[jira] [Commented] (NUTCH-2993) ScoringDepth plugin to skip depth check based on URL Pattern

["Sebastian Nagel (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/NUTCH-2993?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17733442#comment-17733442 ] 

Sebastian Nagel commented on NUTCH-2993:
----------------------------------------

Hi [~markus17], the patch actually applies to master. So no need to create a specific one. Few comments:
- because both pattern matching and URL.toString() are computationally not trivial (an integer comparison is): maybe do the skip depth check only if {{curDepth >= curMaxDepth}}. This saves computation time for all pages which are below {{curMaxDepth}}.
- [Configuration.get(name)|https://hadoop.apache.org/docs/stable/api/org/apache/hadoop/conf/Configuration.html#get-java.lang.String-] returns null as default value, no need to pass it. But maybe pass an empty string as default value and adapt the check.
- add the property to nutch-default.xml with a description
- maybe catch potential exceptions in {{Pattern.compile(...)}} and log them as error. It comes from the configuration and a invalid pattern would otherwise stop the crawl.
- typo in variable name {{depthOverrridePattern}}


> ScoringDepth plugin to skip depth check based on URL Pattern
> ------------------------------------------------------------
>
>                 Key: NUTCH-2993
>                 URL: https://issues.apache.org/jira/browse/NUTCH-2993
>             Project: Nutch
>          Issue Type: Improvement
>            Reporter: Markus Jelsma
>            Assignee: Markus Jelsma
>            Priority: Minor
>             Fix For: 1.20
>
>         Attachments: NUTCH-2993-1.15.patch
>
>
> We do not want some crawl to go deep and broad, but instead focus it on a narrow section of sites. This patch skips the depth check if the current URL matches some regular expression.
>  
> Initially we tried to set a custom maxDepth based on a Pattern match, but this didn't work. The crawler still managed to creep too deep due to having links everywhere.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)