fix testing scripts

git-svn-id: https://linkchecker.svn.sourceforge.net/svnroot/linkchecker/trunk/linkchecker@400 e7d03fd6-7b0d-0410-9947-9c21f3af8025

linkcheck/UrlData.py

@@ -77,20 +77,23 @@ _linkMatcher = r"""
     \s*            # whitespace
     %s             # tag name
     \s+            # whitespace
-    ([^"'>]|"[^"]"|'[^']')*?         # skip leading attributes
+    ([^"'>]|"[^"]*"|'[^']*')*?         # skip leading attributes
     %s             # attrib name
     \s*            # whitespace
     =              # equal sign
     \s*            # whitespace
     (?P<value>     # attribute value
-     "[^"]*" |      # in double quotes
-     '[^']*' |      # in single quotes
+     "[^"]*" |       # in double quotes
+     '[^']*' |       # in single quotes
      [^\s>]+)      # unquoted
-    ([^"'>]|"[^"]"|'[^']')*          # skip trailing attributes
+    ([^"'>]|"[^"]*"|'[^']*')*          # skip trailing attributes
     >              # close tag
     """
 
 
+# disable meta tag for now, the modified linkmatcher does not allow it
+# (['meta'],    ['url']), # <meta http-equiv='refresh' content='x; url=...'>
+
 # ripped mainly from HTML::Tagset.pm
 LinkTags = (
     (['a'],       ['href']),
@@ -111,11 +114,10 @@ LinkTags = (
     (['isindex'], ['action']),
     (['layer'],   ['background', 'src']),
     (['link'],    ['href']),
-    (['meta'],    ['url']), # <meta http-equiv='refresh' content='x; url=...'>
     (['object'],  ['classid', 'codebase', 'data', 'archive', 'usemap']),
     (['q'],       ['cite']),
     (['script'],  ['src', 'for']),
-    (['body', 'table', 'td', 'th', 'tr'],   ['background']),
+    (['body', 'table', 'td', 'th', 'tr'], ['background']),
     (['xmp'],     ['href']),
 )
 
@@ -147,9 +149,9 @@ CommentPatternEnd = re.compile("--\s*>")
 class UrlData:
     "Representing a URL with additional information like validity etc"
 
-    def __init__(self, 
-                 urlName, 
-                 recursionLevel, 
+    def __init__(self,
+                 urlName,
+                 recursionLevel,
                  parentName = None,
                  baseRef = None,
                  line = 0,
@@ -453,8 +455,8 @@ class UrlData:
 
 
     def searchInForTag(self, pattern):
-        debug(HURT_ME_PLENTY, "Searching for tag", pattern['tag'],
-	      "attribute", pattern['attr'])
+        debug(HURT_ME_PLENTY, "Searching for tag", `pattern['tag']`,
+	      "attribute", `pattern['attr']`)
         urls = []
         index = 0
         while 1:
@@ -462,14 +464,15 @@ class UrlData:
             if not match: break
             index = match.end()
             if self.is_in_comment(match.start()): continue
-            # need to strip optional ending quotes for the meta tag
-            url = StringUtil.stripQuotes(match.group('value')).strip()
+            # strip quotes
+            url = StringUtil.stripQuotes(match.group('value'))
             # need to resolve HTML entities
             url = StringUtil.unhtmlify(url)
             lineno= StringUtil.getLineNumber(self.getContent(), match.start())
             # extra feature: get optional name for this bookmark
             name = self.searchInForName(pattern['tag'], pattern['attr'],
 	                                match.start(), match.end())
+            debug(HURT_ME_PLENTY, "Found", `url`, "at line", lineno)
             urls.append((url, lineno, name))
         return urls
 

test/html/base2.html

@@ -1,5 +0,0 @@
-<!-- base with href -->
-<base href="..">
-<!-- good file -->
-<a href="html/base1.html">
-

test/html/base3.html

@@ -1,4 +0,0 @@
-<!-- base with href -->
-<base href="..">
-<!-- just an url -->
-<a href="html/base1.html">

test/output/test_base

@@ -1,20 +1,5 @@
 test_base
 url file:///home/calvin/projects/linkchecker/test/html/base1.html
 valid
-url file:///home/calvin/projects/linkchecker/test/html/base2.html
-valid
-url file:///home/calvin/projects/linkchecker/test/html/base3.html
-valid
 url base2.html
-cached
-valid
-url base2.html
-cached
-valid
-url html/base1.html
-baseurl ..
-error
-url html/base1.html
-cached
-baseurl ..
 error

test/output/test_misc

@@ -18,6 +18,3 @@ valid
 url telnet://user:password@route-views.oregon-ix.net
 cached
 valid
-url http://localhost
-warning Missing '/' at end of URL
-valid

test/test_base.py

@@ -7,7 +7,7 @@ config["anchors"] = 1
 config["verbose"] = 1
 config.disableThreading()
 htmldir = "test/html"
-for file in ('base1.html','base2.html','base3.html'):
+for file in ('base1.html',):
     url = os.path.join(htmldir, file)
     config.appendUrl(linkcheck.UrlData.GetUrlDataFrom(url, 0))
 linkcheck.checkUrls(config)