[TYPO3-core] indexed_search patches

[Kasper Skårhøj]

Hi Michael,

I'm not going to look at indexed search right now but your changes look
interesting. However I have to make a close review sometime before 4.0.

- kasper


On Wed, 2005-09-28 at 03:22, Michael Stucki wrote:
> Hello,
> 
> during the last months I have made many changes to indexed search. This
> results in quite a big patch. I have tested the critical parts for a long
> time now and think that everything should be fine. However it's in any case
> better if anybody reviews my changes.
> 
> If there's something we need to discuss (e.g. changed default values), I
> suggest we do it right here.
> 
> 
> So here is the list of changes:
> 
> - Fixed the page counter which was not working correct under some
> circumstances. Needs review and probably some more testing...
> See http://bugs.typo3.org/view.php?id67
> Path was initiated by Chi Hoang
> 
> 
> - Template support for indexed_search.
> This works great, I already moved two of our customers to this new version.
> See http://bugs.typo3.org/view.php?id1
> Big thanks to Udo von Eynern
> 
> 
> - Fixed a problem with continuing indexing of external documents. This just
> did not work - if a file is fetched from the cache, it's content was not
> indexed anymore. Fixed this by moving the indexer hook out of the page
> generation function, right after it is stored/fetched from the cache.
> 
> 
> - Completely removed the Ruby/OOoExtract functionality. Use "unzip" instead.
> In my opinion, this is much simpler than using a tool that is not part of
> any well-known Linux distribution <=== Are there any drawbacks with this??
> 
> 
> - Replaced is_file calls with a new function: Test if safe_mode or
> open_basedir are set and return true if a custom test has worked.
> 
> 
> - The behavior of the result browser has been changed. It works more
> Google-like now.
> 
> 
> - If an external file is too large (e.g. TSref), then the lexer needs too
> long for processing the words. To work around this, I have hardcoded that
> indexing of documents with $contentArr['body'] longer than 100000
> characters will be skipped. Of course the better solution is to make the
> lexer class (much!) faster...
> 
> 
> Many smaller changes:
> - The default number of results per page will be changed from 20 to 10
> - Fix the hit counter. See http://bugs.typo3.org/view.php?id67
> - Cut HTTP references after the question mark (too long URLs can break a
>   design otherwise!)
> - Generally: preg_replace is faster than ereg_replace! - thanks to the
>   Austrian UG (can't remember who it was) for reporting this.
> - New TS properties: See section "02_setup_txt.diff" below
> - New TLO "config" properties "index_keywords" and "index_description". Both
>   will be disabled by default (which is not the same like the current
>   behavior, though I think that nobody really wants this).
> - Renamed procesWordsInArrays() to processWordsInArrays()
> 
> 
> =====
> 
> Attached are the patched files, split up file by file which makes it a
> little easier to follow:
> 
> 01, 03 and 04 are very minor things.
> 
> 02 adds some new TS properties. See comments in 05_plugin.diff for details.
>    show.forbiddenRecords = 0     !! different behavior than now !!
>    show.alwaysShowPageLinks = 0  (do we need to have this optional?)
>    show.advancedSearchLink = 1
>    show.mediaList    show.resultNumber = 0         !! different behavior than now !!
>    _DEFAULT_PI_VARS.type = 1     !! different behavior than now !!
>    _DEFAULT_PI_VARS.flat = 1     !! different behavior than now !!
> 
>    Do you agree with these changes?
> 
> 05 is the FE plugin. For details, watch the comments in the diff file.
> 
> 06 are the indexer classes.
> 
> 10 contains changes outside of indexed_search, needed for fixing the
> continuing stuff...
> 
> *tmpl are two new templates, written by Udo. The traditional template looks
> exactly like the current one and should be used by default.
> 
> 
> Regards, michael
-- 
- kasper

TYPO3 Association Chairman
-----

"Don't Mess With Texas!"