Re: GSA Content feed for MS Word files - jdandrea - com.googlegroups.google-enterprise-developer

9 messages in com.googlegroups.google-enterprise-developerRe: GSA Content feed for MS Word files

From	Sent On	Attachments
ekovalev	05 Dec 2006 06:43
jdandrea	06 Dec 2006 08:47
jdandrea	06 Dec 2006 09:03
ekovalev	06 Dec 2006 09:51
Jeff Ragusa	07 Dec 2006 12:29
ekovalev	08 Dec 2006 07:41
jdandrea	22 Dec 2006 06:05
ekovalev	29 Dec 2006 07:18
jdandrea	13 Feb 2007 08:41

Subject:	Re: GSA Content feed for MS Word files
From:	jdandrea (jdan...@gmail.com)
Date:	12/22/2006 06:05:15 AM
List:	com.googlegroups.google-enterprise-developer

Greetings! Been away for a li'l bit, but I'm catching up now ...

ekovalev wrote:

I just tried test content feed with embedded ms word encoding with mmencode. Works fine.

Terrific!

One thing when I tried to include three word documents to one feed I got internal error on GSA, but when I upload each document in separate feed it worked without problem.

Hmm. Can you post the XML on a web site (private link) for download/inspection? I'm curious.

One more question. If ever my files become web accessible I have another problem, I don't have file extensions on the server side.

Ahh. That could pose a challenge when saving the files to disk - and in some cases when viewing them within the browser (depending on the server-side MIME Type or Content Type settings).

[Reads more of the thread.] Yep, what JeffR said.

It seems GSA uses mimetype from feed only to represent doc type in front of title on result page.

Ahh. Interesting. So then if you use this:

<record url="..." mimetype="text/plain"> <content encoding="base64binary">Zm9vIGJhcgo</content> </record>

You're saying there's no distinction of "is this Word or PDF" ... interesting.

So Jeff - what's happening is we have a feed posting content that isn't available via HTTP, is encoded using base64, and doesn't have an explicit mime type (other than text/plain which I presume is used along the base64binary encoding). The question then becomes ... how is the _real_ MIME type determined? Since I've never crawled content this way, hmm ... I'm gonna have to try this. :)