Fixed Parser Where Queries Improperly Parsed As Alias

integ-test/src/test/resources/correctness/expressions/arithmetics.txt

@@ -1,3 +1,23 @@
+6 * 2
+6 *2
+6* 2
+6*2
+6 + 2
+6 +2
+6+ 2
+6+2
+6 - 2
+6 -2
+6- 2
+6-2
+6 / 2
+6 /2
+6/ 2
+6/2
+6 % 2
+6 %2
+6% 2
+6%2
 1 + 2
 -1.0 + 1.234
 10 - 5

sql/src/main/antlr/OpenSearchSQLIdentifierParser.g4

@@ -36,7 +36,7 @@ options { tokenVocab=OpenSearchSQLLexer; }
 //    Identifiers
 
 tableName
-    : qualifiedName
+    : indexName
     ;
 
 columnName
@@ -47,6 +47,10 @@ alias
     : ident
     ;
 
+indexName
+    : index (DOT index)*
+    ;
+
 qualifiedName
     : ident (DOT ident)*
     ;
@@ -57,6 +61,12 @@ ident
     | keywordsCanBeId
     ;
 
+ index
+    : FROM_CLAUSE_DOT? IND
+    | BACKTICK_QUOTE_ID
+    | keywordsCanBeId
+    ;
+
 keywordsCanBeId
     : FULL
     | FIELD | D | T | TS // OD SQL and ODBC special

sql/src/main/antlr/OpenSearchSQLLexer.g4

@@ -69,7 +69,7 @@ EXISTS:                             'EXISTS';
 FALSE:                              'FALSE';
 FLOAT:                              'FLOAT';
 FIRST:                              'FIRST';
-FROM:                               'FROM';
+FROM:                               'FROM' -> pushMode (FROM_CLAUSE);
 GROUP:                              'GROUP';
 HAVING:                             'HAVING';
 IN:                                 'IN';
@@ -438,7 +438,7 @@ BACKTICK_QUOTE_ID:                  BQUOTA_STRING;
 
 // Fragments for Literal primitives
 fragment EXPONENT_NUM_PART:         'E' [-+]? DEC_DIGIT+;
-fragment ID_LITERAL:                [@*A-Z]+?[*A-Z_\-0-9]*;
+fragment ID_LITERAL:                [A-Z]+?[A-Z_\-0-9]*;
 fragment DQUOTA_STRING:             '"' ( '\\'. | '""' | ~('"'| '\\') )* '"';
 fragment SQUOTA_STRING:             '\'' ('\\'. | '\'\'' | ~('\'' | '\\'))* '\'';
 fragment BQUOTA_STRING:             '`' ( '\\'. | '``' | ~('`'|'\\'))* '`';
@@ -449,3 +449,48 @@ fragment BIT_STRING_L:              'B' '\'' [01]+ '\'';
 // Last tokens must generate Errors
 
 ERROR_RECOGNITION:                  .    -> channel(ERRORCHANNEL);
+
+
+// FROM_CLAUSE table names are identifiers that have extra characters beyond what MySQL accepts.
+// We use the FROM_CLAUSE mode to enter a separate rule set (defined below).
+// In this rule set, table names can be defined according to what is required by OpenSearch without affecting rules used for lexing the rest of a SQL query.
+// Once the table name is defined, we can exit from this FROM_CLAUSE and return to the default set of tokens
+// see: https://opensearch.org/docs/latest/search-plugins/sql/identifiers/
+mode FROM_CLAUSE;
+FROM_CLAUSE_SPACE:                  [ \t\r\n]+    -> channel(HIDDEN);
+FROM_CLAUSE_SPEC_SQL_COMMENT:       '/*!' .+? '*/' -> channel(SQLCOMMENT);
+FROM_CLAUSE_COMMENT_INPUT:          '/*' .*? '*/' -> channel(HIDDEN);
+FROM_CLAUSE_LINE_COMMENT:           (
+                                      ('-- ' | '#') ~[\r\n]* ('\r'? '\n' | EOF)
+                                      | '--' ('\r'? '\n' | EOF)
+                                    ) -> channel(HIDDEN);
+
+FROM_CLAUSE_DOT:                    '.';
+FROM_CLAUSE_LT_SQR_PRTHS:           '[';
+FROM_CLAUSE_RT_SQR_PRTHS:           ']';
+FROM_CLAUSE_SEMI:                   ';';
+FROM_CLAUSE_AT_SIGN:                '@';
+FROM_CLAUSE_ZERO_DECIMAL:           '0';
+FROM_CLAUSE_ONE_DECIMAL:            '1';
+FROM_CLAUSE_TWO_DECIMAL:            '2';
+FROM_CLAUSE_SINGLE_QUOTE_SYMB:      '\'';
+FROM_CLAUSE_DOUBLE_QUOTE_SYMB:      '"';
+FROM_CLAUSE_REVERSE_QUOTE_SYMB:     '`';
+FROM_CLAUSE_COLON_SYMB:             ':';
+
+//TODO: Implement the following
+//FROM_CLAUSE_LR_BRACKET:             '(';
+//FROM_CLAUSE_RR_BRACKET:             ')';
+//FROM_CLAUSE_COMMA:                  ',';
+
+FROM_CLAUSE_DECIMAL_LITERAL:        DECIMAL_LITERAL;
+
+IND:                                INDEX -> popMode;
+
+INDEX:                              INDEX_FRAGMENT | BACKTICK_QUOTE_IND;
+
+fragment INDEX_FRAGMENT:                     [*@A-Z]+?[*A-Z_\-0-9]*;
+fragment BACKTICK_QUOTE_IND:             '`' ( '\\'. | '``' | ~('`'|'\\'))* '`';
+
+
+FROM_CLAUSE_ERROR_RECOGNITION:      .    -> channel(ERRORCHANNEL);

sql/src/main/java/org/opensearch/sql/sql/parser/AstExpressionBuilder.java

@@ -98,7 +98,7 @@ public class AstExpressionBuilder extends OpenSearchSQLParserBaseVisitor<Unresol
 
   @Override
   public UnresolvedExpression visitTableName(TableNameContext ctx) {
-    return visit(ctx.qualifiedName());
+    return visit(ctx.indexName());
   }
 
   @Override

sql/src/test/java/org/opensearch/sql/sql/antlr/SQLSyntaxParserTest.java

@@ -298,9 +298,9 @@ public void can_parse_query_string_relevance_function() {
     assertNotNull(parser.parse(
         "SELECT id FROM test WHERE query_string([address], 'query')"));
     assertNotNull(parser.parse(
-        "SELECT id FROM test WHERE query_string([addr*], 'query')"));
+        "SELECT id FROM test WHERE query_string(['addr*'], 'query')"));
     assertNotNull(parser.parse(
-        "SELECT id FROM test WHERE query_string([*ss], 'query')"));
+        "SELECT id FROM test WHERE query_string(['*ss'], 'query')"));
     assertNotNull(parser.parse(
         "SELECT id FROM test WHERE query_string([address, notes], 'query')"));
 

sql/src/test/java/org/opensearch/sql/sql/parser/AstQualifiedNameBuilderTest.java

@@ -32,9 +32,8 @@ public void canBuildRegularIdentifierForSQLStandard() {
   @Test
   public void canBuildRegularIdentifierForOpenSearch() {
     buildFromTableName(".opensearch_dashboards").expectQualifiedName(".opensearch_dashboards");
-    buildFromIdentifier("@timestamp").expectQualifiedName("@timestamp");
     buildFromIdentifier("logs-2020-01").expectQualifiedName("logs-2020-01");
-    buildFromIdentifier("*logs*").expectQualifiedName("*logs*");
+    buildFromIdentifier("`*logs*`").expectQualifiedName("*logs*");
   }
 
   @Test