L' Effet Pommier 3/L'Effet Pommier - Volume 03.iso/Disk & Fichiers/FileIndexer 1.3/FileIndexer 1.3.doc

Text File | 1996-01-26 | 7KB | 163 lines

FileIndexer 1.3
---------------

I really miss OnLocation. (It doesn't run reliably under system 7.5). They have recently
released a server version, but I want a personal indexer.

This is my simplistic attempt to knock up some of the key functionality I'm looking for.

In a multi-user environment, where you have budget for a server, I recommend
AppleSearch for indexing your documents. I wanted to have a way to quickly retrieve
a list of documents on a mounted disk that contain a given word.

FileIndexer trades memory for speed. It builds a sorted list of unique words
(not case sensitive), and then keeps a list of the files that contain each word.

This list of lists is kept in memory. This makes it VERY fast to do the binary search.
You can save the index to a file and load it reasonably quickly.

Run FileIndexer, choose Index folder, and select the top folder in a tree of
text files to be indexed. Claris XTND can be used (if available) to read documents.
You need to turn it on in the preferences dialog to use it. Note that some documents,
in my experience, will hang XTND. (I have no control over this). If the beach ball
cursor times out and stops, you are probably stuck. The offending document name
is left displayed in the status field.

When done, type a single word into the find field and press return. A list of all
the files that contain that word will be displayed. The first occurence of the word
will be shown in context. An info checkbox lets you see file mod date and the full
path to the file as an alternative.

Double-click on the file to open the first 32k of it in a text window. You can choose
to view the document in it's original app or the app of your choice.

From the text window you can use Find or Find again (command-G) to locate the word
you searched for.

I built a PowerPC native version but it was slower! Probably due to my heavy use
of the non-native file manager.

Limitations
-----------

o Only searches for one word at a time.
o Can't currently index read only volumes.
o 68020 and above. (Contact me if you want 68000)
o Requires System 7.

To Do
-----

o Index the "sound" of words (use soundex etc).
o Move the in memory list to a disk based database!
o Handle multiple words with AND and OR.
o Able to exclude certain folders containing a given string.

Tools
-----

o Metrowerks Codewarrior 8
o MacApp 3.3
o ObjectMaster

History    (DD/MM/YY)
-------

22/12/95    Turned off qTheDebugger & qInspector
27/12/95    Ditched Insert() when reading files. Load speed 21s down to 3s!
27/12/95    Sped up indexing thanks to Metrowerks code profiler.
            Release of version 1.0
28/12/95    Started on 1.1
            Indexing done in a cohandler so time is given to other apps.
            It's a bit slower but is nicer to other apps and allows context switches etc.
            Plays a sound when indexing is done.
            Reduced RAM overhead of index slightly.
            Added a "stop words" feature to exclude common non-interesting words.
            Added printing of the document window.
30/12/95    Release of version 1.1
            Started on 1.2b1
06/01/96    Fixed write to nil during startup.
            Added extra preferences: stem, xtnd, view options. (Doc version to 3).
07/01/96    Added support for Claris XTND to index and display documents.
07/01/96    Shipped 1.2b1 to beta testers, started 1.2b2
            Added arrowKeyselection behavior to file list view.
            Return and enter now simulate a double click.
            Did some work on the printing, headers & footers etc.
            Fixed home & end keys by applying patch to UScroller.
            Fixed the open in chosen app feature.
            Made the document dirty after preference change.
            Switched to a Hash index rather than storing words.
            - Old test index: 275k, now is 176k
            - This produces some bogus hits, will look for better hash functions
            Bumped file version as no longer compatable.
08/1/96        - Bumped to 1.2b3
            - Found a problem with errors during XTND reads.
            - Added user option to not use XTND, Changed file version.
            - Switched to MacApp's memory allocator.
            - Added a filter to remove TEXT files that shouldn't be indexed,
              such as acrobat '.pdf' files. (Others??)
            - Reduced index size by removing hit count field. 275k -> 176k -> 154k
            - Fixed a bug when indexing a single file.
            - Fixed a bug relating to the stop words compare function.
            - Removed the show word list functionality (no point with hashing).
            - Added a warning if you try to look for more than one word.
09/01/96    - Cleaned up what happens if you close a doc that's indexing.
            - Fixed up out of memory failure crash. Now gives message.
            - Improved the open in app etc functions that sometimes didn't correctly
              find the Finder process.
10/01/96    - Bumped to 1.2b4
            - Changed doc version number
            - Tried Dave Abraham's better hash function
            - Found a bug in AddWord(). (Wrong compare parameter structure used).
            - Implemented word stemming
            - Shipped a version to Al.
            - Improved the about box.
11/01/96    - Bumped to 1.2 final ready for testing and release.
            - Released to Info-mac archives.
12/10/96    - Bumbed to 1.3b1
            - Added barber pole display to thermometers during unknown length operations
            - Improved memory allocator for the TMiniFile class (thanks Jim Coplien!)
              My test folder has 24,000 of these objects to delete.
13/01/96    - Show found word in context feature added.
            - Found small memory leak when closing document.
            - Improved user feedback during file and folder count phase
            - Distributed to test list.
14/01/96    - Made the text drag and dropable
            - Added a drop small icon for indexing folders by dropping them on it
              Clicking it shows the choose index folder dialog.
            - Removed all shorts to get rid of 32k limits. Bumped file version.
            - Added appleScript support (index folder not working yet).
            - Made the barber pole animate the other way.
            - Took out my operator new as it leaks.
16/01/96    - Bumbed to 1.3b2
            - Switched to Metrowerks CW 8 (code got smaller by about 80k!).
19/01/96    - Enter now searches same as return (thanks geo)
            - Added scroll to found word preference, bumped file version
            - After a successful search, select the first hit in the list
            - You can now drop a folder or file anywhere in the window to index.
            - Optimised indexing slightly by removing an object new/delete each time
21/01/96    - Sent to beta testers before general release.
            - Bumped to 1.3b3
            - Removed appleScript support for now due to overhead
              (this will come back in the database version).
            - Removed some unused resources.
            - Built with symbols off to reduce size.
26/01/96    - Release of 1.3 to InfoMac


(c) 1995,1996 Peter Marks
Suggestions for future enhancements most welcome, but I make no guarantee of support.
This freeware is supplied as is, use at your own risk.
Please contact me for permission to include on commercial collections.

mailto: [e-mail]

Acknowledgements
----------------

Thanks for support to:

Gregory Kearney, Allan Herman, Alastair Rankine, Geert Jan Bex, Keith F. Batchelder,
tony stevenson, Federico Cavaglia', Farhad Anklesaria, Dave Abrahams, MetaTheory,
Jim Coplien, George Bray, you...