1 # Copyright (C) 2015-2020 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::Smsg;
16 use PublicInbox::Over;
18 our %X = map { $_ => 0 } qw(BoolWeight Database Enquire
19 NumberValueRangeProcessor QueryParser Stem);
20 our $Xap; # 'Search::Xapian' or 'Xapian'
24 return 1 if defined $Xap;
25 for my $x (qw(Search::Xapian Xapian)) {
29 $x->import(qw(:standard));
31 $X{$_} = $Xap.'::'.$_ for (keys %X);
33 # ENQ_ASCENDING doesn't seem exported by SWIG Xapian.pm,
34 # so lets hope this part of the ABI is stable because it's
36 $ENQ_ASCENDING = $x eq 'Xapian' ?
37 1 : Search::Xapian::ENQ_ASCENDING();
40 *PublicInbox::Smsg::sortable_unserialise =
41 $Xap.'::sortable_unserialise';
42 # n.b. FLAG_PURE_NOT is expensive not suitable for a public
43 # website as it could become a denial-of-service vector
44 # FLAG_PHRASE also seems to cause performance problems chert
45 # (and probably earlier Xapian DBs). glass seems fine...
46 # TODO: make this an option, maybe?
47 # or make indexlevel=medium as default
48 $QP_FLAGS = FLAG_PHRASE() | FLAG_BOOLEAN() | FLAG_LOVEHATE() |
55 # This is English-only, everything else is non-standard and may be confused as
56 # a prefix common in patch emails
57 our $LANG = 'english';
60 # SCHEMA_VERSION history
62 # 1 - subject_path is lower-cased
63 # 2 - subject_path is id_compress in the index, only
64 # 3 - message-ID is compressed if it includes '%' (hack!)
65 # 4 - change "Re: " normalization, avoid circular Reference ghosts
66 # 5 - subject_path drops trailing '.'
67 # 6 - preserve References: order in document data
68 # 7 - remove references and inreplyto terms
69 # 8 - remove redundant/unneeded document data
70 # 9 - disable Message-ID compression (SHA-1)
71 # 10 - optimize doc for NNTP overviews
72 # 11 - merge threads when vivifying ghosts
73 # 12 - change YYYYMMDD value column to numeric
74 # 13 - fix threading for empty References/In-Reply-To
75 # (commit 83425ef12e4b65cdcecd11ddcb38175d4a91d5a0)
76 # 14 - fix ghost root vivification
77 # 15 - see public-inbox-v2-format(5)
78 # further bumps likely unnecessary, we'll suggest in-place
79 # "--reindex" use for further fixes and tweaks
81 # public-inbox v1.5.0 adds (still SCHEMA_VERSION=15):
82 # * "lid:" and "l:" for List-Id searches
86 # note: the non-X term prefix allocations are shared with
87 # Xapian omega, see xapian-applications/omega/docs/termprefixes.rst
88 my %bool_pfx_external = (
89 mid => 'Q', # Message-ID (full/exact), this is mostly uniQue
90 lid => 'G', # newsGroup (or similar entity), just inside <>
93 dfblob => 'XDFPRE XDFPOST',
96 my $non_quoted_body = 'XNQ XDFN XDFA XDFB XDFHH XDFCTX XDFPRE XDFPOST';
98 # for mairix compatibility
100 m => 'XM', # 'mid:' (bool) is exact, 'm:' (prob) can do partial
101 l => 'XL', # 'lid:' (bool) is exact, 'l:' (prob) can do partial
108 b => $non_quoted_body . ' XQUOT',
109 bs => $non_quoted_body . ' XQUOT S',
113 nq => $non_quoted_body,
121 '' => 'XM S A XQUOT XFN ' . $non_quoted_body,
124 # not documenting m: and mid: for now, the using the URLs works w/o Xapian
125 # not documenting lid: for now, either, it is probably redundant with l:,
126 # especially since we don't offer boolean searches for To/Cc/From
129 's:' => 'match within Subject e.g. s:"a quick brown fox"',
131 date range as YYYYMMDD e.g. d:19931002..20101002
132 Open-ended ranges such as d:19931002.. and d:..20101002
136 date-time range as YYYYMMDDhhmmss (e.g. dt:19931002011000..19931002011200)
138 'b:' => 'match within message body, including text attachments',
139 'nq:' => 'match non-quoted text within message body',
140 'q:' => 'match quoted text within message body',
141 'n:' => 'match filename of attachment(s)',
142 't:' => 'match within the To header',
143 'c:' => 'match within the Cc header',
144 'f:' => 'match within the From header',
145 'a:' => 'match within the To, Cc, and From headers',
146 'tc:' => 'match within the To and Cc headers',
147 'l:' => 'match contents of the List-Id header',
148 'bs:' => 'match within the Subject and body',
149 'dfn:' => 'match filename from diff',
150 'dfa:' => 'match diff removed (-) lines',
151 'dfb:' => 'match diff added (+) lines',
152 'dfhh:' => 'match diff hunk header context (usually a function name)',
153 'dfctx:' => 'match diff context lines',
154 'dfpre:' => 'match pre-image git blob ID',
155 'dfpost:' => 'match post-image git blob ID',
156 'dfblob:' => 'match either pre or post-image git blob ID',
161 my ($self, $rdonly) = @_;
162 if ($self->{ibx_ver} == 1) {
163 "$self->{inboxdir}/public-inbox/xapian" . SCHEMA_VERSION;
165 my $dir = "$self->{inboxdir}/xap" . SCHEMA_VERSION;
166 return $dir if $rdonly;
168 my $shard = $self->{shard};
169 defined $shard or die "shard not given";
176 my $dir = xdir($self, 1);
177 my ($xdb, $slow_phrase);
178 my $qpf = \($self->{qp_flags} ||= $QP_FLAGS);
179 if ($self->{ibx_ver} >= 2) {
180 foreach my $shard (<$dir/*>) {
181 -d $shard && $shard =~ m!/[0-9]+\z! or next;
182 my $sub = $X{Database}->new($shard);
184 $xdb->add_database($sub);
188 $slow_phrase ||= -f "$shard/iamchert";
191 $slow_phrase = -f "$dir/iamchert";
192 $xdb = $X{Database}->new($dir);
194 $$qpf |= FLAG_PHRASE() unless $slow_phrase;
200 $self->{xdb} ||= do {
207 my ($class, $ibx) = @_;
208 ref $ibx or die "BUG: expected PublicInbox::Inbox object: $ibx";
210 inboxdir => $ibx->{inboxdir},
211 altid => $ibx->{altid},
212 ibx_ver => $ibx->version,
214 my $dir = xdir($self, 1);
215 $self->{over_ro} = PublicInbox::Over->new("$dir/over.sqlite3");
221 if (my $xdb = $self->{xdb}) {
224 $self; # make chaining easier
229 my ($self, $query_string, $opts) = @_;
231 if ($query_string eq '' && !$opts->{mset}) {
232 $self->{over_ro}->recent($opts);
235 my $qp_flags = $self->{qp_flags};
236 my $query = $qp->parse_query($query_string, $qp_flags);
237 $opts->{relevance} = 1 unless exists $opts->{relevance};
238 _do_enquire($self, $query, $opts);
243 my ($self, $cb, $arg) = @_;
247 eval { @ret = $cb->($arg) };
248 return @ret unless $@;
251 eval { $ret = $cb->($arg) };
252 return $ret unless $@;
254 # Exception: The revision being read has been discarded -
255 # you should call Xapian::Database::reopen()
256 if (ref($@) =~ /\bDatabaseModifiedError\b/) {
257 warn "reopen try #$i on $@\n";
260 # let caller decide how to spew, because ExtMsg queries
261 # get wonky and trigger:
262 # "something terrible happened at .../Xapian/Enquire.pm"
266 die "Too many Xapian database modifications in progress\n";
270 my ($self, $query, $opts) = @_;
271 retry_reopen($self, \&_enquire_once, [ $self, $query, $opts ]);
274 sub _enquire_once { # retry_reopen callback
275 my ($self, $query, $opts) = @{$_[0]};
276 my $xdb = xdb($self);
277 my $enquire = $X{Enquire}->new($xdb);
278 $enquire->set_query($query);
280 my $desc = !$opts->{asc};
281 if (($opts->{mset} || 0) == 2) {
282 $enquire->set_docid_order($ENQ_ASCENDING);
283 $enquire->set_weighting_scheme($X{BoolWeight}->new);
284 } elsif ($opts->{relevance}) {
285 $enquire->set_sort_by_relevance_then_value(TS, $desc);
287 $enquire->set_sort_by_value_then_relevance(TS, $desc);
289 my $offset = $opts->{offset} || 0;
290 my $limit = $opts->{limit} || 50;
291 my $mset = $enquire->get_mset($offset, $limit);
292 return $mset if $opts->{mset};
293 my @msgs = map { PublicInbox::Smsg::from_mitem($_) } $mset->items;
294 return \@msgs unless wantarray;
296 ($mset->get_matches_estimated, \@msgs)
300 sub stemmer { $X{Stem}->new($LANG) }
306 my $qp = $self->{query_parser};
308 my $xdb = xdb($self);
310 $qp = $X{QueryParser}->new;
311 $qp->set_default_op(OP_AND());
312 $qp->set_database($xdb);
313 $qp->set_stemmer($self->stemmer);
314 $qp->set_stemming_strategy(STEM_SOME());
315 $qp->set_max_wildcard_expansion(100);
316 my $nvrp = $X{NumberValueRangeProcessor};
317 $qp->add_valuerangeprocessor($nvrp->new(YYYYMMDD, 'd:'));
318 $qp->add_valuerangeprocessor($nvrp->new(DT, 'dt:'));
320 while (my ($name, $prefix) = each %bool_pfx_external) {
321 $qp->add_boolean_prefix($name, $_) foreach split(/ /, $prefix);
324 # we do not actually create AltId objects,
325 # just parse the spec to avoid the extra DB handles for now.
326 if (my $altid = $self->{altid}) {
327 my $user_pfx = $self->{-user_pfx} = [];
329 # $_ = 'serial:gmane:/path/to/gmane.msgmap.sqlite3'
330 # note: Xapian supports multibyte UTF-8, /^[0-9]+$/,
331 # and '_' with prefixes matching \w+
332 /\Aserial:(\w+):/ or next;
334 push @$user_pfx, "$pfx:", <<EOF;
335 alternate serial number e.g. $pfx:12345 (boolean)
338 $qp->add_boolean_prefix($pfx, 'X'.uc($pfx));
343 while (my ($name, $prefix) = each %prob_prefix) {
344 $qp->add_prefix($name, $_) foreach split(/ /, $prefix);
347 $self->{query_parser} = $qp;
352 $self->qp; # parse altids
354 if (my $user_pfx = $self->{-user_pfx}) {
355 push @ret, @$user_pfx;