feat: Support Presto quoted_identifier semantics in parseTypeSignature (facebookincubator#13658)

Summary:
Presto supports
```
QUOTED_IDENTIFIER
    : '"' ( ~'"' | '""' )* '"'
    ;
```
The semantics are
```
Any character other than double quote is allowed between double quotes
And double double quote can be used to "escape" double quote
```

Fixes: facebookincubator#13648

Pull Request resolved: facebookincubator#13658

Reviewed By: mbasmanova, kagamiori

Differential Revision: D76071088

Pulled By: kKPulla

fbshipit-source-id: c1af839d997092fbaa075911d426359e236ecaf2

Author: majetideepak

velox/expression/signature_parser/SignatureParser.ll

@@ -5,6 +5,30 @@
 #include "velox/expression/signature_parser/SignatureParser.yy.h"  // @manual
 #include "velox/expression/signature_parser/Scanner.h"
 #define YY_DECL int facebook::velox::exec::Scanner::lex(facebook::velox::exec::Parser::semantic_type *yylval)
+
+std::string unescape_doublequote(const char* yytext) {
+    size_t len = strlen(yytext);
+    std::string output;
+    output.resize(len);
+
+    int i = 0;
+    int j = 0;
+
+    while (i < len - 1) {
+        if (yytext[i] == '"' && yytext[i+1] == '"') {
+            output[j++] = '"';
+            i += 2;
+        } else {
+            output[j++] = yytext[i++];
+        }
+    }
+    // Check if the last character needs to be added.
+    if (i < len) {
+       output[j++] = yytext[i++];
+    }
+    output.resize(j);
+    return output;
+}
 %}
 
 %option c++ noyywrap noyylineno nodefault caseless
@@ -34,7 +58,7 @@ Y   [Y|y]
 Z   [Z|z]
 
 WORD              ([[:alnum:]_]*)
-QUOTED_ID         (['"'][[:alnum:][:space:]_]*['"'])
+QUOTED_ID         (\"([^\"\n]|\"\")*\")
 ROW               (ROW|STRUCT)
 
 %%
@@ -48,7 +72,7 @@ ROW               (ROW|STRUCT)
 (DECIMAL)          yylval->build<std::string>(YYText()); return Parser::token::DECIMAL;
 {ROW}              return Parser::token::ROW;
 {WORD}             yylval->build<std::string>(YYText()); return Parser::token::WORD;
-{QUOTED_ID}        yylval->build<std::string>(YYText()); return Parser::token::QUOTED_ID;
+{QUOTED_ID}        {auto val = unescape_doublequote(YYText()); yylval->build<std::string>(val.c_str()); return Parser::token::QUOTED_ID;}
 <<EOF>>            return Parser::token::YYEOF;
 .               /* no action on unmatched input */
 

velox/expression/signature_parser/tests/ParseTypeSignatureTest.cpp

@@ -230,6 +230,35 @@ TEST_F(ParseTypeSignatureTest, row) {
     ASSERT_EQ(rowfield.rowFieldName(), "bla");
     ASSERT_EQ(rowfield.parameters().size(), 0);
   }
+
+  {
+    auto signature = parseTypeSignature("row(\"a (b)\" INTEGER)");
+    ASSERT_EQ(signature.baseName(), "row");
+    ASSERT_EQ(signature.parameters().size(), 1);
+    auto field0 = signature.parameters()[0];
+    ASSERT_EQ(field0.baseName(), "INTEGER");
+    ASSERT_EQ(field0.rowFieldName(), "a (b)");
+  }
+
+  // Test double double quote escape
+  {
+    auto signature = parseTypeSignature(
+        "row(\"a\"\"b\" INTEGER, \"\"\"ab\" INTEGER, \"ab\"\"\"\"\" INTEGER)");
+    ASSERT_EQ(signature.baseName(), "row");
+    ASSERT_EQ(signature.parameters().size(), 3);
+    auto field0 = signature.parameters()[0];
+    ASSERT_EQ(field0.baseName(), "INTEGER");
+    ASSERT_EQ(field0.rowFieldName(), "a\"b");
+    auto field1 = signature.parameters()[1];
+    ASSERT_EQ(field1.baseName(), "INTEGER");
+    ASSERT_EQ(field1.rowFieldName(), "\"ab");
+    auto field2 = signature.parameters()[2];
+    ASSERT_EQ(field2.baseName(), "INTEGER");
+    ASSERT_EQ(field2.rowFieldName(), "ab\"\"");
+  }
+
+  // Single double quote is an error.
+  EXPECT_THROW(parseTypeSignature("row(\"a\"b\" INTEGER)"), VeloxRuntimeError);
 }
 
 TEST_F(ParseTypeSignatureTest, tdigest) {