[
https://issues.apache.org/jira/browse/NUTCH-2554?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16430436#comment-16430436
]
Sebastian Nagel commented on NUTCH-2554:
----------------------------------------
Please have a look at [PR #310|https://github.com/apache/nutch/pull/310] which
includes a fix for the problem. Note that URLs are now read from stdin:
{noformat}
% echo 'http://example.com/?q=%2F' \
| nutch parsechecker
-Dplugin.includes='protocol-http|parse-html|urlnormalizer-basic' -normalize
-followRedirects
fetching: http://example.com/?q=%2F
...
{noformat}
> parserchecker can't fetch some URLs
> -----------------------------------
>
> Key: NUTCH-2554
> URL: https://issues.apache.org/jira/browse/NUTCH-2554
> Project: Nutch
> Issue Type: Bug
> Reporter: Gerard Bouchar
> Priority: Major
>
> The parserchecker (org.apache.nutch.parse.ParserChecker) calls
> _URLUtil.toASCII_ on the url it is given, reencoding already percent-encoded
> URLs.
> For instance, let's say we want to query
> [http://example.com|http://example.com_/], passing a GET parameter with name
> 'q' and value '/'. '/' is a special character, and thus has to be encoded
> before being sent.
> If we pass '[_http://example.com/?q=/_'|http://example.com/?q=/%27] to the
> parserchecker, then it doesn't encode the '/', and tries to fetch the URL as
> is, which is invalid.
> If we try to encode the parameter beforehand, and call the parsechecker with
> 'http://example.com/?q=%2F', then it encodes the '%' sign to '%25', and thus
> fetches '[http://example.com/?q=%252F'.|http://example.com/?q=%252F%27.]
> This actually makes it impossible to fetch the correct URL
> ([http://example.com/?q=%2F]) from the parserchecker.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
