X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FSearchView.pm;h=39400d6d4d16c7c0edfe6e97bf988fc3ab7c1540;hb=398b85e2a27718ba9eb0cc1d022660aadb184add;hp=3623a78b4406e1d90bad1d4b46c536c60c85b0f5;hpb=9d1e5fadd7d18f4c96ab0509d673040e34225a04;p=public-inbox.git diff --git a/lib/PublicInbox/SearchView.pm b/lib/PublicInbox/SearchView.pm index 3623a78b..39400d6d 100644 --- a/lib/PublicInbox/SearchView.pm +++ b/lib/PublicInbox/SearchView.pm @@ -5,14 +5,16 @@ package PublicInbox::SearchView; use strict; use warnings; +use URI::Escape qw(uri_unescape uri_escape); use PublicInbox::SearchMsg; -use PublicInbox::Hval qw/ascii_html/; +use PublicInbox::Hval qw/ascii_html obfuscate_addrs/; use PublicInbox::View; -use PublicInbox::MID qw(mid2path mid_mime mid_clean mid_escape); -use Email::MIME; +use PublicInbox::WwwAtomStream; +use PublicInbox::MID qw(mid2path mid_mime mid_clean mid_escape MID_ESC); +use PublicInbox::MIME; require PublicInbox::Git; -require PublicInbox::Thread; -our $LIM = 50; +require PublicInbox::SearchThread; +our $LIM = 200; sub noop {} @@ -28,27 +30,35 @@ sub sres_top_html { mset => 1, relevance => $q->{r}, }; - my ($mset, $total); + my ($mset, $total, $err, $cb); +retry: eval { $mset = $ctx->{srch}->query($q->{'q'}, $opts); $total = $mset->get_matches_estimated; }; - my $err = $@; + $err = $@; ctx_prepare($q, $ctx); - my $cb; if ($err) { $code = 400; $ctx->{-html_tip} = '
'.err_txt($ctx, $err).'
\n[No results found]
'; + my $initial_q = $ctx->{-uxs_retried}; + if (defined $initial_q) { + my $rewritten = $q->{'q'}; + utf8::decode($initial_q); + utf8::decode($rewritten); + $initial_q = ascii_html($initial_q); + $rewritten = ascii_html($rewritten); + $rv .= " Warning: Initial query:\n $initial_q\n"; + $rv .= " returned no results, used:\n"; + $rv .= " $rewritten\n instead\n\n"; + } - my $rv = "Search results ordered by ["; + $rv .= 'Search results ordered by ['; if ($q->{r}) { my $d = $q->qs_html(r => 0); $rv .= qq{date|relevance}; @@ -110,10 +157,10 @@ sub search_nav_top { my $x = $q->{x}; if ($x eq '') { my $t = $q->qs_html(x => 't'); - $rv .= qq{summary|threaded} + $rv .= qq{summary|nested} } elsif ($q->{x} eq 't') { my $s = $q->qs_html(x => ''); - $rv .= qq{summary|threaded}; + $rv .= qq{summary|nested}; } my $A = $q->qs_html(x => 'A', r => undef); $rv .= qq{|Atom feed]}; @@ -126,15 +173,21 @@ sub search_nav_bot { my $o = $q->{o}; my $end = $o + $nr; my $beg = $o + 1; - my $rv = "Results $beg-$end of $total"; + my $rv = ''; + if ($beg <= $end) { + $rv .= "Results $beg-$end of $total"; + $rv .= ' (estimated)' if $end != $total; + } else { + $rv .= "No more results, only $total"; + } my $n = $o + $LIM; if ($n < $total) { my $qs = $q->qs_html(o => $n); - $rv .= qq{, next} + $rv .= qq{ next} } if ($o > 0) { - $rv .= $n < $total ? '/' : ', '; + $rv .= $n < $total ? '/' : ' '; my $p = $o - $LIM; my $qs = $q->qs_html(o => ($p > 0 ? $p : 0)); $rv .= qq{prev}; @@ -142,29 +195,29 @@ sub search_nav_bot { $rv .= ''; } +sub sort_relevance { + my ($pct) = @_; + sub { + [ sort { (eval { $pct->{$b->topmost->{id}} } || 0) + <=> + (eval { $pct->{$a->topmost->{id}} } || 0) + } @{$_[0]} ] }; +} + sub mset_thread { my ($ctx, $mset, $q) = @_; my %pct; - my @m = map { + my $srch = $ctx->{srch}; + my $msgs = $srch->retry_reopen(sub { [ map { my $i = $_; - my $m = PublicInbox::SearchMsg->load_doc($i->get_document); - $pct{$m->mid} = $i->get_percent; - $m = $m->mini_mime; - $m; - } ($mset->items); - - my $th = PublicInbox::Thread->new(@m); - $th->thread; - if ($q->{r}) { # order by relevance - $th->order(sub { - sort { (eval { $pct{$b->topmost->messageid} } || 0) - <=> - (eval { $pct{$a->topmost->messageid} } || 0) - } @_; - }); - } else { # order by time (default for threaded view) - $th->order(*PublicInbox::View::sort_ts); - } + my $smsg = PublicInbox::SearchMsg->load_doc($i->get_document); + $pct{$smsg->mid} = $i->get_percent; + $smsg; + } ($mset->items) ]}); + my $r = $q->{r}; + my $rootset = PublicInbox::SearchThread::thread($msgs, + $r ? sort_relevance(\%pct) : *PublicInbox::View::sort_ts, + $srch); my $skel = search_nav_bot($mset, $q). ""; my $inbox = $ctx->{-inbox}; $ctx->{-upfx} = ''; @@ -176,21 +229,22 @@ sub mset_thread { $ctx->{prev_attr} = ''; $ctx->{prev_level} = 0; $ctx->{seen} = {}; - $ctx->{s_nr} = scalar(@m).'+ results'; + $ctx->{s_nr} = scalar(@$msgs).'+ results'; - PublicInbox::View::walk_thread($th, $ctx, + # reduce hash lookups in skel_dump + $ctx->{-obfuscate} = $ctx->{-inbox}->{obfuscate}; + PublicInbox::View::walk_thread($rootset, $ctx, *PublicInbox::View::pre_thread); - my $msgs = \@m; + @$msgs = reverse @$msgs if $r; my $mime; sub { return unless $msgs; - while ($mime = shift @$msgs) { - my $mid = mid_clean(mid_mime($mime)); - $mime = $inbox->msg_by_mid($mid) and last; + while ($mime = pop @$msgs) { + $mime = $inbox->msg_by_smsg($mime) and last; } if ($mime) { - $mime = Email::MIME->new($mime); + $mime = PublicInbox::MIME->new($mime); return PublicInbox::View::index_entry($mime, $ctx, scalar @$msgs); } @@ -201,7 +255,9 @@ sub mset_thread { sub ctx_prepare { my ($q, $ctx) = @_; - my $qh = ascii_html($q->{'q'}); + my $qh = $q->{'q'}; + utf8::decode($qh); + $qh = ascii_html($qh); $ctx->{-q_value_html} = $qh; $ctx->{-atom} = '?'.$q->qs_html(x => 'A', r => undef); $ctx->{-title_html} = "$qh - search results"; @@ -216,30 +272,26 @@ sub ctx_prepare { sub adump { my ($cb, $mset, $q, $ctx) = @_; - my $fh = $cb->([ 200, ['Content-Type' => 'application/atom+xml']]); my $ibx = $ctx->{-inbox}; - my $feed_opts = PublicInbox::Feed::get_feedopts($ctx); - my $x = ascii_html($q->{'q'}); - $x = qq{$x - search results}; - $feed_opts->{atomurl} = $feed_opts->{url} . '?'. $q->qs_html; - $feed_opts->{url} .= '?'. $q->qs_html(x => undef); - $x = PublicInbox::Feed::atom_header($feed_opts, $x); - $fh->write($x. PublicInbox::Feed::feed_updated()); - - for ($mset->items) { - $x = PublicInbox::SearchMsg->load_doc($_->get_document)->mid; - $x = mid2path($x); - my $s = PublicInbox::Feed::feed_entry($feed_opts, $x, $ibx); - $fh->write($s) if defined $s; - } - PublicInbox::Feed::end_feed($fh); + my @items = $mset->items; + $ctx->{search_query} = $q; + my $srch = $ctx->{srch}; + PublicInbox::WwwAtomStream->response($ctx, 200, sub { + while (my $x = shift @items) { + $x = load_doc_retry($srch, $x); + $x = $ibx->msg_by_smsg($x) and + return PublicInbox::MIME->new($x); + } + return undef; + }); } package PublicInbox::SearchQuery; use strict; use warnings; +use URI::Escape qw(uri_escape); use PublicInbox::Hval; -use PublicInbox::MID qw(mid_escape); +use PublicInbox::MID qw(MID_ESC); sub new { my ($class, $qp) = @_; @@ -264,7 +316,7 @@ sub qs_html { $self = $tmp; } - my $q = mid_escape($self->{'q'}); + my $q = uri_escape($self->{'q'}, MID_ESC); $q =~ s/%20/+/g; # improve URL readability my $qs = "q=$q";