SERVER-39771 TextMatchExpression crashes instead of serializing in some cases

(cherry picked from commit 8a9a6071b9aac873689cd777dfb4ed6bbad74e04)

3 files changed, 111 insertions, 104 deletions

diff --git a/jstests/core/fts_querylang.js b/jstests/core/fts_querylang.js
index 5fb1b8e606c..de27b65ba5b 100644
--- a/jstests/core/fts_querylang.js
+++ b/jstests/core/fts_querylang.js
@@ -1,104 +1,86 @@
+// Test the $text query operator.
 // @tags: [requires_non_retryable_writes]
-
-// Test $text query operator.
-
-var t = db.getSiblingDB("test").getCollection("fts_querylang");
-var cursor;
-var results;
-
-t.drop();
-
-t.insert({_id: 0, unindexedField: 0, a: "textual content"});
-t.insert({_id: 1, unindexedField: 1, a: "additional content"});
-t.insert({_id: 2, unindexedField: 2, a: "irrelevant content"});
-t.ensureIndex({a: "text"});
-
-// Test text query with no results.
-assert.eq(false, t.find({$text: {$search: "words"}}).hasNext());
-
-// Test basic text query.
-results = t.find({$text: {$search: "textual content -irrelevant"}}).toArray();
-assert.eq(results.length, 2);
-assert.neq(results[0]._id, 2);
-assert.neq(results[1]._id, 2);
-
-// Test sort with basic text query.
-results =
-    t.find({$text: {$search: "textual content -irrelevant"}}).sort({unindexedField: 1}).toArray();
-assert.eq(results.length, 2);
-assert.eq(results[0]._id, 0);
-assert.eq(results[1]._id, 1);
-
-// Test skip with basic text query.
-results = t.find({$text: {$search: "textual content -irrelevant"}})
-              .sort({unindexedField: 1})
-              .skip(1)
-              .toArray();
-assert.eq(results.length, 1);
-assert.eq(results[0]._id, 1);
-
-// Test limit with basic text query.
-results = t.find({$text: {$search: "textual content -irrelevant"}})
-              .sort({unindexedField: 1})
-              .limit(1)
-              .toArray();
-assert.eq(results.length, 1);
-assert.eq(results[0]._id, 0);
-
-// TODO Test basic text query with sort, once sort is enabled in the new query framework.
-
-// TODO Test basic text query with projection, once projection is enabled in the new query
-// framework.
-
-// Test $and of basic text query with indexed expression.
-results = t.find({$text: {$search: "content -irrelevant"}, _id: 1}).toArray();
-assert.eq(results.length, 1);
-assert.eq(results[0]._id, 1);
-
-// Test $and of basic text query with indexed expression, and bad language
-assert.throws(function() {
-    t.find({$text: {$search: "content -irrelevant", $language: "spanglish"}, _id: 1}).itcount();
-});
-
-// Test $and of basic text query with unindexed expression.
-results = t.find({$text: {$search: "content -irrelevant"}, unindexedField: 1}).toArray();
-assert.eq(results.length, 1);
-assert.eq(results[0]._id, 1);
-
-// TODO Test invalid inputs for $text, $search, $language.
-
-// Test $language.
-cursor = t.find({$text: {$search: "contents", $language: "none"}});
-assert.eq(false, cursor.hasNext());
-
-cursor = t.find({$text: {$search: "contents", $language: "EN"}});
-assert.eq(true, cursor.hasNext());
-
-cursor = t.find({$text: {$search: "contents", $language: "spanglish"}});
-assert.throws(function() {
-    cursor.next();
-});
-
-// TODO Test $and of basic text query with geo expression.
-
-// Test update with $text.
-t.update({$text: {$search: "textual content -irrelevant"}}, {$set: {b: 1}}, {multi: true});
-assert.eq(2, t.find({b: 1}).itcount(), 'incorrect number of documents updated');
-
-// TODO Test remove with $text, once it is enabled with the new query framework.
-
-// TODO Test count with $text, once it is enabled with the new query framework.
-
-// TODO Test findAndModify with $text, once it is enabled with the new query framework.
-
-// TODO Test aggregate with $text, once it is enabled with the new query framework.
-
-// TODO Test that old query framework rejects $text queries.
-
-// TODO Test that $text fails without a text index.
-
-// TODO Test that $text accepts a hint of the text index.
-
-// TODO Test that $text fails if a different index is hinted.
-
-// TODO Test $text with {$natural:1} sort, {$natural:1} hint.
+(function() {
+    "use strict";
+
+    const coll = db.getCollection("fts_querylang");
+    coll.drop();
+
+    assert.commandWorked(coll.insert({_id: 0, unindexedField: 0, a: "textual content"}));
+    assert.commandWorked(coll.insert({_id: 1, unindexedField: 1, a: "additional content"}));
+    assert.commandWorked(coll.insert({_id: 2, unindexedField: 2, a: "irrelevant content"}));
+    assert.commandWorked(coll.createIndex({a: "text"}));
+
+    // Test text query with no results.
+    assert.eq(false, coll.find({$text: {$search: "words"}}).hasNext());
+
+    // Test basic text query.
+    let results = coll.find({$text: {$search: "textual content -irrelevant"}}).toArray();
+    assert.eq(results.length, 2, results);
+    assert.neq(results[0]._id, 2, results);
+    assert.neq(results[1]._id, 2, results);
+
+    // Test sort with basic text query.
+    results = coll.find({$text: {$search: "textual content -irrelevant"}})
+                  .sort({unindexedField: 1})
+                  .toArray();
+    assert.eq(results.length, 2, results);
+    assert.eq(results[0]._id, 0, results);
+    assert.eq(results[1]._id, 1, results);
+
+    // Test skip with basic text query.
+    results = coll.find({$text: {$search: "textual content -irrelevant"}})
+                  .sort({unindexedField: 1})
+                  .skip(1)
+                  .toArray();
+    assert.eq(results.length, 1, results);
+    assert.eq(results[0]._id, 1, results);
+
+    // Test limit with basic text query.
+    results = coll.find({$text: {$search: "textual content -irrelevant"}})
+                  .sort({unindexedField: 1})
+                  .limit(1)
+                  .toArray();
+    assert.eq(results.length, 1, results);
+    assert.eq(results[0]._id, 0, results);
+
+    // Test $and of basic text query with indexed expression.
+    results = coll.find({$text: {$search: "content -irrelevant"}, _id: 1}).toArray();
+    assert.eq(results.length, 1, results);
+    assert.eq(results[0]._id, 1, results);
+
+    // Test $and of basic text query with indexed expression and bad language.
+    assert.commandFailedWithCode(assert.throws(function() {
+        coll.find({$text: {$search: "content -irrelevant", $language: "spanglish"}, _id: 1})
+            .itcount();
+    }),
+                                 ErrorCodes.BadValue);
+
+    // Test $and of basic text query with unindexed expression.
+    results = coll.find({$text: {$search: "content -irrelevant"}, unindexedField: 1}).toArray();
+    assert.eq(results.length, 1, results);
+    assert.eq(results[0]._id, 1, results);
+
+    // Test $language.
+    let cursor = coll.find({$text: {$search: "contents", $language: "none"}});
+    assert.eq(false, cursor.hasNext());
+
+    cursor = coll.find({$text: {$search: "contents", $language: "EN"}});
+    assert.eq(true, cursor.hasNext());
+
+    cursor = coll.find({$text: {$search: "contents", $language: "spanglish"}});
+    assert.commandFailedWithCode(assert.throws(function() {
+        cursor.next();
+    }),
+                                 ErrorCodes.BadValue);
+
+    // Test update with $text.
+    coll.update({$text: {$search: "textual content -irrelevant"}}, {$set: {b: 1}}, {multi: true});
+    assert.eq(2, coll.find({b: 1}).itcount(), 'incorrect number of documents updated');
+
+    // $text cannot be contained within a $nor.
+    assert.commandFailedWithCode(assert.throws(function() {
+        coll.find({$nor: [{$text: {$search: 'a'}}]}).itcount();
+    }),
+                                 ErrorCodes.BadValue);
+}());
diff --git a/src/mongo/db/matcher/expression_serialization_test.cpp b/src/mongo/db/matcher/expression_serialization_test.cpp
index 63f238593ae..74a8d3b8f74 100644
--- a/src/mongo/db/matcher/expression_serialization_test.cpp
+++ b/src/mongo/db/matcher/expression_serialization_test.cpp
@@ -1377,6 +1377,22 @@ TEST(SerializeBasic, ExpressionTextSerializesCorrectly) {
     ASSERT_BSONOBJ_EQ(*reserialized.getQuery(), serialize(reserialized.getMatchExpression()));
 }
 
+TEST(SerializeBasic, ExpressionNorWithTextSerializesCorrectly) {
+    boost::intrusive_ptr<ExpressionContextForTest> expCtx(new ExpressionContextForTest());
+    Matcher original(fromjson("{$nor: [{$text: {$search: 'x'}}]}"),
+                     expCtx,
+                     ExtensionsCallbackNoop(),
+                     MatchExpressionParser::kAllowAllSpecialFeatures);
+    Matcher reserialized(serialize(original.getMatchExpression()),
+                         expCtx,
+                         ExtensionsCallbackNoop(),
+                         MatchExpressionParser::kAllowAllSpecialFeatures);
+    ASSERT_BSONOBJ_EQ(*reserialized.getQuery(),
+                      fromjson("{$nor: [{$text: {$search: 'x', $language: '', $caseSensitive: "
+                               "false, $diacriticSensitive: false}}]}"));
+    ASSERT_BSONOBJ_EQ(*reserialized.getQuery(), serialize(reserialized.getMatchExpression()));
+}
+
 TEST(SerializeBasic, ExpressionTextWithDefaultLanguageSerializesCorrectly) {
     boost::intrusive_ptr<ExpressionContextForTest> expCtx(new ExpressionContextForTest());
     Matcher original(fromjson("{$text: {$search: 'a', $caseSensitive: false}}"),
diff --git a/src/mongo/db/matcher/expression_tree.cpp b/src/mongo/db/matcher/expression_tree.cpp
index c24749ea7b3..2229f602895 100644
--- a/src/mongo/db/matcher/expression_tree.cpp
+++ b/src/mongo/db/matcher/expression_tree.cpp
@@ -37,6 +37,7 @@
 #include "mongo/bson/bsonobjbuilder.h"
 #include "mongo/db/matcher/expression_always_boolean.h"
 #include "mongo/db/matcher/expression_path.h"
+#include "mongo/db/matcher/expression_text_base.h"
 
 namespace mongo {
 
@@ -324,6 +325,14 @@ void NotMatchExpression::debugString(StringBuilder& debug, int level) const {
 boost::optional<StringData> NotMatchExpression::getPathIfNotWithSinglePathMatchExpressionTree(
     MatchExpression* exp) {
     if (auto pathMatch = dynamic_cast<PathMatchExpression*>(exp)) {
+        if (dynamic_cast<TextMatchExpressionBase*>(exp)) {
+            // While TextMatchExpressionBase derives from PathMatchExpression, text match
+            // expressions cannot be serialized in the same manner as other PathMatchExpression
+            // derivatives. This is because the path for a TextMatchExpression is embedded within
+            // the $text object, whereas for other PathMatchExpressions it is on the left-hand-side,
+            // for example {x: {$eq: 1}}.
+            return boost::none;
+        }
         return pathMatch->path();
     }
 
@@ -331,7 +340,7 @@ boost::optional<StringData> NotMatchExpression::getPathIfNotWithSinglePathMatchE
         boost::optional<StringData> path;
         for (size_t i = 0; i < exp->numChildren(); ++i) {
             auto pathMatchChild = dynamic_cast<PathMatchExpression*>(exp->getChild(i));
-            if (!pathMatchChild) {
+            if (!pathMatchChild || dynamic_cast<TextMatchExpressionBase*>(exp->getChild(i))) {
                 return boost::none;
             }