SERVER-19639 Add Unicode phrase matcher and tokenizer for FTS

1 files changed, 10 insertions, 0 deletions

diff --git a/src/mongo/db/fts/unicode/string.h b/src/mongo/db/fts/unicode/string.h
index 1fa77af2f3f..ddfa6f93870 100644
--- a/src/mongo/db/fts/unicode/string.h
+++ b/src/mongo/db/fts/unicode/string.h
@@ -62,6 +62,11 @@ public:
     explicit String(StringData utf8_src);
 
     /**
+     * Reset the String with the new UTF-8 source data, reusing the underlying buffer when possible.
+     */
+    void resetData(const StringData utf8_src);
+
+    /**
      * Return a lowercased version of the String instance using the Unicode data in u_data.h.
      */
     String toLower(CaseFoldMode mode = CaseFoldMode::kNormal) const;
@@ -130,6 +135,11 @@ private:
     String(std::u32string&& src);
 
     /**
+     * Helper method for converting a UTF-8 string to a UTF-32 string.
+     */
+    void setData(const StringData utf8_src);
+
+    /**
      * The underlying UTF-32 data.
      */
     std::u32string _data;