@documentencoding and TeX

[Akim Demaille]

It is my understanding that all the outputs properly understand
@documentencoding except TeX.  Since TeX is behind texi2dvi, and since
recode is fairly common, at least on the machine where LC_ALL != C, I
suggest the following patch.  It does considerably simplify my file,
until proper TeX support is implemented.

Index: ChangeLog
from  Akim Demaille  <address@hidden>

        Let DVI and PDF files honor the document encoding in TeX.

        * util/texi2dvi ($language): Be sure to set its value.
        (get_xref_files): Turn this external shell script into a function.
        (filename_rcd, $tmpdir_rcd): New.
        If the file is in Texinfo, and has @documentencoding, and recode
        is available, then run it.

Index: util/texi2dvi
===================================================================
RCS file: /cvsroot/texinfo/texinfo/util/texi2dvi,v
retrieving revision 1.31
diff -u -u -r1.31 texi2dvi
--- util/texi2dvi 14 Aug 2004 00:54:34 -0000 1.31
+++ util/texi2dvi 25 Aug 2004 08:21:35 -0000
@@ -114,7 +114,7 @@
 : ${EGREP=egrep}
 
 # return true if program $1 is somewhere in PATH, else false.
-# 
+#
 findprog () {
   foundprog=false
   for dir in `echo $PATH | tr "$path_sep" " "`; do
@@ -229,7 +229,7 @@
 
 
 # We can't do much without tex.
-# 
+#
 if findprog ${TEX:-tex}; then :; else cat <<EOM
 You don't have a working TeX binary (${TEX:-tex}) installed anywhere in
 your PATH, and texi2dvi cannot proceed without one.  If you want to use
@@ -249,7 +249,7 @@
 # didn't explicitly specify.  We don't check for elatex and pdfelatex
 # because (as of 2003), the LaTeX team has asked that new distributions
 # use etex by default anyway.
-# 
+#
 # End up with the TEX and PDFTEX variables set to what we are going to use.
 if test -z "$TEX"; then
   if findprog etex; then TEX=etex; else TEX=tex; fi
@@ -310,34 +310,31 @@
 s/address@hidden texi2dvi//
 EOF
 
-# A shell script that computes the list of xref files.
+# Compute the list of xref files.
 # Takes the filename (without extension) of which we look for xref
 # files as argument.  The index files must be reported last.
-get_xref_files=$utildir/get_xref.sh
-cat <<\EOF >$get_xref_files
-#! /bin/sh
-
-# Get list of xref files (indexes, tables and lists).
-# Find all files having root filename with a two-letter extension,
-# saves the ones that are really Texinfo-related files.  .?o? catches
-# many files: .toc, .log, LaTeX tables and lists, FiXme's .lox, maybe more.
-for this_file in "$1".?o? "$1".aux "$1".?? "$1".idx; do
-  # If file is empty, skip it.
-  test -s "$this_file" || continue
-  # If the file is not suitable to be an index or xref file, don't
-  # process it.  It's suitable if the first character is a
-  # backslash or right quote, as long as the first line isn't
-  # \input texinfo.
-  first_character=`sed -n '1s/^\(.\).*$/\1/p;q' $this_file`
-  if (test "x$first_character" = "x\\" \
-      && sed 1q $this_file | grep -v '^\\input *texinfo' >/dev/null) \
-     || test "x$first_character" = "x'"; then
-    xref_files="$xref_files ./$this_file"
-  fi
-done
-echo "$xref_files"
-EOF
-chmod 500 $get_xref_files
+get_xref_files ()
+{
+  # Get list of xref files (indexes, tables and lists).
+  # Find all files having root filename with a two-letter extension,
+  # saves the ones that are really Texinfo-related files.  .?o? catches
+  # many files: .toc, .log, LaTeX tables and lists, FiXme's .lox, maybe more.
+  for this_file in "$1".?o? "$1".aux "$1".?? "$1".idx; do
+    # If file is empty, skip it.
+    test -s "$this_file" || continue
+    # If the file is not suitable to be an index or xref file, don't
+    # process it.  It's suitable if the first character is a
+    # backslash or right quote, as long as the first line isn't
+    # \input texinfo.
+    first_character=`sed -n '1s/^\(.\).*$/\1/p;q' $this_file`
+    if (test "x$first_character" = "x\\" \
+        && sed 1q $this_file | grep -v '^\\input *texinfo' >/dev/null) \
+       || test "x$first_character" = "x'"; then
+      xref_files="$xref_files ./$this_file"
+    fi
+  done
+  echo "$xref_files"
+}
 
 # File descriptor usage:
 # 0 standard input
@@ -404,12 +401,16 @@
   tmpdir_xtr=$tmpdir/xtr
   filename_xtr=$tmpdir_xtr/$filename_noext.$ext
 
+  # _rcd.  The Texinfo file recoded in 7bit.
+  tmpdir_rcd=$tmpdir/rcd
+  filename_rcd=$tmpdir_rcd/$filename_noext.$ext
+
   # _bak.  Copies of the previous xref files (another round is run if
   # they differ from the new one).
   tmpdir_bak=$tmpdir/bak
 
   # Make all those directories and give up if we can't succeed.
-  mkdir $tmpdir_src $tmpdir_xtr $tmpdir_bak || exit 1
+  mkdir $tmpdir_src $tmpdir_xtr $tmpdir_rcd $tmpdir_bak || exit 1
 
   # Source file might include additional sources.
   # We want `.:$orig_pwd' before anything else.  (We'll add `.:' later
@@ -488,6 +489,7 @@
     [lL]a[tT]e[xX] | *.ltx | *.tex)
       # Assume a LaTeX file.  LaTeX needs bibtex and uses latex for
       # compilation.  No makeinfo.
+      language=latex
       bibtex=${BIBTEX:-bibtex}
       makeinfo= # no point in running makeinfo on latex source.
       texindex=${MAKEINDEX:-makeindex}
@@ -502,6 +504,7 @@
 
     *)
       # Assume a Texinfo file.  Texinfo files need makeinfo, texindex and tex.
+      language=texinfo
       bibtex=
       texindex=${TEXINDEX:-texindex}
       textra_cmd='/address@hidden/a'
@@ -595,6 +598,22 @@
     filename_input=$filename_xtr
   fi
 
+  # If this is a Texinfo file with a specified input encoding, and
+  # recode is available, then recode to plain 7 bit Texinfo.
+  if test $language = texinfo; then
+    encoding=`sed -ne '                                                \
+                 /@documentencoding/{                          \
+                    s/address@hidden \([^ ]*\).*/\1/p; \
+                    q                                          \
+                 }'                                            \
+                 <$filename_input`
+    if test -n "$encoding" && findprog recode; then
+      $verbose "Recoding from $encoding to Texinfo."
+      recode "$encoding"..texinfo <$filename_input > $filename_rcd
+      filename_input=$filename_rcd
+    fi
+  fi
+
   # If clean mode was specified, then move to the temporary directory.
   if test "$clean" = t; then
     $verbose "cd $tmpdir_src"
@@ -602,7 +621,7 @@
   fi
 
   while :; do # will break out of loop below
-    orig_xref_files=`$get_xref_files "$filename_noext"`
+    orig_xref_files=`get_xref_files "$filename_noext"`
 
     # Save copies of originals for later comparison.
     if test -n "$orig_xref_files"; then
@@ -687,7 +706,7 @@
     fi
 
     # Check if xref files changed.
-    new_xref_files=`$get_xref_files "$filename_noext"`
+    new_xref_files=`get_xref_files "$filename_noext"`
     $verbose "Original xref files = `echo $orig_xref_files | sed 's|\./||g'`"
     $verbose "New xref files      = `echo $new_xref_files | sed 's|\./||g'`"