[
https://issues.apache.org/jira/browse/NUTCH-2064?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sebastian Nagel updated NUTCH-2064:
-----------------------------------
Attachment: NUTCH-2064-v5.patch
Run the last patch over the URL set referenced in NUTCH-1990:
* there is an ArrayIndexOutOfBoundsException (fixed in attached patch, unit
test added): should be {{letter < 128}} in
{code}
... boolean[] unescapedCharacters = new boolean[128];
...
if (letter <= 128 && unescapedCharacters[letter])
{code}
* the normalized URLs look good to me. I've checked a couple of 100
differences, all of them are plausible improvements compared to current trunk
* we should change the title/description of this issue since
** also ASCII characters are encoded and
** unreserved characters are decoded (%41 -> A)
* (as expected and acceptable) the encoding/decoding takes extra time: +11%
(17s -> 19s) when running the whole test set via
{noformat}
cat urls.test | bin/nutch plugin urlnormalizer-basic
org.apache.nutch.net.urlnormalizer.basic.BasicURLNormalizer
{noformat}
> URLNormalizer basic to properly encode non-ASCII characters
> -----------------------------------------------------------
>
> Key: NUTCH-2064
> URL: https://issues.apache.org/jira/browse/NUTCH-2064
> Project: Nutch
> Issue Type: Bug
> Affects Versions: 1.10
> Reporter: Markus Jelsma
> Fix For: 1.12
>
> Attachments: NUTCH-1098.patch, NUTCH-1098.patch, NUTCH-2064-v3.patch,
> NUTCH-2064-v5.patch, NUTCH-2064.patch
>
>
> NUTCH-1098 rewritten to work on trunk. Unit test is identical to 1098.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
