1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # reader for mbox variants we support
5 package PublicInbox::MboxReader;
9 $Data::Dumper::Useqq = 1; # should've been the default, for bad data
12 qr/^From \S+ +\S+ \S+ +\S+ [^\n:]+:[^\n:]+:[^\n:]+ [^\n:]+\n/sm;
15 my ($mbfh, $from_re, $eml_cb, @arg) = @_;
18 while (defined(my $r = read($mbfh, $buf, 65536, length($buf)))) {
19 if ($r == 0) { # close here to check for "curl --fail"
20 close($mbfh) or die "error closing mbox: \$?=$? $!";
23 @raw = split(/$from_strict/mos, $buf, -1);
24 next if scalar(@raw) == 0;
25 $buf = pop(@raw); # last bit may be incomplete
27 @raw = grep /[^ \t\r\n]/s, @raw; # skip empty messages
28 while (defined(my $raw = shift @raw)) {
29 $raw =~ s/^\r?\n\z//ms;
30 $raw =~ s/$from_re/$1/gms;
31 my $eml = PublicInbox::Eml->new(\$raw);
32 $eml_cb->($eml, @arg);
34 return if $r == 0; # EOF
36 die "error reading mboxo/mboxrd handle: $!";
40 my (undef, $mbfh, $eml_cb, @arg) = @_;
41 _mbox_from($mbfh, qr/^>(>*From )/ms, $eml_cb, @arg);
45 my (undef, $mbfh, $eml_cb, @arg) = @_;
46 _mbox_from($mbfh, qr/^>(From )/ms, $eml_cb, @arg);
50 my ($mbfh, $bref, $cl) = @_;
51 my $body = substr($$bref, 0, $cl, '');
52 my $need = $cl - length($body);
54 $mbfh or die "E: needed $need bytes after EOF";
55 defined(my $r = read($mbfh, $body, $need, length($body))) or
56 die "E: read error: $!\n";
57 $r == $need or die "E: read $r of $need bytes\n";
64 if (index($$ref, "\r\n") < 0 && (my $pos = index($$ref, "\n\n")) >= 0) {
66 \substr($$ref, 0, $pos + 2, ''); # sv_chop on $$ref
67 } elsif ($$ref =~ /\r?\n\r?\n/s) {
68 \substr($$ref, 0, $+[0], ''); # sv_chop on $$ref
74 sub _mbox_cl ($$$;@) {
75 my ($mbfh, $uxs_from, $eml_cb, @arg) = @_;
77 while (defined(my $r = read($mbfh, $buf, 65536, length($buf)))) {
78 if ($r == 0) { # detect "curl --fail"
80 die "error closing mboxcl/mboxcl2: \$?=$? $!";
83 while (my $hdr = _extract_hdr(\$buf)) {
84 $$hdr =~ s/\A[\r\n]*From [^\n]*\n//s or
85 die "E: no 'From ' line in:\n", Dumper($hdr);
86 my $eml = PublicInbox::Eml->new($hdr);
87 my @cl = $eml->header_raw('Content-Length');
89 $n == 0 and die "E: Content-Length missing in:\n",
90 Dumper($eml->as_string);
91 $n == 1 or die "E: multiple ($n) Content-Length in:\n",
92 Dumper($eml->as_string);
93 $cl[0] =~ /\A[0-9]+\z/ or die
94 "E: Content-Length `$cl[0]' invalid\n",
95 Dumper($eml->as_string);
96 if (($eml->{bdy} = _cl_body($mbfh, \$buf, $cl[0]))) {
98 ${$eml->{bdy}} =~ s/^>From /From /sgm;
100 $eml_cb->($eml, @arg);
103 $buf =~ /[^ \r\n\t]/ and
104 warn "W: leftover at end of mboxcl/mboxcl2:\n",
109 die "error reading mboxcl/mboxcl2 handle: $!";
113 my (undef, $mbfh, $eml_cb, @arg) = @_;
114 _mbox_cl($mbfh, 1, $eml_cb, @arg);
118 my (undef, $mbfh, $eml_cb, @arg) = @_;
119 _mbox_cl($mbfh, undef, $eml_cb, @arg);
122 sub new { bless \(my $x), __PACKAGE__ }