#!/usr/bin/env perl # $Id: xindy.pl,v 1.8 2006/07/30 10:30:42 jschrod Exp $ #------------------------------------------------------------ # (history at end) =head1 NAME xindy - create sorted and tagged index from raw index =head1 SYNOPSIS xindy [-V?h] [-qv] [-d magic] [-o outfile.ind] [-t log] \ [-L lang] [-C codepage] [-M module] [-I input] \ [--interactive] [--mem-file=xindy.mem] \ [idx0 idx1 ...] =head2 GNU-Style Long Options for Short Options: -V / --version -? / -h / --help -q / --quiet -v / --verbose -d / --debug (multiple times) -o / --out-file -t / --log-file -L / --language -C / --codepage -M / --module (multiple times) -I / --input-markup (supported: latex, omega, xindy) =head1 DESCRIPTION B is the formatter-indepedent command of xindy, the flexible indexing system. It takes a raw index as input, and produces a merged, sorted and tagged index. Merging, sorting, and tagging is controlled by xindy style files. Files with the raw index are passed as arguments. If no arguments are passed, the raw index will be read from standard input. B is completely described in its manual that you will find on its Web Site, http://www.xindy.org/. A good introductionary description appears in the indexing chapter of the LaTeX Companion (2nd ed.) If you want to produce an index for LaTeX documents, the command texindy(1) is probably more of interest for you. It is a wrapper for B that turns on many LaTeX conventions by default. =head1 OPTIONS =over =item C<--version> / B<-V> output version numbers of all relevant components and exit. =item C<--help> / B<-h> / B<-?> output usage message with options explanation. =item C<--quiet> / B<-q> Don't output progress messages. Output only error messages. =item C<--version> / B<-v> Output verbose progress messages. =item C<--debug> I / B<-d> I Output debug messages, this option may be specified multiple times. I determines what is output: magic remark ------------------------------------------------------------ script internal progress messages of driver scripts keep_tmpfiles don't discard temporary files markup output markup trace, as explained in xindy manual level=n log level, n is 0 (default), 1, 2, or 3 =item C<--out-file> F / B<-o> F Output index to file F. If this option is not passed, the name of the output file is the base name of the first argument and the file extension F. If the raw index is read from standard input, this option is mandatory. =item C<--log-file> F / B<-t> F Output log messages to file F. These log messages are independent from the progress messages that you can influence with C<--debug> or C<--verbose>. =item C<--language> I / B<-L> I The index is sorted according to the rules of language I. These rules are encoded in a xindy module created by I. If no input encoding is specified via C<--codepage>, a xindy module for that language is searched with a latin, a cp, an iso, or ascii encoding, in that order. =item C<--codepage> I / B<-C> I The raw input is in input encoding I. This information is used to select the correct xindy sort module and also the I target encoding for C input markup. When C input markup is used, C is always used as codepage, this option is then ignored. =item C<--module> I / B<-M> I Load the xindy module F. This option may be specified multiple times. The modules are searched in the xindy search path that can be changed with the environment variable C. =item C<--input-markup> I / B<-I> I Specifies the input markup of the raw index. Supported values for I are C, C, and C. C input markup is the one that is emitted by default from the LaTeX kernel, or by the C macro package of David Jones. ^^-notation of single byte characters is supported. Usage of LaTeX's I package is assumed as well. C input markup is like C input markup, but with Omega's ^^-notation as encoding for non-ASCII characters. LaTeX I encoding is not used then, and C is enforced to be the codepage. C input markup is specified in the xindy manual. =item C<--interactive> Start xindy in interactive mode. You will be in a xindy read-eval-loop where xindy language expressions are read and evaluated interactively. =item C<--mem-file> I This option is only usable for developers or in very rare situations. The compiled xindy kernel is stored in a so-called I, canonically named F, and located in the xindy library directory. This option allows to use another xindy kernel. =back =head1 SUPPORTED LANGUAGES / CODEPAGES The following languages are supported: =head2 Latin scripts albanian gypsy portuguese croatian hausa romanian czech hungarian russian-iso danish icelandic slovak-small english italian slovak-large esperanto kurdish-bedirxan slovenian estonian kurdish-turkish spanish-modern finnish latin spanish-traditional french latvian swedish general lithuanian turkish german-din lower-sorbian upper-sorbian german-duden norwegian vietnamese greek-iso polish German recognizes two different sorting schemes to handle umlauts: normally, C<ä> is sorted like C, but in phone books or dictionaries, it is sorted like C. The first scheme is known as I, the second as I. C<*-iso> language names assume that the raw index entries are in ISO 8859-9 encoding. C is a northern Russian dialect. =head2 Cyrillic scripts belarusian mongolian serbian bulgarian russian ukrainian macedonian =head2 Other scripts greek klingon =head2 Available Codepages This is not yet written. You can look them up in your xindy distribution, in the F directory (where I is your language). They are named F, where F is most often empty (for german, it's C and C; for spanish, it's C and C, etc.) < Describe available codepages for each language > < Describe relevance of codepages (as internal representation) for LaTeX inputenc > =head1 ENVIRONMENT =over =item C A list of directories where the xindy modules are searched in. No subtree searching is done (as in TDS-conformant TeX). If this environment variable is not set, the default is used: C<.:>IC<:>IC. I is determined at run time, relative to the B command location: Either it's F<../modules>, that's the case for F-installations. Or it's F<../lib/xindy/modules>, that's the case for F-installations. =item C Library directory where F and F are located. The modules directory may be a subdirectory, too. =back =head1 KNOWN BUGS Option B<-q> also prevents output of error messages. Error messages should be output on stderr, progress messages on stdout. There should be a way to output the final index to stdout. This would imply B<-q>, of course. Codepage C should be supported for all languages, and should be used as internal codepage for LaTeX inputenc re-encoding. =head1 SEE ALSO texindy(1), tex2xindy(1) =head1 AUTHOR Joachim Schrod =head1 LEGALESE Copyright (c) 2004-2006 by Joachim Schrod. B is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. =for Emacs #' =cut use 5.006; use strict; BEGIN { use vars qw($Revision $VERSION); q$Revision: 1.8 $ =~ /: (\d+)\.(\d+)/ ; # q wg. Emacs indent! my ($major, $minor) = ($1, $2); $VERSION = "$major." . ($minor<10 ? '0' : '') . $minor; } # Some common variables. # Determine environment. Where is our library directory, and our modules? use File::Basename; use Cwd; our ($cmd_dir, $cmd, $xindy_run, $lib_dir, $modules_dir, $path_sep); BEGIN { $cmd_dir = dirname($0); $cmd = basename($0); if ($^O eq "MSWin32" || $^O eq "cygwin") { $xindy_run = "xindy-lisp.exe"; $path_sep = ";"; } else { $xindy_run = "xindy.run"; $path_sep = ":"; } # library directory if ( $ENV{XINDY_LIBDIR} ) { $lib_dir = $ENV{XINDY_LIBDIR}; die "$cmd: Invalid XINDY_LIBDIR setting: cannot find $lib_dir/$xindy_run" if ! -f "$lib_dir/$xindy_run"; } elsif ( -f "$cmd_dir/$xindy_run" ) { # texlive unix $lib_dir = $cmd_dir; } elsif ( exists $ENV{"SELFAUTOLOC"} && -f "$ENV{SELFAUTOLOC}/$xindy_run" ) { # texlive woe32 $lib_dir = $ENV{"SELFAUTOLOC"}; $cmd_dir = $ENV{"SELFAUTOLOC"}; } elsif ( '/home/vvv/src/tex/texlive/source/utils/01/lib' ne '@libdir' . '@' ) { # GNU configure at work? if ( -d '/home/vvv/src/tex/texlive/source/utils/01/lib/xindy' ) { # /usr style $lib_dir = '/home/vvv/src/tex/texlive/source/utils/01/lib/xindy'; } else { $lib_dir = '/home/vvv/src/tex/texlive/source/utils/01/lib'; # /opt style } die "$cmd: Cannot locate $lib_dir/$xindy_run" if ! -f "$lib_dir/$xindy_run"; } elsif ( -f "$cmd_dir/../lib/$xindy_run" ) { # /opt style $lib_dir = "$cmd_dir/../lib"; } elsif ( -f "$cmd_dir/../lib/xindy/$xindy_run" ) { # /usr style $lib_dir = "$cmd_dir/../lib/xindy"; } else { die "$cmd: Cannot locate xindy library directory"; } # modules directory if ( -d "$cmd_dir/../modules" ) { # /opt style $modules_dir = "$cmd_dir/../modules"; } elsif ( -d "$lib_dir/modules" ) { # /usr style $modules_dir = "$lib_dir/modules"; } elsif ( -d "$lib_dir/../../texmf/xindy" ) { # texlive multiplatform style $modules_dir = "$lib_dir/../../texmf/xindy"; } elsif ( -d "$lib_dir/../texmf/xindy" ) { # texlive non-multiplatform style $modules_dir = "$lib_dir/../texmf/xindy"; } elsif ( -d "/usr/share/xindy" ) { # FSH style $modules_dir = "/usr/share/xindy"; } else { die "$cmd: Cannot locate xindy modules directory"; } # if $cmd_dir or $lib_dir is a symlink, clisp may get confused # because of relative paths above, so we resolve the path $modules_dir = Cwd::realpath($modules_dir); } # Used modules. use Getopt::Long qw(:config bundling); use File::Temp qw(tempfile tmpnam); use File::Spec; use POSIX qw(uname); # Check arguments, store them in proper variables. # # Do also something for backward compatibility: Check if this is an # old-style call. If it is, we have two arguments at the end, and the # second-to-last has the extension ".xdy". Then, call the old driver # script with the original arguments... my $usage_msg = <<_EOT_ usage: $cmd [-V?h] [-qv] [-d magic] [-o outfile.ind] [-t log] \\ [-L lang] [-C codepage] [-M module] [-I input] \\ [--interactive] [--mem-file xindy.mem] \\ [idx0 idx1 ...] GNU-STYLE LONG OPTIONS FOR SHORT OPTIONS: -V / --version -? / -h / --help -q / --quiet -v / --verbose -d / --debug (multiple times) (supported: script, keep_tmpfiles, markup, level=n) -o / --out-file -t / --log-file -L / --language -C / --codepage -M / --module (multiple times) -I / --input-markup (supported: latex, omega, xindy) _EOT_ ; sub usage () { print STDERR $usage_msg; exit 1; } sub parse_options(); sub handle_signals(); sub create_raw_index(); sub filter_index($$); sub xindy_expression(); sub call_xindy($$); our ($quiet, $verbose, %debug, $outfile, $logfile, $language, @codepages, @modules, $input_markup, $interactive, $mem_file); $input_markup = 'latex'; $mem_file = "$lib_dir/xindy.mem"; if ($^O eq 'darwin' && ! -e $mem_file) { # support universal binary on mac my @uname = POSIX::uname(); if ($uname[4] eq 'Power Macintosh') { $mem_file = "$lib_dir/xindy-ppc.mem"; } else { $mem_file = "$lib_dir/xindy-i386.mem"; } } die "$cmd: Cannot locate $mem_file" if ! -e $mem_file; parse_options(); if ( @ARGV == 2 && $ARGV[0] =~ /\.xdy$/ ) { print STDERR <devnull; unless ( $interactive ) { for my $f (@ARGV) { die "$f does not exist" if ! -f $f } $raw_index = create_raw_index(); # processes @ARGV my $filter_cmd = ''; if ( $input_markup eq 'latex' ) { $filter_cmd = "$cmd_dir/tex2xindy"; } elsif ( $input_markup eq 'omega' ) { $filter_cmd = "$cmd_dir/tex2xindy -o"; } $raw_index = filter_index ($filter_cmd, $raw_index) if $filter_cmd; } # Execution: Obey environment variables, create xindy start # expression, and eventually call it. my $xindy_expression = xindy_expression(); # accesses global option vars my $exit_code = call_xindy ($mem_file, $xindy_expression); # Finished: exit, clean up in END handler. exit ($exit_code); # ------------------------------------------------------------ sub parse_options() { my (@debug); GetOptions( 'version|V' => sub { output_version(0); }, 'internal-version' => sub { output_version(1); }, 'help|h|?' => sub { print $usage_msg; exit 0; }, 'quiet|q' => \$quiet, 'verbose|v' => \$verbose, 'debug|d=s' => \@debug, 'out-file|o=s' => \$outfile, 'log-file|t=s' => \$logfile, 'language|L=s' => \$language, 'codepage|C=s' => \@codepages, 'module|M=s' => \@modules, 'input-markup|I=s' => \$input_markup, 'interactive' => \$interactive, 'mem-file=s' => \$mem_file, ) or usage(); # Debug option values are easier to test in a hash. Clean up trace # level options, too. %debug = map { $_ => 1 } @debug; my @trace_level = grep /^level=/, @debug; if ( @trace_level > 1 ) { print STDERR "You can only specify one trace level.\n\n"; exit (1); } delete $debug{$trace_level[0]}; $trace_level[0] =~ s/^level=// ; $debug{trace_level} = $trace_level[0]; # Check for unsupported debug option values. my %debug_check = %debug; foreach my $magic ( qw(script keep_tmpfiles markup trace_level) ) { delete $debug_check{$magic}; } if ( %debug_check ) { my @magic = keys(%debug_check); print STDERR "Unsupported argument for --debug: @magic\n"; usage(); } # Script debugging implies running it verbose and not quiet. if ( $debug{script} ) { $verbose = 1; $quiet = 0; } # Default for the output file: first argument, with extension replaced # by ".ind". unless ( $outfile || $interactive ) { if ( @ARGV == 0 ) { print STDERR "You need to specify --out-file if the raw index is read from standard input.\n\n"; usage(); } my ($name, $path, $suffix) = fileparse ($ARGV[0], '\.[^\.]+'); $outfile = "$path$name.ind"; # FIXME opening "nul" multiple times causes a problem on woe32 $logfile = "$path$name.ilg" if $^O eq "MSWin32" && ! defined $logfile; } # FIXME: xindy wants a log file. Really? $logfile = File::Spec->devnull unless $logfile; # Modules fixup: If they have no .xdy suffix, they get one. @modules = map { /\.xdy$/ ? $_ : "$_.xdy" } @modules; # Check that the input markup is known. # omega markup implies codepage utf8. if ( $input_markup && $input_markup ne 'latex' && $input_markup ne 'omega' && $input_markup ne 'xindy' ) { print STDERR "Unsupported input markup $input_markup.\n\n"; usage(); } if ( $input_markup eq 'omega' ) { @codepages = qw(utf8); } } # # SIGNAL HANDLING # # FIXME: This is not good enough. We also need to kill subprocesses, # i.e., filter or xindy, if they are running. # Our signal handler function just exits. Temporary files are deleted # by the END section above. Actually, the exit code looses the # information about the received signal, that's not good but shouldn't # hurt either... sub signal_exit() { exit (2); } sub handle_signals () { $SIG{'HUP'} = \&signal_exit; # 1 $SIG{'INT'} = \&signal_exit; # 2 $SIG{'QUIT'} = \&signal_exit; # 3 $SIG{'TERM'} = \&signal_exit; # 15 } # # CREATION OF RAW XINDY INDEX # # Handle input files Perl-style. Returns file name that contains # concatenated input file contents. sub create_raw_index () { my ($output, $outfile) = tempfile(); push (@temp_files, $outfile); while ( <> ) { print $output $_; } close ($output); print "concatenated xindy input file: $outfile\n" if $debug{script}; return $outfile; } # Run a filter over raw index. Returns file name with filtered raw # index, supposed to be in xindy input format. # # We assume the file names to be safe from shell meta characters since # they were computed by File::Temp. sub filter_index ( $$ ) { my ($filter, $input) = @_; my $output = tmpnam(); push (@temp_files, $output); print "Running filter: $filter <$input >$output\n" if $verbose; system "$filter <$input >$output"; print "filtered xindy input file: $output\n" if $debug{script}; return $output; } # # XINDY EXECUTION # # Construct final xindy expression, from options. sub xindy_expression () { my ($logging, $tracing, $trace_level, $searchpath); # Determine language module of make-rules framework. Part of the # complexity below is from compatibility with the TLC2 # description. We need to support the language names listed there, # even though they are not current. In addition, the codepage # option was introduced later and we need to guess it. This # guesswork will often be wrong, sadly. # # FIXME: I didn't see all languages. What's on with gypsy and # hausa? if ( $language ) { my $ld = "$modules_dir/lang"; my $variant; # If there is no language directory, this might be a variant. # Language names and variants are separated by hyphens. The # variant name "din" is an abbreviation for "din5007". The # variant name "iso" is ignored, that is actually a codepage # name. # # FIXME: Or is "iso" the variant "translit"?! if ( ! -d "$ld/$language" ) { $language =~ /^([^-]*)-(.*)/ ; # language name ends with 1st hyphen if ( $2 && -d "$ld/$1" ) { # $2 is not set if the regex didn't match $language = $1; $variant = "$2-" unless ( $2 eq 'iso' ); $variant =~ s/din/din5007/ ; } } # Let's guess the codepage. We take any that starts with # "latin", "cp", "iso8859", or "ascii". @codepages = qw(latin cp iso8859 ascii) unless @codepages; my @styles; foreach my $cp ( @codepages ) { @styles = glob("$ld/$language/$variant$cp*-lang.xdy"); last if @styles; } unless ( @styles ) { print STDERR "Cannot locate xindy module for language $language"; print STDERR " in codepage $codepages[0]" if ( @codepages == 1 ); print STDERR ".\n"; exit (1); } # Extract language module name: It's the relative part after # the module directory. Put it at the front of the list of # needed modules. It's important that the language module is # loaded first, it defines the sort rulesets, and subsequent # modules shall be able to add sort rules. unshift (@modules, substr($styles[0], length("$modules_dir/"))); print "Found language module $styles[0]\n" if $debug{script}; } # If there is more than one xindy module, construct a style file. # This is bad, of course; xindy should handle a list of style # files itself. my $style_file = $modules[0]; # will be undef if @modules is not set if ( @modules > 1 ) { my $sf; ($sf, $style_file) = tempfile(); push (@temp_files, $style_file); foreach my $module ( @modules ) { print $sf "(require \"$module\")\n"; } close ($sf); } $style_file = quotify($style_file); my $outfile_q = quotify($outfile); $raw_index = quotify($raw_index); $logging = ':logfile ' . quotify($logfile) if $logfile; $tracing = ':markup-trace :on' if $debug{markup}; $trace_level = ":trace-level $debug{trace_level}" if $debug{trace_level}; $searchpath = quotify(join($path_sep, ".", $modules_dir, "$modules_dir/base")); my $exp = <<_EOT_ (progn (searchpath $searchpath) (xindy:startup :idxstyle $style_file :rawindex $raw_index :output $outfile_q $logging $tracing $trace_level) (exit)) _EOT_ ; return $exp; } # Actual xindy call. Returns exit code. sub call_xindy ( $$ ) { my ($mem_file, $xindy_exp) = @_; my @command = ("$lib_dir/$xindy_run", '-q', '-B', $lib_dir, '-M', $mem_file, '-E', 'iso-8859-1'); if ( $interactive ) { print "Proposed xindy expression:\n\n$xindy_exp\n" unless $quiet; } else { #push (@command, '-x', $xindy_exp); my ($output, $outfile) = tempfile(); push (@temp_files, $outfile); print $output $xindy_exp; close ($output); print "xindy startup file: $outfile\n" if $debug{script}; push (@command, $outfile); } if ( $debug{script} ) { print "modules directory: $modules_dir.\n"; print "command: @command\n"; } if ( $quiet && ! $interactive ) { open (STDOUT, '>', File::Spec->devnull); } unlink($outfile) if $^O eq "MSWin32" && -f $outfile; system @command; if ( $? == -1 ) { print STDERR "$cmd: Could not execute xindy kernel: $!\n"; } elsif ( $? & 127 ) { return 4; } else { return $? >> 8; } } sub output_version ( ;$ ) { # optional arg: internal-version flag my $internal = shift; output_xindy_release() unless $internal; print "$cmd script version: $VERSION\n"; my $exit_code = call_xindy($mem_file, '(xindy:startup :show-version t)'); exit ($exit_code); } sub output_xindy_release () { my $version = 'unknown'; my $version_file; if ( -f "$cmd_dir/../VERSION" ) { $version_file = "$cmd_dir/../VERSION"; } elsif ( -f "$lib_dir/VERSION" ) { $version_file = "$lib_dir/VERSION"; } if ( $version_file ) { if ( open(VERSION, "<$version_file") ) { while ( $version = ) { chomp ($version); $version =~ s/\#.*// ; $version =~ s/^\s+// ; $version =~ s/\s+$// ; last if $version; } close (VERSION); } } print "xindy release: $version\n"; } # Helper function: Make a proper quoted Lisp string. sub quotify ( $ ) { my $s = shift; $s =~ s:([\\\"]):\\$1:g ; # quote double-quote and backslash return "\"$s\""; } #====================================================================== # # $Log: xindy.pl,v $ # Revision 1.8 2006/07/30 10:30:42 jschrod # Check if an exec() error happened and output an error message. # (Ticket 1230801) # # Revision 1.7 2006/07/19 00:29:56 jschrod # Support for omega input markup. # # Revision 1.6 2005/05/02 19:16:26 jschrod # Support new RTE 2.2, built with CLISP 2.33.2, that needs -E option # to specify the default charset. # xindy also has proper version output in Lisp now, no need for # workarounds in script any more. # # Revision 1.5 2004/11/01 22:48:51 jschrod # Locate xindy script. # Terminate on option error. # Fix up version output. # # Revision 1.4 2004/08/05 14:10:54 jschrod # Language variant names may have hyphens now. Language names must # not have hyphens -- the first hyphen of the -L option argument # separates language and variant name. # Revision 1.3 was completely off -- I committed a copy of texindy. # I really don't know what happened there. # # Revision 1.2 2004/05/26 21:30:11 jschrod # Added POD documentation. # # Revision 1.1 2004/05/24 19:47:13 jschrod # Introduce new driver script, as part of the "Companion Release". #