feat: add lexcial and syntax support for rlike (#41)

* add lexcial and syntax support for rlike

* add standlone_expression test and modify the output of the test related

* fix typo and sign commit

zetasql/parser/bison_parser.y

@@ -556,7 +556,7 @@ class DashedIdentifierTmpNode final : public zetasql::ASTNode {
 %left "AND"
 %left "XOR"
 %left UNARY_NOT_PRECEDENCE
-%nonassoc "=" "==" "<>" ">" "<" ">=" "<=" "!=" "LIKE" "ILIKE" "IN" "DISTINCT" "BETWEEN" "IS" "NOT_SPECIAL"
+%nonassoc "=" "==" "<>" ">" "<" ">=" "<=" "!=" "LIKE" "ILIKE" "RLIKE" "IN" "DISTINCT" "BETWEEN" "IS" "NOT_SPECIAL"
 %nonassoc "ESCAPE"
 %left "|"
 %left "^"
@@ -714,6 +714,7 @@ using zetasql::ASTDropStatement;
 %token KW_LEFT "LEFT"
 %token KW_LIKE "LIKE"
 %token KW_ILIKE "ILIKE"
+%token KW_RLIKE "RLIKE"
 %token KW_LIMIT "LIMIT"
 %token KW_LOOKUP "LOOKUP"
 %token KW_MERGE "MERGE"
@@ -1385,6 +1386,7 @@ using zetasql::ASTDropStatement;
 %type <not_keyword_presence> is_operator
 %type <not_keyword_presence> like_operator
 %type <not_keyword_presence> ilike_operator
+%type <not_keyword_presence> rlike_operator
 %type <not_keyword_presence> distinct_operator
 
 %type <preceding_or_following_keyword> preceding_or_following
@@ -5405,6 +5407,15 @@ ilike_operator:
       } %prec "ILIKE"
     ;
 
+rlike_operator:
+    "RLIKE" { $$ = NotKeywordPresence::kAbsent; } %prec "RLIKE"
+    | "NOT_SPECIAL" "RLIKE"
+      {
+        @$ = @2;  // Error messages should point at the "RLIKE".
+        $$ = NotKeywordPresence::kPresent;
+      } %prec "RLIKE"
+    ;
+
 // Returns NotKeywordPresence to indicate whether NOT was present.
 between_operator:
     "BETWEEN"
@@ -5630,6 +5641,27 @@ expression:
         {
           $$ = MAKE_NODE(ASTEscapedExpression, @$, {$1, $3})
         }
+    | expression rlike_operator expression %prec "RLIKE"
+        {
+          // NOT has lower precedence but can be parsed unparenthesized in the
+          // rhs because it is not ambiguous. This is not allowed.
+          if (IsUnparenthesizedNotExpression($3)) {
+            YYERROR_UNEXPECTED_AND_ABORT_AT(@3);
+          }
+          // Bison allows some cases like IN on the left hand side because it's
+          // not ambiguous. The language doesn't allow this.
+          if (!$1->IsAllowedInComparison()) {
+            YYERROR_AND_ABORT_AT(
+                @2,
+                "Syntax error: "
+                "Expression to the left of RIKE must be parenthesized");
+          }
+          auto* binary_expression =
+              MAKE_NODE(ASTBinaryExpression, @1, @3, {$1, $3});
+          binary_expression->set_is_not($2 == NotKeywordPresence::kPresent);
+          binary_expression->set_op(zetasql::ASTBinaryExpression::RLIKE);
+          $$ = binary_expression;
+        }
     | expression distinct_operator expression %prec "DISTINCT"
         {
           if (parser->language_options() == nullptr
@@ -7353,6 +7385,7 @@ reserved_keyword_rule:
     | "RECURSIVE"
     | "RESPECT"
     | "RIGHT"
+    | "RLIKE"
     | "ROLLUP"
     | "ROWS"
     | "ROWS_RANGE"

zetasql/parser/flex_tokenizer.l

@@ -512,6 +512,7 @@ left { return BisonParserImpl::token::KW_LEFT; }
 level { return BisonParserImpl::token::KW_LEVEL; }
 like { return BisonParserImpl::token::KW_LIKE; }
 ilike { return BisonParserImpl::token::KW_ILIKE; }
+rlike { return BisonParserImpl::token::KW_RLIKE; }
 limit { return BisonParserImpl::token::KW_LIMIT; }
 load { return BisonParserImpl::token::KW_LOAD; }
 lookup { return BisonParserImpl::token::KW_LOOKUP; }
@@ -539,14 +540,14 @@ no { return BisonParserImpl::token::KW_NO; }
     file will cause bad error messages. There is no situation where that is
     valid syntax, so there will never be any rejections as a result.
  */
-not{whitespace}(between|in|like|ilike|distinct)[^A-Z_0-9] {
+not{whitespace}(between|in|like|ilike|rlike|distinct)[^A-Z_0-9] {
   SET_RETURN_PREFIX_LENGTH(3);
   if (mode_ == BisonParserMode::kTokenizer) {
     return BisonParserImpl::token::KW_NOT;
   }
   return BisonParserImpl::token::KW_NOT_SPECIAL;
 }
-"!"{whitespace}(between|in|like|ilike|distinct)[^A-Z_0-9] {
+"!"{whitespace}(between|in|like|ilike|rlike|distinct)[^A-Z_0-9] {
   SET_RETURN_PREFIX_LENGTH(1);
   if (mode_ == BisonParserMode::kTokenizer) {
     return BisonParserImpl::token::KW_NOT;

zetasql/parser/keywords.cc

@@ -189,6 +189,7 @@ constexpr KeywordInfoPOD kAllKeywords[] = {
     {"level", KW_LEVEL},
     {"like", KW_LIKE, KeywordInfo::kReserved},
     {"ilike", KW_ILIKE, KeywordInfo::kReserved},
+    {"rlike", KW_RLIKE, KeywordInfo::kReserved},
     {"limit", KW_LIMIT, KeywordInfo::kReserved},
     {"load", KW_LOAD},
     {"lookup", KW_LOOKUP, KeywordInfo::kReserved},

zetasql/parser/keywords_test.cc

@@ -237,7 +237,7 @@ TEST(ParserTest, DontAddNewReservedKeywords) {
   // allows new queries to work that will not work on older code.
   // Before changing this, co-ordinate with all engines to make sure the change
   // is done safely.
-  EXPECT_EQ(105 /* CAUTION */, num_reserved);
+  EXPECT_EQ(106 /* CAUTION */, num_reserved);
 }
 
 }  // namespace

zetasql/parser/parse_tree.cc

@@ -770,6 +770,8 @@ std::string ASTBinaryExpression::GetSQLForOperator() const {
       return is_not_ ? "IS NOT DISTINCT FROM" : "IS DISTINCT FROM";
     case ILIKE:
       return is_not_ ? "NOT ILIKE" : "ILIKE";
+    case RLIKE:
+      return is_not_ ? "NOT RLIKE" : "RLIKE";
   }
 }
 

zetasql/parser/parse_tree_manual.h

@@ -2030,6 +2030,7 @@ class ASTBinaryExpression final : public ASTExpression {
     XOR,          // "XOR"
     DISTINCT,     // "IS DISTINCT FROM"
     ILIKE,        // "ILIKE"
+    RLIKE,        // "RLIKE"
   };
 
   void set_op(Op op) { op_ = op; }
@@ -2048,6 +2049,7 @@ class ASTBinaryExpression final : public ASTExpression {
     switch (op()) {
       case LIKE:
       case ILIKE:
+      case RLIKE:
       case IS:
       case EQ:
       case NE:

zetasql/parser/testdata/is_distinct.test

@@ -154,7 +154,7 @@ SELECT
 # Syntax error
 SELECT 1 NOT DISTINCT FROM 2;
 --
-ERROR: Syntax error: Expected keyword BETWEEN or keyword ILIKE or keyword IN or keyword LIKE but got keyword DISTINCT [at 1:14]
+ERROR: Syntax error: Unexpected keyword DISTINCT [at 1:14]
 SELECT 1 NOT DISTINCT FROM 2;
              ^
 ==

zetasql/parser/testdata/standalone_expression.test

@@ -48,6 +48,16 @@ BinaryExpression(NOT ILIKE) [0-30]
 col1 NOT ILIKE '&_' ESCAPE '&'
 ==
 
+# rlike
+col1 NOT RLIKE '.a'
+--
+BinaryExpression(NOT RLIKE) [0-19]
+  PathExpression [0-4]
+    Identifier(col1) [0-4]
+  StringLiteral('.a') [15-19]
+--
+col1 NOT RLIKE '.a'
+==
 
 abc + @param + @@sysvar + count(*) + (select 1) + (select x from y)
 --