Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!wuarchive!udel!princeton!phoenix!tvz
From: tvz@phoenix.Princeton.EDU (Timothy Van Zandt)
Newsgroups: comp.sys.next
Subject: Re: Spellchecking LaTeX files in emacs
Keywords: Spellchecking LaTeX files in emacs
Message-ID: <6169@idunno.Princeton.EDU>
Date: 11 Feb 91 23:08:55 GMT
References: <1991Feb10.015245.29408@athena.mit.edu>
Sender: news@idunno.Princeton.EDU
Organization: Princeton University, Princeton, New Jersey
Lines: 77

In article <1991Feb10.015245.29408@athena.mit.edu> lastole@athena.mit.edu (Lars A Stole) writes:
>I currently use emacs to write my LaTeX files.  In the past (when I used
>a VAXstation3100) I used ispell (a program on MIT's athena system) to
>spellcheck my latex files, using "-t" as an option so that commands
>beginning with a backslash would be ignored.  Does such a program exist
>on the NeXT?  I am currently using "spell" which is on the NeXT machine
>to do my spell checking, but there doesn't seem to be a simple option to
>ignore TeX commands as in ispell.  Is there a simple way get around this?
>Is there a way to get around this problem using the spell checking
>command in Edit?

I'm not sure about using ispell on the NeXT, but an alternative is to 
convert backslashes to <space>zzz or some other string that never appears
at the beginning a word in your files. You can then spell check not only
English words, but also tex commands. A command like \foobar appears as
zzzfoobar in your dictionary. This works with just about spell checker (that
I have tried at least); in particular, it works with Edit.

For example, the script "tspell" works like spell, but it produces a list
of misspelled words and tex commands, using a dictionary with filename
~/Library/Dict/TexHash (or whatever). Start by copying /usr/dict/hlista to
~/Library/Dict/TexHash. The dictionary is maintained with tspellin, described
below. One warning about "spell" on the Next: It chokes on
lines longer than 512K, whereas on the other systems I use, "spell" can
handle lines up to 1024K. "tspell" breaks lines so this is not a problem. You
can add a few lines to the script so that tspell uses tex as the 
default extension.

-----------------BEGIN tspell-------------------------------------
#!	/bin/sh
/bin/sed 's/\\/\ zzz/g' $infile |
tr -cs A-Za-z '\012' |
spell -d ~/Library/Dict/TexHash |
/bin/sed 's/zzz/\\/g'
-------------------END tspell-----------------------------------

"tspellin" is an adaptation of spellin. It takes a list of words,
including tex commands beginning with backslash, and adds them
to ~/Library/Dict/TexWords and the hashed list ~/Library/Dict/TexHash for use
with "tspell". The unix command "spellout" could also be used in
the script below. You should start by running "tspell" on your favorite
tex macro files to get a list of tex commands to add to your dictionary
using "tspellin."

------------------BEGIN tspellin --------------------------------
#! 	/bin/sh
/bin/sed 's/[\\]/zzz/g' $1 >> ~/Library/Dict/TexWords
spellin < ~/Library/Dict/TexWords /usr/dict/hlista > \
		~/Library/Dict/TexHash
--------------------END tspellin-----------------------------------


In Edit, use the pipe
    sed 's/\\/\ zzz/g'
to convert \ to <space>zzz, then spell check, and then convert the
file back using the pipe
	sed 's/\ zzz/\\/g'
	
For some reason, you cannot just add these pipes to .pipedict. Instead,
define them as shell scripts. I added the lines

TeX2Spell	>	tex2spell
Spell2TeX	<	spell2tex

to .pipedict, where tex2spell and spell2tex are the shell scripts, so that
command-> and command-< do the conversions. Real simple.

Edit and Mail use a user dictionary ~/.NeXT/LocalDictionary. Words are
separated by null rather then newline, which makes it hard to add a list
of words. I took my list TexWords, converted newlines to null with some
difficulty, appended it to my LocalDictionary, and since then, I have
no idea what local dictionary Edit and Mail are using, but its not 
~/.NeXT/LocalDictionary anymore (and it is something). Weird! You can
always just add tex commands to the dictionary as you go along, and avoid
any hassle.

You can use a similar technique with emacs.