Revert "SERVER-20888: Defer $language check to FTSQuery::Parse"

This reverts commit 4c38ff13739e60d33d52b0c944d55249e7d9cc68.

8 files changed, 84 insertions, 56 deletions

diff --git a/src/mongo/db/exec/stagedebug_cmd.cpp b/src/mongo/db/exec/stagedebug_cmd.cpp
index 8ee9f8b0e36..a3f5d29765e 100644
--- a/src/mongo/db/exec/stagedebug_cmd.cpp
+++ b/src/mongo/db/exec/stagedebug_cmd.cpp
@@ -457,11 +457,13 @@ public:
 
             params.spec = fam->getSpec();
 
-            params.query.parse(search,
-                               fam->getSpec().defaultLanguage().str().c_str(),
-                               fts::FTSQuery::caseSensitiveDefault,
-                               fts::FTSQuery::diacriticSensitiveDefault,
-                               fam->getSpec().getTextIndexVersion());
+            if (!params.query.parse(search,
+                                    fam->getSpec().defaultLanguage().str().c_str(),
+                                    fts::FTSQuery::caseSensitiveDefault,
+                                    fts::FTSQuery::diacriticSensitiveDefault,
+                                    fam->getSpec().getTextIndexVersion()).isOK()) {
+                return NULL;
+            }
 
             return new TextStage(txn, params, workingSet, matcher);
         } else if ("delete" == nodeName) {
diff --git a/src/mongo/db/fts/fts_matcher_test.cpp b/src/mongo/db/fts/fts_matcher_test.cpp
index 7694732e2e1..246510a9e70 100644
--- a/src/mongo/db/fts/fts_matcher_test.cpp
+++ b/src/mongo/db/fts/fts_matcher_test.cpp
@@ -38,7 +38,7 @@ namespace fts {
 
 TEST(FTSMatcher, NegWild1) {
     FTSQuery q;
-    q.parse("foo -bar", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("foo -bar", "english", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("$**"
                                                              << "text")))));
@@ -52,7 +52,7 @@ TEST(FTSMatcher, NegWild1) {
 // Regression test for SERVER-11994.
 TEST(FTSMatcher, NegWild2) {
     FTSQuery q;
-    q.parse("pizza -restaurant", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("pizza -restaurant", "english", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("$**"
                                                              << "text")))));
@@ -65,7 +65,7 @@ TEST(FTSMatcher, NegWild2) {
 
 TEST(FTSMatcher, Phrase1) {
     FTSQuery q;
-    q.parse("foo \"table top\"", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("foo \"table top\"", "english", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("$**"
                                                              << "text")))));
@@ -87,7 +87,7 @@ TEST(FTSMatcher, Phrase1) {
 
 TEST(FTSMatcher, Phrase2) {
     FTSQuery q;
-    q.parse("foo \"table top\"", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("foo \"table top\"", "english", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -98,7 +98,7 @@ TEST(FTSMatcher, Phrase2) {
 // language.
 TEST(FTSMatcher, ParsesUsingDocLanguage) {
     FTSQuery q;
-    q.parse("-glad", "none", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("-glad", "none", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -112,7 +112,7 @@ TEST(FTSMatcher, ParsesUsingDocLanguage) {
 // Test the matcher does not filter out stop words from positive terms
 TEST(FTSMatcher, MatcherDoesNotFilterStopWordsNeg) {
     FTSQuery q;
-    q.parse("-the", "none", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("-the", "none", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -124,7 +124,7 @@ TEST(FTSMatcher, MatcherDoesNotFilterStopWordsNeg) {
 // Test the matcher does not filter out stop words from negative terms
 TEST(FTSMatcher, MatcherDoesNotFilterStopWordsPos) {
     FTSQuery q;
-    q.parse("the", "none", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse("the", "none", false, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -137,7 +137,7 @@ TEST(FTSMatcher, MatcherDoesNotFilterStopWordsPos) {
 // case-sensitive text query 'search'.
 static bool docHasPositiveTermWithCase(const std::string& doc, const std::string& search) {
     FTSQuery q;
-    q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -164,7 +164,7 @@ TEST(FTSMatcher, HasPositiveTermCaseSensitive) {
 // case-sensitive text query 'search'.
 static bool docHasNegativeTermWithCase(const std::string& doc, const std::string& search) {
     FTSQuery q;
-    q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -191,7 +191,7 @@ TEST(FTSMatcher, HasNegativeTermCaseSensitive) {
 // from case-sensitive text query 'search'.
 static bool docPositivePhrasesMatchWithCase(const std::string& doc, const std::string& search) {
     FTSQuery q;
-    q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
@@ -214,7 +214,7 @@ TEST(FTSMatcher, PositivePhrasesMatchWithCase) {
 // from case-sensitive text query 'search'.
 static bool docNegativePhrasesMatchWithCase(const std::string& doc, const std::string& search) {
     FTSQuery q;
-    q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT_OK(q.parse(search, "english", true, false, TEXT_INDEX_VERSION_3));
     FTSMatcher m(q,
                  FTSSpec(FTSSpec::fixSpec(BSON("key" << BSON("x"
                                                              << "text")))));
diff --git a/src/mongo/db/fts/fts_query.cpp b/src/mongo/db/fts/fts_query.cpp
index 2dc9b222d3a..f162481066b 100644
--- a/src/mongo/db/fts/fts_query.cpp
+++ b/src/mongo/db/fts/fts_query.cpp
@@ -52,14 +52,15 @@ using std::vector;
 const bool FTSQuery::caseSensitiveDefault = false;
 const bool FTSQuery::diacriticSensitiveDefault = false;
 
-void FTSQuery::parse(const string& query,
-                     StringData language,
-                     bool caseSensitive,
-                     bool diacriticSensitive,
-                     TextIndexVersion textIndexVersion) {
+Status FTSQuery::parse(const string& query,
+                       StringData language,
+                       bool caseSensitive,
+                       bool diacriticSensitive,
+                       TextIndexVersion textIndexVersion) {
     StatusWithFTSLanguage swl = FTSLanguage::make(language, textIndexVersion);
-    uassertStatusOK(swl.getStatus());
-
+    if (!swl.getStatus().isOK()) {
+        return swl.getStatus();
+    }
     _language = swl.getValue();
     _caseSensitive = caseSensitive;
     _diacriticSensitive = diacriticSensitive;
@@ -130,6 +131,8 @@ void FTSQuery::parse(const string& query,
 
     _addTerms(tokenizer.get(), positiveTermSentence, false);
     _addTerms(tokenizer.get(), negativeTermSentence, true);
+
+    return Status::OK();
 }
 
 void FTSQuery::_addTerms(FTSTokenizer* tokenizer, const string& sentence, bool negated) {
diff --git a/src/mongo/db/fts/fts_query.h b/src/mongo/db/fts/fts_query.h
index 3b7fbaa16f3..ea1882e4baf 100644
--- a/src/mongo/db/fts/fts_query.h
+++ b/src/mongo/db/fts/fts_query.h
@@ -50,11 +50,11 @@ public:
     // version 1 (see fts_language.cpp for a list of language strings specific to version
     // 1). Note that the diacritic sensitive option has no effect on FTS queries below index version
     // 3.
-    void parse(const std::string& query,
-               StringData language,
-               bool caseSensitive,
-               bool diacriticSensitive,
-               TextIndexVersion textIndexVersion);
+    Status parse(const std::string& query,
+                 StringData language,
+                 bool caseSensitive,
+                 bool diacriticSensitive,
+                 TextIndexVersion textIndexVersion);
 
     const std::set<std::string>& getPositiveTerms() const {
         return _positiveTerms;
diff --git a/src/mongo/db/fts/fts_query_test.cpp b/src/mongo/db/fts/fts_query_test.cpp
index dfdeeba68e7..bcf9e537142 100644
--- a/src/mongo/db/fts/fts_query_test.cpp
+++ b/src/mongo/db/fts/fts_query_test.cpp
@@ -37,7 +37,7 @@ namespace fts {
 
 TEST(FTSQuery, Basic1) {
     FTSQuery q;
-    q.parse("this is fun", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("this is fun", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(false, q.getCaseSensitive());
     ASSERT_EQUALS(1U, q.getPositiveTerms().size());
@@ -50,7 +50,7 @@ TEST(FTSQuery, Basic1) {
 
 TEST(FTSQuery, ParsePunctuation) {
     FTSQuery q;
-    q.parse("hello.world", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("hello.world", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(false, q.getCaseSensitive());
     ASSERT_EQUALS(2U, q.getPositiveTerms().size());
@@ -64,7 +64,7 @@ TEST(FTSQuery, ParsePunctuation) {
 
 TEST(FTSQuery, Neg1) {
     FTSQuery q;
-    q.parse("this is -really fun", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("this is -really fun", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(1U, q.getPositiveTerms().size());
     ASSERT_EQUALS("fun", *q.getPositiveTerms().begin());
@@ -75,7 +75,8 @@ TEST(FTSQuery, Neg1) {
 
 TEST(FTSQuery, Phrase1) {
     FTSQuery q;
-    q.parse("doing a \"phrase test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("doing a \"phrase test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3)
+               .isOK());
 
     ASSERT_EQUALS(3U, q.getPositiveTerms().size());
     ASSERT_EQUALS(0U, q.getNegatedTerms().size());
@@ -89,26 +90,29 @@ TEST(FTSQuery, Phrase1) {
 
 TEST(FTSQuery, Phrase2) {
     FTSQuery q;
-    q.parse("doing a \"phrase-test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("doing a \"phrase-test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3)
+               .isOK());
     ASSERT_EQUALS(1U, q.getPositivePhr().size());
     ASSERT_EQUALS("phrase-test", q.getPositivePhr()[0]);
 }
 
 TEST(FTSQuery, NegPhrase1) {
     FTSQuery q;
-    q.parse("doing a -\"phrase test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(
+        q.parse("doing a -\"phrase test\" for fun", "english", false, false, TEXT_INDEX_VERSION_3)
+            .isOK());
     ASSERT_EQUALS("fun||||||phrase test", q.debugString());
 }
 
 TEST(FTSQuery, CaseSensitiveOption) {
     FTSQuery q;
-    q.parse("this is fun", "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("this is fun", "english", true, false, TEXT_INDEX_VERSION_3).isOK());
     ASSERT_EQUALS(true, q.getCaseSensitive());
 }
 
 TEST(FTSQuery, CaseSensitivePositiveTerms) {
     FTSQuery q;
-    q.parse("This is Positively fun", "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("This is Positively fun", "english", true, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(2U, q.getTermsForBounds().size());
     ASSERT_EQUALS(1,
@@ -124,7 +128,8 @@ TEST(FTSQuery, CaseSensitivePositiveTerms) {
 
 TEST(FTSQuery, CaseSensitiveNegativeTerms) {
     FTSQuery q;
-    q.parse("-This -is -Negatively -miserable", "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("-This -is -Negatively -miserable", "english", true, false, TEXT_INDEX_VERSION_3)
+               .isOK());
 
     ASSERT_EQUALS(0U, q.getPositiveTerms().size());
     ASSERT_EQUALS(0U, q.getTermsForBounds().size());
@@ -137,7 +142,8 @@ TEST(FTSQuery, CaseSensitiveNegativeTerms) {
 
 TEST(FTSQuery, CaseSensitivePositivePhrases) {
     FTSQuery q;
-    q.parse("doing a \"Phrase Test\" for fun", "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("doing a \"Phrase Test\" for fun", "english", true, false, TEXT_INDEX_VERSION_3)
+               .isOK());
 
     ASSERT_EQUALS(1U, q.getPositivePhr().size());
     ASSERT_EQUALS(0U, q.getNegatedPhr().size());
@@ -146,7 +152,8 @@ TEST(FTSQuery, CaseSensitivePositivePhrases) {
 
 TEST(FTSQuery, CaseSensitiveNegativePhrases) {
     FTSQuery q;
-    q.parse("doing a -\"Phrase Test\" for fun", "english", true, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q.parse("doing a -\"Phrase Test\" for fun", "english", true, false, TEXT_INDEX_VERSION_3)
+               .isOK());
 
     ASSERT_EQUALS(0U, q.getPositivePhr().size());
     ASSERT_EQUALS(1U, q.getNegatedPhr().size());
@@ -155,15 +162,17 @@ TEST(FTSQuery, CaseSensitiveNegativePhrases) {
 
 TEST(FTSQuery, Mix1) {
     FTSQuery q;
-    q.parse("\"industry\" -Melbourne -Physics", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(
+        q.parse("\"industry\" -Melbourne -Physics", "english", false, false, TEXT_INDEX_VERSION_3)
+            .isOK());
     ASSERT_EQUALS("industri||melbourn|physic||industry||", q.debugString());
 }
 
 TEST(FTSQuery, NegPhrase2) {
     FTSQuery q1, q2, q3;
-    q1.parse("foo \"bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
-    q2.parse("foo \"-bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
-    q3.parse("foo \" -bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q1.parse("foo \"bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
+    ASSERT(q2.parse("foo \"-bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
+    ASSERT(q3.parse("foo \" -bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(2U, q1.getPositiveTerms().size());
     ASSERT_EQUALS(2U, q2.getPositiveTerms().size());
@@ -184,9 +193,9 @@ TEST(FTSQuery, NegPhrase2) {
 
 TEST(FTSQuery, NegPhrase3) {
     FTSQuery q1, q2, q3;
-    q1.parse("foo -\"bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
-    q2.parse("foo -\"-bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
-    q3.parse("foo -\" -bar\"", "english", false, false, TEXT_INDEX_VERSION_3);
+    ASSERT(q1.parse("foo -\"bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
+    ASSERT(q2.parse("foo -\"-bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
+    ASSERT(q3.parse("foo -\" -bar\"", "english", false, false, TEXT_INDEX_VERSION_3).isOK());
 
     ASSERT_EQUALS(1U, q1.getPositiveTerms().size());
     ASSERT_EQUALS(1U, q2.getPositiveTerms().size());
@@ -209,7 +218,7 @@ TEST(FTSQuery, NegPhrase3) {
 // stemmer and stopword list.
 TEST(FTSQuery, TextIndexVersion1LanguageEnglish) {
     FTSQuery q;
-    q.parse("the running", "english", false, false, TEXT_INDEX_VERSION_1);
+    ASSERT(q.parse("the running", "english", false, false, TEXT_INDEX_VERSION_1).isOK());
     ASSERT_EQUALS(1U, q.getPositiveTerms().size());
     ASSERT_EQUALS("run", *q.getPositiveTerms().begin());
     ASSERT_EQUALS(0U, q.getNegatedTerms().size());
@@ -221,7 +230,7 @@ TEST(FTSQuery, TextIndexVersion1LanguageEnglish) {
 // no stopword list.
 TEST(FTSQuery, TextIndexVersion1LanguageEng) {
     FTSQuery q;
-    q.parse("the running", "eng", false, false, TEXT_INDEX_VERSION_1);
+    ASSERT(q.parse("the running", "eng", false, false, TEXT_INDEX_VERSION_1).isOK());
     ASSERT_EQUALS(2U, q.getPositiveTerms().size());
     ASSERT_EQUALS(1, std::count(q.getPositiveTerms().begin(), q.getPositiveTerms().end(), "the"));
     ASSERT_EQUALS(1, std::count(q.getPositiveTerms().begin(), q.getPositiveTerms().end(), "run"));
@@ -234,7 +243,7 @@ TEST(FTSQuery, TextIndexVersion1LanguageEng) {
 // and no stopword list will be used.
 TEST(FTSQuery, TextIndexVersion1LanguageInvalid) {
     FTSQuery q;
-    q.parse("the running", "invalid", false, false, TEXT_INDEX_VERSION_1);
+    ASSERT(q.parse("the running", "invalid", false, false, TEXT_INDEX_VERSION_1).isOK());
     ASSERT_EQUALS(2U, q.getPositiveTerms().size());
     ASSERT_EQUALS(1, std::count(q.getPositiveTerms().begin(), q.getPositiveTerms().end(), "the"));
     ASSERT_EQUALS(1,
diff --git a/src/mongo/db/matcher/expression_parser_text.cpp b/src/mongo/db/matcher/expression_parser_text.cpp
index d8dd49f4b85..6968dc6c0cb 100644
--- a/src/mongo/db/matcher/expression_parser_text.cpp
+++ b/src/mongo/db/matcher/expression_parser_text.cpp
@@ -60,8 +60,11 @@ StatusWithMatchExpression expressionParserTextCallbackReal(const BSONObj& queryO
                                              "$language requires a string value");
         }
         language = languageElt.String();
-        // NOTE: the language parameter is validated during FTSQuery::Parse when we have the index
-        // spec and know which version of the index to use for the language list
+        Status status = fts::FTSLanguage::make(language, fts::TEXT_INDEX_VERSION_2).getStatus();
+        if (!status.isOK()) {
+            return StatusWithMatchExpression(ErrorCodes::BadValue,
+                                             "$language specifies unsupported language");
+        }
     }
     string query = queryObj["$search"].String();
 
diff --git a/src/mongo/db/matcher/expression_parser_text_test.cpp b/src/mongo/db/matcher/expression_parser_text_test.cpp
index 5aae8f0acd0..3d51604f7ef 100644
--- a/src/mongo/db/matcher/expression_parser_text_test.cpp
+++ b/src/mongo/db/matcher/expression_parser_text_test.cpp
@@ -54,6 +54,13 @@ TEST(MatchExpressionParserText, Basic) {
     ASSERT_EQUALS(textExp->getDiacriticSensitive(), fts::FTSQuery::diacriticSensitiveDefault);
 }
 
+TEST(MatchExpressionParserText, LanguageError) {
+    BSONObj query = fromjson("{$text: {$search:\"awesome\", $language:\"spanglish\"}}");
+
+    StatusWithMatchExpression result = MatchExpressionParser::parse(query);
+    ASSERT_FALSE(result.isOK());
+}
+
 TEST(MatchExpressionParserText, CaseSensitiveTrue) {
     BSONObj query = fromjson("{$text: {$search:\"awesome\", $caseSensitive: true}}");
 
diff --git a/src/mongo/db/query/stage_builder.cpp b/src/mongo/db/query/stage_builder.cpp
index 5395294c1c6..b971c44288a 100644
--- a/src/mongo/db/query/stage_builder.cpp
+++ b/src/mongo/db/query/stage_builder.cpp
@@ -279,11 +279,15 @@ PlanStage* buildStages(OperationContext* txn,
         const std::string& language =
             ("" == node->language ? fam->getSpec().defaultLanguage().str() : node->language);
 
-        params.query.parse(node->query,
-                           language,
-                           node->caseSensitive,
-                           node->diacriticSensitive,
-                           fam->getSpec().getTextIndexVersion());
+        Status parseStatus = params.query.parse(node->query,
+                                                language,
+                                                node->caseSensitive,
+                                                node->diacriticSensitive,
+                                                fam->getSpec().getTextIndexVersion());
+        if (!parseStatus.isOK()) {
+            warning() << "Can't parse text search query";
+            return NULL;
+        }
 
         return new TextStage(txn, params, ws, node->filter.get());
     } else if (STAGE_SHARDING_FILTER == root->getType()) {