From 419f8077ae0f3a86985ada77426b26ebb8a7ae64 Mon Sep 17 00:00:00 2001
From: Robert Stepanek <rsto@fastmailteam.com>
Date: Tue, 16 Apr 2024 12:12:51 +0200
Subject: [PATCH] xapian_wrap: ignore punct-only terms only in unstructed
 search parts

This fixes a regression in 0c282379154a776099a83f120311a973c628eb55
Punctuation-only terms only should be ignored in unstructured search
parts, such as bodies and subjects.
---
 .../JMAPEmail/email_query_punct_no_text       | 56 +++++++++++++++++++
 imap/xapian_wrap.cpp                          | 36 +++++++-----
 2 files changed, 79 insertions(+), 13 deletions(-)
 create mode 100644 cassandane/tiny-tests/JMAPEmail/email_query_punct_no_text

diff --git a/cassandane/tiny-tests/JMAPEmail/email_query_punct_no_text b/cassandane/tiny-tests/JMAPEmail/email_query_punct_no_text
new file mode 100644
index 0000000000..f6c65b7c27
--- /dev/null
+++ b/cassandane/tiny-tests/JMAPEmail/email_query_punct_no_text
@@ -0,0 +1,56 @@
+#!perl
+use Cassandane::Tiny;
+
+sub test_email_query_punct_no_text
+    :needs_component_sieve :needs_component_jmap :JMAPExtensions
+{
+    my ($self) = @_;
+    my $imap = $self->{store}->get_client();
+
+    $imap->create("matches") or die;
+
+    # Assert that punctuation-only terms in non-text criteria
+    # match nothing. Also see email_query_utf8punct_term.
+
+    $self->{instance}->install_sieve_script(<<'EOF'
+require ["x-cyrus-jmapquery", "x-cyrus-log", "variables", "fileinto"];
+# Search: "from:\"=\""
+if allof(
+  not string :is "${stop}" "Y",
+  jmapquery text:
+  {
+     "conditions" : [
+        {
+           "from" : "\"=\""
+        }
+     ],
+     "operator" : "OR"
+  }
+.
+) {
+  fileinto "matches";
+  set "stop" "Y";
+}
+EOF
+    );
+
+    my $mime = <<'EOF';
+From: from@local
+To: to@local
+Subject: test
+Date: Mon, 13 Apr 2020 15:34:03 +0200
+MIME-Version: 1.0
+Content-Type: text/plain;charset=us-ascii
+Content-Transfer-Encoding: 7bit
+
+hello
+EOF
+    $mime =~ s/\r?\n/\r\n/gs;
+    my $msg = Cassandane::Message->new();
+    $msg->set_lines(split /\n/, $mime);
+    $self->{instance}->deliver($msg);
+
+    xlog "Assert that message did not match";
+    $self->assert_num_equals(0, $imap->message_count('matches'));
+    $self->assert_num_equals(1, $imap->message_count('INBOX'));
+}
diff --git a/imap/xapian_wrap.cpp b/imap/xapian_wrap.cpp
index df1c8c4b50..5911a7e914 100644
--- a/imap/xapian_wrap.cpp
+++ b/imap/xapian_wrap.cpp
@@ -1945,18 +1945,33 @@ xapian_query_new_match_internal(const xapian_db_t *db, int partnum, const char *
             return query_new_type(db, prefix, str);
         }
 
-        // Don't stem queries for Thaana codepage (0780) or higher.
+        // Match unstructured search parts
+
+        static Xapian::Query *q = NULL;
+
+        int need_word_break = 0;
         for (const unsigned char *p = (const unsigned char *)str; *p; p++) {
-            if (*p > 221) //has highbit
-                return xapian_query_new_match_word_break(db, str, prefix);
+            // Use ICU word break for Thaana codepage (0780) or higher.
+            if (*p > 221) {
+                need_word_break = 1;
+                break;
+            }
         }
 
-        // Stemable codepage.
-        Xapian::TermGenerator::stem_strategy stem_strategy =
-            get_stem_strategy(XAPIAN_DB_CURRENT_VERSION, partnum);
-
-        return query_new_textmatch(db, str, prefix, stem_strategy);
+        if (need_word_break) {
+            q = xapian_query_new_match_word_break(db, str, prefix);
+        }
+        else {
+            Xapian::TermGenerator::stem_strategy stem_strategy =
+                get_stem_strategy(XAPIAN_DB_CURRENT_VERSION, partnum);
+            q = query_new_textmatch(db, str, prefix, stem_strategy);
+        }
+        if (q && q->get_type() == Xapian::Query::LEAF_MATCH_NOTHING) {
+            delete q;
+            q = NULL;
+        }
 
+        return q;
     } catch (const Xapian::Error &err) {
         xsyslog(LOG_ERR, "IOERROR: caught exception",
                          "exception=<%s>",
@@ -2004,11 +2019,6 @@ xapian_query_new_match(const xapian_db_t *db, int partnum, const char *str)
         charset_free(&utf8);
     }
 
-    if (q && q->get_type() == Xapian::Query::LEAF_MATCH_NOTHING) {
-        delete q;
-        q = NULL;
-    }
-
     return (xapian_query_t*) q;
 }