-# Copyright (C) 2015 all contributors <meta@public-inbox.org>
-# License: AGPLv3 or later (https://www.gnu.org/licenses/agpl-3.0.txt)
+# Copyright (C) 2015-2020 all contributors <meta@public-inbox.org>
+# License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
# based on notmuch, but with no concept of folders, files or flags
+#
+# Read-only search interface for use by the web and NNTP interfaces
package PublicInbox::Search;
use strict;
use warnings;
-use constant TS => 0;
-use Search::Xapian qw/:standard/;
-use PublicInbox::SearchMsg;
-use Email::MIME;
-use PublicInbox::MID qw/mid_clean mid_compress/;
+
+# values for searching
+use constant TS => 0; # Received: header in Unix time
+use constant YYYYMMDD => 1; # Date: header for searching in the WWW UI
+use constant DT => 2; # Date: YYYYMMDDHHMMSS
+
+use PublicInbox::Smsg;
+use PublicInbox::Over;
+my $QP_FLAGS;
+our %X = map { $_ => 0 } qw(BoolWeight Database Enquire
+ NumberValueRangeProcessor QueryParser Stem);
+our $Xap; # 'Search::Xapian' or 'Xapian'
+my $ENQ_ASCENDING;
+
+sub load_xapian () {
+ return 1 if defined $Xap;
+ for my $x (qw(Search::Xapian Xapian)) {
+ eval "require $x";
+ next if $@;
+
+ $x->import(qw(:standard));
+ $Xap = $x;
+ $X{$_} = $Xap.'::'.$_ for (keys %X);
+
+ # ENQ_ASCENDING doesn't seem exported by SWIG Xapian.pm,
+ # so lets hope this part of the ABI is stable because it's
+ # just an integer:
+ $ENQ_ASCENDING = $x eq 'Xapian' ?
+ 1 : Search::Xapian::ENQ_ASCENDING();
+
+ # for Smsg:
+ *PublicInbox::Smsg::sortable_unserialise =
+ $Xap.'::sortable_unserialise';
+ # n.b. FLAG_PURE_NOT is expensive not suitable for a public
+ # website as it could become a denial-of-service vector
+ # FLAG_PHRASE also seems to cause performance problems chert
+ # (and probably earlier Xapian DBs). glass seems fine...
+ # TODO: make this an option, maybe?
+ # or make indexlevel=medium as default
+ $QP_FLAGS = FLAG_PHRASE() | FLAG_BOOLEAN() | FLAG_LOVEHATE() |
+ FLAG_WILDCARD();
+ return 1;
+ }
+ undef;
+}
# 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 {
# SCHEMA_VERSION history
# 0 - initial
# 1 - subject_path is lower-cased
- # 2 - subject_path is mid_compress in the index, only
+ # 2 - subject_path is id_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,
+ # 9 - disable Message-ID compression (SHA-1)
+ # 10 - optimize doc for NNTP overviews
+ # 11 - merge threads when vivifying ghosts
+ # 12 - change YYYYMMDD value column to numeric
+ # 13 - fix threading for empty References/In-Reply-To
+ # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
+ # 14 - fix ghost root vivification
+ # 15 - see public-inbox-v2-format(5)
+ # further bumps likely unnecessary, we'll suggest in-place
+ # "--reindex" use for further fixes and tweaks
+ SCHEMA_VERSION => 15,
};
-# setup prefixes
-my %bool_pfx_internal = (
- type => 'T', # "mail" or "ghost"
- thread => 'G', # newsGroup (or similar entity - e.g. a web forum name)
-);
-
my %bool_pfx_external = (
- path => 'XPATH',
- mid => 'Q', # uniQue id (Message-ID)
+ mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
+ dfpre => 'XDFPRE',
+ dfpost => 'XDFPOST',
+ dfblob => 'XDFPRE XDFPOST',
);
+my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
my %prob_prefix = (
- subject => 'S',
- s => 'S', # for mairix compatibility
- m => 'Q', # 'mid' is exact, 'm' can do partial
+ # for mairix compatibility
+ s => 'S',
+ m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
+ f => 'A',
+ t => 'XTO',
+ tc => 'XTO XCC',
+ c => 'XCC',
+ tcf => 'XTO XCC A',
+ a => 'XTO XCC A',
+ b => $non_quoted_body . ' XQUOT',
+ bs => $non_quoted_body . ' XQUOT S',
+ n => 'XFN',
+
+ q => 'XQUOT',
+ nq => $non_quoted_body,
+ dfn => 'XDFN',
+ dfa => 'XDFA',
+ dfb => 'XDFB',
+ dfhh => 'XDFHH',
+ dfctx => 'XDFCTX',
+
+ # default:
+ '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
);
-my %all_pfx = (%bool_pfx_internal, %bool_pfx_external, %prob_prefix);
+# not documenting m: and mid: for now, the using the URLs works w/o Xapian
+our @HELP = (
+ 's:' => 'match within Subject e.g. s:"a quick brown fox"',
+ 'd:' => <<EOF,
+date range as YYYYMMDD e.g. d:19931002..20101002
+Open-ended ranges such as d:19931002.. and d:..20101002
+are also supported
+EOF
+ 'dt:' => <<EOF,
+date-time range as YYYYMMDDhhmmss (e.g. dt:19931002011000..19931002011200)
+EOF
+ 'b:' => 'match within message body, including text attachments',
+ 'nq:' => 'match non-quoted text within message body',
+ 'q:' => 'match quoted text within message body',
+ 'n:' => 'match filename of attachment(s)',
+ 't:' => 'match within the To header',
+ 'c:' => 'match within the Cc header',
+ 'f:' => 'match within the From header',
+ 'a:' => 'match within the To, Cc, and From headers',
+ 'tc:' => 'match within the To and Cc headers',
+ 'bs:' => 'match within the Subject and body',
+ 'dfn:' => 'match filename from diff',
+ 'dfa:' => 'match diff removed (-) lines',
+ 'dfb:' => 'match diff added (+) lines',
+ 'dfhh:' => 'match diff hunk header context (usually a function name)',
+ 'dfctx:' => 'match diff context lines',
+ 'dfpre:' => 'match pre-image git blob ID',
+ 'dfpost:' => 'match post-image git blob ID',
+ 'dfblob:' => 'match either pre or post-image git blob ID',
+);
+chomp @HELP;
-sub xpfx { $all_pfx{$_[0]} }
+sub xdir ($;$) {
+ my ($self, $rdonly) = @_;
+ if ($self->{ibx_ver} == 1) {
+ "$self->{inboxdir}/public-inbox/xapian" . SCHEMA_VERSION;
+ } else {
+ my $dir = "$self->{inboxdir}/xap" . SCHEMA_VERSION;
+ return $dir if $rdonly;
-our %PFX2TERM_RMAP;
-my %meta_pfx = (mid => 1, thread => 1, path => 1);
-while (my ($k, $v) = each %all_pfx) {
- $PFX2TERM_RMAP{$v} = $k if $meta_pfx{$k};
+ my $shard = $self->{shard};
+ defined $shard or die "shard not given";
+ $dir .= "/$shard";
+ }
}
-my $mail_query = Search::Xapian::Query->new(xpfx('type') . 'mail');
+sub _xdb ($) {
+ my ($self) = @_;
+ my $dir = xdir($self, 1);
+ my ($xdb, $slow_phrase);
+ my $qpf = \($self->{qp_flags} ||= $QP_FLAGS);
+ if ($self->{ibx_ver} >= 2) {
+ foreach my $shard (<$dir/*>) {
+ -d $shard && $shard =~ m!/[0-9]+\z! or next;
+ my $sub = $X{Database}->new($shard);
+ if ($xdb) {
+ $xdb->add_database($sub);
+ } else {
+ $xdb = $sub;
+ }
+ $slow_phrase ||= -f "$shard/iamchert";
+ }
+ } else {
+ $slow_phrase = -f "$dir/iamchert";
+ $xdb = $X{Database}->new($dir);
+ }
+ $$qpf |= FLAG_PHRASE() unless $slow_phrase;
+ $xdb;
+}
-sub xdir {
- my (undef, $git_dir) = @_;
- "$git_dir/public-inbox/xapian" . SCHEMA_VERSION;
+sub xdb ($) {
+ my ($self) = @_;
+ $self->{xdb} ||= do {
+ load_xapian();
+ _xdb($self);
+ };
}
sub new {
- my ($class, $git_dir) = @_;
- my $dir = $class->xdir($git_dir);
- my $db = Search::Xapian::Database->new($dir);
- bless { xdb => $db, git_dir => $git_dir }, $class;
+ my ($class, $ibx) = @_;
+ ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
+ my $self = bless {
+ inboxdir => $ibx->{inboxdir},
+ altid => $ibx->{altid},
+ ibx_ver => $ibx->version,
+ }, $class;
+ my $dir = xdir($self, 1);
+ $self->{over_ro} = PublicInbox::Over->new("$dir/over.sqlite3");
+ $self;
}
-sub reopen { $_[0]->{xdb}->reopen }
+sub reopen {
+ my ($self) = @_;
+ if (my $xdb = $self->{xdb}) {
+ $xdb->reopen;
+ }
+ $self; # make chaining easier
+}
# read-only
sub query {
my ($self, $query_string, $opts) = @_;
- my $query;
-
$opts ||= {};
- unless ($query_string eq '') {
- $query = $self->qp->parse_query($query_string, QP_FLAGS);
+ if ($query_string eq '' && !$opts->{mset}) {
+ $self->{over_ro}->recent($opts);
+ } else {
+ my $qp = qp($self);
+ my $qp_flags = $self->{qp_flags};
+ my $query = $qp->parse_query($query_string, $qp_flags);
$opts->{relevance} = 1 unless exists $opts->{relevance};
+ _do_enquire($self, $query, $opts);
}
-
- $self->do_enquire($query, $opts);
}
-sub get_subject_path {
- my ($self, $path, $opts) = @_;
- my $q = Search::Xapian::Query->new(xpfx("path").mid_compress($path));
- $self->do_enquire($q, $opts);
+sub retry_reopen {
+ my ($self, $cb, $arg) = @_;
+ for my $i (1..10) {
+ if (wantarray) {
+ my @ret;
+ eval { @ret = $cb->($arg) };
+ return @ret unless $@;
+ } else {
+ my $ret;
+ eval { $ret = $cb->($arg) };
+ return $ret unless $@;
+ }
+ # Exception: The revision being read has been discarded -
+ # you should call Xapian::Database::reopen()
+ if (ref($@) =~ /\bDatabaseModifiedError\b/) {
+ warn "reopen try #$i on $@\n";
+ reopen($self);
+ } else {
+ # let caller decide how to spew, because ExtMsg queries
+ # get wonky and trigger:
+ # "something terrible happened at .../Xapian/Enquire.pm"
+ die;
+ }
+ }
+ die "Too many Xapian database modifications in progress\n";
}
-sub get_thread {
- my ($self, $mid, $opts) = @_;
- my $smsg = eval { $self->lookup_message($mid) };
-
- 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, $opts);
+sub _do_enquire {
+ my ($self, $query, $opts) = @_;
+ retry_reopen($self, \&_enquire_once, [ $self, $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 = $mail_query;
- }
+sub _enquire_once { # retry_reopen callback
+ my ($self, $query, $opts) = @{$_[0]};
+ my $xdb = xdb($self);
+ my $enquire = $X{Enquire}->new($xdb);
$enquire->set_query($query);
- if ($opts->{relevance}) {
- $enquire->set_sort_by_relevance_then_value(TS, 1);
+ $opts ||= {};
+ my $desc = !$opts->{asc};
+ if (($opts->{mset} || 0) == 2) {
+ $enquire->set_docid_order($ENQ_ASCENDING);
+ $enquire->set_weighting_scheme($X{BoolWeight}->new);
+ } elsif ($opts->{relevance}) {
+ $enquire->set_sort_by_relevance_then_value(TS, $desc);
} else {
- $enquire->set_sort_by_value_then_relevance(TS, 1);
+ $enquire->set_sort_by_value_then_relevance(TS, $desc);
}
- $opts ||= {};
my $offset = $opts->{offset} || 0;
my $limit = $opts->{limit} || 50;
my $mset = $enquire->get_mset($offset, $limit);
return $mset if $opts->{mset};
- my @msgs = map {
- PublicInbox::SearchMsg->load_doc($_->get_document);
- } $mset->items;
+ my @msgs = map { PublicInbox::Smsg::from_mitem($_) } $mset->items;
+ return \@msgs unless wantarray;
- { total => $mset->get_matches_estimated, msgs => \@msgs }
+ ($mset->get_matches_estimated, \@msgs)
}
# read-write
-sub stemmer { Search::Xapian::Stem->new($LANG) }
+sub stemmer { $X{Stem}->new($LANG) }
# read-only
sub qp {
my $qp = $self->{query_parser};
return $qp if $qp;
-
+ my $xdb = xdb($self);
# new parser
- $qp = Search::Xapian::QueryParser->new;
- $qp->set_default_op(OP_AND);
- $qp->set_database($self->{xdb});
+ $qp = $X{QueryParser}->new;
+ $qp->set_default_op(OP_AND());
+ $qp->set_database($xdb);
$qp->set_stemmer($self->stemmer);
- $qp->set_stemming_strategy(STEM_SOME);
- $qp->add_valuerangeprocessor($self->ts_range_processor);
- $qp->add_valuerangeprocessor($self->date_range_processor);
+ $qp->set_stemming_strategy(STEM_SOME());
+ $qp->set_max_wildcard_expansion(100);
+ my $nvrp = $X{NumberValueRangeProcessor};
+ $qp->add_valuerangeprocessor($nvrp->new(YYYYMMDD, 'd:'));
+ $qp->add_valuerangeprocessor($nvrp->new(DT, 'dt:'));
while (my ($name, $prefix) = each %bool_pfx_external) {
- $qp->add_boolean_prefix($name, $prefix);
+ $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
}
- while (my ($name, $prefix) = each %prob_prefix) {
- $qp->add_prefix($name, $prefix);
+ # we do not actually create AltId objects,
+ # just parse the spec to avoid the extra DB handles for now.
+ if (my $altid = $self->{altid}) {
+ my $user_pfx = $self->{-user_pfx} = [];
+ for (@$altid) {
+ # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
+ # note: Xapian supports multibyte UTF-8, /^[0-9]+$/,
+ # and '_' with prefixes matching \w+
+ /\Aserial:(\w+):/ or next;
+ my $pfx = $1;
+ push @$user_pfx, "$pfx:", <<EOF;
+alternate serial number e.g. $pfx:12345 (boolean)
+EOF
+ # gmane => XGMANE
+ $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
+ }
+ chomp @$user_pfx;
}
- $self->{query_parser} = $qp;
-}
-
-sub ts_range_processor {
- $_[0]->{tsrp} ||= Search::Xapian::NumberValueRangeProcessor->new(TS);
-}
-
-sub date_range_processor {
- $_[0]->{drp} ||= Search::Xapian::DateValueRangeProcessor->new(TS);
-}
-
-sub lookup_message {
- my ($self, $mid) = @_;
- $mid = mid_clean($mid);
-
- my $doc_id = $self->find_unique_doc_id('mid', $mid);
- my $smsg;
- if (defined $doc_id) {
- # raises on error:
- my $doc = $self->{xdb}->get_document($doc_id);
- $smsg = PublicInbox::SearchMsg->wrap($doc, $mid);
- $smsg->doc_id($doc_id);
+ while (my ($name, $prefix) = each %prob_prefix) {
+ $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
}
- $smsg;
-}
-
-sub find_unique_doc_id {
- my ($self, $term, $value) = @_;
-
- my ($begin, $end) = $self->find_doc_ids($term, $value);
-
- return undef if $begin->equal($end); # not found
-
- my $rv = $begin->get_docid;
-
- # sanity check
- $begin->inc;
- $begin->equal($end) or die "Term '$term:$value' is not unique\n";
- $rv;
-}
-
-# returns begin and end PostingIterator
-sub find_doc_ids {
- my ($self, $term, $value) = @_;
- $self->find_doc_ids_for_term(xpfx($term) . $value);
-}
-
-# returns begin and end PostingIterator
-sub find_doc_ids_for_term {
- my ($self, $term) = @_;
- my $db = $self->{xdb};
-
- ($db->postlist_begin($term), $db->postlist_end($term));
-}
-
-# normalize subjects so they are suitable as pathnames for URLs
-sub subject_path {
- my $subj = pop;
- $subj = subject_normalized($subj);
- $subj =~ s![^a-zA-Z0-9_\.~/\-]+!_!g;
- lc($subj);
-}
-
-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;
-}
-
-# 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;
+ $self->{query_parser} = $qp;
}
-sub enquire {
+sub help {
my ($self) = @_;
- $self->{enquire} ||= Search::Xapian::Enquire->new($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}} ];
+ $self->qp; # parse altids
+ my @ret = @HELP;
+ if (my $user_pfx = $self->{-user_pfx}) {
+ push @ret, @$user_pfx;
+ }
+ \@ret;
}
1;