SERVER-20792 Only negate term if directly proceeded by hyphen.

For example, the term "documentation" should be negated in the text
query "MongoDB -documentation", but should not be negated in the text
query "MongoDB - documentation".

3 files changed, 130 insertions, 7 deletions

diff --git a/src/mongo/db/fts/fts_query_impl.cpp b/src/mongo/db/fts/fts_query_impl.cpp
index 674c8e84ab2..2c4b5016f90 100644
--- a/src/mongo/db/fts/fts_query_impl.cpp
+++ b/src/mongo/db/fts/fts_query_impl.cpp
@@ -75,6 +75,13 @@ Status FTSQueryImpl::parse(TextIndexVersion textIndexVersion) {
             if (inPhrase && inNegation) {
                 // don't add term
             } else {
+                // A negation should only continue until the next whitespace character. For example,
+                // "-foo" should negate "foo", "- foo" should not negate "foo", and "-foo-bar"
+                // should negate both "foo" and "bar".
+                if (inNegation && t.previousWhiteSpace) {
+                    inNegation = false;
+                }
+
                 if (inNegation) {
                     negativeTermSentence.append(s);
                     negativeTermSentence.push_back(' ');
@@ -83,9 +90,6 @@ Status FTSQueryImpl::parse(TextIndexVersion textIndexVersion) {
                     positiveTermSentence.push_back(' ');
                 }
             }
-
-            if (inNegation && !inPhrase)
-                inNegation = false;
         } else if (t.type == QueryToken::DELIMITER) {
             char c = t.data[0];
             if (c == '-') {
@@ -105,11 +109,20 @@ Status FTSQueryImpl::parse(TextIndexVersion textIndexVersion) {
                     } else {
                         _positivePhrases.push_back(phrase.toString());
                     }
-                    inNegation = false;
+
+                    // Do not reset 'inNegation' here, since a negation should continue until the
+                    // next whitespace character. For example, '-"foo bar"-"baz quux"' should negate
+                    // both the phrase "foo bar" and the phrase "baz quux".
+
                     inPhrase = false;
                 } else {
                     // start of a phrase
                     inPhrase = true;
+                    // A "-" should only be treated as a negation if there is no whitespace between
+                    // the "-" and the start of the phrase.
+                    if (inNegation && t.previousWhiteSpace) {
+                        inNegation = false;
+                    }
                     quoteOffset = t.offset;
                 }
             }
diff --git a/src/mongo/db/fts/fts_query_impl.h b/src/mongo/db/fts/fts_query_impl.h
index 888cec08e13..a3933a39806 100644
--- a/src/mongo/db/fts/fts_query_impl.h
+++ b/src/mongo/db/fts/fts_query_impl.h
@@ -67,8 +67,25 @@ public:
 
     std::string toString() const;
 
+    /**
+     * Returns a string with the following format:
+     * <positive terms>||<negative terms>||<positive phrases>||<negative phrases>
+     *
+     * Each set of terms or phrases is separated by '|' characters. For example, if we had positive
+     * terms 'foo' and 'bar', negative term 'baz', and no phrases, the debug string would be
+     * 'foo|bar||baz||||'.
+     */
     std::string debugString() const;
 
+    /**
+     * Returns a BSON object with the following format:
+     * {
+     *   terms: <array of positive terms>,
+     *   negatedTerms: <array of negative terms>,
+     *   phrases: <array of positive phrases>,
+     *   negatedPhrases: <array of negative phrases>
+     * }
+     */
     BSONObj toBSON() const;
 
 private:
diff --git a/src/mongo/db/fts/fts_query_impl_test.cpp b/src/mongo/db/fts/fts_query_impl_test.cpp
index 538ee3755bc..0465fff71d3 100644
--- a/src/mongo/db/fts/fts_query_impl_test.cpp
+++ b/src/mongo/db/fts/fts_query_impl_test.cpp
@@ -70,9 +70,9 @@ TEST(FTSQueryImpl, ParsePunctuation) {
     ASSERT_TRUE(q.getTermsForBounds() == q.getPositiveTerms());
 }
 
-TEST(FTSQueryImpl, Neg1) {
+TEST(FTSQueryImpl, HyphenBeforeWordShouldNegateTerm) {
     FTSQueryImpl q;
-    q.setQuery("this is -really fun");
+    q.setQuery("-really fun");
     q.setLanguage("english");
     q.setCaseSensitive(false);
     q.setDiacriticSensitive(false);
@@ -85,6 +85,69 @@ TEST(FTSQueryImpl, Neg1) {
     ASSERT_TRUE(q.getTermsForBounds() == q.getPositiveTerms());
 }
 
+TEST(FTSQueryImpl, HyphenFollowedByWhitespaceShouldNotNegate) {
+    FTSQueryImpl q;
+    q.setQuery("- really fun");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+
+    auto positiveTerms = q.getPositiveTerms();
+    ASSERT_EQUALS(2U, positiveTerms.size());
+    ASSERT_EQUALS(1U, positiveTerms.count("fun"));
+    ASSERT_EQUALS(1U, positiveTerms.count("realli"));
+    ASSERT_EQUALS(0U, q.getNegatedTerms().size());
+    ASSERT_TRUE(q.getTermsForBounds() == q.getPositiveTerms());
+}
+
+TEST(FTSQueryImpl, TwoHyphensShouldNegate) {
+    FTSQueryImpl q;
+    q.setQuery("--really fun");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+
+    ASSERT_EQUALS(1U, q.getPositiveTerms().size());
+    ASSERT_EQUALS("fun", *q.getPositiveTerms().begin());
+    ASSERT_EQUALS(1U, q.getNegatedTerms().size());
+    ASSERT_EQUALS("realli", *q.getNegatedTerms().begin());
+    ASSERT_TRUE(q.getTermsForBounds() == q.getPositiveTerms());
+}
+
+TEST(FTSQueryImpl, HyphenWithNoSurroundingWhitespaceShouldBeTreatedAsDelimiter) {
+    FTSQueryImpl q;
+    q.setQuery("really-fun");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+
+    auto positiveTerms = q.getPositiveTerms();
+    ASSERT_EQUALS(2U, positiveTerms.size());
+    ASSERT_EQUALS(1U, positiveTerms.count("fun"));
+    ASSERT_EQUALS(1U, positiveTerms.count("realli"));
+    ASSERT_EQUALS(0U, q.getNegatedTerms().size());
+    ASSERT_TRUE(q.getTermsForBounds() == q.getPositiveTerms());
+}
+
+TEST(FTSQueryImpl, HyphenShouldNegateAllSucceedingTermsSeparatedByHyphens) {
+    FTSQueryImpl q;
+    q.setQuery("-really-fun-stuff");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+
+    auto negatedTerms = q.getNegatedTerms();
+    ASSERT_EQUALS(3U, negatedTerms.size());
+    ASSERT_EQUALS(1U, negatedTerms.count("realli"));
+    ASSERT_EQUALS(1U, negatedTerms.count("fun"));
+    ASSERT_EQUALS(1U, negatedTerms.count("stuff"));
+    ASSERT_EQUALS(0U, q.getPositiveTerms().size());
+}
+
 TEST(FTSQueryImpl, Phrase1) {
     FTSQueryImpl q;
     q.setQuery("doing a \"phrase test\" for fun");
@@ -114,7 +177,7 @@ TEST(FTSQueryImpl, Phrase2) {
     ASSERT_EQUALS("phrase-test", q.getPositivePhr()[0]);
 }
 
-TEST(FTSQueryImpl, NegPhrase1) {
+TEST(FTSQueryImpl, HyphenDirectlyBeforePhraseShouldNegateEntirePhrase) {
     FTSQueryImpl q;
     q.setQuery("doing a -\"phrase test\" for fun");
     q.setLanguage("english");
@@ -124,6 +187,36 @@ TEST(FTSQueryImpl, NegPhrase1) {
     ASSERT_EQUALS("fun||||||phrase test", q.debugString());
 }
 
+TEST(FTSQueryImpl, HyphenSurroundedByWhitespaceBeforePhraseShouldNotNegateEntirePhrase) {
+    FTSQueryImpl q;
+    q.setQuery("doing a - \"phrase test\" for fun");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+    ASSERT_EQUALS("fun|phrase|test||||phrase test||", q.debugString());
+}
+
+TEST(FTSQueryImpl, HyphenBetweenTermAndPhraseShouldBeTreatedAsDelimiter) {
+    FTSQueryImpl q;
+    q.setQuery("doing a-\"phrase test\" for fun");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+    ASSERT_EQUALS("fun|phrase|test||||phrase test||", q.debugString());
+}
+
+TEST(FTSQueryImpl, HyphenShouldNegateAllSucceedingPhrasesSeparatedByHyphens) {
+    FTSQueryImpl q;
+    q.setQuery("-\"really fun\"-\"stuff here\" \"another phrase\"");
+    q.setLanguage("english");
+    q.setCaseSensitive(false);
+    q.setDiacriticSensitive(false);
+    ASSERT(q.parse(TEXT_INDEX_VERSION_3).isOK());
+    ASSERT_EQUALS("anoth|phrase||||another phrase||really fun|stuff here", q.debugString());
+}
+
 TEST(FTSQueryImpl, CaseSensitiveOption) {
     FTSQueryImpl q;
     q.setQuery("this is fun");