Very Computer

Quote:> Hi All,

> Any idea how to create a so called "inverted index".

> dBASE/FoxPro code or pseudo code would be great. Also a
> shareware/PD/freeware program will do.

Larry

-----------------------------------------------------------------------------

Colebrook, NH  03576-0158             CIS 72427,2567    fax   +1 603 237 8430

Quote:>I believe that you can create an inverted index by replacing each character
>of the index expression with CHR(255-ASC(m.the_character)).  This will
>'invert' the expression.  Numbers, Dates, Logicals, etc will have to be
>handled differently...  The following function should do the trick.  

Either I have no idea what ya'll mean by an 'inverted' index OR there
is a much easier way to do this.  The INDEX command in FoxPro 2.5
contains the ASCENDING | DESCENDING clause for changing the order
of the index.  If that's not what you meant, please explain 'inverted'
to me.

Bill

: > Hi All,
: >
: >
: > Any idea how to create a so called "inverted index".
: >
: > dBASE/FoxPro code or pseudo code would be great. Also a
: > shareware/PD/freeware program will do.

: Don't know if it will help, but in Clipper you do it by using the descend()
: function.  I don't believe the function exists in dBase III+; not sure about
: Fox.

: Larry

: -----------------------------------------------------------------------------

: Colebrook, NH  03576-0158             CIS 72427,2567    fax   +1 603 237 8430

Yes, DESCEND will work in foxpro aswell. Just add 'DESCENDING' to the end of
your index statement. ie:
                     SET ORDER TO TAG index_name DESCENDING

*******************************************************************
*  Philip Miles              *                                    *
*  Dalhousie University      *  Write a piece of code that does   *
*  Computing Science         *  what ?!?                          *

*******************************************************************

In dBase III+ you can always try something like

USE DFILE
INDEX ON -RECNO() TO REVERSED
USE DFILE INDEX REVERSED
LIST ALL

This should display the DBF file in reverse order.

Armando R. Gonzalez
The University of Iowa

******************************************************************

FUNCTION DESCEND
PARAMETERS cOldKey
PRIVATE cNewKey
cNewKey = ""
FOR X = 1 TO LEN(cOldKey)
        cNewKey = cNewKey + CHR(255-ASC(SUBSTR(cOldKey,X,1)))
ENDFOR
RETURN cNewKey

You could also use a CHRTRAN() function in FoxPro, something like...

FUNCTION Descend
PARAMETERS ctheExpr
RETURN CHRTRAN(ctheExpr,'ABCDEFGHI','IHGFEDCBA')

The second and third parameters of CHRTRAN may need to be 256 characters
long with the entire set inverted in one of the two expressions.  (I used a
simple set for purposes of illustration).

One benefit of using the second example over the first is that it can be
incorporated as part of an index without having to call an external
function.  It will operate significantly faster as a result, however, I'm
not sure how the expression will work if you include the entire ASCII set
within the string.  You will need to consider that the string delimiters
are also part of the ASCII set and you will probably have to split the
expression and add to get a working function.

Any suggestions or comments are welcome...

--
Mark T. Boyer              


Oak Ridge, Tennessee

 You can order back issues of FoxTalk from:

 Pinnacle Publishing INC,
 P.O. Box 888,
 Kent, WA 98035-0888.
 Phone: (206) 251-1900
 Fax: (206) 251-5057

 Cost is $US15.00 outside the US, and source code disks are an extra
$US10.

 Also Xitech do a text indexer for memo files that seems to be very fast.

5515 Southwyck Blvd
Toledo OH 43614

Tel: 419 867 3622
Fax: 419 865 9365

 or Xitech Europe:

 Phone +44 707 276 637
 Fax   +44 707 267 777

 Regards,
 Dave.

 P.S. for those posting methods of making descending indexes, an
"inverted" index actually indexes every word in a text database. In a
FoxPro context you would usually use it to index every word in every
memo field in a file.

--

INDEX ON tag NAME tag DESCENDING

Actually, I think what they're looking for is a routine where you can
put in

"Keith Cochran"

But when you sort it, it sorts like you had put in:

"Cochran Keith"

The simplest way I can think to do that is to have two input fields, and
then:

INDEX ON lname+fname TAG invert DESCENDING
--

=My thoughts, my posts, my ideas, my responsibility, my beer, my pizza.  OK???=
=        "I had to wait 300 years for a * to light the candle..."        =

Quote:>>   Any idea how to create a so called "inverted index".

> Don't know if it will help, but in Clipper you do it by using the
> descend() function.  I don't believe the function exists in dBase III+;
> not sure about Fox.

/Paul
--
+--------------------------+----------------------------------------------+
| Paul Russell             | Phone:     (902) 481-3680                    |
| Software Kinetics        | Fax:       (902) 468-3679                    |

| Dartmouth                | "Use technology because it's appropriate...  |
| Nova Scotia, Canada      |     not just because it's cool!"             |
| B3B 1W2                  |             - E. Davidson, Software Kinetics |
+--------------------------+----------------------------------------------+

--
Daiv Stoner                   ||  Busy life / Not enough time to look in

index on -(asc (substr (x_field, 1, 1))) to x_ndx

the trick is to use the unary negative operator when the ascii value has
been returned.  of course if the field you wish to index on is numeric,
then the asc() and substr() functions aren't even used.  

of course the limitation on this is that only one character of the field
is used.  but then again, you can concatenate the expression:

index on -(expr1) + -(expr2) + ... + -(expr n) to x_ndx
   where  expr n is the thing above.

later days...

leo d.

--
------------------------------------------------------------------------------
 \ The above does not represent OIT, UNC-CH, laUNChpad, or its other users. /
   ------------------------------------------------------------------------

USE DFILE
INDEX ON -RECNO() TO REVERSED
USE DFILE INDEX REVERSED

This also works fine If the field is numeric. If PRICE is a numeric
field you can always do something like:

INDEX ON -PRICE TO REV_PRICE

I'm sure something can be done for dates with a function
like CTOD(). But the best way to solve this problem is to
use dbest xbase language   C l i p p e r.

Armando

: I'm sure something can be done for dates with a function
: like CTOD(). But the best way to solve this problem is to
: use dbest xbase language   C l i p p e r.

In dBASE III PLUS, you can create a descending order index based on a date
field by, in the index expression, subtracting the date field from a very
large date -- preferrably one that would never appear in the table. For
example:

INDEX ON CTOD('12/31/3099') - DateFld TO DateDesc

--
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
_/ Steve Koterski               _/   The opinions expressed here are    _/

_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/

A quite good technique for reversing record order.

AG>I'm sure something can be done for dates with a function
  >like CTOD(). But the best way to solve this problem is to
  >use dbest xbase language   C l i p p e r.

Use the Big M method:  subtract the date from a very large
date, e.g.,

INDEX ON {12/31/2999} - <date> TO <etc>

Of course, that will create a numeric type index rather than
date type, but it will reverse the dates.  You can do
something similar for indexing character type fields, but it
involves a UDF in the index.  Hope this gives you some ideas.

---
 * SLMR 2.0 #1193 * My other car is a modem!