[
https://issues.apache.org/jira/browse/DROIDS-58?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12843318#action_12843318
]
Bertil Chapuis commented on DROIDS-58:
--------------------------------------
In my opinion the filter mecanism makes sense straight after the phase 1 and
before the phase 4 you described.
However, for the phase 2 and 3 the hierarchical approach of the Decorator
pattern may provide a simpler solution.
Let's see the following process:
1. taskQueue.poll():link
2. filter.isAllowed(link):boolean
3. fetcherDecorator.fetch(link):resource
3 bis. fetcher.fetch(link):resource
4. parserDecorator.parse(link, resource)
4 bis. parserDecorator.parse(link, resource)
4 bis bis. parser.parse(link, resource)
5. extractorDecorator.extract(link, resource):outlinks
5 bis. extractor.extract(link, resource):outlinks
6. filter.isAllowed(outlink)
7. taskQueue.add(outlink)
> Implement a filter mechanism that allow intecepting every stage of a crawling
> process
> -------------------------------------------------------------------------------------
>
> Key: DROIDS-58
> URL: https://issues.apache.org/jira/browse/DROIDS-58
> Project: Droids
> Issue Type: New Feature
> Affects Versions: 0.01
> Reporter: Mingfai Ma
>
> refer to this:
> http://mail-archives.apache.org/mod_mbox/incubator-droids-dev/200906.mbox/%[email protected]%3e
> assume the process is
> 1. poll a link from queue
> 2. fetch entity
> 3. parse entity
> 4. extract outlinks
> we provide a mechanism to intercept the process in every stage. e.g. a
> LinkFilter has a "public T polled(T link);" interface, any filter may reject
> or transform a Link polled from the queue. similar logic applies to fetching,
> parsing, and extracting (outlink)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
