1 # Copyright (C) 2015-2019 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
3 # based on notmuch, but with no concept of folders, files or flags
5 # Read-only search interface for use by the web and NNTP interfaces
6 package PublicInbox::Search;
10 # values for searching
11 use constant TS => 0; # Received: header in Unix time
12 use constant YYYYMMDD => 1; # Date: header for searching in the WWW UI
13 use constant DT => 2; # Date: YYYYMMDDHHMMSS
15 use PublicInbox::SearchMsg;
16 use PublicInbox::MIME;
17 use PublicInbox::MID qw/id_compress/;
18 use PublicInbox::Over;
20 our %X = map { $_ => 0 } qw(BoolWeight Database Enquire
21 NumberValueRangeProcessor QueryParser Stem);
22 our $Xap; # 'Search::Xapian' or 'Xapian'
26 return 1 if defined $Xap;
27 for my $x (qw(Search::Xapian Xapian)) {
31 $x->import(qw(:standard));
33 $X{$_} = $Xap.'::'.$_ for (keys %X);
35 # ENQ_ASCENDING doesn't seem exported by SWIG Xapian.pm,
36 # so lets hope this part of the ABI is stable because it's
38 $ENQ_ASCENDING = $x eq 'Xapian' ?
39 1 : Search::Xapian::ENQ_ASCENDING();
42 *PublicInbox::SearchMsg::sortable_unserialise =
43 $Xap.'::sortable_unserialise';
44 # n.b. FLAG_PURE_NOT is expensive not suitable for a public
45 # website as it could become a denial-of-service vector
46 # FLAG_PHRASE also seems to cause performance problems chert
47 # (and probably earlier Xapian DBs). glass seems fine...
48 # TODO: make this an option, maybe?
49 # or make indexlevel=medium as default
50 $QP_FLAGS = FLAG_PHRASE() | FLAG_BOOLEAN() | FLAG_LOVEHATE() |
57 # This is English-only, everything else is non-standard and may be confused as
58 # a prefix common in patch emails
59 our $LANG = 'english';
62 # SCHEMA_VERSION history
64 # 1 - subject_path is lower-cased
65 # 2 - subject_path is id_compress in the index, only
66 # 3 - message-ID is compressed if it includes '%' (hack!)
67 # 4 - change "Re: " normalization, avoid circular Reference ghosts
68 # 5 - subject_path drops trailing '.'
69 # 6 - preserve References: order in document data
70 # 7 - remove references and inreplyto terms
71 # 8 - remove redundant/unneeded document data
72 # 9 - disable Message-ID compression (SHA-1)
73 # 10 - optimize doc for NNTP overviews
74 # 11 - merge threads when vivifying ghosts
75 # 12 - change YYYYMMDD value column to numeric
76 # 13 - fix threading for empty References/In-Reply-To
77 # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
78 # 14 - fix ghost root vivification
79 # 15 - see public-inbox-v2-format(5)
80 # further bumps likely unnecessary, we'll suggest in-place
81 # "--reindex" use for further fixes and tweaks
85 my %bool_pfx_external = (
86 mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
89 dfblob => 'XDFPRE XDFPOST',
92 my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
94 # for mairix compatibility
96 m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
103 b => $non_quoted_body . ' XQUOT',
104 bs => $non_quoted_body . ' XQUOT S',
108 nq => $non_quoted_body,
116 '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
119 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
121 's:' => 'match within Subject e.g. s:"a quick brown fox"',
123 date range as YYYYMMDD e.g. d:19931002..20101002
124 Open-ended ranges such as d:19931002.. and d:..20101002
128 date-time range as YYYYMMDDhhmmss (e.g. dt:19931002011000..19931002011200)
130 'b:' => 'match within message body, including text attachments',
131 'nq:' => 'match non-quoted text within message body',
132 'q:' => 'match quoted text within message body',
133 'n:' => 'match filename of attachment(s)',
134 't:' => 'match within the To header',
135 'c:' => 'match within the Cc header',
136 'f:' => 'match within the From header',
137 'a:' => 'match within the To, Cc, and From headers',
138 'tc:' => 'match within the To and Cc headers',
139 'bs:' => 'match within the Subject and body',
140 'dfn:' => 'match filename from diff',
141 'dfa:' => 'match diff removed (-) lines',
142 'dfb:' => 'match diff added (+) lines',
143 'dfhh:' => 'match diff hunk header context (usually a function name)',
144 'dfctx:' => 'match diff context lines',
145 'dfpre:' => 'match pre-image git blob ID',
146 'dfpost:' => 'match post-image git blob ID',
147 'dfblob:' => 'match either pre or post-image git blob ID',
152 my ($self, $rdonly) = @_;
153 if ($self->{version} == 1) {
154 "$self->{inboxdir}/public-inbox/xapian" . SCHEMA_VERSION;
156 my $dir = "$self->{inboxdir}/xap" . SCHEMA_VERSION;
157 return $dir if $rdonly;
159 my $shard = $self->{shard};
160 defined $shard or die "shard not given";
167 my $dir = xdir($self, 1);
168 my ($xdb, $slow_phrase);
169 my $qpf = \($self->{qp_flags} ||= $QP_FLAGS);
170 if ($self->{version} >= 2) {
171 foreach my $shard (<$dir/*>) {
172 -d $shard && $shard =~ m!/[0-9]+\z! or next;
173 my $sub = $X{Database}->new($shard);
175 $xdb->add_database($sub);
179 $slow_phrase ||= -f "$shard/iamchert";
182 $slow_phrase = -f "$dir/iamchert";
183 $xdb = $X{Database}->new($dir);
185 $$qpf |= FLAG_PHRASE() unless $slow_phrase;
191 $self->{xdb} ||= do {
198 my ($class, $ibx) = @_;
199 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
201 inboxdir => $ibx->{inboxdir},
202 altid => $ibx->{altid},
203 version => $ibx->{version} // 1,
205 my $dir = xdir($self, 1);
206 $self->{over_ro} = PublicInbox::Over->new("$dir/over.sqlite3");
212 if (my $xdb = $self->{xdb}) {
215 $self; # make chaining easier
220 my ($self, $query_string, $opts) = @_;
222 if ($query_string eq '' && !$opts->{mset}) {
223 $self->{over_ro}->recent($opts);
226 my $qp_flags = $self->{qp_flags};
227 my $query = $qp->parse_query($query_string, $qp_flags);
228 $opts->{relevance} = 1 unless exists $opts->{relevance};
229 _do_enquire($self, $query, $opts);
234 my ($self, $cb, $arg) = @_;
238 eval { @ret = $cb->($arg) };
239 return @ret unless $@;
242 eval { $ret = $cb->($arg) };
243 return $ret unless $@;
245 # Exception: The revision being read has been discarded -
246 # you should call Xapian::Database::reopen()
247 if (ref($@) =~ /\bDatabaseModifiedError\b/) {
248 warn "reopen try #$i on $@\n";
251 # let caller decide how to spew, because ExtMsg queries
252 # get wonky and trigger:
253 # "something terrible happened at .../Xapian/Enquire.pm"
257 die "Too many Xapian database modifications in progress\n";
261 my ($self, $query, $opts) = @_;
262 retry_reopen($self, \&_enquire_once, [ $self, $query, $opts ]);
265 sub _enquire_once { # retry_reopen callback
266 my ($self, $query, $opts) = @{$_[0]};
267 my $xdb = xdb($self);
268 my $enquire = $X{Enquire}->new($xdb);
269 $enquire->set_query($query);
271 my $desc = !$opts->{asc};
272 if (($opts->{mset} || 0) == 2) {
273 $enquire->set_docid_order($ENQ_ASCENDING);
274 $enquire->set_weighting_scheme($X{BoolWeight}->new);
275 } elsif ($opts->{relevance}) {
276 $enquire->set_sort_by_relevance_then_value(TS, $desc);
278 $enquire->set_sort_by_value_then_relevance(TS, $desc);
280 my $offset = $opts->{offset} || 0;
281 my $limit = $opts->{limit} || 50;
282 my $mset = $enquire->get_mset($offset, $limit);
283 return $mset if $opts->{mset};
284 my @msgs = map { PublicInbox::SearchMsg::from_mitem($_) } $mset->items;
285 return \@msgs unless wantarray;
287 ($mset->get_matches_estimated, \@msgs)
291 sub stemmer { $X{Stem}->new($LANG) }
297 my $qp = $self->{query_parser};
299 my $xdb = xdb($self);
301 $qp = $X{QueryParser}->new;
302 $qp->set_default_op(OP_AND());
303 $qp->set_database($xdb);
304 $qp->set_stemmer($self->stemmer);
305 $qp->set_stemming_strategy(STEM_SOME());
306 $qp->set_max_wildcard_expansion(100);
307 my $nvrp = $X{NumberValueRangeProcessor};
308 $qp->add_valuerangeprocessor($nvrp->new(YYYYMMDD, 'd:'));
309 $qp->add_valuerangeprocessor($nvrp->new(DT, 'dt:'));
311 while (my ($name, $prefix) = each %bool_pfx_external) {
312 $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
315 # we do not actually create AltId objects,
316 # just parse the spec to avoid the extra DB handles for now.
317 if (my $altid = $self->{altid}) {
318 my $user_pfx = $self->{-user_pfx} ||= [];
320 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
321 /\Aserial:(\w+):/ or next;
323 push @$user_pfx, "$pfx:", <<EOF;
324 alternate serial number e.g. $pfx:12345 (boolean)
327 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
332 while (my ($name, $prefix) = each %prob_prefix) {
333 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
336 $self->{query_parser} = $qp;
340 my ($self, $num) = @_;
341 $self->{over_ro}->get_art($num);
346 $self->qp; # parse altids
348 if (my $user_pfx = $self->{-user_pfx}) {
349 push @ret, @$user_pfx;