Loadstar 128 19/q19.d81/t.dictionary

Text File | 2022-08-28 | 7KB | 149 lines

                    ZERO PAGE: THE LOADSTAR 128 DICTIONARY

                        by Fender Tucker and Jeff Jones


     I remember the first computer word game I came across.  It was called
WORDSMITH and it was on one of the very first LOADSTARs.  The rules were
simple: the computer provided a four-letter word and you were to change one
of the letters in the word to make another four-letter word.  Then it was
the computer's turn to do the same.  As soon as one of you couldn't come up
with a word that hadn't been used yet, the game was over.

     The program came with 26 SEQ files full of words for the computer to
use -- one for each letter of the alphabet.  The computer had to search
through these files every time a word was entered so you can imagine what a
slow game it was.  The only reason it was reasonably fast was that the word
lists weren't very big.  Consequently, we haven't published any more
programs that required the computer to recognize if an entered sequence of
letters was actually a word.

     Until now.  Recently I was given a dictionary disk by the Apple
department here at Softdisk.  I had them convert it to IBM ASCII and from
there I used Big Blue Reader to get it in Commodore PETASCII format.  It was
huge -- the five-letter word list was over 300 blocks long!  It had a
carriage return after each word and contained every five-letter sequence
ever uttered by William Buckley.  I needed to do something to get this into
a format we could use in our programs.

     First I wrote a little program that displayed each word on the screen
one at a time and asked me if I wanted it saved in another file.  I spent a
very boring few hours going through the 10,000+ words deciding which ones
are "normal" enough and which ones are only known to crossword fans.
(Luckily there was a "Get Smart" marathon on Nickelodeon at the time.)  I
tried to be fairly liberal with my decisions (for instance, I allowed
"abaca" because I figured it meant two abacuses).  I undoubtedly made
mistakes and bypassed a few words that should be in the list, and allowed a
few that shouldn't.  You try looking at 10,000 words, one at a time, in one
sitting.  A few plurals got past me, but in general, there are no plurals in
the list.

     So now I had a 63-block long list of five-letter words attached to each
other with no carriage returns.  Just one word after another.  There are
3403 words in the list.  Then I asked Jeff Jones to come up with an ML
routine that, when given a five-letter word, would zip through the list and
let me know if it's in it.  It had to be fast.  He surprised me by making
his ML even more powerful by allowing ? wildcards.  This means that you can
pass the ML something like "ab?ca" and it will find all words that fit that
pattern, one at a time.

     It's fast.  Play BLOTTO and you'll see that the computer takes less
than a second to determine if a word is in the dictionary.  BLOTTO didn't
require the wildcard feature, but WORDSMITH would.

     I imagine there are hundreds of nifty word games waiting to be written
now that there's a way to quickly (and painlessly) determine if a
five-letter sequence is really a word.  I'm hoping that the excellent 128
programmers out there will be inspired to write some and send them to
LOADSTAR 128 for publication and big bucks.

     Here is another bonus: I also went through the four-letter list I
received from the Apple guys and Jeff added it to the dictionary.  So
four-letter word games are possible -- or games that can use both four- and
five-letter words.  The same ML will handle either word length.

     Everything you need is found in the 96-block file called "dictionary
9400".  It resides in BANK 1 at page 148 so you need to use these lines to
load it into place:

 10 poke58,148:clr:dv=peek(186):ifdv<8thendv=8
 20 bload"dictionary 9400",u(dv),b1,p37888

where dv is the current drive.  Line 10 protects the dictionary from being
overwritten by BASIC variables and sets the drive number (dv) to the current
drive.  Line 20 BLOADs the dictionary into place in BANK 1.

     Then you write your program.  When you come to the place where you want
to check if a sequence is really a word, use this set of commands:

 2000 w$="board":hb=int(pointer(w$)/256):lb=int(pointer(w$)-hb*256)
 2010 bank1:sys37888,lb,hb:bank15

If the word "board" is really a word, w$ will equal "board" when the program
comes back to BASIC.  If it's not in the dictionary, w$ will equal a null
string, ("").  You may use any variable name in place of w$.  Just use the
same string variable throughout the commands.

     If w$ is a four-letter word, the program will even more quickly check
just the four-letter word dictionary.

     To use a wildcard, just use a question mark (?) in w$.  If w$="cr??e"
then after line 2010 above, w$ will equal "crane".  To find the next word
that fits the pattern use this command:

 2020 bank1:sys37891:bank15

and w$ will equal "crate".  To find all of the words that fit the pattern,
use a loop like this:

 2020 i = 1 : bank1
 2030 sys37891
 2040 if w$ = "" then 2080
 2050 wd$(i) = w$
 2060 i = i + 1
 2070 goto 2030
 2080 bank15:program continues with all matching words in the array wd$(x)

Make sure to DIMension wd$(x) large enough to hold all of the matches you
anticipate.  Since perverse users will try to enter "?????" if you allow
them to use wildcards, you probably will need to TRAP the program to take
care of the "bad subscript" error you'll get if you don't DIMension to at
least 3403.  Obviously, "?????" would find all 3403 five-letter words.

     That's about all there is to know about using "dictionary 9400".  We
will probably come up with an improved dictionary that has two- and
three-letter words, as well as some new ML tools for using the lists.  For
instance, if you want to have the computer choose a word from its dictionary
at random, you would just do a SYS.  In the meantime, here's a short BASIC
routine that will pick a word at random from the five-letter list:

 3000 v$ = "" : bank1
 3010 r = int(rnd(1) * 3403)
 3020 for i = 0 to 4
 3030 v$ = v$ + chr$(peek(45122 + r * 5 + i))
 3040 next : bank15

The word will be found in v$.  For four-letter words, substitute:

 3010 r = int(rnd(1) * 1679)
 3020 for i = 0 to 3
 3030 v$ = v$ + chr$(peek(38400 + r * 4 + i))

From the above numbers you can tell that there are 1679 words in the
four-letter list and that they begin in memory at 38400 in BANK 1.

     I found that returning the program back to the default bank, BANK 15,
is a good idea, since I use some ROM routines in BANK 15.  I'm sure that
programmers who are more conversant with BANKing on the 128 than I am will
know what to do.

     Use the dictionary as you wish in your programs, but please mention
that you got it from LOADSTAR 128 if you upload or sell your program.  Of
course the most profitable thing to do with your programs is to send them to
LOADSTAR 128.

FT

                           **** RETURN - Menu ****