#472: linkcheck builder: Check links in parallel, use HTTP HEAD requests and allow configuring the timeout.

New config values: :confval:`linkcheck_timeout` and :confval:`linkcheck_workers`.
This commit is contained in:
Georg Brandl
2011-01-04 11:27:42 +01:00
parent 3642b521ed
commit 51852c0e87
4 changed files with 127 additions and 75 deletions

View File

@@ -24,6 +24,10 @@ Release 1.1 (in development)
* #443: Allow referencing external graphviz files.
* #472: linkcheck builder: Check links in parallel, use HTTP HEAD
requests and allow configuring the timeout. New config values:
:confval:`linkcheck_timeout` and :confval:`linkcheck_workers`.
* #221: Add Swedish locale.
* Added ``inline`` option to graphviz directives, and fixed the

View File

@@ -1132,6 +1132,21 @@ Options for the linkcheck builder
.. versionadded:: 1.1
.. confval:: linkcheck_timeout
A timeout value, in seconds, for the linkcheck builder. **Only works in
Python 2.6 and higher.** The default is to use Python's global socket
timeout.
.. versionadded:: 1.1
.. confval:: linkcheck_workers
The number of worker threads to use when checking links. Default is 5
threads.
.. versionadded:: 1.1
.. rubric:: Footnotes

View File

@@ -10,9 +10,12 @@
"""
import re
import sys
import Queue
import socket
import threading
from os import path
from urllib2 import build_opener, HTTPError
from urllib2 import build_opener, Request
from docutils import nodes
@@ -24,6 +27,12 @@ opener = build_opener()
opener.addheaders = [('User-agent', 'Mozilla/5.0')]
class HeadRequest(Request):
"""Subclass of urllib2.Request that sends a HEAD request."""
def get_method(self):
return 'HEAD'
class CheckExternalLinksBuilder(Builder):
"""
Checks for broken external links.
@@ -40,6 +49,83 @@ class CheckExternalLinksBuilder(Builder):
# create output file
open(path.join(self.outdir, 'output.txt'), 'w').close()
# create queues and worker threads
self.wqueue = Queue.Queue()
self.rqueue = Queue.Queue()
self.workers = []
for i in range(self.app.config.linkcheck_workers):
thread = threading.Thread(target=self.check_thread)
thread.setDaemon(True)
thread.start()
self.workers.append(thread)
def check_thread(self):
kwargs = {}
if sys.version_info > (2, 5) and self.app.config.linkcheck_timeout:
kwargs['timeout'] = self.app.config.linkcheck_timeout
def check():
# check for various conditions without bothering the network
if len(uri) == 0 or uri[0:7] == 'mailto:' or uri[0:4] == 'ftp:':
return 'unchecked', ''
elif not (uri[0:5] == 'http:' or uri[0:6] == 'https:'):
return 'local', ''
elif uri in self.good:
return 'working', ''
elif uri in self.broken:
return 'broken', self.broken[uri]
elif uri in self.redirected:
return 'redirected', self.redirected[uri]
for rex in self.to_ignore:
if rex.match(uri):
return 'ignored', ''
# need to actually check the URI
try:
f = opener.open(HeadRequest(uri), **kwargs)
f.close()
except Exception, err:
self.broken[uri] = str(err)
return 'broken', str(err)
if f.url.rstrip('/') == uri.rstrip('/'):
self.good.add(uri)
return 'working', 'new'
else:
self.redirected[uri] = f.url
return 'redirected', f.url
while True:
uri, docname, lineno = self.wqueue.get()
if uri is None:
break
status, info = check()
self.rqueue.put((uri, docname, lineno, status, info))
def process_result(self, result):
uri, docname, lineno, status, info = result
if status == 'unchecked':
return
if status == 'working' and info != 'new':
return
if lineno:
self.info('(line %3d) ' % lineno, nonl=1)
if status == 'ignored':
self.info(uri + ' - ' + darkgray('ignored'))
elif status == 'local':
self.info(uri + ' - ' + darkgray('local'))
self.write_entry('local', docname, lineno, uri)
elif status == 'working':
self.info(uri + ' - ' + darkgreen('working'))
elif status == 'broken':
self.info(uri + ' - ' + red('broken: ') + info)
self.write_entry('broken', docname, lineno, uri + ': ' + info)
if self.app.quiet:
self.warn('broken link: %s' % uri,
'%s:%s' % (self.env.doc2path(docname), lineno))
elif status == 'redirected':
self.info(uri + ' - ' + purple('redirected') + ' to ' + info)
self.write_entry('redirected', docname, lineno, uri + ' to ' + info)
def get_target_uri(self, docname, typ=None):
return ''
@@ -51,65 +137,25 @@ class CheckExternalLinksBuilder(Builder):
def write_doc(self, docname, doctree):
self.info()
n = 0
for node in doctree.traverse(nodes.reference):
try:
self.check(node, docname)
except KeyError:
if 'refuri' not in node:
continue
def check(self, node, docname):
uri = node['refuri']
if '#' in uri:
uri = uri.split('#')[0]
if uri in self.good:
return
lineno = None
while lineno is None:
node = node.parent
if node is None:
break
lineno = node.line
if len(uri) == 0 or uri[0:7] == 'mailto:' or uri[0:4] == 'ftp:':
return
if lineno:
self.info('(line %3d) ' % lineno, nonl=1)
for rex in self.to_ignore:
if rex.match(uri):
self.info(uri + ' - ' + darkgray('ignored'))
return
if uri[0:5] == 'http:' or uri[0:6] == 'https:':
self.info(uri, nonl=1)
if uri in self.broken:
(r, s) = self.broken[uri]
elif uri in self.redirected:
(r, s) = self.redirected[uri]
else:
(r, s) = self.resolve(uri)
if r == 0:
self.info(' - ' + darkgreen('working'))
self.good.add(uri)
elif r == 2:
self.info(' - ' + red('broken: ') + s)
self.write_entry('broken', docname, lineno, uri + ': ' + s)
self.broken[uri] = (r, s)
if self.app.quiet:
self.warn('broken link: %s' % uri,
'%s:%s' % (self.env.doc2path(docname), lineno))
else:
self.info(' - ' + purple('redirected') + ' to ' + s)
self.write_entry('redirected', docname,
lineno, uri + ' to ' + s)
self.redirected[uri] = (r, s)
else:
self.info(uri + ' - ' + darkgray('local'))
self.write_entry('local', docname, lineno, uri)
uri = node['refuri']
if '#' in uri:
uri = uri.split('#')[0]
lineno = None
while lineno is None:
node = node.parent
if node is None:
break
lineno = node.line
self.wqueue.put((uri, docname, lineno), False)
n += 1
done = 0
while done < n:
self.process_result(self.rqueue.get())
done += 1
if self.broken:
self.app.statuscode = 1
@@ -120,21 +166,6 @@ class CheckExternalLinksBuilder(Builder):
line, what, uri))
output.close()
def resolve(self, uri):
try:
f = opener.open(uri)
f.close()
except HTTPError, err:
#if err.code == 403 and uri.startswith('http://en.wikipedia.org/'):
# # Wikipedia blocks requests from urllib User-Agent
# return (0, 0)
return (2, str(err))
except Exception, err:
return (2, str(err))
if f.url.rstrip('/') == uri.rstrip('/'):
return (0, 0)
else:
return (1, f.url)
def finish(self):
return
for worker in self.workers:
self.wqueue.put((None, None, None), False)

View File

@@ -168,6 +168,8 @@ class Config(object):
# linkcheck options
linkcheck_ignore = ([], None),
linkcheck_timeout = (None, None),
linkcheck_workers = (5, None),
)
def __init__(self, dirname, filename, overrides, tags):