From 1c0dce28a8604b16488ea7416dcf1bddd66a5d42 Mon Sep 17 00:00:00 2001 From: Eric Wong Date: Sat, 24 Apr 2021 10:23:30 +0000 Subject: extindex: --gc: use escape pathnames for SQL LIKE properly This allows us to handle odd inboxes w/o a newsgroup configured if they also make the strange choice of having backslashes in their path name. Also, ensure we use case-sensitive LIKE, since case-insensitive FSes are not worth supporting. --- lib/PublicInbox/ExtSearchIdx.pm | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'lib/PublicInbox') diff --git a/lib/PublicInbox/ExtSearchIdx.pm b/lib/PublicInbox/ExtSearchIdx.pm index a17e7579..9d6b3b9d 100644 --- a/lib/PublicInbox/ExtSearchIdx.pm +++ b/lib/PublicInbox/ExtSearchIdx.pm @@ -336,9 +336,11 @@ sub eidx_gc { $self->idx_init($opt); # acquire lock via V2Writable::_idx_init my $dbh = $self->{oidx}->dbh; + $dbh->do('PRAGMA case_sensitive_like = ON'); # only place we use LIKE my $x3_doc = $dbh->prepare('SELECT docid FROM xref3 WHERE ibx_id = ?'); my $ibx_ck = $dbh->prepare('SELECT ibx_id,eidx_key FROM inboxes'); - my $lc_i = $dbh->prepare('SELECT key FROM eidx_meta WHERE key LIKE ?'); + my $lc_i = $dbh->prepare(<<''); +SELECT key FROM eidx_meta WHERE key LIKE ? ESCAPE ? $ibx_ck->execute; while (my ($ibx_id, $eidx_key) = $ibx_ck->fetchrow_array) { @@ -354,8 +356,8 @@ DELETE FROM inboxes WHERE ibx_id = ? # drop last_commit info my $pat = $eidx_key; - $pat =~ s/([_%])/\\$1/g; - $lc_i->execute("lc-%:$pat//%"); + $pat =~ s/([_%\\])/\\$1/g; + $lc_i->execute("lc-%:$pat//%", '\\'); while (my ($key) = $lc_i->fetchrow_array) { next if $key !~ m!\Alc-v[1-9]+:\Q$eidx_key\E//!; warn "I: removing $key\n"; -- cgit v1.2.3-24-ge0c7