#!/hgdv/software/IRIX6/bin/perl # CVSHTMLDIFF # # Does a 'cvs diff' on a HTML file # and produces a new HTML file showing the differences # highlighted by coloring. # # Synopsis: # cvshtmldiff [options..] file.html # -o out.html output html file (default is file_diff.html) # -r rev do the diff against this revision (default: head of branch) # -q don't give feedback about which revision has been retrieved # # Notice: # It can be arbitrarily complex to produce correct HTML from a diff! # Therefore, no attempt is being made to do that. ;-) # If the results are unacceptable, you might want to try 'tidy' # on the resulting HTML file (can be obtained from w3.org). # # Note also, that some changes in HTML files may not be visible! # (such as inserting an anchor) # Such changes will not be visible in the resulting HTML file either. # # A legend ist inserted right after the . # If there is no tag, then there will be no legend ;-) # # Author: # Gabriel Zachmann (zach@igd.fhg.de,Gabriel.Zachmann@gmx.net) # Feb 2000. # # TODO: # - es geht nur zeilenweise, d.h. es wird mehr markiert als sich wirklich # geaendert hat. Kann man das fixen dass er da noch Anfang und Ende # mitvergleicht? # #################################### # Things you might want to customize %regionstart = ( "d" => "\n", "a" => "\n", "c" => "\n" ); # "d" => "\n", %regionend = ( "d" => "\n", "a" => "\n", "c" => "\n" ); # Warning: regionstart and regionend will be inserted inside .., too! $legend = "

red     = deleted<\/FONT>
green   = appended<\/FONT>
purple  = changed<\/FONT>

\n"; # #################################### require 5.004; require "getopts.pl"; # parse options / parameters &Getopts('hqo:r:'); $infile = shift; if ( $opt_h || $infile eq "" ) { print "usage: cvshtmldiff [options..] file.html\n"; print "-o out.html output html file (default is file_diff.html)\n"; print "-r rev do the diff against this revision (default: head of branch)\n"; print "-q don't give feedback about which revision has been retrieved\n"; print "-h this help\n"; exit 0; } if ( $opt_r ) { $rev = "-r$opt_r"; } else { $rev = ""; } if ( $opt_o ) { $outfile = $opt_o; } else { ($outfile,$dummy) = $infile =~ /^(.*?)(.html?)?$/; $outfile .= "_diff.html"; } if ( length($outfile) == 0 ) { print STDERR "cvshtmldiff: no output file!\n"; exit 1; } $tmpdir = $ENV{"TMPDIR"}; if ( ! $tmpdir ) { $tmpdir = "/tmp"; } # do diff $difffile = $tmpdir . "/cvshtmldiff-$$"; $err = system( "cvs diff $opt_r $infile > $difffile" ); # parse 'cvs diff' and merge it into htmlfile if ( ! open(OUTFILE, "> $outfile") ) { die "\ncvshtmldiff: could not open output file $outfile: $!\n\n"; } if ( ! open(INFILE, "< $infile") ) { die "\ncvshtmldiff: could not open input file $infile: $!\n\n"; } if ( ! open(DIFFFILE, "< $difffile") ) { die "\ncvshtmldiff: could not open diff file $difffile: $!\n\n"; } # main $line = 0; # $line = # lines read from infile and copied to outfile DIFF: while ( $chunk = ) { # read difffile up to next diff chunk while ( ! ( ($m1,$m2,$op,$n1,$n2) = $chunk =~ m/^(\d+)(?:,(\d+))?([adc])(\d+)(?:,(\d+))?$/o ) ) { if ( $chunk =~ m/^retrieving revision/o && ! $opt_q ) { # feedback print $chunk; } if ( ! ($chunk = ) ) { last DIFF; } } # sanity check if ( $m1 eq "" || $n1 eq "" || $op eq "" ) { print STDERR "\ncvshtmldiff: Error: couldn't parse chunk\n"; print STDERR $chunk; print STDERR "from diff file $difffile!\n"; next; } if ( $op eq "a" || $op eq "c" ) { # append or change (text ist already in infile) if ( $op eq "a" && $m2 ) { print STDERR "\ncvshtmldiff: Warning: append chunk $chunk\n"; print STDERR "has 2 line numbers w/ respect to file 1!\n"; print STDERR "Don't know how to handle that.\n"; } $err = xfer_upto( $n1-1 ); if ( $err ) { last; } # copy changed/appended text if ( ! $n2 ) { $n2 = $n1; } $err = xfer_upto( $n2, $regionstart{$op}, $regionend{$op} ); if ( $err ) { last; } } elsif ( $op eq "d" ) { # delete (deleted text is in diff file) if ( $n2 ) { print STDERR "\ncvshtmldiff: Warning: delete chunk $chunk\n"; print STDERR "has 2 line numbers w/ respect to file 2!\n"; print STDERR "Don't know how to handle that.\n"; } $err = xfer_upto( $n1 ); if ( $err ) { last; } # copy deleted text to output file if ( ! $m2 ) { $m2 = $m1; } $err = copy_diff( $m2-$m1+1, $regionstart{$op}, $regionend{$op} ); if ( $err ) { last; } } else { print STDERR "\ncvshtmldiff: Error: don't know about operator "; print STDERR "\'$op\'\nin chunk $chunk of diff file $difffile!\n"; next; } } # transfer rest of infile to outfile while ( ) { print OUTFILE $_; } close OUTFILE; close INFILE; close DIFFFILE; exit 0; # # Copy lines from infile to outfile # up to, and including, line $1. # The first line in a file is # 1. # Enclose block of copied lines in $rs .. $re. # Global variables: $line # sub xfer_upto { my ($upto,$rs,$re) = @_; if ( $line > $upto ) # = is ok, can happen { print STDERR "\ncvshtmldiff: xfer_upto: Bug: upto ($upto) < "; print STDERR "line ($line) on entry!\n"; return 1; } if ( $rs ) { print OUTFILE $rs; } while ( $line < $upto ) { $_ = ; if ( $_ eq "" ) { print STDERR "\ncvshtmldiff: Error: input file $infile "; print STDERR "exhausted!\n"; print STDERR "while trying to process chunk $chunk\n"; print STDERR "from diff file $difffile\n"; return 1; } # insert legend s/(]*>)/$1\n$legend\n/oi; if ( $rs ) { # insert regionstart/-end inside s/(]*>)/$1$rs/ogi; s/(<\/TD>)/$re$1/ogi; } print OUTFILE $_; $line ++ ; } # print region-end twice, just in case we got out of sync if ( $re ) { print OUTFILE $re; print OUTFILE $re; } return 0; } # copy $1 many lines from difffile to outfile # Enclose block of copied lines in $rs .. $re. sub copy_diff { my ($l,$rs,$re) = @_; if ( $l <= 0 ) { print STDERR "\ncvshtmldiff: copy_upto: Bug: #lines ($l) <= 0!\n "; return 1; } if ( $rs ) { print OUTFILE $rs; } while ( $l > 0 ) { $_ = ; if ( $_ eq "" ) { print STDERR "\ncvshtmldiff: Error: diff file $difffile "; print STDERR "exhausted!\n"; print STDERR "while trying to append deleted lines $m1-$m2!\n"; print STDERR "of chunk $chunk from diff file $difffile\n"; return 1; } s/< //o; if ( $rs ) { # insert regionstart/-end inside s/(]*>)/$1$rs/ogi; s/(<\/TD>)/$re$1/ogi; } print OUTFILE $_; $l -- ; } # print region-end twice, just in case we got out of sync if ( $re ) { print OUTFILE $re; print OUTFILE $re; } return 0; }