hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Noah Levitt (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HTTPCORE-325) support custom implementations of SessionInputBuffer and SessionOutputBuffer
Date Fri, 04 Jan 2013 03:20:12 GMT

    [ https://issues.apache.org/jira/browse/HTTPCORE-325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13543568#comment-13543568
] 

Noah Levitt commented on HTTPCORE-325:
--------------------------------------

> I have not thoroughly studied Heritrix classes you are referring to but I cant help thinking
that you would be much better off decorating Socket input stream the session buffer binds
to, rather than session buffer itself.

That was the approach I tried first, if I understand you correctly. Marking the end of http
headers / beginning of the body turned out to be a problem. Heritrix needs to do this so it
knows when to start digesting (e.g. sha1), among other things. I overrode HttpClientConnection.receiveResponseEntity()
to do that. But by the time receiveResponseEntity() is called, SessionInputBufferImpl.fillBuffer()
has already blown right through the start of the body.

To prevent this from happening in fillBuffer(), I had to override readLine(), and to do that
I needed access to those 3 private variables and the one method. See https://github.com/nlevitt/heritrix3/blob/hc43/modules/src/main/java/org/archive/modules/fetcher/RecordingSessionInputBuffer.java
(SessionOutputBufferImpl on the other hand didn't need any visibility changes, now that I
go back and look. My mistake.)

Is there a better way?
                
> support custom implementations of SessionInputBuffer and SessionOutputBuffer
> ----------------------------------------------------------------------------
>
>                 Key: HTTPCORE-325
>                 URL: https://issues.apache.org/jira/browse/HTTPCORE-325
>             Project: HttpComponents HttpCore
>          Issue Type: Bug
>    Affects Versions: 4.3-alpha2
>            Reporter: Noah Levitt
>         Attachments: httpcore-325-20121231182846.diff
>
>
> In heritrix we have a set of classes that wrap streams and record them verbatim for replay.
One of the things it needs to do is make a note of where the http headers end and the message
body begins. In order to make this work with httpcomponents I found I needed custom implementations
of Session*Buffer. 

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

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


Mime
View raw message