Comments on: Inside the Dojo Toolbox

By: John Wood

John Wood — Tue, 16 Sep 2008 20:04:27 +0000

Thanks I found it.

By: Kevin Dangoor

Kevin Dangoor — Tue, 16 Sep 2008 19:15:42 +0000

I think Paver is convenient and powerful, but I’m biased :)

The index database is packaged up in the API doc zip file. That is downloaded by the Toolbox and unzipped into the “Application Storage” directory (which is located in a platform-appropriate space by Air).

Kevin

By: John Wood

John Wood — Tue, 16 Sep 2008 18:41:56 +0000

I never used Paver before but it looks like it is very powerful.

John

By: John Wood

John Wood — Tue, 16 Sep 2008 18:40:28 +0000

Hi Kevin,

Where do you store the index database on windows?

Thanks,
John

By: Kevin Dangoor

Kevin Dangoor — Tue, 16 Sep 2008 13:06:59 +0000

Hi John,

ft3 is an optional part of sqlite, and AIR does not ship with it. If it did, I probably would never have created my own full-text search solution.

One of the great features of sqlite is that its databases are just single, platform-portable files. After the Python code generates the database, I just package up the index database along with the documentation itself. Then, the API viewer just opens a connection to that database as needed.

The pavement.py file creates a script called “apigen”. That script runs the main function in apigen/transform.py, which is what uses the indexer code. All of that runs completely independently of the AIR code. apigen generates the search index database and it also massages the API docs that we get to better tune it for what we need in the Toolbox.

Kevin

By: John Wood

John Wood — Tue, 16 Sep 2008 01:14:27 +0000

Hi Kevin,

I forgot to ask where in the code do you use index.py?

Thanks,
John

By: John Wood

John Wood — Tue, 16 Sep 2008 01:05:32 +0000

Hi Kevin,
Thanks for quick reply. I see that index.py generates new db and index tables using words parsed by sphinx’s beautifulsoup parser. I really like how u use your own ft tables!

Did you try to use ft3 that comes with sqlite3?

How did you package the generated db with air file?

By: Kevin Dangoor

Kevin Dangoor — Mon, 15 Sep 2008 14:33:33 +0000

Hi John,

I don’t believe it would be possible to modify the builtin sqlite. The AIR runtime really doesn’t provide any way to call native code.

Creating the fulltext search DB was quite easy. The indexer script was already generating “tables” of information in JSON form. I just converted those tables to a relational DB form.

The code for that is here:

https://projects.sitepen.com/svn/toolbox/trunk/apigen/index.py

The database just needs to match up words with the files the words appear in.

I’d be happy to go into more detail on any part of that, if it would be useful.

Kevin

By: John Wood

John Wood — Mon, 15 Sep 2008 10:24:48 +0000

Hi Kevin,

Thanks for posting this article!
I have couple questions:

1) Did you modify builtin Sqlite?

2) If it not a problem could you please elaborate how you used python was used to modify indexer script to generate FT Sqlite DB?

Thanks,
John

By: Kevin Dangoor

Kevin Dangoor — Wed, 10 Sep 2008 14:18:52 +0000

@micha: Flash Magazine mentioned this:

“AIR 1.5 does not have a date and it’s also a kind of maintenence release, but it will bring two important updates – Flash Player 10 and a more recent release of WebKit (with video-tag support). There will be some bugfixes as well with this release.”

http://www.flashmagazine.com/news/detail/air_roadmap/

WebKit has been evolving very quickly. For example, Squirrelfish landed after AIR 1.0 was released. Just keeping up with WebKit changes will be an evolution of the AIR platform.

@Psy: Cool looking widget!