2010-07-22 23:09:39 +02:00
|
|
|
#!/usr/bin/perl
|
|
|
|
|
|
|
|
|
|
use warnings;
|
|
|
|
|
use strict;
|
|
|
|
|
|
|
|
|
|
use HTML::Entities;
|
|
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
my $debug = 9999;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
open FH, "<n1256.txt" or die "Could not open n1256.txt: $!";
|
2012-01-16 16:34:32 +01:00
|
|
|
#open FH, "<n1570.txt" or die "Could not open n1570.txt: $!";
|
2010-07-22 23:09:39 +02:00
|
|
|
my @contents = <FH>;
|
|
|
|
|
close FH;
|
|
|
|
|
|
|
|
|
|
my $text = join '', @contents;
|
|
|
|
|
$text =~ s/\r//g;
|
|
|
|
|
|
|
|
|
|
print "<html>\n<body>\n";
|
|
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
my ($section_title, $this_section);
|
2010-07-22 23:09:39 +02:00
|
|
|
|
|
|
|
|
my $last_section_number = 0;
|
|
|
|
|
my $section_number = 0;
|
2012-01-17 01:08:26 +01:00
|
|
|
my $last_section;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
while($text =~ m/^\s{0,4}([0-9A-Z]+\.[0-9\.]*)/msg) {
|
2010-07-22 23:09:39 +02:00
|
|
|
$last_section_number = $section_number;
|
2012-01-17 01:08:26 +01:00
|
|
|
$last_section = $this_section;
|
2010-07-22 23:09:39 +02:00
|
|
|
$this_section = $1;
|
|
|
|
|
|
|
|
|
|
($section_number) = $this_section =~ /([^.]+)\./;
|
|
|
|
|
|
2012-01-16 16:34:32 +01:00
|
|
|
print STDERR "----------------------------------\n" if $debug >= 2;
|
|
|
|
|
print STDERR "Processing section [$this_section]; number [$section_number]\n" if $debug;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
print STDERR "this_section: [$this_section]; last_section: [$last_section]\n";
|
|
|
|
|
print STDERR "Section diff: ", ($this_section - $last_section), "\n";
|
2010-07-22 23:09:39 +02:00
|
|
|
|
|
|
|
|
my $diff = $section_number - $last_section_number;
|
2012-01-16 16:34:32 +01:00
|
|
|
print STDERR "Diff: $diff\n" if $debug >= 2;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
|
|
|
|
if($section_number > 0 and $diff < 0 or $diff > 1) {
|
2012-01-17 01:08:26 +01:00
|
|
|
print STDERR "Diff out of bounds: $diff\n";
|
|
|
|
|
last;
|
2010-07-22 23:09:39 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
my $section_text;
|
|
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
if($text =~ m/(.*?)^(?=\s{0,4}[0-9A-Z]+\.)/msg) {
|
2010-07-22 23:09:39 +02:00
|
|
|
$section_text = $1;
|
|
|
|
|
} else {
|
2012-01-16 16:34:32 +01:00
|
|
|
print STDERR "No section text, end of file marker found.\n" if $debug >= 4;
|
2010-07-22 23:09:39 +02:00
|
|
|
last;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if($section_text =~ m/(.*?)$/msg) {
|
|
|
|
|
$section_title = $1 if length $1;
|
|
|
|
|
$section_title =~ s/^\s+//;
|
|
|
|
|
$section_title =~ s/\s+$//;
|
|
|
|
|
}
|
|
|
|
|
|
2012-01-16 16:34:32 +01:00
|
|
|
print STDERR "$this_section [$section_title]\n" if $debug >= 2;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
|
|
|
|
print "<hr>\n";
|
|
|
|
|
print "<a name='$this_section'>";
|
|
|
|
|
print "<h3>$this_section [$section_title]</h3>";
|
|
|
|
|
print "</a>\n";
|
|
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
print STDERR "section text: [$section_text]\n" if $debug >= 2;
|
|
|
|
|
|
|
|
|
|
if(not $section_text =~ m/^(?=\d+\s)/msg) {
|
|
|
|
|
print "<pre>$section_text</pre>\n";
|
|
|
|
|
} else {
|
|
|
|
|
while($section_text =~ m/^(\d+)\s(.*?)^(?=\d)/msgc or $section_text =~ m/^(\d+)\s(.*)/msg) {
|
|
|
|
|
my $p = $1;
|
|
|
|
|
my $t = $2;
|
|
|
|
|
|
|
|
|
|
print STDERR "paragraph $p: [$t]\n" if $debug >= 3;
|
2010-07-22 23:09:39 +02:00
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
$t = encode_entities($t);
|
2010-07-22 23:09:39 +02:00
|
|
|
|
2012-01-17 01:08:26 +01:00
|
|
|
print "<a name='$this_section" . "p$p'>";
|
|
|
|
|
print "<pre>$p $t</pre>\n";
|
|
|
|
|
print "</a>\n";
|
|
|
|
|
}
|
2010-07-22 23:09:39 +02:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
print "\n</body>\n</html>\n";
|
