Merge branch 'master' into do-1985

getnikola · Sep 1, 2015 · e4b7806 · e4b7806
2 parents 4edd6e0 + 4e0be1b
commit e4b7806
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 3 deletions.
diff --git a/CHANGES.txt b/CHANGES.txt
@@ -28,6 +28,7 @@ Features
 Bugfixes
 --------
 
+* Link checker should check all absolute URLs to self (Issue #1991) 
 * Check ``img|source[@srcset]`` as part of ``check -l``  (Issue #1989)
 * Clean up translations for third party components
 * ``pagekind["main_index"]`` set on the main indexes to differentiate

diff --git a/nikola/nikola.py b/nikola/nikola.py
@@ -1106,7 +1106,7 @@ def rewrite_links(self, doc, src, lang):
         doc.rewrite_links(lambda dst: self.url_replacer(src, dst, lang), resolve_base_href=False)
 
         # lxml ignores srcset in img and source elements, so do that by hand
-        objs = list(doc.findall('(*//img|*//source)'))
+        objs = list(doc.xpath('(*//img|*//source)'))
         for obj in objs:
             if 'srcset' in obj.attrib:
                 urls = [u.strip() for u in obj.attrib['srcset'].split(',')]

diff --git a/nikola/plugins/command/check.py b/nikola/plugins/command/check.py
@@ -217,6 +217,11 @@ def analyze(self, fname, find_sources=False, check_remote=False):
                 if target == "#":
                     continue
                 target, _ = urldefrag(target)
+
+                # absolute URL to root-relative
+                if target.startswith(base_url.geturl()):
+                    target = target.replace(base_url.geturl(), '/')
+
                 parsed = urlparse(target)
 
                 # Warn about links from https to http (mixed-security)
@@ -229,8 +234,6 @@ def analyze(self, fname, find_sources=False, check_remote=False):
                         ((parsed.scheme or target.startswith('//')) and url_type in ('rel_path', 'full_path')):
                     if not check_remote or parsed.scheme not in ["http", "https"]:
                         continue
-                    if parsed.netloc == base_url.netloc:  # absolute URL to self.site
-                        continue
                     if target in self.checked_remote_targets:  # already checked this exact target
                         if self.checked_remote_targets[target] in [301, 307]:
                             self.logger.warn("Remote link PERMANENTLY redirected in {0}: {1} [Error {2}]".format(filename, target, self.checked_remote_targets[target]))