hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Nick Trown (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HTTPCLIENT-1641) CircularRedirect due to rebuilding URI after decoding path, etc.
Date Fri, 17 Apr 2015 15:48:28 GMT

    [ https://issues.apache.org/jira/browse/HTTPCLIENT-1641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14500075#comment-14500075
] 

Nick Trown commented on HTTPCLIENT-1641:
----------------------------------------

I've also been trying to debug the function and you are right about URI constructor of course.
 The problem is rather subtle.  

The redirect is insisting that the parens be encoded to %28 and %29.  However, the parens
are part of the "unreserved" character categories.  The URI constructor specifies that for
path anything not in "unreserved", "punct", "escaped", or "other" categories are encoded.
 Therefore "unreserved" is not encoded, and hence the parens are not encoded which the redirect
is insisting on.

Using the raw path does not work and you mention as when URI constructor is called with the
encoded characters it encodes the % that are present within.

> CircularRedirect due to rebuilding URI after decoding path, etc.
> ----------------------------------------------------------------
>
>                 Key: HTTPCLIENT-1641
>                 URL: https://issues.apache.org/jira/browse/HTTPCLIENT-1641
>             Project: HttpComponents HttpClient
>          Issue Type: Bug
>          Components: HttpClient
>    Affects Versions: 4.3.5
>            Reporter: Nick Trown
>            Priority: Minor
>
> When I use HttpClient for the following URL the DefaultRedirectStrategy fails to build
a redirected URI properly.
> http://hotpads.com/apartments/Camelot-Apartments-(-Drexel-&-UPenn-)-Philadelphia-PA-19104--3phq8mahfzya3
> The reason is that the redirect URL returned is:
> /apartments/Camelot-Apartments-%28-Drexel-%26-UPenn-%29-Philadelphia-PA-19104--3phq8mahfzya3
> ** Note: the only difference is that the redirect URL has the parens encoded.
> The DefaultDirectEncoder does get the redirect URL, and since it is not absolute eventually
calls URIUtils.resolve().  This then calls normalizeSyntax() where the bug is.
> In normalizeSyntax() the URI is reconstructed.  However, it incorrectly uses getPath(),
getFragment(), etc.  instead of the raw equivalents such as getRawPath(), getRawFragment(),
etc.
> In this case, the getPath() returned is therefore:
> apartments/Camelot-Apartments-(-Drexel-&-UPenn-)-Philadelphia-PA-19104--3phq8mahfzya3
> Where the parens are back.  The DefaultRedirectStrategy detects this as a circular redirect.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Mime
View raw message