New Scott Adams format interpreter "PerlScott"

There are several errors in that basic listing!

do you have a scan of the original?

some obvious bad lines are; 140,470,1260,1300. Probably a lot more too!

I agree with that as I have compared it to the original listing in Byte magazine, which I still have after all these years.
It looks like some of the : have been replace with = and , replaced with . to mention 2 of the errors I can see among others.
Possibly the OCR didn’t interpret the scan as well as it should.

Thanks for the valuable comments, and for taking a look! Honestly, i never thought anybody had any sort of interest in something like this. The Scott Adams system and all the various things around it developed into a bit of a personal obsession at some point.

I’ve added the scanned image version of the code to the repository, for comparison. Corrections have been made to the basic listing on lines 80, 120, 140, 250, 470, 590, 720, 1170, 1260, 1300, so far. Probably some more errors remain, and i’ll need to go through it in more detail in the future. OCR has been anything but perfect so the basic listing is a combination of OCR with corrections for more obvious things and manually typed in bits. It proved to be enough at the time, for automatic translation into skeleton Perl code to work.

Pull requests are always welcome, as well, if somebody doesn’t want to wait for me to make the required changes :slight_smile:

I also converted and reformatted two old sources of information to Markdown, for easy reference:
The ADVENTURE Data Base Format (1980) by Alan Moluf
Adventure Editor Manual (1981) by Bruce Hansen

Thanks for the updates!

I’m looking at line 1290 in the listing,

1290 FORX=0TONLSTEP10:FORY=0TO1:INPUT#D,NV$(X,Y),NV$(X+1,Y),NV$(X+2,Y),NV$(X+3,Y),NV$(X+4,Y),NV$(X+5,Y),NV$(X+6,Y),NV$(X+7,Y),NV$(X+8,Y),NV$(X+9,Y):NEXTY,X

But it seems that the noun/verb list is interleaved. From your perl version, you have;

# Words { my $word = 0; while ( $word < ( ( $number_of_words + 1 ) * 2 ) ) { my $input = read_string($handle); $list_of_verbs_and_nouns[ int( $word / 2 ) ][ $word % 2 ] = $input; $word++; } }

How does line 1290 do that? I was expecting something like:


Was the basic listing for an earlier version of the data format?

Your little translation arrived at the perfect time for me, and testing it certainly opened doors for my own personal obsession with the system, though as an archivist instead of as a programmer.


This is an interesting line that i had forgotten about. The best way i can describe it, is that it’s a “loop unrolling” optimization. It has the slightly unfortunate side-effect of only allowing you to define words in the data files in groups of 10. You may notice in some game data files, that there are empty words in the end of the word list. It’s because of this. If you were to write an editor this would be a very good thing to keep in mind, for the sake of backward compatibility.

For the Perl version, just in case, i didn’t want to adhere to this limitation for the sake of robustness. My thinking was that some newer game data files may not adhere to this arbitrary convention and cause breakage.

The Basic listing was for an earlier version of the data format, which was basically Basic DATA statements. The underlying data structure wasn’t too different though. I haven’t bothered to transcribe the two games available in the original format, since it’s a bit too much of a time-consuming and error-prone activity at the moment. (It would be a fun project, at some point.)

Not sure if this answers the questions properly :slight_smile:

That’s fun to hear. I think there are probably a lot of undiscovered and obscure games that use the Scott Adams format still out there. It’s a simple system to get your head around, and given the availability of authoring systems and documentation, I find it hard to believe that we’ve seen all there is already.

I’ve been toying with a user friendly authoring system for Scott Adams format games for what seems like forever. Perhaps, it’ll be done one day. The basic structure of a Scott Adams interpreter could also lend itself quite well to non-typing user interfaces on smartphones or a voice recognition assistant. Perhaps there could be an interesting future for the format as a whole. I’d like to think so :slight_smile:

CASA - The Classic Adventures Solution Archive, is the right place for that.

Some game databases would need to be cleaned and completed: some header and trailer are incomplete. There is a data structure problem in recent games, they cannot be played with the Adams interpreter 8.5 for TRS-80.

I agree with you. But what the Adams system may lack is the ability to intercept the player’s movements.

I had decompiled adv05 “The Count”, and other games, with The Adventure System instruction set. It’s simple and readable and very useful to solve and fixe game.
In my post in SPOILER:SHOW

Ok. Here is a trickier problem. If I have an action with more than 4 instructions it’s split with 3 instructions + “CONT” in the action and the rest is pushed to an “occur 0%” directly after with the rest of the instructions.


ident 1
version 416
wordlen 3
maxload 6
lighttime 125
unknown1 5953
unknown2 819
start forest
treasury forest

room forest "forest"

action SAY HEL
	set_flag 2
	set_flag 2
	set_flag 2
	set_flag 2
	print "Hello Sailor!"

Plays like this (with debug):

I'm in a forest

DEBUG: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
DEBUG: 0 0 0 0 0 0 0 0 125
Tell me what to do
say hello

DEBUG: Action 0. verb 2 (SAY), noun 7 (HEL) (CONT 0), ""
DEBUG: Condition 0 Par with parameter 2
DEBUG: Condition 0 Par with parameter 2
DEBUG: Condition 0 Par with parameter 2
DEBUG: Condition 0 Par with parameter 0
DEBUG: Condition 0 Par with parameter 0
DEBUG: Command code 6 SETz
DEBUG: Command code 6 SETz
DEBUG: Command code 6 SETz
DEBUG: Command code 21 CONT
DEBUG: 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
DEBUG: 0 0 0 0 0 0 0 0 125
Tell me what to do

If you try adventure #3 “Mission Impossible” and frisk the dead saboteur he never drops the map, the case or the leaflet and you never get the message!

Source (decompiled with scottkit):

action FRI SAB when present saboteur and !exists envelope
	drop envelope
	drop yarn
	drop maintance

occur 0%
	drop map
	drop case
	print "Something fell to the floor."
	drop leaflet

Show like this in perlscott:

Tell me what to do
fri sab

Tell me what to do

I see nothing special.
I'm in a yellow room. Visible items here:
Yellow door with tv camera over it. Empty manila envelope. Piece of yarn. Picture of saboteur stamped -window maintance-. Dead saboteur.
Obvious exits: SOUTH

Tell me what to do

should this not work in 4.6? But my copy of Adventureland have a “CONT”?

Anyhow, the “CONT” instruction doesn’t seem to work.

I have a possible solution to the “CONT”-problem. I have logged a ticket at the project - PerlScott

1 Like

That issue is now fixed. Thanks for the feedback! The fix isn’t very elegant but it works, and it’s hard to tell how it was solved originally.

It seems likely that the “continue” command may have been introduced with adventure #3. (Even though commonly available files for adventure #1 and #2 use continue, they didn’t do so originally.) If the first release of adventure #3 was bundled with a Basic interpreter on TRS-80 that supports “continue”, the exact mechanism could be determined, but I haven’t tried to verify that.

Update: I’m clearly full of it, when it comes to this. Continue seems to be part of the original implementation, and i just failed to translate the functionality correctly, somehow, and my memory was a bit hazy here.

1 Like

I’ll try it now. I’m testing “Ghost King” in it now. Let’s see if I’ll can finish it in PerlScott!

Ok. I played a little bit and I’m sorry to report that the fix broke something.

Not every item is “gettable”. If you test Adv #3 and move west, west from your starting point and try to pick up the pail it’ll fail.

I think it’s the line 1254 that is the problem because if I remove that line everything seems to work.

if ( $word_action_done == $TRUE ) { return 1; }

I’m of course not certin what this line is there for so removing it might cause other problems…

Another problem is command 23 BYx<-x (put_with). It takes two items as parameters and both ScottFree and SkottKit put item1 at item2:s location. PerlScott does it the other way around. In Adv #2 there is:

occur 30% when present Parrot and !present crackers and !present pirate
	print "Bird flys off looking very unhappy"
	put_with Parrot crackers

I’m also logging this in GitHub.

1 Like

Thanks a lot for this! Sometimes maintaining old code is like playing whack-a-mole :slight_smile:
Corrected the two problems you mention above, and fixed another problem that was caused by the “continue” fix.

1 Like

Thanks! I’ll try it tomorrow.

Ok. You got another mole to whack.

It’s still the “continue”-command. The continue doesn’t work if your action only have a verb and not a noun.

I think the return-statement on line 1272 should have a condition just like the return-statement on line 1280.


Haha, thanks a lot - again! That should be a quick one line fix at least. :sweat:

I agree, and it’s been corrected just like you said. I do accept pull requests, if you feel that there’s anything else that needs correcting and you don’t feel like waiting for me to wrap my head around it. :slight_smile:
Thanks again!

I’m happy to report that I played “Ghost King” all the way through in PerlScott yesterday. I found one minor bug and have a couple of suggestions.

First the minor bug. The game uses “GET” in an unsual way (not picking anything up but doing other things instead). As it is now I do the right tings but get an “You can’t see that here!”. I think the check on line 1296 to before the call to the subroutine “handle_carry_and_drop_verb”. If I try it it behaves as excepted.

And some suggestions:

  1. You might consider implementing the on letter abbreviations for directions, inventory and look. It probably wasn’t in the 4.6 release but it’s quite tedious to type “GO” before efery direction. I tried the following code on line 624-
	if ( length( $keyboard_input_2 ) == 1 ) {
		if ( lc( $keyboard_input_2 ) eq "n" ) { $keyboard_input_2 = "go north" };
		if ( lc( $keyboard_input_2 ) eq "s" ) { $keyboard_input_2 = "go south" };
		if ( lc( $keyboard_input_2 ) eq "e" ) { $keyboard_input_2 = "go east" };
		if ( lc( $keyboard_input_2 ) eq "w" ) { $keyboard_input_2 = "go west" };
		if ( lc( $keyboard_input_2 ) eq "u" ) { $keyboard_input_2 = "go up" };
		if ( lc( $keyboard_input_2 ) eq "d" ) { $keyboard_input_2 = "go down" };
		if ( lc( $keyboard_input_2 ) eq "o" ) { $keyboard_input_2 = "go out" };
		if ( lc( $keyboard_input_2 ) eq "i" ) { $keyboard_input_2 = "inventory" };
		if ( lc( $keyboard_input_2 ) eq "l" ) { $keyboard_input_2 = "look" };

There probably is a more elegant way to do this in perl (regexp?) but this worked.

  1. Looking at the original TRS-80 screenshots ( with it’s 64 characters screen it think the startmessage should look something like this:
                 *** Welcome ***

 Unless told differently you must find *TREASURES* 

I'm your puppet. Give me english commands that
consist of a noun and verb. Some examples...

To find out what you're carrying you might say: TAKE INVENTORY 
to go into a hole you might say: GO HOLE 
to save current game: SAVE GAME

You will at times need special items to do things: But I'm 
sure you'll be a good adventurer and figure these things out.

     Happy adventuring... Hit enter to start
  1. Change the prompt to "Tell me what to do? " with a question mark and no CR (TRS-80 look & feel).

That’s it (for now)!

(I’m not comfortable enough with Perl yet to mess with the GitHub-project so I let you do the changes that you feel is relevant. But I will create issues for this.)

That’s great!

Thanks for the GitHub issues. I’ll have to meditate on the GET behavior a bit.

My comments on the suggestions:

  1. Compass direction abbreviations are a good addition. In my next interpreter I’m going to structure this in a more flexible way that permits localizations in the games (so that you can play translated Scott Adams games in Swedish or whatever). Your solution works well for what I want PerlScott to accomplish though.
  2. Agreed.
  3. This is a duplicate of Issue #6, which has been sitting unhandled for too long :roll_eyes: