[Raion]

We've been working on it off and on. I can't afford anything until my locksmithing school tuition is over with tbh, then I'll start paying my devs more.

[jwhat]

Hi JPS & Raion,

yes I found out about the pre-rendered cat, which appears to have terminal control (ANSI ??) stuff in it.

There are quite a few "ansi to html" renderers out there, can you confirm that the SGI cat pages are ANSI ?

This seems like a pretty simple case of:

Code:

if (manX dir) {
  do nroff -> html stuff
} else if (catX dir) {
  do ansi -> html stuff
}

Cheers from Oz,


jwhat/John

[Raion]

They're not. It's most than just putting up HTML and I didn't find any out of the box solutions.

[jwhat]

HI Raion,

I can see from looking at the uncompressed file all the codes (using vi) and it is un-readable, but when you cat the same file in xterm window it renders fine, so it must be something pretty standard...

Doing some searchs on ANSI -> HTML engines now to see how its looks piped through one of these.

EDIT: Tried:
  -- aha - rubbish default rendering, but if you play with options it looks better, so maybe starting point (no hyperlinks)
  -- ansi2html (shell script + gawk + gsed) - renders page simply and faithfully, no hyperlinks
  -- lots of go, python, ruby, rust and other variants, but I don't want to download a GB of stuff just to do something simple, where did the C programmers go ;-)


Cheers from Oz,

jwhat/John

[jwhat]

HI Raion,

I may have saved you some money/time on programming effort ;-)

I have been testing with "sh" man page and it is rendering quite nicely with "dummy" URL links (as I have not got driver programming going yet).
This is a good test case, as it long entry and has a lot of links which are presented in different ways.

To get "nice" output I used, pretty standard tools (on FreeBSD):
- ul - underline cleanup
- aha - ANSI HTML Adaptor
- xstlproc - to do a little bit of XML/HTML style sheet cleaning up
- sed - to add HTML links into the page.

The result is a page that renders on par with original SGI Infosearch: https://www.graphica.com.au/sh.ul.aha.xslt.sed.html

I have put code here: https://github.com/zebity/freebsd-doc/tree/main/IRIX

What needs to be done is:
1. Write a driver script that:

Code:

- Looks to see if it has pre-rendered HTML page and
    if so serve it up
  Otherwise
    Find requested page in IRIX cat/man directories
    Run through HTML rendering pipeline
    Save result to cached pages
    Go back to start ...

Now that all the futzing around with the IRIX formating is solved this should be pretty easy.
Would you be able to host pages ?

Cheers from Oz,


John.

[Raion]

We need to do it in PHP, and I want the pages indexed in a db which makes it more difficult. I'm probably going to just restart the code base when I have the money completely all over because the guy who was doing it didn't really follow what I asked for. What I ideally want is a PHP script and a cron that will collectively:

1. Take the awf formatted pages and put them into a database.

2. Index them into a comprehensive search.

3. Preserve all formatting and output it.

I'm not looking for an out of the box solution because nothing out there really is going to work for what I was going for. I don't really have any more time to mess with this. Locksmithing school ends this week and maybe I'll have more time after that but no guarantees.

[Raion]

Apologies for the abruptness of my last one but what I'm trying to do here especially with my new disposable income that will be around is making sure that we can create tools that are reusable but yet aren't as basic as say what the bsds or others are using. With your method we could easily create an IRIX 6.5.x manpage set, but I want to go beyond that and cover all major versions.

To that end I have to be aware of and account for those things. This will cost money and time but ultimately I think the end result will be much better than trying to use an out-of-the-box tool that was not designed to work with what we are looking at here. Especially since the Cartman paths vary from version to version

[Raion]

Yeah after looking over the codebase we're starting over. Sucks, but that's life for you. I'll advise when work begins and link the repo.

[jwhat]

Hi Raion,

np, I have been slowly getting HTML render cleaned up.

I believe to make this searchable simplest way is to feed in via web based HTML pages.

I have now got these cleaned up by removing extra headers/footers/empty lines & links would work if pages were on web server.

I stopped doing via sed/awk and just did the final cleanup via perl (awk’s regex does not cut it).

Here is link to new cleaned up sh man page (as example): https://www.graphica.com.au/sh.ul.aha.xslt.perl.html

Processing pipeline is now: cat | ul | aha | xslt | perl

Using what I have posted on github, you could just do "find" of all cat.z pages and feed them into the "irix-catman2html.sh" script to generate a full set of html cross-referenced pages...

Once the pages (via web server) are available you can point an indexing/search engine at them to get fully searchable result.

That is my simple view of how to get something ASAP.

BTW: What is awf ?

Cheers from Oz,

jwhat/john

[Raion]

The format of IRIX manpage is awf