ADVENT OF COBOL: Day 2
I cheated. Two days in and the desire was already overwhelming.
I spent most of the day on day 1’s writeup, and so I didn’t have much time for day 2. I guess that means that the future writeups will have to be a lot shorter, which will give you more time for doomscrolling; You can thank me later. However, that did mean that I didn’t have the time to read input in its original form, and had to do a bit of preprocessing on it before I could load it into a table:
cat advent2.txt |sed -e "s/-/,/" -e 's/ /,/' -e 's/: /,/' >advent2.csv
I copied that to PASE, just like yesterday, and then went about defining my record format.
Now, we have four fields in our input:
The minimum number of matchingcharacters, a number between 1 and 20
The maximum number of matching characters, also a number between 1 and 20
The character to match, which is a single character
The password that supposedly fits those criteria, which is a string of up to 20 characters
This should be relatively straightforward to fit into DDS, and in fact it is:
A R ADVDTA021 A MINCOUNT 2S A MAXCOUNT 2S A LIMITCHR 1A A PASSWD 20A
We have a new value for the type field:
A, which is an alphabetic
character; aside from that, this is no different from last week. Even
loading our CSV into the data table is exactly the same as last week;
`CPYFRMIMPF’s defaults are to load a CSV file.
Alas, we have no choice but to move on to the program.
I’ll skip over the identification and environment divisions, because they’re much the same as before. I’m just reading from a different file.
Our data division, though, is totally different. There aren’t a lot of records here, but I didn’t want to retype the whole thing. Fortunately, IBM i has the concept of an externally described file, which allows you to just import declarations from your DDS and the compiler will generate your record definition for you.
DATA DIVISION. FILE SECTION. FD SCAN1. 01 DATA-RECORD. COPY DD-ADVDTA021-I OF ADVENT2020-ADVDTA021.
Let’s break that last line down a little, because, for a four word line of code, there’s a lot going on.
COPY is COBOL’s equivalent of
#include; it just textually pastes
code from another source. Its general form is
COPY member [OF file].
So, if we had another source member called
INCLUDES, we could
include it using
COPY INCLUDES OF ADVENTSRC.
file part of that is optional, but we include it because the
default is not to copy from the current file, but rather
I guess I’ve been going against the grain by putting all my source in
a single file, and thus losing out on convenience features like this.
Oh well, I’ll try to do better tomorrow.
As an IBM extension, though, if the member name starts with
translates from DDS source; you then give the name of the DDS member
and what type of record you want to import. We’re only reading from
the database, so we can use
-I as the record type.
At first, it wasn’t completely clear what I’d get out of that, but I just threw the compiler at it and read the listing. There were naturally a ton of compiler errors, but I did find what I needed.
As always, full details are available in IBM’s docs
We can move on to our local variables:
LOCAL-STORAGE SECTION. 01 CHAR-COUNT PIC 9(2). 01 BADPWD PIC 9(4) VALUE IS 0. 01 GOODPWD PIC 9(4) VALUE IS 0. 01 NREC PIC 9(5) VALUE IS 0.
This is still pretty straightforward. I did learn that I can use
9(4) as a shorthand for
9999 in the
PICTURE clause, which is a
Finally, we have our code. This is still pretty short, but we’ll still break it up into sections to examine it more closely.
PROCEDURE DIVISION. MAIN-PROCESSING SECTION. SETUP. OPEN INPUT SCAN1. READ-DATA. READ SCAN1 NEXT RECORD AT END GO TO PRINT-REPORT END-READ ADD 1 TO NREC
This is still pretty much the same as our first program, so no need to explain anything here. The other major language on IBM i, RPG, has a nice feature where it implicitly includes this outer loop for you. Maybe it’ll make a guest appearance soon.
MOVE 0 TO CHAR-COUNT INSPECT PASSWD TALLYING CHAR-COUNT FOR ALL LIMITCHR
I expected COBOL to be absolutely terrible at character handling,
but this is it. This counts the number of
puts the result into
CHAR-COUNT. Quite frankly, I’m impressed.
INSPECT command can do more than this; it can count leading
characters, replace one character with another, etc. It’s actually
pretty powerful, but I will admit that I have a hard time seeing how
to use it in any situation that’s not this contrived. I guess
IF CHAR-COUNT IS LESS THAN MINCOUNT OR CHAR-COUNT IS GREATER THAN MAXCOUNT THEN ADD 1 TO BADPWD ELSE ADD 1 TO GOODPWD END-IF GO TO READ-DATA.
That condition is a helluva lot more verbose than it would be in, say, C, but it’s completely clear what it’s doing. There’s really nothing surprising here, except that I expected to need to type a lot more.
PRINT-REPORT. DISPLAY "PROCESSED " NREC " RECORDS" DISPLAY "FOUND " BADPWD " FAULTY RECORDS" DISPLAY "FOUND " GOODPWD " ACCEPTABLE RECORDS" CLOSE SCAN1 STOP RUN.
And here’s the final output bit. Again, nothing of any interest here, and it gives the right result, so we’ll move straight on to part 2.
(Well, it took me a few tries, because I misread the spec as wanting a count of the bad passwords. Oops)
OK, now we need to look at two specific character positions, not just count the total mistakes.
This actually took me a long time, because I couldn’t find any documentation on how to take a substring of a string in COBOL. To be honest, even now that I know how to do it, it’s just not documented in ILE COBOL. It was in COBOL for z/OS, and while the two are very different products, the z/OS code ended up working on i.
Anyway, we’ll need another variable in our local storage section:
01 TESTCHR PIC X.
X is a picture of any character, you see.
We also need to change our lovely inspect statement to something much worse:
MOVE PASSWD(MINCOUNT:1) TO TESTCHR IF TESTCHR IS EQUAL TO LIMITCHR THEN ADD 1 TO CHAR-COUNT END-IF MOVE PASSWD(MAXCOUNT:1) TO TESTCHR IF TESTCHR IS EQUAL TO LIMITCHR THEN ADD 1 TO CHAR-COUNT END-IF
Here you can see how subscripting works. You can only do it as part of
a MOVE statement, and the syntax seems to be
Whatever; it works.
Then we just need to change the if statement to check whether
CHAR-COUNT is 1, but I won’t bore you with that.
That’s all for today, folks!