Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!uunet!ginosko!aplcen!bink
From: bink@aplcen.apl.jhu.edu (Ubben Greg)
Newsgroups: comp.unix.questions
Subject: Re: How to make nroff boldface on a terminal that can highlight?
Summary: LEX easily converts the default NROFF "typefaces"
Message-ID: <2536@aplcen.apl.jhu.edu>
Date: 5 Aug 89 16:34:01 GMT
References: <20477@adm.BRL.MIL> <537@anasaz.UUCP>
Reply-To: bink@aplcen.apl.jhu.edu (Greg Ubben)
Organization: The Johns Hopkins University, Baltimore MD
Lines: 58

In article <537@anasaz.UUCP> duane@anasaz.UUCP (Duane Morse) writes:
> In article <20477@adm.BRL.MIL>, justice@dao.nrc.ca (Gerald Justice) writes:
> > I am writing some local man pages and was wondering if it was possible
> > to have nroff pass information about boldfaced text to a user's terminal
> > to have it appear as highlighted text.
>...
> Some terminals are smart enough to take the standard char-backspace-char
> sequence and do boldface.  The public domain 'less' program is smart
> enough to take these sequences and use inverse video for highlighting.
> We keep formatted man pages under /usr/catman, like a lot of other
> systems, to avoid the overhead of formatting the pages every time someone
> wants to look something up.
>...

Jan Anderson also mentions a BSD filter called "ul" which converts the
italics and bolding overstrikes used in default (-T37) nroff output into
highlighting commands for a particular output device.  If you don't have
access to this program (System V users), I've found that one can easily
create a filter for any particular device using a simple lex(1) program.
As an example, the following lex program (eproff.l) will create a filter
to format nroff output for an Epson printer:

%start ITALIC BOLD
%%
<INITIAL>_\b			    { printf("\0334"); BEGIN ITALIC; }
<INITIAL>.\b			    { printf("\033E"); BEGIN BOLD;   }
.\b				    ;
<ITALIC>./[^_]			    { ECHO; printf("\0335"); BEGIN INITIAL; }
<BOLD>./[\t\n ]*([!-~][^\b]|_\b)    { ECHO; printf("\033F"); BEGIN INITIAL; }

You would compile it by
	lex eproff.l && cc lex.yy.c -ll -o eproff
and might use it as
	man lex | eproff | lp -dep ...

If the input were:
	It does _^Hi_^Ht_^Ha_^Hl_^Hi_^Hc_^Hs
	_^Ht_^He_^Hx_^Ht and b^Hb^Hb^Hbo^Ho^Ho^Hol^Hl^Hl^Hld^Hd^Hd^Hd
	t^Ht^Ht^Hte^He^He^Hex^Hx^Hx^Hxt^Ht^Ht^Ht correctly.
The output would be:
	It does ^[4italics^[5
	^[4text^[5 and ^[Ebold
	text^[F correctly.

Notice the program maximizes the text between bolding commands in order to
simplify the output -- this might be undesirable for reverse-video on display
devices.  It doesn't do so for italics because this was converted from a
version which changed italics to continuous underlining for a Xerox 2700-II
laser printer.

You can create a filter for other devices by simply changing the printf()
arguments.  Or make it more general by adding a main() which looks up TERM
or a -T argument in the terminfo to get the appropriate codes.  My laser
printer version also recognizes the line motions ^[7, ^[8, and ^[9.
E-mail questions or feedback are welcome.

						-- Greg Ubben
						   bink@aplcen.apl.jhu.edu