From the Model 100 BASIC poetry generator. (Zip file, 4.0 KB).

At Tandy Assembly 2018, I bought a DWP-230 daisywheel printer, and I bought the December 1982 issue of 80 micro. That issue included the fifth installment of Jay Chidsey’s “Bit Smitten” series.

This month Jay Chidsey shows you how to gain access to string data for use in programs which require random selection of words.

It was written for the Model I or III and only 16K of RAM. There was nothing in it, however, that used the specific capabilities of the Model I/III. Because of the ever-present variations on different versions of BASIC, even though both the I and 100 use Microsoft BASIC, some conversions were necessary.

The biggest change was generating random numbers. The Model I/III used RND(N) to generate random integers from 1 to N. The Model 100 only generates random numbers between 0 and 1. It’s up to the programmer to convert that to the desired integer range. The standard way of doing this is to multiply the random fraction by N, take the integer from that, and add 1.

Further, while Chidsey didn’t use it, the Model I/III had the ability to use a semi-random seed¹ so that the same results don’t come up every time you run the program. The Model 100 lacks that ability, which means the programmer has to seed the random number generator randomly. The standard means of doing this was to poll the clock. If you poll the seconds, you have a semi-reasonable random number from 0 to 59, giving you sixty seeds. I chose, instead to poll the ones digit of the seconds twice, in order to get a potentially more random seed from 0 to 99. There are two places in the program where the time could conceivably provide random seconds: the time the program starts up, and the time that the user says, start generating poetry.

Even this would only mean a hundred different permutations of poetry, all down the line. Since there’s a keypress wait at the end of each poem, I also have it randomize a bit more based on the ones digit at that point.

At this point, there are a hundred poems to start with, then a thousand poems for the second poem, and ten thousand potentials for the third—depending on how random the randomizer is in the Model 100.

The other major change involved adjusting the PRINT@ locations. The Model I/III had 48 lines where the Model 100 has only 8.

I had a lot of help from David A. Lien’s learner’s manual in doing these conversions, as well as adding new features.

The original version of the program generated only one form of poem:

• • How madly despairs the plum!
  • Plums seldom despair.
• • More beautiful are the dogs,
  • as they play.

The first two lines used one set of adjective, verb, and noun, and the second two lines used a different set.

I added several more forms, such as

• • Manors brood.
  • Moons dance.
  • Soldiers shout with soldiers.
  • Run sweetly!

Each line uses its own set of nouns, verbs, and/or adjectives; no line uses a noun, verb, and an adjective, and one line uses two nouns. So I separated the subroutine that generates a random noun, verb, and adjective into three subroutines that generate one each. I kept the subroutine that generates the three-at-a-time noun, verb, and adjective but had it call the three new subroutines to get each of those parts of speech.

Because there were more formats, I also moved the code for turning an adjective into an adverb, pluralizing a noun, and matching the verb to the noun, into the subroutine for that part of speech. Each subroutine generates two variables. For nouns, the singular and plural form (NS$, NP$). For adjectives, the adjective and adverb (AJ$, AV$). And for verbs, the they form and the it form (VT$, VI$).

There was also a grammatical bug in the original. Chidsey appears to have chosen his words carefully to avoid nouns and verbs that can’t have ’S’ added to them, and to avoid adjectives that cannot be turned into an adverb by adding ‘LY’ to them. But he did include the noun ‘FLY’ and he pluralized it into ‘FLYS’. I added a quick check to see if the final letter was ‘Y’ and turned ‘FLY’ into ‘FLIES’.

And also ended up with ‘PLAIES’ as the it form of ‘PLAY’. Thus the long line 6130 to make sure that verbs ending in ‘Y’ get ‘IES’ as their it form unless the second-to-last letter is a vowel.

The adjectives contain three issues that I chose not to change. “Purplely” is perfectly valid as far as I can tell, it’s just not a word that gets used. “SCENICLY” is an archaic spelling of “SCENICALLY” which itself is not exactly the most common adverb for that purpose. And ‘GOODLY’ is a common adverb that isn’t usually the adverb form of ‘GOOD’. I chose to punt on fixing those issues until and unless enough examples come up to indicate a good solution for a goodly number of cases.

Here, then, is the full code. You can add nouns, verbs, and adjectives by adding more DATA statements to the appropriate section of code in section 2000 (nouns), 3000 (verbs) or 4000 (adjectives). Nouns need to be able to be pluralized by adding an ’s’ to them. Verbs need to be able to be transformed to the it form by adding an ’s’ to them or, if they end in ‘y’ but now vowel-‘y’, replace the ‘y’ with ‘ies’. Adjectives must be able to be turned into adverbs by adding ‘ly’ to them.

Unless, of course, you add the appropriate code in section 6000 to catch special cases.

I’ve also set it so that when a poem is displayed, you can hit the letter ‘s’ to append it to the file poems.do. And you can hit the letter ‘p’ to print it, the same as using the PRINT function key. The escape key exits the program.

Here’s the full code:

If you add more words you may have to increase the memory cleared for strings. Line seventy sets up the end-of-line character, which, for these computers, needs to be a carriage return (13) and a line feed (10).

In order to DIM the noun, verb, and adjective arrays exactly, the program reads all the words in just to count them up, and then resets the DATA pointer back to the beginning.

According to Tony B. Anderson in Programming Tips, Peeks, and Pokes for the Tandy Portable Computers, memory address 1 contains a code that indicates whether this is a Model 100, 102, or 200. I can’t verify the 102, but my 100 and 200 match his numbers.

This is the major randomizer. It takes the ones digit of the seconds from when the user ran the program, multiplies it by ten and adds the current ones digit of the seconds, after they’ve said they’re ready to make some poetry.

It doesn’t use a FOR…NEXT loop because zero is an option, and this particular BASIC does not allow going through a FOR…NEXT loop zero times.

F is a random number from 1 to 5; there are five subroutines for generating poetry, and the program branches to one of those subroutines based on F.

And, these are the formats for the poems. Each subroutines constructs PO$, including a centered title (using the subroutine in 5100) and often capitalizing the first word of the sentence (using the subroutine in 5300). Chidsey’s original program just printed each poem to the screen. I wanted to be able to save an interesting poem to a file, so switched to creating a string that is more easily saved. Here, Anderson’s tips book failed me. He lists 65024 as the start of the LCD buffer for the Model 100, which worked, but 64048 for the start of the Model 200’s buffer, which rarely worked, and usually just provided a bunch of spaces. Constructing out of strings and saving that creates a much cleaner file, anyway.

If you add nouns, verbs, or adjectives to the appropriate section, you may have to increase the amount of memory CLEARed for string space. Also, unless you want to program a smarter adverbizer or pluralizer, use nouns that can have “s” added to them; verbs that can have “s” added to them unless they end in a consonant plus a “y”, in which case the “y” is replaced by “ies”; and adjectives that can be turned into adverbs by adding “ly” to them. It will help if the adverb is the adverb form of the adjective it is constructed from. As opposed to, for example, “good” and “goodly” or “lone” and “lonely”.

These are the various subroutines used to do something other than get an individual word. Because many lines require one each of noun, verb, and adjective, the first subroutine provides them. The second subroutine takes whatever is in TI$ and (a) makes it all-caps while (b) centering it for a 40-character-width display.

The keypress wait subroutine checks for the escape key (and quits), semi-invisible keys such as the PRINT key, which return a numeric value of zero and shouldn’t move to the next poem anyway, and then the letters “s” and “p” for saving to “poems.do” and printing to an attached printer, respectively. Otherwise, it returns to where it came from.

The capitalization subroutine capitalizes the first letter of ST$, which is likely to be a word beginning a line of poetry.

Subroutine 5400 saves the current poem (PO$) to “poems.do” for future reference.

Subroutine 5500 prints a centered title. This is so that the variable length “The MODEL Poet” title at the beginning can be centered.

Subroutine 5600 calls the keypress wait subroutine, and then does a short randomization based on the seconds. Since it’s a number from 0 to 9, it doesn’t take as long as the major randomization that goes from 0 to 99.

And, finally, the subroutines that provide a random word—either a noun, a verb, or an adjective. The subroutines each provide two forms of their word. In the case of nouns, the singular and plural; in the case of verbs, the they form and the it form; and for adjectives the adjective and the adverb form of the adjective.