intlist.pl: Terminal application to view interrupt list

[ecm-pushbx]

Current revision is https://hg.pushbx.org/ecm/tractest/file/1e1ae3f40538/intlist.pl

You specify list files using ./intlist.pl --listing INTERRUP.LST

It can search with / or ? (only usefully jumps to entries in list view mode) and move the list using Down, Up, Page Down, Page Up, g, G. Enter drills down into a more detailed mode, while Esc or space returns to the less detailed mode. q quits the program.

The modes are list view mode, line select mode, detail view with list (about half of the window lists the entry), and detail view focus (most of the window lists the entry, and movement affects the entry rather than list).

No hyperlinks implemented yet, I think I will eventually add those to the detail view focus mode.

[shidel]

If you think it is something we should include in releases of The List, we have a couple of options.

One of such, would be to simply keep that viewer's development where it is now. Then, just drop a zip file with it, the manual and any related files into the Miscellaneous sub-directory. Then, makelist will copy it as-is along with the other files in there when creating a release.

Alternatively or additionally, we me be better off to include a VIEWERS section in the INTERRUP.1ST file. Using a file called _Viewers.txt, it could list the known current viewers and where to get them. That would keep the release size as small as possible. The exception would be any official viewers (if any) we provide in the release. Actually, I think a VIEWERS section right after the FILELIST section might be a great idea.

But honestly, the more I think about it, the less I think we need any official viewers.

For a little while, I was thinking of using my new cross platform application framework that I built under Lazarus for use in the Codepage Witch and other projects. But, soon realized that would likely just be a waste of time. Why wouldn't they just use the HTML version once I get the new parser built. Ether that generated site or where I put the Nightly builds, will include a archive file where the user can download the entire HTML conversion for usage offline.

[ecm-pushbx]

I think it's fine to keep it in its current repo, or maybe spin off a new repo only for this tool. (The tractest repo is a bit of a kitchen sink collecting different tools.)

I added a bunch of features to it today, https://hg.pushbx.org/ecm/tractest/shortlog/0fdff56ebe2f

It can now cycle (using Tab or Backtick) through hyperlinks visible in an entry, and when you press Enter while in detail view focus mode it will display in the "error line" what the hyperlink should load. Actually loading the referenced entry is still tbd.

I found that there are some register statements misinterpreted as hyperlinks by my detection. Question is do we modify the list to insert blanks (blanks make my detection not detect these as links) or should I add something so only register-only strings with AH=, AX=, or (at least for INT 2D) AL= at the beginning are detected as hyperlinks?

Even with the latter choice, there's some uses like returns CF set/AX=0001h which would still be misdetected, so insert blanks at least for those?

[shidel]

It has been many years since I wrote that first parser to convert RBIL into HTML.
(I really need to start working on the new one soon.)

But from what I recall, it would handle things a couple different ways.

• First there are the references to tables in the text. These were always (supposed to be) referenced like (See #P101). Always in parenthesis with "see #" something. The something being the table number. Except for those in the INTERRUP.LST file, all other tables start with some sort of letter.
• Then, there are the SeeAlso: lines. These are all references to something somewhere. Could be and interrupt, table, port, memory location, etc.
• Then there is the text in the body or tables. For these, it would search for things starting with INT, MEM, etc. When it found those, it would attempt to decipher the following text to determine how much of it was part of the reference. It would then perform its best guess on where to link and wrap that text up in an HREF.

You cannot really rely on AH, AL, or AX. I swear I just saw a listing somewhere that was an INT ?? and used CL. I do not recall which one, and it is possible it just got stored in my head that way. But, I think I saw one link that.

But, linking for Ports, Memory, CMOS, etc, definitely do need to work differently.

:-)

[ecm-pushbx]

You cannot really rely on AH, AL, or AX. I swear I just saw a listing somewhere that was an INT ?? and used CL. I do not recall which one, and it is possible it just got stored in my head that way. But, I think I saw one link that.

But there's many references to AH= and AX= throughout the INT 21 entries. I would like to support these as hyperlinks too. May be lots of work but I'd rather insert blanks where such register statements appear that aren't meant as links.

[ecm-pushbx]

I fully implemented hyperlinks in the interrupt list now, for INT and tables. (Not yet for others like the memory list.) Like I suggested I limited the reg-only hyperlink matching to require AX=, AH=, or AL= initially.

I also applied the special case for "INT 13 Ext" as a name like I said in #37

There's a new command (L) and a command line switch (--link or -l) which immediately jump to the detail focus view of a particular entry.

I also fixed things so lines like the following will accept the register constant value as a part of the summary/signature:

TheList/source/Interrupt List/INT 21 DOS Function Calls/INT 212B01CX444D Quarterdeck DOSUPSYS v200 INSTALLATION CHECK.txt

Line 9 in 0e5ab37

CX = 444Dh ('DM')

(This particular entry is linked by the 21.52 entry.)

Current revision at https://hg.pushbx.org/ecm/tractest/file/8c0830f94493/intlist.pl

[ecm-pushbx]

You cannot really rely on AH, AL, or AX. I swear I just saw a listing somewhere that was an INT ?? and used CL. I do not recall which one, and it is possible it just got stored in my head that way. But, I think I saw one link that.

But there's many references to AH= and AX= throughout the INT 21 entries. I would like to support these as hyperlinks too. May be lots of work but I'd rather insert blanks where such register statements appear that aren't meant as links.

I may have misinterpreted you here. You're suggesting we should not limit hyperlinking with reg-only references to A[XHL] because there's INTs that use CL or other registers for the function number?

In that case, it would be nice to detect those too but the A register ones are more important to me.

[shidel]

@ecm-pushbx, yes. That is what I meant. But, it is hard to know for certain. There is a lot of information involved and it could be something that I had in impression of without being an actual fact. I do not know the exact requirements for all INTs.

I will have some hard data on this topic once I create the one-off utility to improve Uniqueness. It will be a simple matter to have it spit out some statistics on what parameters are used and how often. Will keep you posted.

I wanted to do that fix-flags program yesterday. But alas, I was sidetracked by IRL. Going to work on that one now. :-)

[ecm-pushbx]

Updated intlist.pl again: https://hg.pushbx.org/ecm/tractest/shortlog/fcb75830a94a

It now supports going Back and Forwards in a simple history that's written whenever you load a hyperlink (either from the -l switch, L command, or by selecting it with Enter from the detail view focus mode). MEM, CALL, and PORT hyperlinks are supported as well, and so are table references with a capital letter after the # rather than a decit. PORT hyperlinks use a relaxed name match because many of the names do not lead the summary line but rather are found later in the summary line.

Also, INT 2D"AMIS" is special cased to link to the spec overview entry rather than 2D.mm00 as it would before.

[shidel]

@ecm-pushbx,

While I'm am just starting on the ID Uniqueness, I saw a couple of those odd IDs that use something other than A[XHL]. So, I won't need to do any statics.

Here are a couple examples that popped up:

https://github.com/LoopZ/TheList/blob/main/source/Interrupt%20List/INT%20E0%20IBM%20ROM%20BASIC%20used%20while%20in%20interpreter%20multiple%20purposes/INT%20E0CL3B%20REAL32%20PLOAD%20LOAD%20CMD%20FILE%20INTO%20MEMORY.txt

https://github.com/LoopZ/TheList/blob/main/source/Interrupt%20List/INT%20EF%20BASIC%20ORIGINAL%20INT%2009%20VECTOR/INT%20EFCX00C8%20GEM%20AES%20INTERFACE.txt

There could be others elsewhere, these were easy to spot when I was scrolling back threw some console output because they were near the end.

[shidel]

@ecm-pushbx Scrolling back further in that output, I see ones just using BL, BX, SI, etc.

[shidel]

After much messing about, it will just be easier to Fix the Uniqueness issue by hand. Ugh.

[ecm-pushbx]

Added a new mode where you pass --list or --list --broken and it just dumps all / broken hyperlinks in the files given (don't forget --multi): https://hg.pushbx.org/ecm/tractest/rev/6820a69e7677

I added this because it may be useful to find broken links, and I wanted to re-use the hyperlinking code that I already made for this program.