Resolve parsing error for CHARACTER SET and COLLATE in MySQL ALTER TABLE (issue 2027) (#2045)

* fix: parsing alter text character set

* style: remove unnecessary comment

* fix: TEXT as a data type token

---------

Co-authored-by: mj-db <mj.db@kakaocorp.com>

Author: minleejae

src/main/java/net/sf/jsqlparser/parser/ParserKeywordsUtils.java

@@ -312,7 +312,7 @@ public static void buildGrammarForRelObjectNameWithoutValue(File file) throws Ex
                 + "{    Token tk = null; }\n"
                 + "{\n"
                 // @todo: find a way to avoid those hardcoded compound tokens
-                + "    ( tk=<DATA_TYPE> | tk=<S_IDENTIFIER> | tk=<S_QUOTED_IDENTIFIER> |  tk=<K_DATE_LITERAL> | tk=<K_DATETIMELITERAL> | tk=<K_STRING_FUNCTION_NAME> | tk=<K_ISOLATION> | tk=<K_TIME_KEY_EXPR> \n"
+                + "    ( tk=<DATA_TYPE> | tk=<S_IDENTIFIER> | tk=<S_QUOTED_IDENTIFIER> |  tk=<K_DATE_LITERAL> | tk=<K_DATETIMELITERAL> | tk=<K_STRING_FUNCTION_NAME> | tk=<K_ISOLATION> | tk=<K_TIME_KEY_EXPR> | tk=<K_TEXT_LITERAL> \n"
                 + "      ");
 
         for (String keyword : allKeywords) {

src/main/jjtree/net/sf/jsqlparser/parser/JSqlParserCC.jjt

@@ -451,6 +451,7 @@ TOKEN: /* SQL Keywords. prefixed with K_ to avoid name clashes */
 |   <K_TRIGGER: "TRIGGER">
 |   <K_THEN:"THEN">
 |   <K_TEMP:"TEMP">
+|   <K_TEXT_LITERAL: "TEXT" | "TINYTEXT" | "MEDIUMTEXT" | "LONGTEXT" >
 |   <K_TEMPORARY:"TEMPORARY">
 |   <K_TIME_KEY_EXPR : (
                          ("CURRENT" ( "_" | (" ")+ ) "TIMESTAMP")
@@ -569,7 +570,7 @@ TOKEN : /* Data Types */
     | <#TYPE_UHUGEINT: "UHUGEINT" >
     | <#TYPE_REAL: "REAL" | "FLOAT4" | "FLOAT">
     | <#TYPE_DOUBLE: "DOUBLE" | "PRECISION"  | "FLOAT8" | "FLOAT64">
-    | <#TYPE_VARCHAR: "NVARCHAR" | "VARCHAR" | "NCHAR" | <K_CHAR> | "BPCHAR" | "STRING" | "TEXT" | <K_CHARACTER> | "VARYING">
+    | <#TYPE_VARCHAR: "NVARCHAR" | "VARCHAR" | "NCHAR" | <K_CHAR> | "BPCHAR" | "TEXT" | "STRING" | <K_CHARACTER> | "VARYING">
     | <#TYPE_TIME: "TIMETZ" >
     | <#TYPE_TIMESTAMP: "TIMESTAMP_NS" | "TIMESTAMP_MS" | "TIMESTAMP_S"  >
 
@@ -1957,8 +1958,8 @@ The following tokens are allowed as Names for Schema, Table, Column and Aliases
 String RelObjectNameWithoutValue() :
 {    Token tk = null; }
 {
-    ( tk=<DATA_TYPE> | tk=<S_IDENTIFIER> | tk=<S_QUOTED_IDENTIFIER> |  tk=<K_DATE_LITERAL> | tk=<K_DATETIMELITERAL> | tk=<K_STRING_FUNCTION_NAME> | tk=<K_ISOLATION> | tk=<K_TIME_KEY_EXPR> 
-       | tk="ACTION" | tk="ACTIVE" | tk="ADD" | tk="ADVANCE" | tk="ADVISE" | tk="AGAINST" | tk="ALGORITHM" | tk="ALTER" | tk="ANALYZE" | tk="APPLY" | tk="APPROXIMATE" | tk="ARCHIVE" | tk="ARRAY" | tk="ASC" | tk="AT" | tk="AUTHORIZATION" | tk="AUTO" | tk="BASE64" | tk="BEGIN" | tk="BERNOULLI" | tk="BINARY" | tk="BIT" | tk="BLOCK" | tk="BROWSE" | tk="BUFFERS" | tk="BY" | tk="BYTE" | tk="BYTES" | tk="CACHE" | tk="CALL" | tk="CASCADE" | tk="CASE" | tk="CAST" | tk="CHANGE" | tk="CHANGES" | tk="CHAR" | tk="CHARACTER" | tk="CHECKPOINT" | tk="CLOSE" | tk="COLLATE" | tk="COLUMN" | tk="COLUMNS" | tk="COMMENT" | tk="COMMIT" | tk="CONCURRENTLY" | tk="CONFLICT" | tk="CONSTRAINTS" | tk="CONVERT" | tk="COSTS" | tk="CS" | tk="CYCLE" | tk="DATA" | tk="DATABASE" | tk="DATETIME" | tk="DBA_RECYCLEBIN" | tk="DDL" | tk="DECLARE" | tk="DEFAULT" | tk="DEFERRABLE" | tk="DELAYED" | tk="DELETE" | tk="DESC" | tk="DESCRIBE" | tk="DISABLE" | tk="DISCONNECT" | tk="DIV" | tk="DML" | tk="DO" | tk="DOMAIN" | tk="DROP" | tk="DUMP" | tk="DUPLICATE" | tk="ELEMENTS" | tk="EMIT" | tk="ENABLE" | tk="END" | tk="ESCAPE" | tk="EXCLUDE" | tk="EXEC" | tk="EXECUTE" | tk="EXPLAIN" | tk="EXPLICIT" | tk="EXTENDED" | tk="EXTRACT" | tk="FALSE" | tk="FILTER" | tk="FIRST" | tk="FLUSH" | tk="FN" | tk="FOLLOWING" | tk="FORMAT" | tk="FULLTEXT" | tk="FUNCTION" | tk="GRANT" | tk="GROUP_CONCAT" | tk="GUARD" | tk="HASH" | tk="HIGH_PRIORITY" | tk="HISTORY" | tk="HOPPING" | tk="INCLUDE" | tk="INCLUDE_NULL_VALUES" | tk="INCREMENT" | tk="INDEX" | tk="INSERT" | tk="INTERLEAVE" | tk="INTERPRET" | tk="INVALIDATE" | tk="ISNULL" | tk="JSON" | tk="JSON_ARRAY" | tk="JSON_ARRAYAGG" | tk="JSON_OBJECT" | tk="JSON_OBJECTAGG" | tk="KEEP" | tk="KEY" | tk="KEYS" | tk="LAST" | tk="LEADING" | tk="LINK" | tk="LOCAL" | tk="LOCKED" | tk="LOG" | tk="LOOP" | tk="LOW_PRIORITY" | tk="MATCH" | tk="MATCHED" | tk="MATERIALIZED" | tk="MAX" | tk="MAXVALUE" | tk="MEMBER" | tk="MERGE" | tk="MIN" | tk="MINVALUE" | tk="MODIFY" | tk="MOVEMENT" | tk="NEXT" | tk="NO" | tk="NOCACHE" | tk="NOKEEP" | tk="NOLOCK" | tk="NOMAXVALUE" | tk="NOMINVALUE" | tk="NOORDER" | tk="NOTHING" | tk="NOTNULL" | tk="NOVALIDATE" | tk="NOWAIT" | tk="NULLS" | tk="OF" | tk="OFF" | tk="OPEN" | tk="OVER" | tk="OVERLAPS" | tk="PARALLEL" | tk="PARENT" | tk="PARTITION" | tk="PATH" | tk="PERCENT" | tk="PLACING" | tk="PRECEDING" | tk="PRIMARY" | tk="PRIOR" | tk="PURGE" | tk="QUERY" | tk="QUICK" | tk="QUIESCE" | tk="RANGE" | tk="RAW" | tk="READ" | tk="RECURSIVE" | tk="RECYCLEBIN" | tk="REFERENCES" | tk="REFRESH" | tk="REGEXP" | tk="REGISTER" | tk="REMOTE" | tk="RENAME" | tk="REPEATABLE" | tk="REPLACE" | tk="RESET" | tk="RESPECT" | tk="RESTART" | tk="RESTRICT" | tk="RESTRICTED" | tk="RESUMABLE" | tk="RESUME" | tk="RETURN" | tk="RLIKE" | tk="ROLLBACK" | tk="ROLLUP" | tk="ROOT" | tk="ROW" | tk="ROWS" | tk="RR" | tk="RS" | tk="SAFE_CAST" | tk="SAVEPOINT" | tk="SCHEMA" | tk="SECURE" | tk="SEED" | tk="SEPARATOR" | tk="SEQUENCE" | tk="SESSION" | tk="SETS" | tk="SHARE" | tk="SHOW" | tk="SHUTDOWN" | tk="SIBLINGS" | tk="SIGNED" | tk="SIMILAR" | tk="SIZE" | tk="SKIP" | tk="STORED" | tk="STRING" | tk="STRUCT" | tk="SUSPEND" | tk="SWITCH" | tk="SYNONYM" | tk="SYSTEM" | tk="TABLE" | tk="TABLESPACE" | tk="TEMP" | tk="TEMPORARY" | tk="THEN" | tk="TIMEOUT" | tk="TIMESTAMPTZ" | tk="TIMEZONE" | tk="TO" | tk="TRIGGER" | tk="TRUE" | tk="TRUNCATE" | tk="TRY_CAST" | tk="TUMBLING" | tk="TYPE" | tk="UNLOGGED" | tk="UNQIESCE" | tk="UNSIGNED" | tk="UPDATE" | tk="UPSERT" | tk="UR" | tk="USER" | tk="VALIDATE" | tk="VERBOSE" | tk="VIEW" | tk="VOLATILE" | tk="WAIT" | tk="WITHIN" | tk="WITHOUT" | tk="WITHOUT_ARRAY_WRAPPER" | tk="WORK" | tk="XML" | tk="XMLAGG" | tk="XMLDATA" | tk="XMLSCHEMA" | tk="XMLTEXT" | tk="XSINIL" | tk="YAML" | tk="YES" | tk="ZONE" )
+    ( tk=<DATA_TYPE> | tk=<S_IDENTIFIER> | tk=<S_QUOTED_IDENTIFIER> |  tk=<K_DATE_LITERAL> | tk=<K_DATETIMELITERAL> | tk=<K_STRING_FUNCTION_NAME> | tk=<K_ISOLATION> | tk=<K_TIME_KEY_EXPR> | tk=<K_TEXT_LITERAL> 
+       | tk="ACTION" | tk="ACTIVE" | tk="ADD" | tk="ADVANCE" | tk="ADVISE" | tk="AGAINST" | tk="ALGORITHM" | tk="ALTER" | tk="ANALYZE" | tk="APPLY" | tk="APPROXIMATE" | tk="ARCHIVE" | tk="ARRAY" | tk="ASC" | tk="AT" | tk="AUTHORIZATION" | tk="AUTO" | tk="BASE64" | tk="BEGIN" | tk="BERNOULLI" | tk="BINARY" | tk="BIT" | tk="BLOCK" | tk="BROWSE" | tk="BUFFERS" | tk="BY" | tk="BYTE" | tk="BYTES" | tk="CACHE" | tk="CALL" | tk="CASCADE" | tk="CASE" | tk="CAST" | tk="CHANGE" | tk="CHANGES" | tk="CHAR" | tk="CHARACTER" | tk="CHECKPOINT" | tk="CLOSE" | tk="COLLATE" | tk="COLUMN" | tk="COLUMNS" | tk="COMMENT" | tk="COMMIT" | tk="CONCURRENTLY" | tk="CONFLICT" | tk="CONSTRAINTS" | tk="CONVERT" | tk="COSTS" | tk="CS" | tk="CYCLE" | tk="DATA" | tk="DATABASE" | tk="DATETIME" | tk="DBA_RECYCLEBIN" | tk="DDL" | tk="DECLARE" | tk="DEFAULT" | tk="DEFERRABLE" | tk="DELAYED" | tk="DELETE" | tk="DESC" | tk="DESCRIBE" | tk="DISABLE" | tk="DISCONNECT" | tk="DIV" | tk="DML" | tk="DO" | tk="DOMAIN" | tk="DROP" | tk="DUMP" | tk="DUPLICATE" | tk="ELEMENTS" | tk="EMIT" | tk="ENABLE" | tk="END" | tk="ESCAPE" | tk="EXCLUDE" | tk="EXEC" | tk="EXECUTE" | tk="EXPLAIN" | tk="EXPLICIT" | tk="EXTENDED" | tk="EXTRACT" | tk="FALSE" | tk="FILTER" | tk="FIRST" | tk="FLUSH" | tk="FN" | tk="FOLLOWING" | tk="FORMAT" | tk="FULLTEXT" | tk="FUNCTION" | tk="GRANT" | tk="GROUP_CONCAT" | tk="GUARD" | tk="HASH" | tk="HIGH_PRIORITY" | tk="HISTORY" | tk="HOPPING" | tk="INCLUDE" | tk="INCLUDE_NULL_VALUES" | tk="INCREMENT" | tk="INDEX" | tk="INSERT" | tk="INTERLEAVE" | tk="INTERPRET" | tk="INVALIDATE" | tk="ISNULL" | tk="JSON" | tk="JSON_ARRAY" | tk="JSON_ARRAYAGG" | tk="JSON_OBJECT" | tk="JSON_OBJECTAGG" | tk="KEEP" | tk="KEY" | tk="KEYS" | tk="LAST" | tk="LEADING" | tk="LINK" | tk="LOCAL" | tk="LOCKED" | tk="LOG" | tk="LONGTEXT" | tk="LOOP" | tk="LOW_PRIORITY" | tk="MATCH" | tk="MATCHED" | tk="MATERIALIZED" | tk="MAX" | tk="MAXVALUE" | tk="MEDIUMTEXT" | tk="MEMBER" | tk="MERGE" | tk="MIN" | tk="MINVALUE" | tk="MODIFY" | tk="MOVEMENT" | tk="NEXT" | tk="NO" | tk="NOCACHE" | tk="NOKEEP" | tk="NOLOCK" | tk="NOMAXVALUE" | tk="NOMINVALUE" | tk="NOORDER" | tk="NOTHING" | tk="NOTNULL" | tk="NOVALIDATE" | tk="NOWAIT" | tk="NULLS" | tk="OF" | tk="OFF" | tk="OPEN" | tk="OVER" | tk="OVERLAPS" | tk="PARALLEL" | tk="PARENT" | tk="PARTITION" | tk="PATH" | tk="PERCENT" | tk="PLACING" | tk="PRECEDING" | tk="PRIMARY" | tk="PRIOR" | tk="PURGE" | tk="QUERY" | tk="QUICK" | tk="QUIESCE" | tk="RANGE" | tk="RAW" | tk="READ" | tk="RECURSIVE" | tk="RECYCLEBIN" | tk="REFERENCES" | tk="REFRESH" | tk="REGEXP" | tk="REGISTER" | tk="REMOTE" | tk="RENAME" | tk="REPEATABLE" | tk="REPLACE" | tk="RESET" | tk="RESPECT" | tk="RESTART" | tk="RESTRICT" | tk="RESTRICTED" | tk="RESUMABLE" | tk="RESUME" | tk="RETURN" | tk="RLIKE" | tk="ROLLBACK" | tk="ROLLUP" | tk="ROOT" | tk="ROW" | tk="ROWS" | tk="RR" | tk="RS" | tk="SAFE_CAST" | tk="SAVEPOINT" | tk="SCHEMA" | tk="SECURE" | tk="SEED" | tk="SEPARATOR" | tk="SEQUENCE" | tk="SESSION" | tk="SETS" | tk="SHARE" | tk="SHOW" | tk="SHUTDOWN" | tk="SIBLINGS" | tk="SIGNED" | tk="SIMILAR" | tk="SIZE" | tk="SKIP" | tk="STORED" | tk="STRING" | tk="STRUCT" | tk="SUSPEND" | tk="SWITCH" | tk="SYNONYM" | tk="SYSTEM" | tk="TABLE" | tk="TABLESPACE" | tk="TEMP" | tk="TEMPORARY" | tk="TEXT" | tk="THEN" | tk="TIMEOUT" | tk="TIMESTAMPTZ" | tk="TIMEZONE" | tk="TINYTEXT" | tk="TO" | tk="TRIGGER" | tk="TRUE" | tk="TRUNCATE" | tk="TRY_CAST" | tk="TUMBLING" | tk="TYPE" | tk="UNLOGGED" | tk="UNQIESCE" | tk="UNSIGNED" | tk="UPDATE" | tk="UPSERT" | tk="UR" | tk="USER" | tk="VALIDATE" | tk="VERBOSE" | tk="VIEW" | tk="VOLATILE" | tk="WAIT" | tk="WITHIN" | tk="WITHOUT" | tk="WITHOUT_ARRAY_WRAPPER" | tk="WORK" | tk="XML" | tk="XMLAGG" | tk="XMLDATA" | tk="XMLSCHEMA" | tk="XMLTEXT" | tk="XSINIL" | tk="YAML" | tk="YES" | tk="ZONE" )
     { return tk.image; }
 }
 
@@ -6150,6 +6151,11 @@ ColDataType DataType():
 }
 {
     (
+        tk=<K_TEXT_LITERAL> {
+            type = tk.image;
+            return new ColDataType(type, precision, scale);
+        }
+        |
 		LOOKAHEAD(2) tk=<K_ARRAY_LITERAL> (
             ("<" arrayType = ColDataType() ">") {
                  colDataType.setDataType("ARRAY<" + arrayType.getDataType() + ">");

src/test/java/net/sf/jsqlparser/statement/alter/AlterTest.java

@@ -1023,4 +1023,22 @@ public void testIssue1875() throws JSQLParserException {
                 "ALTER TABLE IF EXISTS usercenter.dict_surgeries ADD COLUMN IF NOT EXISTS operation_grade_id int8 NULL";
         assertSqlCanBeParsedAndDeparsed(stmt);
     }
+
+    @Test
+    public void testIssue2027() throws JSQLParserException{
+        String sql = "ALTER TABLE `foo_bar` ADD COLUMN `baz` text";
+        assertSqlCanBeParsedAndDeparsed(sql);
+
+        String sqlText = "ALTER TABLE `foo_bar` ADD COLUMN `baz` text CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NULL";
+        assertSqlCanBeParsedAndDeparsed(sqlText);
+
+        String sqlTinyText = "ALTER TABLE `foo_bar` ADD COLUMN `baz` tinytext CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NULL";
+        assertSqlCanBeParsedAndDeparsed(sqlTinyText);
+
+        String sqlMediumText = "ALTER TABLE `foo_bar` ADD COLUMN `baz` mediumtext CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NULL";
+        assertSqlCanBeParsedAndDeparsed(sqlMediumText);
+
+        String sqlLongText = "ALTER TABLE `foo_bar` ADD COLUMN `baz` longtext CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci NULL";
+        assertSqlCanBeParsedAndDeparsed(sqlLongText);
+    }
 }