fix load data parser bug

src/sql/engine/cmd/ob_load_data_impl.cpp

@@ -1264,29 +1264,29 @@ int ObCSVParser::next_line(bool& yield_line)
   bool yield = false;
   int with_back_slash = 0;
 
-  for (; !yield && cur_pos_ != buf_end_pos_; ++cur_pos_) {
-
+  for (; !yield && cur_pos_ != buf_end_pos_; ++cur_pos_, ++cur_field_end_pos_) {
+    bool line_term_matched = false;
     if (*cur_pos_ == formats_.enclose_char_ && !in_enclose_flag_ && cur_pos_ == cur_field_begin_pos_) {
       in_enclose_flag_ = true;
-    }
-
-    if (!is_escaped_flag_ && *cur_pos_ == formats_.escape_char_) {
-      is_escaped_flag_ = true;
-    } else {
-      char escaped_res = *cur_pos_;
-      if (is_escaped_flag_) {
-        escaped_res = escaped_char(*cur_pos_, &with_back_slash);
+      last_end_enclosed_ = NULL;
+    } else if ((*cur_pos_ == formats_.escape_char_ && formats_.escape_char_ != formats_.enclose_char_) ||
+               (in_enclose_flag_ && formats_.enclose_char_ == *cur_pos_ && cur_pos_ < buf_end_pos_ &&
+                   formats_.enclose_char_ == *(cur_pos_ + 1))) {
+      if (cur_pos_ < buf_end_pos_) {
+        cur_pos_++;
+        if (!is_fast_parse_) {
+          *cur_field_end_pos_ = escaped_char(*cur_pos_, &with_back_slash);
+        }
       }
-      bool has_escaped = cur_field_end_pos_ != cur_pos_;
-      if (has_escaped && !is_fast_parse_) {
-        *cur_field_end_pos_ = escaped_res;
+    } else {
+      if (cur_field_end_pos_ != cur_pos_ && !is_fast_parse_) {
+        *cur_field_end_pos_ = *cur_pos_;
       }
-
-      bool line_term_matched = false;
-
-      if (is_terminate_char(*cur_pos_, cur_field_end_pos_, line_term_matched)) {
+      if (formats_.enclose_char_ == *cur_pos_) {
+        last_end_enclosed_ = cur_field_end_pos_;
+      } else if (is_terminate_char(*cur_pos_, cur_field_end_pos_, line_term_matched)) {
         if (!line_term_matched || cur_field_begin_pos_ < cur_pos_) {
-          handle_one_field(cur_field_end_pos_, has_escaped);
+          handle_one_field(cur_field_end_pos_, cur_field_end_pos_ != cur_pos_);
           field_id_++;
         }
         char* next_pos = cur_pos_ + 1;
@@ -1303,12 +1303,6 @@ int ObCSVParser::next_line(bool& yield_line)
           cur_line_begin_pos_ = next_pos;
         }
       }
-
-      if (is_escaped_flag_) {
-        is_escaped_flag_ = false;
-      }
-
-      ++cur_field_end_pos_;
     }
   }
 

src/sql/engine/cmd/ob_load_data_impl.h

@@ -542,6 +542,7 @@ public:
     cur_line_begin_pos_ = NULL;
     buf_begin_pos_ = NULL;
     buf_end_pos_ = NULL;
+    last_end_enclosed_ = NULL;
     field_id_ = 0;
     in_enclose_flag_ = false;
     is_escaped_flag_ = false;
@@ -590,12 +591,13 @@ private:
   common::ObBitSet<> string_type_column_;
   // parsing state variables
   bool is_last_buf_;
-  char* cur_pos_;
-  char* cur_field_begin_pos_;
-  char* cur_field_end_pos_;
-  char* cur_line_begin_pos_;
-  char* buf_begin_pos_;
-  char* buf_end_pos_;
+  char *cur_pos_;
+  char *cur_field_begin_pos_;
+  char *cur_field_end_pos_;
+  char *cur_line_begin_pos_;
+  char *buf_begin_pos_;
+  char *buf_end_pos_;
+  char *last_end_enclosed_;
   int64_t field_id_;
   bool in_enclose_flag_;
   bool is_escaped_flag_;
@@ -643,10 +645,8 @@ OB_INLINE bool ObCSVParser::is_terminate_char(char cur_char, char*& cur_pos, boo
     if (!in_enclose_flag_) {
       ret_bool = true;  // return true
     } else {
-      char* pre_pos = cur_pos - 1;
       // with in_enclose_flag_ = true, a term char is valid only if an enclosed char before it
-      if (static_cast<int64_t>(*pre_pos) == formats_.enclose_char_ &&
-          cur_field_begin_pos_ != pre_pos) {  // 123---->'---->123
+      if (last_end_enclosed_ == cur_pos - 1) {
         remove_enclosed_char(cur_pos);
         ret_bool = true;  // return true
       } else {

src/sql/resolver/cmd/ob_load_data_resolver.cpp

@@ -534,12 +534,14 @@ int ObLoadDataResolver::validate_stmt(ObLoadDataStmt* stmt)
       escape_char = (data_struct_in_file.field_escaped_str_.empty()
                          ? INT64_MAX
                          : static_cast<int64_t>(data_struct_in_file.field_escaped_str_[0]));
+      /*
       if (OB_SUCC(ret)) {
         if (escape_char != ObDataInFileStruct::DEFAULT_FIELD_ESCAPED_CHAR) {
           ret = OB_WRONG_FIELD_TERMINATORS;
           LOG_USER_ERROR(OB_WRONG_FIELD_TERMINATORS);
         }
       }
+      */
       if (OB_SUCC(ret)) {
         const char* is_ambiguous_field_sep = strchr("ntrb0ZN", static_cast<int>(field_sep_char));
         const char* is_unsafe_field_sep = strchr(".0123456789e+-", static_cast<int>(field_sep_char));