#! /usr/bin/perl -w
##############################################################################
# $Id: pstoimg.pin,v 1.11 1999/10/25 21:18:22 MRO Exp $
#
# pstoimg
#
# Accompanies LaTeX2HTML
#
# Script to convert an arbitrary PostScript image to a cropped GIF or PNG
# image suitable for incorporation into HTML documents as inlined images
# to be viewed with WWW browsers.
#
# This software is provided as is without any guarantee.
#
##############################################################################
#
# $Log: pstoimg.pin,v $
# Revision 1.11 1999/10/25 21:18:22 MRO
#
# -- added more configure options (Jens' suggestions)
# -- fixed bug in regexp range reported by Achim Haertel
# -- fixed old references in documentation (related to mail list/archive)
#
# Revision 1.10 1999/10/06 22:04:13 MRO
#
# -- texexpand: latex2html calls texexpand with the -out option instead of
# output redirection: this is safer on non-UNIX platforms
# -- pstoimg: now there's no default cropping (useful for standalone
# conversions). latex2html was changes appropriately
# -- minor cleanups in latex2html script and documentation
#
# Revision 1.9 1999/09/14 22:02:02 MRO
#
# -- numerous cleanups, no new features
#
# Revision 1.8 1999/07/19 09:51:00 RRM
# -- added -aaliastext switch, for easier way to specify anti-aliased
# font characters, without also anti-aliasing other graphics objects.
#
# Revision 1.7 1999/06/24 07:28:59 MRO
#
#
# -- removed L2HMODULE
# -- fixed processing of -info switch
# -- changed option order for dvips on win32 (thanks JCL)
# -- bumped version to 99.2a8
#
# Revision 1.6 1999/06/06 14:24:50 MRO
#
#
# -- many cleanups wrt. to TeXlive
# -- changed $* to /m as far as possible. $* is deprecated in perl5, all
# occurrences should be removed.
#
# Revision 1.5 1999/06/04 20:14:25 MRO
#
#
# -- Reworked option parsing completely. Should behave much the same as before,
# options with -no_* work just like before.
# -- Changed $NOFORK to $CAN_FORK and inverted the logic.
# -- Small debugging enhancement in pstoimg
#
# Revision 1.4 1999/06/04 15:30:15 MRO
#
#
# -- fixed errors introduced by cleaning up TMP*
# -- made pstoimg -quiet really quiet
# -- pstoimg -debug now saves intermediate result files
# -- several fixes for OS/2
#
# Revision 1.3 1999/06/01 06:55:35 MRO
#
#
# - fixed small bug in L2hos/*
# - added some test_mode related output to latex2html
# - improved documentation
# - fixed small bug in pstoimg wrt. OS2
#
# Revision 1.2 1999/05/17 21:30:59 MRO
#
#
# -- make texexpand warning-free and start making it use strict
# compliant
#
# Revision 1.1 1999/05/11 06:10:00 MRO
#
#
# - merged config stuff, did first tries on Linux. Simple document
# passes! More test required, have to ger rid of Warnings in texexpand
#
# Revision 1.18 1999/05/05 19:47:03 MRO
#
#
# - many cosmetic changes
# - final backup before merge
#
# Revision 1.17 1999/03/15 23:00:53 MRO
#
#
# - moved L2hos modules to top level directory, so that no dir-
# delimiter is necessary in the @INC-statement.
# - changed strategy for "shave": Do not rely on STDERR redirection any
# more (caused problems on at least Win32)
#
# Revision 1.16 1999/02/14 23:44:34 MRO
#
#
# -- first attempt to fix Win32 problems
#
# Revision 1.15 1999/02/11 00:18:29 MRO
#
#
# -- cleaned up warppers, TeXlive stuff and Makefile
#
# Revision 1.14 1999/02/10 01:37:12 MRO
#
#
# -- changed os-dependency structure again - now neat OO modules are
# used: portable, extensible, neat!
# -- some minor cleanups and bugfixes
#
# Revision 1.13 1998/12/07 23:19:58 MRO
#
#
# -- added POD documentation to pstoimg and did a general cleanup
# -- some finetuning of config procedure and modules
#
# Revision 1.12 1998/10/31 14:13:05 MRO
# -- changed OS-dependent module loading strategy: Modules are now located in
# different (OS-specific) directories nut have the same name: Easier to
# maintain and cleaner code
# -- Cleaned up config procedure
# -- Extended makefile functionality
#
# Revision 1.11 1998/08/09 20:45:20 MRO
# -- some cleanup
#
# Revision 1.10 1998/06/14 14:10:38 latex2html
# -- Started to implement TeXlive configuration and better OS specific
# handling (Batch files) (Marek)
#
# Revision 1.9 1998/06/07 22:35:24 latex2html
# -- included things I learned from the Win95 port to config procedure:
# GS_LIB, Win32 module calls, directory separator stuff, ... (Marek)
#
# Revision 1.8 1998/06/01 12:57:56 latex2html
# -- Cleanup and cosmetics.
#
# Revision 1.7 1998/05/14 22:27:37 latex2html
# -- more work on config procedure (Makefile, GS_LIB)
# -- tested pstoimg in 98.1 environment successfully on Linux
#
# Revision 1.6 1998/05/06 22:31:09 latex2html
# -- Enhancements to the config procedure: Added a "generic" target
# in the Makefile for the TeXlive CD (not perfect yet)
# -- included test for kpsewhich / Web2C
# -- included latest stuff from Override.pm into os_*.pm
#
# Revision 1.5 1998/04/28 22:18:11 latex2html
# - The platform specific stuff is now kept in a separate perl module. This
# does not introduce significant overhead and enhances maintainability.
#
# Revision 1.4 1998/03/19 23:38:06 latex2html
# -- made pstoimg plug-in compatible with old one (touchwood!)
# -- cleaned up, added some comments
# -- inserted version information output
# -- incorporated patches to make OS/2 run better (thanks Uli)
# -- updated Makefile: make, make test, make install should work now
#
# Revision 1.3 1998/03/11 23:44:00 latex2html
# -- cleaned up config.pl and reworked dvips checks
# -- got pstoimg.pin up to par with the regular pstoimg
# -- cosmetic changes
# -- runs now under Win95 with Fabrice Popineau's Win32 tools (gs, TeX,...)
#
# Revision 1.2 1998/03/02 23:38:40 latex2html
# Reworked configuration procedure substantially. Fixed some killing bugs.
# Should now run on Win32, too.
# The file prefs.pm contains user-configurable stuff for DOS platforms.
# UNIX users can override the settings with the configure utility (preferred).
#
# Revision 1.1 1998/02/14 19:31:55 latex2html
# Preliminary checkin of configuration procedure
#
# (end CVS log)
###############################################################################
# This file has been automatically generated by build.pl from pstoimg.pin
# Do not edit this file as your changes will be lost when reconfiguring.
# If you want to supply patches, please apply them to pstoimg.pin and send
# the diff relative to the original pstoimg.pin. Thank you.
=head1 NAME
pstoimg - Convert a PostScript file to a bitmap image using
Ghostscript and the Netpbm utilities
=cut
use 5.003;
use strict;
#use diagnostics;
use vars qw(*SAVEERR $LATEX2HTMLDIR $SCRIPT);
# This variable points to the DIRECTORY where the latex2html files
# can be found.
use Getopt::Long;
# see below for a description of the environment
BEGIN {
# print "scanning for l2hdir\n";
if($ENV{LATEX2HTMLDIR}) {
$LATEX2HTMLDIR = $ENV{LATEX2HTMLDIR};
} else {
$ENV{LATEX2HTMLDIR} = $LATEX2HTMLDIR = '/usr/share/latex2html';
}
if(-d $LATEX2HTMLDIR) {
push(@INC,$LATEX2HTMLDIR);
} else {
die qq{Fatal: Directory "$LATEX2HTMLDIR" does not exist.\n};
}
}
use L2hos; # load OS-specific stuff
my $RELEASE = '2K.1beta';
my ($VERSION) = q$Revision: 1.11 $ =~ /:\s*(\S+)/;
$| = 1; # unbuffer STDOUT
my $dd = L2hos->dd; # Directory delimiter
my $prompt;
($prompt = $0) =~ s|^.*[/\\]||;
# Configuration as determined by "configure"
#
# Ghostscript
my $GS = '/usr/bin/gs';
my $GSDEVICE = 'pnmraw';
my $GSALIASDEVICE = 'ppmraw';
# Supported format(s)
my @IMAGE_TYPES = qw(png);
# Netpbm
my $PNMCROP = '/usr/bin/pnmcrop';
my $PPMQUANT = '/usr/bin/ppmquant';
my $PNMFLIP = '/usr/bin/pnmflip';
my $PNMCAT = '/usr/bin/pnmcat';
my $PNMFILE = '/usr/bin/pnmfile';
my $PBMMAKE = '/usr/bin/pbmmake';
# PNG support
my $PNMTOPNG = '/usr/bin/pnmtopng';
# Temporary diskspace
my $def_tmp = '/tmp'; # Space for temporary files
# Some lengths used by dvips
# MRO: Is this true for all runs of dvips?
my $PAGE_HEIGHT = 841.889; # dvips page height, in pts.
my $PAGE_WIDTH = 595.275; # dvips page width, in pts.
my $PAGE_HMARGIN = 72; # dvips margin: 1 inch = 72pt
my $PAGE_VMARGIN = 72; # dvips margin: 1 inch = 72pt
# The color to be made transparent
my $trans_color = '#ffffff';
###############################################################################
# Default settings
# Environment overrides defaults, command line options override everything
unless(@ARGV) {
print_help();
exit 0;
}
=head1 SYNOPSIS
B<pstoimg> B<-help> | B<-version>
B<pstoimg>
S<[ B<-antialias> ]>
S<[ B<-aaliastext> ]>
S<[ B<-center> I<num> ]>
S<[ B<-color> I<num> ]>
S<[ B<-crop> I<code> ]>
S<[ B<-debug> ]>
S<[ B<-density> I<num>]>
S<[ B<-depth> I<num> ]>
S<[ B<-discard> ]>
S<[ B<-flip> I<code> ]>
S<[ B<-geometry> I<X>xI<Y> ]>
S<[ B<-interlaced> ]>
S<[ B<-margins> I<X>,I<Y> ]>
S<[ B<-multipage> ]>
S<[ B<-out> I<file> ]>
S<[ B<-quiet> ]>
S<[ B<-rightjustify> I<num> ]>
S<[ B<-scale> I<num> ]>
S<[ B<-tmp> I<path> ]>
S<[ B<-topjustify> [B<x>]I<num> ]>
S<[ B<-transparent> ]>
S<[ B<-type> I<type> ]>
S<[ B<-shoreup> I<num>[B<d>] ]>
S<[ B<-white> ]>
I<file>
S<[ I<file2> ... ]>
=cut
my %opt = ();
unless(&GetOptions(\%opt, qw(-help -version -debug -discard -antialias -aaliastext
-multipage -type=s -gif -png -out=s -depth=i -color=i -flip=s -density=i
-scale=f -geometry=s -margins=s -crop=s -transparent -interlaced
-rightjustify=i -center=i -topjustify=s -shoreup=s -tmp=s -white -quiet))) {
print_usage("$prompt: Error: Invalid option(s) specified.");
exit 1;
}
=head1 OPTIONS
The command line options may be abbreviated to the shortest unique
prefix.
=over 4
=item B<-help>
Show this help page and exit.
=cut
if($opt{help}) {
print_help();
exit 0;
}
=item B<-version>
Show the release and version of pstoimg and exit.
=cut
if($opt{version}) {
print_version();
exit 0;
}
banner() unless($opt{quiet});
=item B<-antialias>
Use Ghostscript's anti-aliasing feature for rendering "softer" images.
This applies to lines and edges of polygonal and oval or circular shapes.
Only valid if Ghostscipt 4.03 or higher is installed.
=item B<-aaliastext>
Use Ghostscript's anti-aliasing feature for "smoother" font characters,
without the jagged edges. Similar to B<-antialias> for graphic components.
Only valid if Ghostscipt 4.03 or higher is installed.
=item B<-center> I<num>
Add the appropriate amount of whitespace to the left of the image so
that the image appears to be centered in a total width of I<num> pixels.
=cut
my $CENTER = 0; # No centering by default
if($opt{center}) {
$CENTER = $opt{center};
die <<"EOF" unless ($CENTER =~ /^\d+$/ && $CENTER > 0);
$prompt: Error: Illegal width for -center specified: "$CENTER"
Value must be a positive integer.
EOF
}
=item B<-crop> I<code>
Crop the bitmap from the given directions. I<code> may be a string of
several cropping instructions, which are executed strictly in the given
order. Possible values are: B<h> (horizontal, i.e. crop top and
bottom), B<v> (vertical), B<tblr> (top, bottom, left, right) and B<a> (all
directions). A special case is B<s>: "shave" the image at the bottom, but only
if a single line of whitespace exists.
=cut
my $EXTRA_CROP = '';
if($opt{crop}) {
$EXTRA_CROP = lc($opt{crop});
die <<"EOF" unless ( $EXTRA_CROP =~ /^([vhtblras]+)$/i );
$prompt: Error: Illegal crop specified: "$EXTRA_CROP"
Crop must be h, v, t, b, l, r, a, s or combination
EOF
}
=item B<-debug>
Turn on debugging output. This can get rather verbose. Any intermediate
files generated are not removed to help debugging.
=cut
if($ENV{DEBUG}) {
$opt{debug} = 1;
}
=item B<-density> I<num>
The density (resolution) in DPI in which to render the bitmap. The default
is 72.
=cut
my $DENSITY = 72;
if($opt{density}) {
$DENSITY = $opt{density};
}
elsif($ENV{DENSITY}) {
$DENSITY = $ENV{DENSITY};
}
die <<"EOF" unless $DENSITY =~ /^\d+$/;
$prompt: Error: Illegal density specified: "$DENSITY"
Density must be an integer value. Default is 72.
EOF
=item B<-depth> I<num> or B<-color> I<num>
Specify the color depth of the bitmap. Legal values are 1 (black & white),
8 (256 colors) and 24 (true color).
=cut
unless($opt{depth}) {
if($opt{color}) {
$opt{depth} = $opt{color};
}
elsif($ENV{DEPTH}) {
$opt{depth} = $ENV{DEPTH};
}
else {
$opt{depth} = 8;
}
}
die <<"EOF" unless $opt{depth} =~ /^(1|8|24)$/;
$prompt: Error: Illegal color depth specified: "$opt{depth}"
Depth must be either 1, 8 or 24.
EOF
=item B<-discard>
Delete the input postscript file if the conversion was successful. Setting
the environment DISCARD to a true value (as perl sees it) has the same
effect.
=cut
if($ENV{DISCARD}) {
$opt{discard} = 1;
}
=item B<-flip> I<code>
Flip all generated output bitmaps. The following codes are recognized:
lr (flip left-right), tb (flip top-bottom), xy (flip bottom/left-top/right),
r90 and ccw (rotate by 90 degrees counterclockwise), r270 and cw (rotate
90 degrees clockwise) and r180 (rotate 180 degrees).
=cut
if($opt{flip}) {
$opt{flip} = lc($opt{flip});
die <<"EOF" unless $opt{flip} =~ /^(lr|tb|xy|r90|ccw|r270|cw|r180)$/;
$prompt: Error: Illegal flip option specified: "$opt{flip}"
Flip must be one of: lr tb xy r90 ccw r270 cw r180
EOF
}
=item B<-geometry> I<X>xI<Y>
Render only this "window" of the PostScript file. If given, this option
can dramatically reduce memory requirements and speed up conversion. The
geometry is automatically detected in case of EPS files (Encapsulated
PostScript).
=cut
my $GEOMETRY = '';
if($opt{geometry}) {
$GEOMETRY = $opt{geometry};
if($GEOMETRY =~ s/-//o ) {
$EXTRA_CROP .= 'bl';
}
die <<"EOF" unless ($GEOMETRY =~ /^\d+x\d+$/i);
$prompt: Error: Illegal geometry specified: "$GEOMETRY"
Geometry must be <width>x<height>
EOF
}
=item B<-interlaced>
Generate an interlaced bitmap. Interlaced images build up from coarse to
fine as they are loaded. This option may not work on every installation
and/or bitmap type, depending of the capabilities of external programs.
=cut
my $INTERLACE = 0; # Do not make interlaced images by default
if($opt{interlaced}) {
if($opt{type} eq 'gif') {
print "$prompt: Warning: Interlaced GIFs not supported.\n";
}
$INTERLACE=1;
}
=item B<-margins> I<X>,I<Y>
The offset of the rectangle in the postscript file that is going to be
rendered from top/left. Can be used together with B<-geometry> to further
reduce the size of the intermediate bitmap file generated by Ghostscript.
=cut
if($opt{margins}) {
die <<"EOF" unless (($opt{margins} =~ /^(\d+),(\d+)$/));
$prompt: Error: Illegal margins specified: "$opt{margins}"
Margins must be <hmargin>,<vmargin>
EOF
$PAGE_HMARGIN = $1;
$PAGE_VMARGIN = $2;
}
=item B<-multipage>
Process a multi-page PostScript file, i.e. create an individual bitmap
for every page. The resulting files are numbered: The decimal number
(starting with 1) is appended to the basename of the PostScript input
file (or the basename of the filename specified with B<-out>), while
keeping the extension.
=item B<-out> I<file>
The file where to write the bitmap. If multiple PostScript files are
supplied on the command line, this option is ignored. The bitmap type
extension is appended automatically if I<file> does not contain a dot.
In connection with B<-multipage> I<file> is extended by the page number
as shown in this example:
-outfile foo.gif --------E<gt> foo1.gif, foo2.gif, ...
=cut
if(!$opt{out} && $ENV{OUTFILE}) {
$opt{out} = $ENV{OUTFILE};
}
=item B<-quiet>
Do not print anything except error messages.
=item B<-rightjustify> I<num>
Add the appropriate amount of whitespace to the left of the image so that
it appears to be aligned to the right in a total width of I<num> pixels.
=cut
my $RIGHT_JUSTIFY = 0; # No right justifying by default
if($opt{rightjustify}) {
$RIGHT_JUSTIFY = $opt{rightjustify};
die <<"EOF" unless ($RIGHT_JUSTIFY =~ /^\d+$/ && $RIGHT_JUSTIFY > 0);
$prompt: Error: Illegal width for -rightjustify specified: "$RIGHT_JUSTIFY"
Value must be a positive integer.
EOF
}
=item B<-scale> I<factor>
Scale the image by I<factor>. Valid choices are any numbers greater than
zero. Useful choices are numbers between 0.1 - 5.
Large numbers may generate very large intermediate files and will take
longer to process. If this option is omitted, the environment SCALE is
considered.
=cut
unless($opt{scale}) {
if($ENV{SCALE}) {
$opt{scale} = $ENV{SCALE};
}
else {
$opt{scale} = 1;
}
}
die <<"EOF" unless ($opt{scale} =~ /^[\d.e]+$/i && $opt{scale} > 0);
$prompt: Error: Illegal scale specified: "$opt{scale}"
Scale must be nonnegative float value.
EOF
=item B<-shoreup> I<num>[B<d>]
Make height and width of the bitmap(s) an exact multiple of I<num>. If
I<num> is followed by a "d", then half the extra vertical space is placed
underneath. This option is useful, if you want to have "blown-up" images
of high quality for print, but downscale them in HTML using
C<E<lt>IMG WIDTH=x HEIGHT=yE<gt>>. If the actual image is is not an
integer multiple of x,y then browsers tend to display distorted images.
=cut
my $SHORE_UP = 0; # No pixel alignment by default
if($opt{shoreup}) {
$SHORE_UP = $opt{shoreup};
die <<"EOF" unless $SHORE_UP =~ /^\d+d?$/i;
$prompt: Error: Illegal shore-up specified: "$SHORE_UP"
Value must be a positive integer, optionally followed by d
EOF
}
=item B<-tmp> I<path>
Use I<path> to store temporary files. Defaults to /tmp on this
installation. This parameter can be set by the environment B<TMP> or
B<TEMP>, too.
=cut
my $TMP = '';
if($opt{tmp}) {
$opt{tmp} =~ s|\Q$dd\E+$||; # remove trailing directory separator(s)
if(-d $opt{tmp} && -r _ && -w _) {
$TMP = $opt{tmp};
} else {
print "$prompt: Warning: Cannot use $opt{tmp} as temporary directory.\n";
}
}
if(!$TMP && ($ENV{TMP} || $ENV{TEMP})) {
($opt{tmp} = $ENV{TMP} || $ENV{TEMP}) =~ s|\Q$dd\E+$||;
if(-d $opt{tmp} && -r _ && -w _) {
$TMP = $opt{tmp};
} else {
print "$prompt: Warning: Cannot use $opt{tmp} as temporary directory.\n";
}
}
if(!$TMP && -d $def_tmp && -r _ && -w _) {
$TMP = $def_tmp;
}
print "$prompt: Temporary directory is $TMP\n" if($opt{debug});
=item B<-topjustify> [B<x>]I<num>
Add padding whitespace to the image so that it gets a defined height.
If an integer value is given, it defines the total height. The whitespace
is added at the bottom. If the number is preceded by "x", then this
multiple of the image height is added as whitespace at the bottom.
=cut
my $TOP_JUSTIFY = 0; # No top justifying by default
if($opt{topjustify}) {
$TOP_JUSTIFY = $opt{topjustify};
die <<"EOF" unless $TOP_JUSTIFY =~ /^x?\d+[.]?\d*$/i;
$prompt: Error: Illegal align specified: "$TOP_JUSTIFY"
Value must be positive numeric, optionally preceded by x
EOF
}
=item B<-transparent>
Generate transparent bitmaps, i.e. the background color (white) is
transparent if viewed with certain viewers (e.g. browsers). This option
may not be available due to missing capabilities of external
programs.
=cut
my $TRANSPARENT = 0; # Do not make make images transparent by default
if($opt{transparent}) {
$TRANSPARENT = 1;
}
=item B<-type> I<type>
Instruct pstoimg to render the bitmap in I<type> format. Depending on
the local installation, pstoimg is capable of generating either GIF or
PNG bitmaps. This site features the following types: png
If omitted, the first type in this list is taken.
=cut
if($opt{type}) {
$opt{type} = lc($opt{type});
die <<"EOF" unless grep($_ eq $opt{type},@IMAGE_TYPES);
$prompt: Error: This version of pstoimg does not support
"$opt{type}" image format.
EOF
}
else {
($opt{type}) = @IMAGE_TYPES; # default image type
}
# Support -gif and -png for a transition period
if($opt{gif}) {
print qq{$prompt: Warning: The -gif switch is deprecated. Use "-type gif" instead.\n};
if(grep($_ eq 'gif',@IMAGE_TYPES)) {
$opt{type} = 'gif';
}
else {
die <<"EOF";
$prompt: Error: This version of pstoimg does not support "gif" format.
EOF
}
}
if($opt{png}) {
print qq{$prompt: Warning: The -png switch is deprecated. Use "-type png" instead.\n};
if(grep($_ eq 'png',@IMAGE_TYPES)) {
$opt{type} = 'png';
}
else {
die <<"EOF";
$prompt: Error: This version of pstoimg does not support "png" format.
EOF
}
}
=item B<-white>
Remove TeX's page color information from the PostScript file before
converting so that a white background is used.
=back
=cut
# do some consistency checks on the options
if($TRANSPARENT && $opt{type} eq 'gif') {
print "$prompt: Warning: Transparent GIFs not supported.\n";
}
die <<"EOF" if($RIGHT_JUSTIFY && $CENTER);
$prompt: Error: Conflicting options -center and -rightjustify.
EOF
# now setup some parameters
# calculate dpi resolution from density and scale
$DENSITY = int($opt{scale} * $DENSITY + .5) if($opt{scale} != 1);
my $reduce_color = '';
if($opt{depth} == 1) {
$reduce_color = "$PPMQUANT 2";
}
elsif ($opt{depth} == 8) {
$reduce_color = "$PPMQUANT 256";
}
my $gs_aalias = '';
if($opt{antialias}) {
$GSDEVICE = $GSALIASDEVICE;
if($opt{depth} == 1) {
$gs_aalias = '-dTextAlphaBits=4 ';
$reduce_color = "$PPMQUANT -floyd 256";
}
else {
$gs_aalias = '-dTextAlphaBits=4 -dGraphicsAlphaBits=4 ';
}
}
elsif ($opt{aaliastext}) {
$GSDEVICE = $GSALIASDEVICE;
$gs_aalias = '-dTextAlphaBits=4 ';
$reduce_color = "$PPMQUANT -floyd 256";
}
my $PAPERSIZE = $ENV{PAPERSIZE} || '';
# This rx matches float values in Bounding Box expressions
my $Brx = '-?\d+(?:\.\d*|)';
##############################################################################
# Main program
=head1 DESCRIPTION
B<pstoimg> iterates over the given input files and runs them through
Ghostscipt. The resulting pnm (portable anymap files) are processed
with different Netpbm tools (cropping, color mapping, aligning, ...)
and finally converted into (currently) either GIF or PNG format. The
bitmaps can now be included e.g. in WWW pages.
The PostScript file is converted as is. If a valid bounding box is
found (EPS format), then only this area is converted. The image is
I<not> cropped by default.
=cut
die "$prompt: Error: No input file(s) specified\n"
unless(@ARGV);
# suppress diagnostics messages if possible
my $NULLFILE = '/dev/null';
open(STDERR, ">$NULLFILE") unless($opt{debug});
my $exit = 0;
$opt{out} = '' if(@ARGV > 1); # disable -out if multiple ps files given
my $psfile;
foreach $psfile (@ARGV) {
unless (-f $psfile) {
print qq{$prompt: Error: Cannot find file "$psfile": $!\n};
exit 1;
}
$exit += (&pstoimg($psfile) ? 0 : 1);
}
=head1 RETURN VALUE
=over 4
=item 0
if everything went all right
=item x
(x != 0) something went wrong. See the message output.
=back
=cut
exit $exit ? 1 : 0;
##############################################################################
# Subroutines
sub pstoimg {
my ($psfile) = @_;
print "$prompt: Processing $psfile\n" unless($opt{quiet});
# remove a trailing suffix the same way a shell would do it
my $base = $psfile;
$base =~ s|[.][^.$dd$dd]*$||;
my $outfile;
if($opt{out}) {
$outfile = $opt{out};
# append the type unless -outfile has a "." in it
$outfile .= ".$opt{type}" unless($outfile =~ /[.]/);
}
else {
$outfile = "$base.$opt{type}";
}
# Invoke Ghostscript
my $pnmdir = $TMP ? "$TMP$dd" : ".$dd";
my $pnmbase = "p$$"; # keep it short for dos
my $pnmfile = $pnmdir .
($opt{multipage} ? "%d_${pnmbase}.pnm" : "$pnmbase.pnm");
ps2pnm($psfile,$pnmfile) || return 0;
my $ok = 1;
if (-f $pnmfile) {
if(crop_scale_etc($pnmfile, $outfile)) {
L2hos->Unlink($pnmfile) unless($opt{debug});
}
else {
return 0;
}
}
elsif($opt{multipage}) {
unless(opendir(DIR,$pnmdir)) {
print qq{$prompt: Error: Could not open directory "$pnmdir": $!\n};
return 0;
}
my @list = grep(/^\d+_\w*\./,readdir(DIR));
closedir(DIR);
if(@list) {
my $i;
foreach $i (@list) {
my ($n) = $i =~ /^(\d+)_/;
my $j = $outfile;
$j =~ s|(\.[^/.]*)$|$n$1|;
if(crop_scale_etc("$pnmdir$i", $j)) {
L2hos->Unlink("$pnmdir$i") unless($opt{debug});
}
else {
$ok = 0;
}
}
}
else {
goto not_found;
}
}
else {
not_found:
print "$prompt: Error: Couldn't find pnm output of $psfile\n";
}
L2hos->Unlink($psfile) if($opt{discard} && !$opt{debug});
$ok;
}
sub ps2pnm {
my ($psfile,$pnmfile) = @_;
my $gs_size = $PAPERSIZE ? "-sPAPERSIZE=$PAPERSIZE" : '';
my $gs_density = ($DENSITY != 72) ? "-r$DENSITY" : '';
my ($bbx, $bby, $bbw, $bbh) = (0,0,0,0);
my $max_lines = 100;
my ($epsf,$have_geometry) = (0,0);
# Parse postscript file for information
unless(open(PS, "<$psfile")) {
print qq{$prompt: Error: Cannot read "$psfile": $!};
return 0;
}
$_ = <PS>; # read one line
if( /^%!.*EPSF/ ) {
# we're in a EPSF file
$epsf = 1;
}
if($GEOMETRY || $epsf) {
while (defined ($_ = <PS>)) {
# Look for bounding box comment
# MRO: add support of precise bounding boxes
if ($epsf && (
/^%+(?:HiRes|Exact)BoundingBox:\s+($Brx)\s+($Brx)\s+($Brx)\s+($Brx)/o ||
/^\%\%BoundingBox:\s+(-?\d+)\s+(-?\d+)\s+(-?\d+)\s+(-?\d+)/)) {
$bbx = 0 - $1; $bby = 0 - $2;
$bbw = $3 + $bbx; $bbh = $4 + $bby;
if(($bbw > 0) && ($bbh > 0)) { # we have a valid bounding box
print "$prompt: EPSF dimensions are ${bbw}x$bbh\n" if($opt{debug});
# this overrides the -geometry switch
if($DENSITY) { # scale the output
my $scale = $DENSITY / 72.0;
$bbw *= $scale;
$bbh *= $scale;
}
$bbw = int($bbw + 0.99);
$bbh = int($bbh + 0.99);
$GEOMETRY = "${bbw}x${bbh}";
$have_geometry = 1;
last;
}
}
# Look for page size information
elsif($GEOMETRY && /TeXDict\s+begin\s+(\d+)\s+(\d+)\s+/) {
$PAGE_WIDTH = int($1 / 65536 * 72 /72.27 +.5);
$PAGE_HEIGHT = int($2 / 65536 * 72 /72.27 +.5);
print "$prompt: Page dimensions are ${PAGE_WIDTH}x$PAGE_HEIGHT\n"
if($opt{debug});
# we don't have to look further for EPSF stuff at this point
last;
}
elsif(!$GEOMETRY && (/^\%\%EndComments/ || --$max_lines == 0)) {
# abort at a certain point to avoid scanning huge ps files
last;
}
}
}
close PS;
if($GEOMETRY && !$have_geometry) { # RRM: overrides $PAPERSIZE
# no geometry info found in the Postscript file
$bbx = $PAGE_HMARGIN;
$bby = $PAGE_HEIGHT - $PAGE_VMARGIN;
unless($GEOMETRY =~ /\s*(\d+)x(\d+)/i) {
print qq{$prompt: Illegal geometry "$GEOMETRY" specified.\n};
return 0;
}
$bbw = $1 + 10; # allow for the side-bars
$bbh = $2;
$bby = int(-$bby + $bbh + 8); # allow small margin for error
$bbx = int(-$bbx + 5); # allow small margin for error
if($DENSITY) {
my $scale = $DENSITY / 72.0;
$bbw = int($scale * $bbw + .99);
$bbh = int($scale * $bbh + .99);
}
$bbw += 10; # add a 5pt margin for safety
$bbh += 40; # add a 20pt margin for safety
$GEOMETRY = "${bbw}x$bbh";
$have_geometry = 1;
}
if($have_geometry) {
$gs_size = "-g$GEOMETRY ";
}
my $ps_changed = 0;
if($have_geometry || $opt{white}) {
# Remove any Postscript commands concerning Papersize if -g switch is used
# thanks to Axel Ramge for identifying the problem and for this code
local($/) = undef;
open(PS,"<$psfile");
my $ps = <PS>;
close(PS);
my $had_papersize;
if($have_geometry) {
$had_papersize = ($ps =~ s/\n%%BeginPaperSize.*?%%EndPaperSize[^\n]*\n/\n/sg);
}
my $had_nonwhite;
if($opt{white}) {
$had_nonwhite = ($ps =~ s/(\n\d+ \d+ bop gsave) \d*\.\d+ (TeXcolorgray clippath fill grestore)/$1 1 $2/s);
}
$ps_changed = $had_papersize || $had_nonwhite;
if($ps_changed) {
my $tmppsfile = $pnmfile; # was "tmpps$$.ps"
$tmppsfile =~ s/\.[^.]*$/.ps/;
unless(open(PS,">$tmppsfile") && (print PS $ps) && (close PS)) {
if($had_papersize) {
print <<"EOF";
$prompt: Warning: Could not write "$tmppsfile": $!
"$psfile" contains %%Papersize comments.
Any of these should be removed else GS will fail.
EOF
}
if($had_nonwhite) {
print <<"EOF";
$prompt: Warning: Could not write "$tmppsfile": $!
"$psfile" has a non-white background.
This may cause ugly images.
EOF
}
}
$psfile = $tmppsfile;
print qq{Debug: Papersize comment in "$psfile" deleted.\n}
if($had_papersize && $opt{debug});
print qq{Debug: Background switched to white in "$psfile".\n}
if($had_nonwhite && $opt{debug});
}
}
my $gs_quiet = $opt{debug} ? '' : '-q -dNOPAUSE -dNO_PAUSE';
my $out_redirect = $opt{debug} ? '' : "> $NULLFILE";
my $gs_out = "-sOutputFile=$pnmfile";
my $gsfile = $psfile;
# Ghostscript understands only '/' as path delimiter!
if($opt{debug}) {
print "$prompt: Running $GS $gs_quiet -sDEVICE=$GSDEVICE $gs_size $gs_density $gs_aalias $gs_out $out_redirect\n";
print "GS>$bbx $bby translate\n" if($have_geometry);
print "GS>($gsfile) run\n";
print "GS>showpage\n" if ($epsf);
print "GS>quit\n";
}
open (GS, "|$GS $gs_quiet -sDEVICE=$GSDEVICE $gs_size $gs_density $gs_aalias $gs_out $out_redirect");
print GS "$bbx $bby translate\n" if ($have_geometry);
print GS "($gsfile) run\n";
print GS "showpage\n" if ($epsf);
print GS "quit\n";
print "\n" if($opt{debug});
unless(close(GS)) {
print "$prompt: Error: Ghostscript returned error status ",$?>>8,"\n";
}
L2hos->Unlink($psfile) if($ps_changed && !$opt{debug});
1;
}
# This sub post-processes the PNM images that come out of Ghostscript.
# The image is cropped, flipped and finally converted to PNG or GIF.
sub crop_scale_etc {
my ($in, $out) = @_;
# create temp filename; should be auto-incrementable
my $tmp = $in;
$tmp =~ s/(\.[^.]*)?$/.t00/;
# save the original Ghostscript result
if($opt{debug}) {
L2hos->Copy($in,$tmp);
&increment_name($tmp);
}
my ($cmd,$type,$width,$height,$just);
my $must_align = 0;
#$EXTRA_CROP = "a$EXTRA_CROP" # hack to ensure first all-over cropping
# unless($EXTRA_CROP =~ /^a/i);
# RRM: Remove justification bars
$EXTRA_CROP =~ s/h/bt/gi; # crop horizontally
$EXTRA_CROP =~ s/v/rl/gi; # crop vertically
while ($EXTRA_CROP =~ /([atblrs])/gi) {
my $edge = $1;
my $croparg = '';
if($edge =~ /b/i) {
$croparg = '-bot -sides ';
} elsif($edge =~ /[tlr]/i) {
$croparg = "-$edge -sides ";
} elsif($edge =~ /s/i) {
#RRM: shave at most 1-2 rows of white from the bottom
if($cmd) {
# Terminate command pipe
&do_cmd($in,$tmp,$cmd) || return 0; # failure
$cmd = '';
}
my ($type,$width,$height) = get_image_geometry($in);
next unless($type); # skip if no geometry
if(&do_cmd_norename("$PNMCROP -bot -sides < $in",$tmp)) {
my ($type,$width,$height2) = get_image_geometry($tmp);
if($type && ($height - $height2) < 3 ) {
# command succeeded and shaved less than 3 rows
if($opt{debug}) {
L2hos->Copy($tmp,$in);
&increment_name($tmp);
} else {
L2hos->Rename($tmp,$in);
}
next;
}
}
# MRO: this shouldn't be necessary: L2hos->Unlink($tmp);
next; # go to next crop argument
} # end switch on crop codes
if($cmd) {
# Continue command pipe
$cmd .= "| $PNMCROP $croparg";
} else {
# start new pipe
$cmd = "$PNMCROP $croparg< $in ";
}
} # end cropping
if($opt{flip}) {
unless($cmd) {
$cmd = "$PNMFLIP -$opt{flip} < $in";
} else {
$cmd .= "| $PNMFLIP -$opt{flip} ";
}
}
if($RIGHT_JUSTIFY || $TOP_JUSTIFY || $CENTER || $SHORE_UP) {
if($cmd) {
# empty command pipe, we need the image's geometry
&do_cmd($in,$tmp,$cmd);
$cmd='';
}
# Get bitmap type and dimensions
($type,$width,$height) = &get_image_geometry($in);
return 0 unless($type);
my ($white_left,$white_right,$white_top,$white_bottom) = (0,0,0,0);
if($RIGHT_JUSTIFY || $CENTER) {
if($RIGHT_JUSTIFY) {
$white_left = int($RIGHT_JUSTIFY-$width);
} else { # CENTER
$white_left = int(($CENTER-$width) / 2);
}
$white_left = 0 unless($white_left > 0);
$width += $white_left;
}
if($TOP_JUSTIFY) {
if($TOP_JUSTIFY =~ /^x([0-9.]+)/io) {
$white_bottom = $1 * $height;
} else {
$white_bottom = $TOP_JUSTIFY - $height;
}
if($white_bottom > 0) {
$white_bottom = int($white_bottom + 0.99); # round up
$height += $white_bottom;
} else {
$white_bottom = 0;
}
}
if($SHORE_UP =~ /(\d+)(d?)/ && $1) {
# RRM: make height and width an exact multiple of $SHORE_UP
my ($shoreup,$depth) = ($1,$2);
my $extra = $height % $shoreup;
if($depth) { # image needs depth, place half the extra space underneath
my $bextra = int($extra/2);
$white_bottom += $bextra;
$white_top += $extra - $bextra;
} else {
$white_top += $extra;
}
$extra = $width % $shoreup;
my $rextra = int($extra/2);
$white_right += $rextra;
$white_left += $extra - $rextra;
$cmd = '';
}
if($white_left) {
if($cmd) {
&do_cmd($in,$tmp,$cmd) || return 0;
}
# Start new command pipe
$cmd = "$PBMMAKE -white $white_left 1 | $PNMCAT -white -lr - $in ";
}
if($white_right) {
if($cmd) {
&do_cmd($in,$tmp,$cmd) || return 0;
}
# Start new command pipe
$cmd = "$PBMMAKE -white $white_right 1 | $PNMCAT -white -lr $in - ";
}
if($white_top) {
if($cmd) {
&do_cmd($in,$tmp,$cmd) || return 0;
}
# Start new command pipe
$cmd = "$PBMMAKE -white 1 $white_top | $PNMCAT -white -tb - $in ";
}
if($white_bottom) {
if($cmd) {
&do_cmd($in,$tmp,$cmd) || return 0;
}
# Start new command pipe
$cmd = "$PBMMAKE -white 1 $white_bottom | $PNMCAT -white -tb $in - ";
}
} # endif must_align
my $pnmtoimg;
if($opt{type} eq 'png') {
$pnmtoimg = $PNMTOPNG;
if($INTERLACE) {
$pnmtoimg .= ' -interlace';
}
if($TRANSPARENT) {
$pnmtoimg .= ' -trans ' . L2hos->quote($trans_color);
}
}
unless($pnmtoimg) {
print qq($prompt: Error: unknown image type "$opt{type}".\n);
exit 2;
}
unless($type) {
($type,$width,$height) = &get_image_geometry($in);
return 0 unless($type);
}
# run ppmquant only on color/gray images
if(!$type || $type =~ /(ppm|pgm)/i) {
if($cmd) {
$cmd .= "| $reduce_color "
} else {
$cmd = "$reduce_color < $in ";
}
}
if($cmd) {
$cmd .= "| $pnmtoimg "
} else {
$cmd = "$pnmtoimg < $in ";
}
&do_cmd_norename($cmd,$out) || return 0;
print qq{$prompt: Written $out\n} unless($opt{quiet});
1;
}
sub banner {
print "$prompt V$RELEASE (Revision $VERSION, Perl $])\n";
}
sub print_version {
my $formats = join(',',@IMAGE_TYPES);
print <<"EOM";
$prompt (Revision $VERSION, perl $]),
part of LaTeX2HTML Release V$RELEASE.
Supported output image format(s): $formats
EOM
1;
}
sub print_help {
L2hos->perldoc($SCRIPT);
1;
}
sub print_usage {
my $start = 0;
my $usage = 'Usage: ';
my $indent = '';
print (@_, "\n") if @_;
my $perldoc = '/usr/bin'.$dd."perldoc";
my $script = $SCRIPT || $0;
open(PIPE, "$perldoc -t $script |")
|| die "Fatal: can't open pipe: $!";
while (<PIPE>)
{
if (/^\s*$/) {
next;
} elsif (/^SYNOPSIS/) {
$start = 1;
} elsif (/^\w/) {
$start = 0;
} elsif ($start == 1) {
($indent) = /^(\s*)/;
s/^$indent/$usage/;
$usage =~ s/./ /g;
$start = 2;
print $_;
} elsif ($start == 2) {
s/^$indent/$usage/;
print $_;
}
}
close PIPE;
1;
}
sub do_cmd {
my ($in,$tmp,$cmd) = @_;
print qq{Running "$cmd > $tmp"\n} if($opt{debug});
my $stat = system("$cmd > $tmp");
if($stat) { # error
print qq{$prompt: Error: "$cmd > $tmp" failed: $!\n};
return 0; # failure
}
elsif(!-s $tmp) { # does not exist or zero size
print qq{$prompt: Error: "$cmd > $tmp" produced empty file\n};
L2hos->Unlink($tmp) if(-e $tmp);
return 0; # failure
}
if($opt{debug}) {
# increase the temporary filename by 1
# this uses perl's magic autoincrement
&increment_name($_[1]);
return L2hos->Copy($tmp,$in);
} elsif(!L2hos->Rename($tmp,$in)) {
print qq{$prompt: Error: rename of "$tmp" to "$in" failed: $!\n};
return 0; # failure
}
1;
}
sub do_cmd_norename {
my ($cmd,$out) = @_;
print qq{Running "$cmd > $out"\n} if($opt{debug});
my $stat = system("$cmd > $out");
if($stat) { # error
print qq{$prompt: Error: "$cmd > $out" failed: $!\n};
return 0; # failure
}
elsif(!-s $out) { # does not exist or zero size
print qq{$prompt: Error: "$cmd > $out" produced empty file\n};
L2hos->Unlink($out) if(-e $out);
return 0; # failure
}
1;
}
sub do_cmd_plain {
my ($cmd) = @_;
print qq{Running "$cmd"\n} if($opt{debug});
my $stat = system($cmd);
if($stat) { # error
print qq{$prompt: Error: "$cmd" failed: $!\n};
return 0; # failure
}
1;
}
sub get_image_geometry {
my ($pnmfile) = @_;
my ($type,$width,$height);
my $out = `$PNMFILE $pnmfile`;
if($? || $out =~ /(P[BGP]M)[^0-9]*(\d+)\s*by\s*(\d+)/i) {
$type = $1;
$width = $2;
$height = $3;
print qq{Image "$pnmfile" is $type, ${width}x$height\n} if($opt{debug});
} else {
print "$prompt: Error: Could not determine image size: $out\n";
return undef;
}
($type,$width,$height);
}
# push the number in the suffix up one notch
sub increment_name {
$_[0] =~ s/(\d+)$/$a=$1;++$a/e;
}
__DATA__
=head1 EXAMPLES
=over 4
=item C<pstoimg foo.ps>
Convert the first page of foo.ps to the default bitmap type.
=item C<pstoimg -type png -crop a -trans -interlace foo.ps>
Same as above, but force png output and crop all the whitespace
around the image and make the color white transparent and
generate an interlaced bitmap.
=item C<pstoimg -multi -out bar -type gif -crop a foo.ps>
Consider foo.ps a multiple page PostScript file and create output
files bar1.gif, bar2.gif, etc.
=back
=head1 ENVIRONMENT
=over 4
=item DENSITY, DEPTH, DEBUG, DISCARD
See B<-density>, B<-depth>, B<-debug>, B<-discard>, respectively.
=item GS_LIB
This variable is set to the path(s) where Ghostscript libraries have
been found on this system during configuration, but only if the built-in
paths are not correct. This fixes the problem of relocation that is quite
common on Win32 installations. This behavior can be overridden by
setting GS_LIB manually before starting pstoimg.
=item LATEX2HTMLDIR
The directory where the LaTeX2HTML library and perl modules are found.
Defaults to "/usr/share/latex2html" on this installation.
=item OUTFILE
Setting this has the same effect as specifying B<-out>. Please do not rely
on this feature any more, it will disappear from the next releases!
=item PAPERSIZE
The papersize to use by Ghostscript to render the image. pstoimg tries
hard to optimize for rendering on the smallest possible bitmap size.
Still this option is there to enable tuning by hand, although it is
deprecated. If pstoimg finds a better setting, this parameter is ignored.
=item SCALE
See the discussion of B<-scale>.
=item TMP and TEMP
Unless overridden by B<-tmp>, these variables denote a directory where
to store temporary files. TMP is considered first, then TEMP.
=back
=head1 SEE ALSO
gs, pnmcrop, pnmquant, pbmmake, pnmcat, pnmfile, pnmflip, ppmtogif,
pnmtopng, giftool, giftrans.
=head1 NOTES
Several people have suggested to use ImageMagick's convert instead of
pstoimg. A few comments on this: convert uses (of course) Ghostscript
for conversion of PostScript to bitmap, so one still needs gs. And
for the special requirements of LaTeX2HTML convert's features are not
sufficient. The ImageMagick toolset has everything in place, but it
has some overhead that can prove killing when processing some 100
images. pstoimg only does what it really has to, so it should be
quite efficient. Don't get me wrong - I like ImageMagick, but not in
the context of LaTeX2HTML.
=head1 CAVEATS
This utility is automatically configured and built to work on the
local setup. If this setup changes (e.g. some of the external commands
are moved), the script has be be reconfigured.
Despite the portability of perl, a pstoimg configured on UNIX will
probably not work on Win32 and vice versa.
=head1 BUGS
This is a major enhancement release, so there may be a few bugs. As
the user inteface changed a bit, some of your tools that were using
pstoimg may not work any more.
Please report bugs to latex2html@tug.org, stating the (debug) output
of pstoimg, your perl version and the versions of the external tools.
Best is to include the cfgcache.pm file from the configuration procedure.
=head1 AUTHOR
Marek Rouchal E<lt>marek@saftsack.fs.uni-bayreuth.deE<gt>
=head1 HISTORY
This script went through a long evolution, beginning with a modification
of Doug Crabill's E<lt>dgc@cs.purdue.eduE<gt> ps2epsi script.
The first perl version was done by Nikos Drakos <nikos@cbl.leeds.ac.uk>.
It was gradually improved by numerous LaTeX2HTML developers:
Ross Moore <ross@mpce.mq.edu.au>, Jens Lippmann
<lippmann@rbg.informatik.tu-darmstadt.de> and others (sorry for not
mentioning everyone and thanks for your contributions).
=cut