Bug 693026

Description of problem:

If the URL in the form "action" is the same as the URL of the web page containing the form, wget assumes it has already downloaded the file at that URL and rewrites the URL in form action to point to the local file, which is obviously not something one would expect. Instead, that URL should be converted to an absolute HTTP URL.

This happens because convert_links_in_hashtable() in convert.c knows that question.asp itself was downloaded and rewrites the action url to point to the local file.

Version-Release number of selected component (if applicable):


How reproducible:

Mine is 1.12.2 but I believe the same issue exists in just about all versions.

Steps to Reproduce:

$ echo "<form action="wget1.asp.html" method="post"/>" > wget.html
$ python -m SimpleHTTPServer

then try to download the file:
$ wget -P. -k http://localhost:8000/wget.asp.html

You'll see that the URL in the form's action gets converted:
$ cat wget.asp.html 
<form action="http://localhost:8000/wget1.asp.html" method="post"/>

Now, change that URL to be the same as the page name, e.g.,

$ echo "<form action="wget.asp.html" method="post"/>" > wget.html
$ python -m SimpleHTTPServer

and try downloading it again with the same wget command. You'll see that the URL is now rewritten to point to a local file:

$ wget -P. -k http://localhost:8000/wget.asp.html
$ $ cat wget.asp.html.1 
<form action="wget.asp.html.1" method="post"/>

Expected results:

The form action URL should be rewritten into an absolute URL consistently.

Additional info:

I've already submitted this upstream and attached a proposed patch there but it would be nice to get this fixed in Fedora as there hasn't been much response there: http://savannah.gnu.org/bugs/?32587