docs/Makefile.am for gnuplot-ja.pdf

Sorry for my delay in looking at this.

I understand the part about including a japanese version of allterm.h.
But I do not understand why a different version of doc2tex is needed. Can you explain?

I see in the Makefile fragment that you attached that doc2tex is compiled with a flag
-DJAPANESE_DOC but that conditional symbol is not used anywhere in doc2tex.c.

What does the awk script do? Can you attach a copy of that here?
Can we do the same manipulation in doc2tex.c and avoid the need for a separate step and a separate script?

Q1

I do not understand why a different version of doc2tex is needed.
Can you explain ?

A1. In fact, doc2tex.c itself doesn't have conditional macros
"#ifdef JAPANESE_DOC", but allterm-ja.h has them per each terminal
help, and doc2tex.c include allterm-ja.h through doc2x.h.
So, the doc2tex binary compiled with the flag -DJAPANESE_DOC is differ
from the binary without the flag.

Q2.

What does the awk script do ?

It does as the doc2tex do for "see [keyword]". In Japanese, we write
"IkaSansyo" ("see" in English) or "IkamoSansyo" ("see also" in
English). However, the doc2tex did convert "[keyword]" part to
"{\bf [keyword]}"
only, not
"{\bf [keyword] (p.~\pageref{[keyword]})\index{[keyword]})"
for Japanese document.

So, I made tex2ja.awk to search the string
"Ika(mo)?Sansyo {\bf [keyword]}"
in the output of "doc2tex gnuplot-ja.doc", and to modify it to
"Ika(mo)?Sansyo {\bf [keyword]
(p.~\pageref{[keyword]})\index{[keyword]})"

But, the script tex2ja.awk is old and may be incorrect, it
doesn't do the process correctly for the current document.

Q3.

Can you attach a copy of that here?
It is included in term-ja.diff.

Q4.

Can we do the same manipulation in doc2tex.c and avoid the
need for separate step and a separate script ?
It may be the correct way. In fact, I found the code about "ja_see" in
current doc2tex.c. So, I try to check whether the code do the process
for Japanese "Ika(mo)?Sansyo" strings.

I have modified the makefiles to build and include allterm-ja.h when making gnuplot-ja.pdf

Please test and let me know if more needs to be done. Is the awk script needed for any of the other build targets?

commit 7d63cb58fa309d1e9496956e09616ade9e717f81 (HEAD -> master, origin/master, origin/HEAD)
Author: Ethan A Merritt <merritt@u.washington.edu>
Date:   Fri Dec 2 14:19:51 2022 -0800

    japanese docs:  Handle terminal help in the Makefile target gnuplot-ja.pdf

    Notes:

     - I'm not great at autotools modifications.  Please check!

     - The awk script in term-ja.diff is no longer needed for doc->tex.
       I do not know if it is used for anything else.

     - allterm-ja.h is now added to the git repository so that it is
       unconditionally available for #include by doc2x.h, but it is
       only used by doc2tex if the command line switch -japanese is present.
       It will be rebuilt if any terminal text changes.

     - I could not get inclusion of translated gnuplot-tikz.help
       to work properly.  The make ordering is wrong.
       However the existing text correctly instructs users to retrieve
       the full help text using 'set term lua tikz help'.

     - term-ja.diff is modified to fix the section patching block.trm
       and to remove inclusion of gnuplot-tikz.help by lua.trm

I think attached patch is need for doc2tex.c.

1) The first strncmp() search the Japanese string "Sansyo". But "Sansho" is also used at not the target of ja_see. Since I use "Sansyo:" for the target of ja_see, ':' should be included in the search string.
2) strncmp() functions searching "Sansyo" and Japanese-period is placed after str is incremented. But it may fail the process of '' for the string "name[Japanese-period]". At the second '', ja_see == FALSE because str is incremented before strncmp() and it matches the Japanese-period. The str should be incremented after strncomp().
3) Because "see" or "See" string may appear in Japanese document, they may fail the process for '`'. The see (not ja_see) variable should not be TRUE for japanese doc. (I think ja_see may become see.)