X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FMID.pm;h=369bb034c8dd9d5c03ce4beb2e35d1c5fc632a9c;hb=9cdeab04ebd95c7f778ba5f258399377b180f2cc;hp=96085399b6d2f2d96ba9bef3d06376a953132953;hpb=4030525cb228eb3837f5260637bd7a5a861e81e2;p=public-inbox.git
diff --git a/lib/PublicInbox/MID.pm b/lib/PublicInbox/MID.pm
index 96085399..369bb034 100644
--- a/lib/PublicInbox/MID.pm
+++ b/lib/PublicInbox/MID.pm
@@ -1,4 +1,4 @@
-# Copyright (C) 2015-2018 all contributors
+# Copyright (C) 2015-2020 all contributors
# License: AGPL-3.0+
#
# Various Message-ID-related functions.
@@ -6,27 +6,33 @@ package PublicInbox::MID;
use strict;
use warnings;
use base qw/Exporter/;
-our @EXPORT_OK = qw/mid_clean id_compress mid2path mid_mime mid_escape MID_ESC
- mids references/;
+our @EXPORT_OK = qw(mid_clean id_compress mid2path mid_escape MID_ESC
+ mids references mids_for_index $MID_EXTRACT);
use URI::Escape qw(uri_escape_utf8);
use Digest::SHA qw/sha1_hex/;
-use constant MID_MAX => 40; # SHA-1 hex length
+require PublicInbox::Address;
+use constant {
+ ID_MAX => 40, # SHA-1 hex length for HTML id anchors
+ MAX_MID_SIZE => 244, # max term size (Xapian limitation) - length('Q')
+};
+
+our $MID_EXTRACT = qr/<([^>]+)>/s;
sub mid_clean {
my ($mid) = @_;
defined($mid) or die "no Message-ID";
# MDA->precheck did more checking for us
- if ($mid =~ /<([^>]+)>/) {
+ if ($mid =~ $MID_EXTRACT) {
$mid = $1;
}
$mid;
}
-# this is idempotent
+# this is idempotent, used for HTML anchor/ids and such
sub id_compress {
my ($id, $force) = @_;
- if ($force || $id =~ /[^\w\-]/ || length($id) > MID_MAX) {
+ if ($force || $id =~ /[^a-zA-Z0-9_\-]/ || length($id) > ID_MAX) {
utf8::encode($id);
return sha1_hex($id);
}
@@ -47,21 +53,34 @@ sub mid2path {
"$x2/$x38";
}
-sub mid_mime ($) { $_[0]->header_obj->header_raw('Message-ID') }
-
-sub mids ($) {
- my ($hdr) = @_;
+# only intended for Message-ID and X-Alt-Message-ID
+sub extract_mids {
my @mids;
- my @v = $hdr->header_raw('Message-Id');
- foreach my $v (@v) {
- my @cur = ($v =~ /<([^>]+)>/sg);
+ for my $v (@_) {
+ my @cur = ($v =~ /$MID_EXTRACT/g);
if (@cur) {
push(@mids, @cur);
} else {
push(@mids, $v);
}
}
- uniq_mids(\@mids);
+ \@mids;
+}
+
+sub mids ($) {
+ my ($hdr) = @_;
+ my @mids = $hdr->header_raw('Message-ID');
+ uniq_mids(extract_mids(@mids));
+}
+
+# we allow searching on X-Alt-Message-ID since PublicInbox::NNTP uses them
+# to placate some clients, and we want to ensure NNTP-only clients can
+# import and index without relying on HTTP endpoints
+sub mids_for_index ($) {
+ my ($hdr) = @_;
+ my @mids = $hdr->header_raw('Message-ID');
+ my @alts = $hdr->header_raw('X-Alt-Message-ID');
+ uniq_mids(extract_mids(@mids, @alts));
}
# last References should be IRT, but some mail clients do things
@@ -72,20 +91,35 @@ sub references ($) {
foreach my $f (qw(References In-Reply-To)) {
my @v = $hdr->header_raw($f);
foreach my $v (@v) {
- push(@mids, ($v =~ /<([^>]+)>/sg));
+ push(@mids, ($v =~ /$MID_EXTRACT/g));
}
}
- uniq_mids(\@mids);
+
+ # old versions of git-send-email would prompt users for
+ # In-Reply-To and users' muscle memory would use 'y' or 'n'
+ # as responses:
+ my %addr = ( y => 1, n => 1 );
+
+ foreach my $f (qw(To From Cc)) {
+ my @v = $hdr->header_raw($f);
+ foreach my $v (@v) {
+ $addr{$_} = 1 for (PublicInbox::Address::emails($v));
+ }
+ }
+ uniq_mids(\@mids, \%addr);
}
-sub uniq_mids ($) {
- my ($mids) = @_;
+sub uniq_mids ($;$) {
+ my ($mids, $seen) = @_;
my @ret;
- my %seen;
- foreach (@$mids) {
- next if $seen{$_};
- push @ret, $_;
- $seen{$_} = 1;
+ $seen ||= {};
+ foreach my $mid (@$mids) {
+ $mid =~ tr/\n\t\r//d;
+ if (length($mid) > MAX_MID_SIZE) {
+ warn "Message-ID: <$mid> too long, truncating\n";
+ $mid = substr($mid, 0, MAX_MID_SIZE);
+ }
+ push(@ret, $mid) unless $seen->{$mid}++;
}
\@ret;
}