-# Copyright (C) 2015, all contributors <meta@public-inbox.org>
+# Copyright (C) 2015 all contributors <meta@public-inbox.org>
# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
# based on notmuch, but with no concept of folders, files or flags
package PublicInbox::Search;
use strict;
use warnings;
-use PublicInbox::SearchMsg;
-use base qw/Exporter/;
-use Digest::SHA qw//;
+use constant TS => 0;
use Search::Xapian qw/:standard/;
-require PublicInbox::View;
-use Date::Parse qw/str2time/;
-use POSIX qw//;
+use PublicInbox::SearchMsg;
use Email::MIME;
+use PublicInbox::MID qw/mid_clean mid_compress/;
-our @EXPORT = qw/xpfx mid_compressed/;
+# This is English-only, everything else is non-standard and may be confused as
+# a prefix common in patch emails
+our $REPLY_RE = qr/^re:\s+/i;
+our $LANG = 'english';
use constant {
- TS => 0,
- SHA1HEX_LEN => 40,
- SCHEMA_VERSION => 0,
- LANG => 'english',
+ # SCHEMA_VERSION history
+ # 0 - initial
+ # 1 - subject_path is lower-cased
+ # 2 - subject_path is mid_compress in the index, only
+ # 3 - message-ID is compressed if it includes '%' (hack!)
+ # 4 - change "Re: " normalization, avoid circular Reference ghosts
+ # 5 - subject_path drops trailing '.'
+ # 6 - preserve References: order in document data
+ # 7 - remove references and inreplyto terms
+ # 8 - remove redundant/unneeded document data
+ # 9 - disable Message-ID compression
+ SCHEMA_VERSION => 9,
+
+ # n.b. FLAG_PURE_NOT is expensive not suitable for a public website
+ # as it could become a denial-of-service vector
QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
};
-use constant MID_MAX => SHA1HEX_LEN;
-
# setup prefixes
my %bool_pfx_internal = (
type => 'T', # "mail" or "ghost"
- mid => 'Q', # uniQue id (Message-ID or mid_compressed)
+ thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
);
my %bool_pfx_external = (
path => 'XPATH',
- thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
- references => 'XREFS',
- inreplyto => 'XIRT',
+ mid => 'Q', # uniQue id (Message-ID)
);
my %prob_prefix = (
subject => 'S',
+ s => 'S', # for mairix compatibility
+ m => 'Q', # 'mid' is exact, 'm' can do partial
);
my %all_pfx = (%bool_pfx_internal, %bool_pfx_external, %prob_prefix);
sub xpfx { $all_pfx{$_[0]} }
our %PFX2TERM_RMAP;
+my %meta_pfx = (mid => 1, thread => 1, path => 1);
while (my ($k, $v) = each %all_pfx) {
- next if $prob_prefix{$k};
- $PFX2TERM_RMAP{$v} = $k;
+ $PFX2TERM_RMAP{$v} = $k if $meta_pfx{$k};
}
my $mail_query = Search::Xapian::Query->new(xpfx('type') . 'mail');
-# this is idempotent
-sub mid_compressed {
- my ($mid) = @_;
- return $mid if (length($mid) <= MID_MAX);
- Digest::SHA::sha1_hex($mid);
+sub xdir {
+ my (undef, $git_dir) = @_;
+ "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
}
sub new {
- my ($class, $git_dir, $writable) = @_;
- # allow concurrent versions for easier rollback:
- my $dir = "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
- my $db;
-
- if ($writable) { # not used by the WWW interface
- require Search::Xapian::WritableDatabase;
- require File::Path;
- File::Path::mkpath($dir);
- $db = Search::Xapian::WritableDatabase->new($dir,
- Search::Xapian::DB_CREATE_OR_OPEN);
- } else {
- $db = Search::Xapian::Database->new($dir);
- }
+ my ($class, $git_dir) = @_;
+ my $dir = $class->xdir($git_dir);
+ my $db = Search::Xapian::Database->new($dir);
bless { xdb => $db, git_dir => $git_dir }, $class;
}
sub reopen { $_[0]->{xdb}->reopen }
-sub add_message {
- my ($self, $mime) = @_; # mime = Email::MIME object
- my $db = $self->{xdb};
-
- my $doc_id;
- my $mid = clean_mid($mime->header('Message-ID'));
- $mid = mid_compressed($mid);
- my $was_ghost = 0;
- my $ct_msg = $mime->header('Content-Type') || 'text/plain';
- my $enc_msg = PublicInbox::View::enc_for($ct_msg);
-
- $db->begin_transaction;
- eval {
- my $smsg = $self->lookup_message($mid);
- my $doc;
-
- if ($smsg) {
- $smsg->ensure_metadata;
- # convert a ghost to a regular message
- # it will also clobber any existing regular message
- $smsg->mime($mime);
- $doc = $smsg->{doc};
-
- my $type = xpfx('type');
- eval {
- $doc->remove_term($type . 'ghost');
- $was_ghost = 1;
- };
-
- # probably does not exist:
- eval { $doc->remove_term($type . 'mail') };
- $doc->add_term($type . 'mail');
- } else {
- $smsg = PublicInbox::SearchMsg->new($mime);
- $doc = $smsg->{doc};
- $doc->add_term(xpfx('mid') . $mid);
- }
-
- my $subj = $mime->header('Subject');
- $subj = '' unless defined $subj;
-
- if (length $subj) {
- $doc->add_term(xpfx('subject') . $subj);
-
- my $path = subject_path($subj);
- $doc->add_term(xpfx('path') . $path);
- }
-
- my $from = $mime->header('From') || '';
- my @from;
-
- if ($from) {
- @from = Email::Address->parse($from);
- $from = $from[0]->name;
- }
-
- my $ts = eval { str2time($mime->header('Date')) } || 0;
- my $date = POSIX::strftime('%Y-%m-%d %H:%M', gmtime($ts));
- $ts = Search::Xapian::sortable_serialise($ts);
- $doc->add_value(PublicInbox::Search::TS, $ts);
-
- # this is what we show in index results:
- $subj =~ tr/\n/ /;
- $from =~ tr/\n/ /;
- $doc->set_data("$mid\n$subj\n$from\n$date");
-
- my $tg = $self->term_generator;
-
- $tg->set_document($doc);
- $tg->index_text($subj, 1, 'S') if $subj;
- $tg->increase_termpos;
- $tg->index_text($subj) if $subj;
- $tg->increase_termpos;
-
- if (@from) {
- $tg->index_text($from[0]->format);
- $tg->increase_termpos;
- }
-
- $mime->walk_parts(sub {
- my ($part) = @_;
- return if $part->subparts; # walk_parts already recurses
- my $ct = $part->content_type || $ct_msg;
-
- # account for filter bugs...
- $ct =~ m!\btext/plain\b!i or return;
-
- my $enc = PublicInbox::View::enc_for($ct, $enc_msg);
- my (@orig, @quot);
- foreach my $l (split(/\n/, $enc->decode($part->body))) {
- if ($l =~ /^\s*>/) {
- push @quot, $l;
- } else {
- push @orig, $l;
- }
- }
- if (@quot) {
- $tg->index_text(join("\n", @quot), 0);
- $tg->increase_termpos;
- }
- if (@orig) {
- $tg->index_text(join("\n", @orig));
- $tg->increase_termpos;
- }
- });
-
- if ($was_ghost) {
- $doc_id = $smsg->doc_id;
- $self->link_message($smsg, 0);
- $db->replace_document($doc_id, $doc);
- } else {
- $self->link_message($smsg, 0);
- $doc_id = $db->add_document($doc);
- }
- };
-
- if ($@) {
- warn "failed to index message <$mid>: $@\n";
- $db->cancel_transaction;
- } else {
- $db->commit_transaction;
- }
- $doc_id;
-}
-
-# returns deleted doc_id on success, undef on missing
-sub remove_message {
- my ($self, $mid) = @_;
- my $db = $self->{xdb};
- my $doc_id;
- $mid = clean_mid($mid);
- $mid = mid_compressed($mid);
-
- $db->begin_transaction;
- eval {
- $doc_id = $self->find_unique_doc_id('mid', $mid);
- $db->delete_document($doc_id) if defined $doc_id;
- };
-
- if ($@) {
- warn "failed to remove message <$mid>: $@\n";
- $db->cancel_transaction;
- } else {
- $db->commit_transaction;
- }
- $db->commit;
- $doc_id;
-}
-
# read-only
sub query {
my ($self, $query_string, $opts) = @_;
- my $query = $self->qp->parse_query($query_string, QP_FLAGS);
+ my $query;
+
+ $opts ||= {};
+ unless ($query_string eq '') {
+ $query = $self->qp->parse_query($query_string, QP_FLAGS);
+ $opts->{relevance} = 1 unless exists $opts->{relevance};
+ }
- $query = Search::Xapian::Query->new(OP_AND, $mail_query, $query);
$self->do_enquire($query, $opts);
}
-# given a message ID, get replies to a message
-sub get_replies {
- my ($self, $mid, $opts) = @_;
- $mid = clean_mid($mid);
- $mid = mid_compressed($mid);
- my $qp = $self->qp;
- my $irt = $qp->parse_query("inreplyto:$mid", 0);
- my $ref = $qp->parse_query("references:$mid", 0);
- my $query = Search::Xapian::Query->new(OP_OR, $irt, $ref);
-
- $self->do_enquire($query);
+sub get_subject_path {
+ my ($self, $path, $opts) = @_;
+ my $q = Search::Xapian::Query->new(xpfx("path").mid_compress($path));
+ $self->do_enquire($q, $opts);
}
sub get_thread {
my ($self, $mid, $opts) = @_;
my $smsg = eval { $self->lookup_message($mid) };
- return { count => 0, msgs => [] } unless $smsg;
- my $qp = $self->qp;
- my $qtid = $qp->parse_query('thread:'.$smsg->thread_id);
- my $qsub = $qp->parse_query('path:'.$smsg->path);
+ return { total => 0, msgs => [] } unless $smsg;
+ my $qtid = Search::Xapian::Query->new(xpfx('thread').$smsg->thread_id);
+ my $path = mid_compress($smsg->path);
+ my $qsub = Search::Xapian::Query->new(xpfx('path').$path);
my $query = Search::Xapian::Query->new(OP_OR, $qtid, $qsub);
- $self->do_enquire($query);
+ $self->do_enquire($query, $opts);
}
# private subs below
sub do_enquire {
my ($self, $query, $opts) = @_;
my $enquire = $self->enquire;
-
+ if (defined $query) {
+ $query = Search::Xapian::Query->new(OP_AND,$query,$mail_query);
+ } else {
+ $query = $mail_query;
+ }
$enquire->set_query($query);
- $enquire->set_sort_by_relevance_then_value(TS, 0);
+ if ($opts->{relevance}) {
+ $enquire->set_sort_by_relevance_then_value(TS, 1);
+ } else {
+ $enquire->set_sort_by_value_then_relevance(TS, 1);
+ }
$opts ||= {};
my $offset = $opts->{offset} || 0;
my $limit = $opts->{limit} || 50;
my $mset = $enquire->get_mset($offset, $limit);
- my @msgs = map { $_->get_document->get_data } $mset->items;
+ return $mset if $opts->{mset};
+ my @msgs = map {
+ PublicInbox::SearchMsg->load_doc($_->get_document);
+ } $mset->items;
- { count => $mset->get_matches_estimated, msgs => \@msgs }
+ { total => $mset->get_matches_estimated, msgs => \@msgs }
}
# read-write
-sub stemmer { Search::Xapian::Stem->new(LANG) }
+sub stemmer { Search::Xapian::Stem->new($LANG) }
# read-only
sub qp {
$self->{query_parser} = $qp;
}
-sub term_generator { # write-only
- my ($self) = @_;
-
- my $tg = $self->{term_generator};
- return $tg if $tg;
-
- $tg = Search::Xapian::TermGenerator->new;
- $tg->set_stemmer($self->stemmer);
-
- $self->{term_generator} = $tg;
-}
-
-sub next_doc_id { $_[0]->{xdb}->get_lastdocid + 1 }
-
-# increments last_thread_id counter
-# returns a 64-bit integer represented as a hex string
-sub next_thread_id {
- my ($self) = @_;
- my $db = $self->{xdb};
- my $last_thread_id = int($db->get_metadata('last_thread_id') || 0);
-
- $db->set_metadata('last_thread_id', ++$last_thread_id);
-
- $last_thread_id;
-}
-
sub ts_range_processor {
$_[0]->{tsrp} ||= Search::Xapian::NumberValueRangeProcessor->new(TS);
}
$_[0]->{drp} ||= Search::Xapian::DateValueRangeProcessor->new(TS);
}
-sub clean_mid {
- my ($mid) = @_;
- defined($mid) or die "no Message-ID";
- # MDA->precheck did more checking for us
- $mid =~ s/\A\s*<?//;
- $mid =~ s/>?\s*\z//;
- $mid;
-}
-
-sub link_message {
- my ($self, $smsg, $is_ghost) = @_;
-
- if ($is_ghost) {
- $smsg->ensure_metadata;
- } else {
- $self->link_message_to_parents($smsg);
- }
-}
-
-sub link_message_to_parents {
- my ($self, $smsg) = @_;
- my $doc = $smsg->{doc};
- my $mid = mid_compressed($smsg->mid);
- my $mime = $smsg->mime;
- my $refs = $mime->header('References');
- my @refs = $refs ? ($refs =~ /<([^>]+)>/g) : ();
- my $irt = $mime->header('In-Reply-To');
- if ($irt) {
- if ($irt =~ /<([^>]+)>/) {
- $irt = $1;
- }
- push @refs, $irt;
- }
-
- my $tid;
- if (@refs) {
- @refs = map { mid_compressed($_) } @refs;
- my %uniq;
- @refs = grep { !$uniq{$_}++ } @refs; # uniq
-
- $doc->add_term(xpfx('inreplyto') . $refs[-1]);
-
- my $ref_pfx = xpfx('references');
-
- # first ref *should* be the thread root,
- # but we can never trust clients to do the right thing
- my $ref = shift @refs;
- $doc->add_term($ref_pfx . $ref);
- $tid = $self->_resolve_mid_to_tid($ref);
-
- # the rest of the refs should point to this tid:
- foreach $ref (@refs) {
- $doc->add_term($ref_pfx . $ref);
- my $ptid = $self->_resolve_mid_to_tid($ref);
- if ($tid ne $ptid) {
- $self->merge_threads($tid, $ptid);
- }
- }
- } else {
- $tid = $self->next_thread_id;
- }
- $doc->add_term(xpfx('thread') . $tid);
-}
-
sub lookup_message {
my ($self, $mid) = @_;
- $mid = clean_mid($mid);
- $mid = mid_compressed($mid);
+ $mid = mid_clean($mid);
my $doc_id = $self->find_unique_doc_id('mid', $mid);
my $smsg;
($db->postlist_begin($term), $db->postlist_end($term));
}
-# this will create a ghost as necessary
-sub _resolve_mid_to_tid {
- my ($self, $mid) = @_;
-
- my $smsg = $self->lookup_message($mid) || $self->create_ghost($mid);
- $smsg->thread_id;
-}
-
-sub create_ghost {
- my ($self, $mid, $tid) = @_;
-
- $mid = mid_compressed($mid);
- $tid = $self->next_thread_id unless defined $tid;
-
- my $doc = Search::Xapian::Document->new;
- $doc->add_term(xpfx('mid') . $mid);
- $doc->add_term(xpfx('thread') . $tid);
- $doc->add_term(xpfx('type') . 'ghost');
-
- my $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
- $self->link_message($smsg, 1);
- $self->{xdb}->add_document($doc);
-
- $smsg;
-}
-
-sub merge_threads {
- my ($self, $winner_tid, $loser_tid) = @_;
- my ($head, $tail) = $self->find_doc_ids('thread', $loser_tid);
- my $thread_pfx = xpfx('thread');
- my $db = $self->{xdb};
-
- for (; $head != $tail; $head->inc) {
- my $docid = $head->get_docid;
- my $doc = $db->get_document($docid);
- $doc->remove_term($thread_pfx . $loser_tid);
- $doc->add_term($thread_pfx . $winner_tid);
- $db->replace_document($docid, $doc);
- }
-}
-
# normalize subjects so they are suitable as pathnames for URLs
sub subject_path {
- my ($subj) = @_;
-
- $subj =~ s/\A\s+//;
- $subj =~ s/\s+\z//;
- $subj =~ s/^(?:re|aw):\s*//i; # remove reply prefix (aw: German)
+ my $subj = pop;
+ $subj = subject_normalized($subj);
$subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
- $subj;
-}
-
-sub do_cat_mail {
- my ($git, $blob) = @_;
- my $mime = eval {
- my $str = $git->cat_file($blob);
- Email::MIME->new($str);
- };
- $@ ? undef : $mime;
+ lc($subj);
}
-sub index_blob {
- my ($self, $git, $blob) = @_;
- my $mime = do_cat_mail($git, $blob) or return;
- eval { $self->add_message($mime) };
- warn "W: index_blob $blob: $@\n" if $@;
+sub subject_normalized {
+ my $subj = pop;
+ $subj =~ s/\A\s+//s; # no leading space
+ $subj =~ s/\s+\z//s; # no trailing space
+ $subj =~ s/\s+/ /gs; # no redundant spaces
+ $subj =~ s/\.+\z//; # no trailing '.'
+ $subj =~ s/$REPLY_RE//igo; # remove reply prefix
+ $subj;
}
-sub unindex_blob {
- my ($self, $git, $blob) = @_;
- my $mime = do_cat_mail($git, $blob) or return;
- my $mid = $mime->header('Message-ID');
- eval { $self->remove_message($mid) } if defined $mid;
- warn "W: unindex_blob $blob: $@\n" if $@;
+# for doc data
+sub subject_summary {
+ my $subj = pop;
+ my $max = 68;
+ if (length($subj) > $max) {
+ my @subj = split(/\s+/, $subj);
+ $subj = '';
+ my $l;
+
+ while ($l = shift @subj) {
+ my $new = $subj . $l . ' ';
+ last if length($new) >= $max;
+ $subj = $new;
+ }
+ if ($subj ne '') {
+ my $r = scalar @subj ? ' ...' : '';
+ $subj =~ s/ \z/$r/s;
+ } else {
+ # subject has one REALLY long word, and NOT spam? wtf
+ @subj = ($l =~ /\A(.{1,72})/);
+ $subj = $subj[0] . ' ...';
+ }
+ }
+ $subj;
}
sub enquire {
$self->{enquire} ||= Search::Xapian::Enquire->new($self->{xdb});
}
+sub mid_prefix {
+ my ($self, $mpfx) = @_;
+ my $query = eval { $self->qp->parse_query("m:$mpfx", FLAG_PARTIAL) };
+ return if $@;
+ my $res = $self->do_enquire($query, { relevance => 1 });
+ return unless $res->{total};
+ [ map { $_->mid } @{$res->{msgs}} ];
+}
+
1;