Blame

Date:
Message:
refactor the parsing into the module change the mscan format string to siplify the parsing, and refactor the parsing code into the module for reuse. pe is an exception in that it doesn't care about the format string, it just need to decide if the current line starts a new thread or not, that's why it doesn't use the newly introduced `parse'. The new format simplifies pe too though.
Actions:
History | Blob | Raw File
 1 7d6d378a 2022-08-24 op #!/usr/bin/env perl

 2 7d6d378a 2022-08-24 op 

 3 7d6d378a 2022-08-24 op use open ":std", ":encoding(UTF-8)";

 4 7d6d378a 2022-08-24 op use utf8;

 5 7d6d378a 2022-08-24 op use strict;

 6 7d6d378a 2022-08-24 op use warnings;

 7 7d6d378a 2022-08-24 op use v5.32;

 8 7d6d378a 2022-08-24 op 

 9 054f3fd4 2022-08-25 op use OpenBSD::Pledge;

10 054f3fd4 2022-08-25 op use OpenBSD::Unveil;

11 054f3fd4 2022-08-25 op 

12 04eab9af 2022-08-25 op use lib ".";

13 de557185 2022-08-26 op use GotMArc qw(parse san $logo initpage endpage);

14 04eab9af 2022-08-25 op 

15 7d6d378a 2022-08-24 op my $outdir = $ENV{'OUTDIR'};

16 7d6d378a 2022-08-24 op die 'Set $OUTDIR' unless defined $outdir;

17 7d6d378a 2022-08-24 op 

18 054f3fd4 2022-08-25 op unveil("/usr/local/bin/mshow", "rx") or die "unveil mshow: $!";

19 054f3fd4 2022-08-25 op unveil($outdir, "rwc") or die "unveil $outdir: $!";

20 054f3fd4 2022-08-25 op unveil(".", "rwc") or die "unveil .: $!";

21 054f3fd4 2022-08-25 op pledge("stdio rpath wpath cpath proc exec") or die "pledge: $!";

22 054f3fd4 2022-08-25 op 

23 7d6d378a 2022-08-24 op my $tid;

24 7d6d378a 2022-08-24 op while (<>) {

25 de557185 2022-08-26 op 	my ($level, $fname, $mid, $date, $from, $subj) = parse;

26 7d6d378a 2022-08-24 op 

27 7d6d378a 2022-08-24 op 	$tid = $mid if $level == 0;

28 f3e4455d 2022-08-25 op 	die "unknown tid" unless defined $tid;

29 7d6d378a 2022-08-24 op 

30 c971fcee 2022-08-25 op 	my $dest = "$outdir/mail/$mid.html";

31 c971fcee 2022-08-25 op 	next if -f $dest;

32 c971fcee 2022-08-25 op 

33 7d6d378a 2022-08-24 op 	open(my $fh, '>', "$dest") or die "can't open $dest: $!";

34 7d6d378a 2022-08-24 op 

35 04eab9af 2022-08-25 op 	initpage $fh, $subj;

36 7d6d378a 2022-08-24 op 

37 f1ceade9 2022-08-24 op 	# prepare the parts listing file

38 f1ceade9 2022-08-24 op 	$ENV{'MESSAGE_ID'} = $mid;

39 f1ceade9 2022-08-24 op 	open(my $parts, '+>', "parts.html")

40 f1ceade9 2022-08-24 op 	    or die "can't create parts.html: $!";

41 f1ceade9 2022-08-24 op 

42 7d6d378a 2022-08-24 op 	open(my $mshow, "-|", "mshow", "-nNA", "text/plain", $fname)

43 7d6d378a 2022-08-24 op 	    or die "can't exec mshow: $!";

44 7d6d378a 2022-08-24 op 

45 84989e10 2022-08-24 op 	open(my $text, '>', "$outdir/text/$mid.txt")

46 84989e10 2022-08-24 op 	    or die "can't open $outdir/text/$mid.txt: $!";

47 84989e10 2022-08-24 op 

48 7d6d378a 2022-08-24 op 	print $fh "<header class='mail-header'>";

49 7d6d378a 2022-08-24 op 	print $fh "<p>";

50 7d6d378a 2022-08-24 op 	print $fh $logo;

51 4a89bbd3 2022-08-25 op 	print $fh "<a href='/'>Index</a>";

52 4a89bbd3 2022-08-25 op 	print $fh " | ";

53 4a89bbd3 2022-08-25 op 	print $fh "<a href='/thread/$tid.html#$mid'>Thread</a>";

54 7d6d378a 2022-08-24 op 	print $fh "</p>";

55 7d6d378a 2022-08-24 op 	print $fh "<dl>";

56 7d6d378a 2022-08-24 op 	while (<$mshow>) {

57 7d6d378a 2022-08-24 op 		chomp;

58 84989e10 2022-08-24 op 		say $text $_;

59 7d6d378a 2022-08-24 op 		last if /^$/;

60 7d6d378a 2022-08-24 op 		my ($h, $v) = m/^([-A-Za-z]+): (.*)/;

61 7d6d378a 2022-08-24 op 		die "bogus line? $fname : $_" unless (defined $h and defined $v);

62 7d6d378a 2022-08-24 op 

63 7d6d378a 2022-08-24 op 		# drop the (1 day ago) string

64 7d6d378a 2022-08-24 op 		$v =~ s/\(.*\)//g if ($h eq "Date");

65 7d6d378a 2022-08-24 op 

66 7d6d378a 2022-08-24 op 		print $fh "<dt>", san($h), ":</dt>";

67 7d6d378a 2022-08-24 op 		print $fh "<dd>", san($v), "</dd>";

68 7d6d378a 2022-08-24 op 	}

69 ed91b1e1 2022-08-24 op 	print $fh "</dl>";

70 a53202df 2022-08-25 op 	print $fh "<p>Download raw <a href='/text/$mid.txt'>body</a>.</p>";

71 ed91b1e1 2022-08-24 op 	print $fh "</header>";

72 7d6d378a 2022-08-24 op 

73 7d6d378a 2022-08-24 op 	print $fh "<pre>";

74 5b01d758 2022-08-25 op 	while (<$mshow>) {

75 5b01d758 2022-08-25 op 		print $text $_;

76 5b01d758 2022-08-25 op 		print $fh san($_);

77 5b01d758 2022-08-25 op 	}

78 7d6d378a 2022-08-24 op 	print $fh "</pre>";

79 7d6d378a 2022-08-24 op 

80 f1ceade9 2022-08-24 op 	# generate the listing for the exported parts

81 f1ceade9 2022-08-24 op 	my $part_seen = 0;

82 f1ceade9 2022-08-24 op 	while (<$parts>) {

83 f1ceade9 2022-08-24 op 		if (!$part_seen) {

84 f1ceade9 2022-08-24 op 			$part_seen = 1;

85 f1ceade9 2022-08-24 op 			say $fh "<ul class='parts'>";

86 f1ceade9 2022-08-24 op 		}

87 f1ceade9 2022-08-24 op 		print $fh $_;

88 f1ceade9 2022-08-24 op 	}

89 f1ceade9 2022-08-24 op 	say $fh "</ul>" if $part_seen;

90 f1ceade9 2022-08-24 op 

91 04eab9af 2022-08-25 op 	endpage $fh;

92 7d6d378a 2022-08-24 op 

93 84989e10 2022-08-24 op 	close($text);

94 7d6d378a 2022-08-24 op 	close($mshow);

95 f1ceade9 2022-08-24 op 	close($parts);

96 7d6d378a 2022-08-24 op 	close($fh);

97 7d6d378a 2022-08-24 op }

98 f1ceade9 2022-08-24 op 

99 f1ceade9 2022-08-24 op unlink "parts.html";