|
||||||
|
|
||||||
def fuzzy_url(url):
|
def fuzzy_url(url):
|
||||||
"""Return a QUrl based on an user input which is URL or search term.
|
"""Return a QUrl based on an user input which is URL or search term.
|
||||||
|
|
||||||
url -- URL to load as QUrl or string.
|
url -- URL to load as QUrl or string.
|
||||||
"""
|
"""
|
||||||
u = qurl(url)
|
u = qurl(url)
|
||||||
urlstr = urlstring(url)
|
urlstr = urlstring(url)
|
||||||
if is_url(u):
|
if is_url(u):
|
||||||
# probably an address
|
# probably an address
|
||||||
logging.debug("url is a fuzzy address")
|
logging.debug("url is a fuzzy address")
|
||||||
newurl = QUrl.fromUserInput(urlstr)
|
newurl = QUrl.fromUserInput(urlstr)
|
||||||
else: # probably a search term
|
else: # probably a search term
|
||||||
logging.debug("url is a fuzzy search term")
|
logging.debug("url is a fuzzy search term")
|
||||||
try:
|
try:
|
||||||
newurl = _get_search_url(urlstr)
|
newurl = _get_search_url(urlstr)
|
||||||
except ValueError: # invalid search engine
|
except ValueError: # invalid search engine
|
||||||
newurl = QUrl.fromUserInput(urlstr)
|
newurl = QUrl.fromUserInput(urlstr)
|
||||||
logging.debug('Converting fuzzy term {} to url -> {}'.format(urlstr,
|
logging.debug('Converting fuzzy term {} to url -> {}'.format(urlstr,
|
||||||
newurl.url()))
|
newurl.url()))
|
||||||
return newurl
|
return newurl
|
||||||
|
|
||||||
|
|
||||||
def _get_search_url(txt):
|
def _get_search_url(txt):
|
||||||
"""Return a search engine URL (QUrl) for a text."""
|
"""Return a search engine URL (QUrl) for a text."""
|
||||||
logging.debug('Finding search engine for "{}"'.format(txt))
|
logging.debug('Finding search engine for "{}"'.format(txt))
|
||||||
r = re.compile(r'(^|\s+)!(\w+)($|\s+)')
|
r = re.compile(r'(^|\s+)!(\w+)($|\s+)')
|
||||||
m = r.search(txt)
|
m = r.search(txt)
|
||||||
if m:
|
if m:
|
||||||
engine = m.group(2)
|
engine = m.group(2)
|
||||||
# FIXME why doesn't fallback work?!
|
# FIXME why doesn't fallback work?!
|
||||||
template = config.config.get('searchengines', engine, fallback=None)
|
template = config.config.get('searchengines', engine, fallback=None)
|
||||||
term = r.sub('', txt)
|
term = r.sub('', txt)
|
||||||
logging.debug('engine {}, term "{}"'.format(engine, term))
|
logging.debug('engine {}, term "{}"'.format(engine, term))
|
||||||
else:
|
else:
|
||||||
template = config.config.get('searchengines', '__default__',
|
template = config.config.get('searchengines', '__default__',
|
||||||
fallback=None)
|
fallback=None)
|
||||||
term = txt
|
term = txt
|
||||||
logging.debug('engine: default, term "{}"'.format(txt))
|
logging.debug('engine: default, term "{}"'.format(txt))
|
||||||
if template is None or not term:
|
if template is None or not term:
|
||||||
raise ValueError
|
raise ValueError
|
||||||
return QUrl.fromUserInput(template.format(urllib.parse.quote(term)))
|
return QUrl.fromUserInput(template.format(urllib.parse.quote(term)))
|
||||||
|
|
||||||
|
|
||||||
def is_about_url(url):
|
def is_about_url(url):
|
||||||
"""Return True if url (QUrl) is an about:... or other special URL."""
|
"""Return True if url (QUrl) is an about:... or other special URL."""
|
||||||
return urlstring(url).replace('http://', '').startswith('about:')
|
return urlstring(url).replace('http://', '').startswith('about:')
|
||||||
|
|
||||||
|
|
||||||
def is_url(url):
|
def is_url(url):
|
||||||
"""Return True if url (QUrl) seems to be a valid URL."""
|
"""Return True if url (QUrl) seems to be a valid URL."""
|
||||||
logging.debug('Checking if "{}" is an URL'.format(url.url()))
|
logging.debug('Checking if "{}" is an URL'.format(url.url()))
|
||||||
if ' ' in urlstring(url):
|
if ' ' in urlstring(url):
|
||||||
# An URL will never contain a space
|
# An URL will never contain a space
|
||||||
logging.debug('Contains space -> no url')
|
logging.debug('Contains space -> no url')
|
||||||
return False
|
return False
|
||||||
elif config.config.getboolean('general', 'addressbar_dns_lookup'):
|
elif config.config.getboolean('general', 'addressbar_dns_lookup'):
|
||||||
logging.debug('Checking via DNS')
|
logging.debug('Checking via DNS')
|
||||||
return _is_url_dns(QUrl.fromUserInput(urlstring(url)))
|
return _is_url_dns(QUrl.fromUserInput(urlstring(url)))
|
||||||
else:
|
else:
|
||||||
logging.debug('Checking via naive check')
|
logging.debug('Checking via naive check')
|
||||||
return _is_url_naive(url)
|
return _is_url_naive(url)
|
||||||
|
|
||||||
|
|
||||||
def _is_url_naive(url):
|
def _is_url_naive(url):
|
||||||
"""Naive check if given url (QUrl) is really an url."""
|
"""Naive check if given url (QUrl) is really an url."""
|
||||||
PROTOCOLS = ['http://', 'https://']
|
PROTOCOLS = ['http://', 'https://']
|
||||||
u = urlstring(url)
|
u = urlstring(url)
|
||||||
return (any([u.startswith(proto) for proto in PROTOCOLS]) or '.' in u or
|
return (any([u.startswith(proto) for proto in PROTOCOLS]) or '.' in u or
|
||||||
is_about_url(url) or u == 'localhost')
|
is_about_url(url) or u == 'localhost')
|
||||||
|
|
||||||
|
|
||||||
def _is_url_dns(url):
|
def _is_url_dns(url):
|
||||||
"""Check if an url (QUrl) is really an url via DNS."""
|
"""Check if an url (QUrl) is really an url via DNS."""
|
||||||
# FIXME we could probably solve this in a nicer way by attempting to open
|
# FIXME we could probably solve this in a nicer way by attempting to open
|
||||||
# the page in the webview, and then open the search if that fails.
|
# the page in the webview, and then open the search if that fails.
|
||||||
host = url.host()
|
host = url.host()
|
||||||
logging.debug("DNS request for {}".format(host))
|
logging.debug("DNS request for {}".format(host))
|
||||||
if not host:
|
if not host:
|
||||||
return False
|
return False
|
||||||
try:
|
try:
|
||||||
socket.gethostbyname(host)
|
socket.gethostbyname(host)
|
||||||
except socket.gaierror:
|
except socket.gaierror:
|
||||||
return False
|
return False
|
||||||
else:
|
else:
|
||||||
return True
|
return True
|
||||||
|
|||||||
Loading…
Reference in New Issue
Block a user