From 250d8e25bb9377ecda130b10e926a6bc8a1ed4b9 Mon Sep 17 00:00:00 2001
From: wangt1xiuyi <13547954130@163.com>
Date: Thu, 8 Feb 2024 13:51:23 +0000
Subject: [PATCH] fix const number of fast parse and normal parse is different
 with mysql sql comment and parser error code refine

---
 src/sql/parser/gen_parser.sh           |  4 ++++
 src/sql/parser/ob_fast_parser.cpp      | 18 ++++++++++++++----
 src/sql/parser/sql_parser_base.c       | 11 ++++++++++-
 src/sql/parser/sql_parser_mysql_mode.l | 13 +++++++++++++
 src/sql/parser/sql_parser_mysql_mode.y |  4 +++-
 5 files changed, 44 insertions(+), 6 deletions(-)

diff --git a/src/sql/parser/gen_parser.sh b/src/sql/parser/gen_parser.sh
index 051f0e02c6..6fe26d54f1 100755
--- a/src/sql/parser/gen_parser.sh
+++ b/src/sql/parser/gen_parser.sh
@@ -40,6 +40,7 @@ flex -Cfa -B -8 -o ../../../src/sql/parser/sql_parser_mysql_mode_lex.c ../../../
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/int i/d}" -i ../../../src/sql/parser/sql_parser_mysql_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/for ( i = 0; i < _yybytes_len; ++i )/d}" -i ../../../src/sql/parser/sql_parser_mysql_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{s/\tbuf\[i\] = yybytes\[i\]/memcpy(buf, yybytes, _yybytes_len)/g}" -i ../../../src/sql/parser/sql_parser_mysql_mode_lex.c
+sed "/obsql_mysql_yylex_init is special because it creates the scanner itself/,/Initialization is the same as for the non-reentrant scanner/{s/return 1/return errno/g}" -i ../../../src/sql/parser/sql_parser_mysql_mode_lex.c
 
 
 if [ -d "../../../close_modules/oracle_parser/sql/parser" ]; then
@@ -82,6 +83,7 @@ flex -o ../../../src/sql/parser/sql_parser_oracle_latin1_mode_lex.c ../../../src
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/int i/d}" -i ../../../src/sql/parser/sql_parser_oracle_latin1_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/for ( i = 0; i < _yybytes_len; ++i )/d}" -i ../../../src/sql/parser/sql_parser_oracle_latin1_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{s/\tbuf\[i\] = yybytes\[i\]/memcpy(buf, yybytes, _yybytes_len)/g}" -i ../../../src/sql/parser/sql_parser_oracle_latin1_mode_lex.c
+sed "/obsql_oracle_latin1_yylex_init is special because it creates the scanner itself/,/Initialization is the same as for the non-reentrant scanner/{s/return 1/return errno/g}" -i ../../../src/sql/parser/sql_parser_oracle_latin1_mode_lex.c
 cat ../../../src/sql/parser/non_reserved_keywords_oracle_mode.c > ../../../src/sql/parser/non_reserved_keywords_oracle_latin1_mode.c
 sed '/#include "ob_non_reserved_keywords.h"/a\#include "sql/parser/sql_parser_oracle_latin1_mode_tab.h\"'  -i ../../../src/sql/parser/non_reserved_keywords_oracle_latin1_mode.c
 sed  "s/non_reserved_keywords_oracle_mode.c is for …/non_reserved_keywords_oracle_latin1_mode.c is auto generated by gen_parser.sh/g" -i ../../../src/sql/parser/non_reserved_keywords_oracle_latin1_mode.c
@@ -139,6 +141,7 @@ flex -o ../../../src/sql/parser/sql_parser_oracle_utf8_mode_lex.c ../../../src/s
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/int i/d}" -i ../../../src/sql/parser/sql_parser_oracle_utf8_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/for ( i = 0; i < _yybytes_len; ++i )/d}" -i ../../../src/sql/parser/sql_parser_oracle_utf8_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{s/\tbuf\[i\] = yybytes\[i\]/memcpy(buf, yybytes, _yybytes_len)/g}" -i ../../../src/sql/parser/sql_parser_oracle_utf8_mode_lex.c
+sed "/obsql_oracle_utf8_yylex_init is special because it creates the scanner itself/,/Initialization is the same as for the non-reentrant scanner/{s/return 1/return errno/g}" -i ../../../src/sql/parser/sql_parser_oracle_utf8_mode_lex.c
 cat ../../../src/sql/parser/non_reserved_keywords_oracle_mode.c > ../../../src/sql/parser/non_reserved_keywords_oracle_utf8_mode.c
 sed '/#include "ob_non_reserved_keywords.h"/a\#include "sql/parser/sql_parser_oracle_utf8_mode_tab.h\"'  -i ../../../src/sql/parser/non_reserved_keywords_oracle_utf8_mode.c
 sed  "s/non_reserved_keywords_oracle_mode.c is for …/non_reserved_keywords_oracle_utf8_mode.c is auto generated by gen_parser.sh/g" -i ../../../src/sql/parser/non_reserved_keywords_oracle_utf8_mode.c
@@ -194,6 +197,7 @@ flex -o ../../../src/sql/parser/sql_parser_oracle_gbk_mode_lex.c ../../../src/sq
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/int i/d}" -i ../../../src/sql/parser/sql_parser_oracle_gbk_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{/for ( i = 0; i < _yybytes_len; ++i )/d}" -i ../../../src/sql/parser/sql_parser_oracle_gbk_mode_lex.c
 sed "/Setup the input buffer state to scan the given bytes/,/}/{s/\tbuf\[i\] = yybytes\[i\]/memcpy(buf, yybytes, _yybytes_len)/g}" -i ../../../src/sql/parser/sql_parser_oracle_gbk_mode_lex.c
+sed "/obsql_oracle_gbk_yylex_init is special because it creates the scanner itself/,/Initialization is the same as for the non-reentrant scanner/{s/return 1/return errno/g}" -i ../../../src/sql/parser/sql_parser_oracle_gbk_mode_lex.c
 cat ../../../src/sql/parser/non_reserved_keywords_oracle_mode.c > ../../../src/sql/parser/non_reserved_keywords_oracle_gbk_mode.c
 sed '/#include "ob_non_reserved_keywords.h"/a\#include "sql/parser/sql_parser_oracle_gbk_mode_tab.h\"'  -i ../../../src/sql/parser/non_reserved_keywords_oracle_gbk_mode.c
 sed  "s/non_reserved_keywords_oracle_mode.c is for …/non_reserved_keywords_oracle_gbk_mode.c is auto generated by gen_parser.sh/g" -i ../../../src/sql/parser/non_reserved_keywords_oracle_gbk_mode.c
diff --git a/src/sql/parser/ob_fast_parser.cpp b/src/sql/parser/ob_fast_parser.cpp
index 9a0eee0f8e..8217c7593d 100644
--- a/src/sql/parser/ob_fast_parser.cpp
+++ b/src/sql/parser/ob_fast_parser.cpp
@@ -2754,16 +2754,26 @@ int ObFastParserMysql::parse_next_token()
       }
       case '-': {
         // need to deal with sql_comment or negative sign
-        int64_t space_len = 0;
         ch = raw_sql_.scan();
-        if ('-' == ch && IS_MULTI_SPACE(raw_sql_.cur_pos_ + 1, space_len)) {
-          // "--"{space}+{non_newline}*
+        if ('-' == ch &&
+            raw_sql_.cur_pos_ + 1 < raw_sql_.raw_sql_len_ &&
+            (raw_sql_.raw_sql_[raw_sql_.cur_pos_ + 1] == ' ' ||
+             raw_sql_.raw_sql_[raw_sql_.cur_pos_ + 1] == '\t')) {
+          // "--"[ \t]+{non_newline}*
           cur_token_type_ = IGNORE_TOKEN;
           // skip the second '-' and space
-          raw_sql_.scan(1 + space_len);
+          raw_sql_.scan(1);
           while (!raw_sql_.is_search_end() && is_non_newline(ch)) {
             ch = raw_sql_.scan();
           }
+        } else if ('-' == ch &&
+                   raw_sql_.cur_pos_ + 1 < raw_sql_.raw_sql_len_ &&
+                   (raw_sql_.raw_sql_[raw_sql_.cur_pos_ + 1] == '\n' ||
+                    raw_sql_.raw_sql_[raw_sql_.cur_pos_ + 1] == '\r')) {
+          // "--"[\n\r]
+          cur_token_type_ = IGNORE_TOKEN;
+          //skip the second '-' and ('\n' or \r)
+          raw_sql_.scan(1);
         } else {
           OZ (process_negative());
         }
diff --git a/src/sql/parser/sql_parser_base.c b/src/sql/parser/sql_parser_base.c
index c360633258..be85268dd2 100644
--- a/src/sql/parser/sql_parser_base.c
+++ b/src/sql/parser/sql_parser_base.c
@@ -62,7 +62,7 @@ int parse_init(ParseResult *p)
   static __thread char error_msg[MAX_ERROR_MSG] = {'\0'};
   p->error_msg_ = error_msg;
   if (OB_UNLIKELY(NULL == p || NULL == p->malloc_pool_)) {
-    ret = -1;
+    ret = OB_PARSER_ERR_UNEXPECTED;
     if (NULL != p) {
       (void)snprintf(p->error_msg_, MAX_ERROR_MSG, "malloc_pool_ must be set");
     }
@@ -108,6 +108,15 @@ int parse_init(ParseResult *p)
 #ifdef OB_BUILD_ORACLE_PARSER
     }
 #endif
+#define	ENOMEM		12	/* Out of memory */
+    //refine parser error code to OB error code
+    if (0 == ret) {
+      ret = OB_PARSER_SUCCESS;
+    } else if (ENOMEM == ret) {
+      ret = OB_PARSER_ERR_NO_MEMORY;
+    } else {
+      ret = OB_PARSER_ERR_PARSE_SQL;
+    }
   }
   return ret;
 }
diff --git a/src/sql/parser/sql_parser_mysql_mode.l b/src/sql/parser/sql_parser_mysql_mode.l
index 266d959d26..4898d325d5 100644
--- a/src/sql/parser/sql_parser_mysql_mode.l
+++ b/src/sql/parser/sql_parser_mysql_mode.l
@@ -135,6 +135,7 @@ NULL {
   check_value(yylval);
   ParseResult *p = (ParseResult *)yyextra;
   malloc_new_node(yylval->node, p->malloc_pool_, T_NULL, 0);
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   COPY_STRING(p->input_sql_ + yylloc->first_column - 1, yyleng, yylval->node->raw_text_);
   yylval->node->text_len_ = yyleng;
   if (IS_FAST_PARAMETERIZE) {
@@ -152,6 +153,7 @@ NULL {
   ParseResult *p = (ParseResult *)yyextra;
   malloc_new_node(node, p->malloc_pool_, T_INT, 0);
   yylval->node = node;
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   int err_no = 0;
   COPY_NUM_STRING(p, node);
   // we treated '-  12' as a const node, that is to say node->str_value_ = '-  12'
@@ -205,6 +207,7 @@ NULL {
   malloc_new_node(node, p->malloc_pool_, T_DOUBLE, 0);
   check_value(yylval);
   yylval->node = node;
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   COPY_NUM_STRING(p, node);
   node->raw_text_ = node->str_value_;
   node->text_len_ = node->str_len_;
@@ -224,6 +227,7 @@ NULL {
   malloc_new_node(node, p->malloc_pool_, T_NUMBER/* should be T_NUMBER,*/, 0);
   check_value(yylval);
   yylval->node = node;
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   COPY_NUM_STRING(p, node);
   node->raw_text_ = node->str_value_;
   node->text_len_ = node->str_len_;
@@ -240,6 +244,7 @@ TRUE {
   check_value(yylval);
   ParseResult *p = (ParseResult *)yyextra;
   malloc_new_node(yylval->node, p->malloc_pool_, T_BOOL, 0);
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   yylval->node->value_ = 1;
   COPY_STRING(p->input_sql_ + yylloc->first_column - 1, yyleng, yylval->node->raw_text_);
   yylval->node->text_len_ = yyleng;
@@ -265,6 +270,7 @@ FALSE {
   check_value(yylval);
   ParseResult *p = (ParseResult *)yyextra;
   malloc_new_node(yylval->node, p->malloc_pool_, T_BOOL, 0);
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   yylval->node->value_ = 0;
   COPY_STRING(p->input_sql_ + yylloc->first_column - 1, yyleng, yylval->node->raw_text_);
   yylval->node->text_len_ = yyleng;
@@ -298,6 +304,7 @@ FALSE {
   }
   check_value(yylval);
   malloc_new_node(yylval->node, p->malloc_pool_, T_NCHAR, 0);
+  CHECK_MYSQL_COMMENT(p, yylval->node);
 
   yylval->node->str_len_ = 0;
   if (IS_FAST_PARAMETERIZE && !IS_NEED_PARAMETERIZE) {
@@ -312,6 +319,7 @@ FALSE {
   p->start_col_ = yylloc->first_column;
   check_value(yylval);
   malloc_new_node(yylval->node, p->malloc_pool_, T_VARCHAR, 0);
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   yylval->node->str_len_ = 0;
   p->last_escape_check_pos_ = 0;
   yylval->node->str_value_ = NULL;
@@ -412,6 +420,7 @@ FALSE {
     }
 
     malloc_new_node(yylval->node->children_[0], p->malloc_pool_, T_CONCAT_STRING, 0);
+    CHECK_MYSQL_COMMENT(p, yylval->node->children_[0]);
     (*yylval->node->children_)->str_value_ = parse_strndup(tmp_literal, yylval->node->str_len_ + 1,
                                             p->malloc_pool_);
     check_malloc((*yylval->node->children_)->str_value_);
@@ -459,6 +468,7 @@ FALSE {
       check_malloc(*tmp_literal);
     }
     malloc_new_node(yylval->node, p->malloc_pool_, T_VARCHAR, 0);
+    CHECK_MYSQL_COMMENT(p, yylval->node);
     yylval->node->str_len_ = 0;
     p->last_escape_check_pos_ = 0;
     if (IS_FAST_PARAMETERIZE && !IS_NEED_PARAMETERIZE) {
@@ -553,6 +563,7 @@ FALSE {
     }
 
     malloc_new_node(yylval->node->children_[0], p->malloc_pool_, T_CONCAT_STRING, 0);
+    CHECK_MYSQL_COMMENT(p, yylval->node->children_[0]);
     (*yylval->node->children_)->str_value_ = parse_strndup(tmp_literal, yylval->node->str_len_ + 1,
                                             p->malloc_pool_);
     check_malloc((*yylval->node->children_)->str_value_);
@@ -730,6 +741,7 @@ X'([0-9A-F])*'|0X([0-9A-F])+ {
     node->str_len_ = 0;
   }
   yylval->node = node;
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   COPY_STRING(p->input_sql_ + yylloc->first_column - 1, yyleng, yylval->node->raw_text_);
   yylval->node->text_len_ = yyleng;
   yylval->node->is_copy_raw_text_ = 1;
@@ -773,6 +785,7 @@ B'([01])*'|0B([01])+ {
     node->str_len_ = 0;
   }
   yylval->node = node;
+  CHECK_MYSQL_COMMENT(p, yylval->node);
   COPY_STRING(p->input_sql_ + yylloc->first_column - 1, yyleng, yylval->node->raw_text_);
   yylval->node->text_len_ = yyleng;
   yylval->node->is_copy_raw_text_ = 1;
diff --git a/src/sql/parser/sql_parser_mysql_mode.y b/src/sql/parser/sql_parser_mysql_mode.y
index a254030326..cb17fc60c0 100644
--- a/src/sql/parser/sql_parser_mysql_mode.y
+++ b/src/sql/parser/sql_parser_mysql_mode.y
@@ -968,6 +968,7 @@ STRING_VALUE %prec LOWER_THAN_COMP
   $$->sql_str_off_ = @1.first_column;
   @$.first_column = @1.first_column;
   @$.last_column = @1.last_column;
+  $$->is_forbid_parameter_ = $1->is_forbid_parameter_;
 }
 | charset_introducer STRING_VALUE
 {
@@ -977,6 +978,7 @@ STRING_VALUE %prec LOWER_THAN_COMP
   $$->raw_text_ = $2->raw_text_;
   $$->text_len_ = $2->text_len_;
   $$->sql_str_off_ = $2->sql_str_off_;
+  $$->is_forbid_parameter_ = $2->is_forbid_parameter_;
 }
 | charset_introducer HEX_STRING_VALUE
 {
@@ -987,6 +989,7 @@ STRING_VALUE %prec LOWER_THAN_COMP
   $$->raw_text_ = $2->raw_text_;
   $$->text_len_ = $2->text_len_;
   $$->sql_str_off_ = $2->sql_str_off_;
+  $$->is_forbid_parameter_ = $2->is_forbid_parameter_;
 }
 | STRING_VALUE string_val_list %prec LOWER_THAN_COMP
 {
@@ -1106,7 +1109,6 @@ literal
 {
   $$ = $1;
   $$->sql_str_off_ = $1->sql_str_off_;
-  CHECK_MYSQL_COMMENT(result, $$);
 }
 | SYSTEM_VARIABLE { $$ = $1; }
 | QUESTIONMARK { $$ = $1; }