hcq/oceanbase
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[FEAT MERGE]4_1_sql_feature

Browse Source 
Co-authored-by: leslieyuchen <leslieyuchen@gmail.com>
Co-authored-by: Charles0429 <xiezhenjiang@gmail.com>
Co-authored-by: raywill <hustos@gmail.com>
This commit is contained in:
obdev
2023-01-28 16:01:26 +08:00
committed by ob-robot
parent 3080f2b66f
commit 2d19a9d8f5
846 changed files with 161957 additions and 116661 deletions
Download Patch File Download Diff File Expand all files Collapse all files

462
src/sql/engine/expr/ob_expr_regexp_replace.cpp
View File

@ -1,3 +1,4 @@
/**
* Copyright (c) 2021 OceanBase
* OceanBase CE is licensed under Mulan PubL v2.
@ -42,13 +43,19 @@ int ObExprRegexpReplace::calc_result_typeN(ObExprResType &type,
int64_t param_num,
common::ObExprTypeCtx &type_ctx) const
{
UNUSED(type_ctx);
int ret = OB_SUCCESS;
ObRawExpr * raw_expr = type_ctx.get_raw_expr();
CK(NULL != type_ctx.get_raw_expr());
int64_t max_allowed_packet = 0;
if (OB_FAIL(ret)) {
} else if (OB_UNLIKELY(param_num < 2 || param_num > 6)) {
ret = OB_ERR_PARAM_SIZE;
LOG_WARN("param number of regexp_replace at least 2 and at most 6", K(ret), K(param_num));
} else if (OB_ISNULL(type_ctx.get_session())) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("get unexpected null", K(ret), K(type_ctx.get_session()));
} else if (OB_FAIL(type_ctx.get_session()->get_max_allowed_packet(max_allowed_packet))) {
LOG_WARN("failed to get max allowed packet", K(ret));
} else {
for (int i = 0; OB_SUCC(ret) && i < param_num; i++) {
if (!types[i].is_null() && !is_type_valid(types[i].get_type())) {
@ -56,76 +63,92 @@ int ObExprRegexpReplace::calc_result_typeN(ObExprResType &type,
LOG_WARN("the parameter is not castable", K(ret), K(i));
}
}
if (lib::is_oracle_mode()) {
//deduce length
ObExprResType &text = types[0];
int64_t to_len = types[2].get_length();
common::ObLength len = text.get_length();
int64_t offset = len * to_len;
len = static_cast<common::ObLength>(len + offset);
CK(len <= INT32_MAX);
type.set_length(static_cast<common::ObLength>(len));
auto input_params = make_const_carray(&types[0]);
OZ(aggregate_string_type_and_charset_oracle(
*type_ctx.get_session(), input_params, type, PREFER_VAR_LEN_CHAR));
OZ(deduce_string_param_calc_type_and_charset(*type_ctx.get_session(), type, input_params));
OX(type.set_length_semantics(type_ctx.get_session()->get_actual_nls_length_semantics()));
} else {
const common::ObLengthSemantics default_length_semantics = (OB_NOT_NULL(type_ctx.get_session())
? type_ctx.get_session()->get_actual_nls_length_semantics()
: common::LS_BYTE);
if (types[0].is_lob()) {
type.set_type(types[0].get_type());
if (OB_SUCC(ret)) {
if (lib::is_oracle_mode()) {
//deduce length
ObExprResType &text = types[0];
int64_t to_len = types[2].get_length();
common::ObLength len = text.get_length();
int64_t offset = len * to_len;
len = static_cast<common::ObLength>(len + offset);
CK(len <= INT32_MAX);
type.set_length(static_cast<common::ObLength>(len));
auto input_params = make_const_carray(&types[0]);
OZ(aggregate_string_type_and_charset_oracle(
*type_ctx.get_session(), input_params, type, PREFER_VAR_LEN_CHAR));
OZ(deduce_string_param_calc_type_and_charset(*type_ctx.get_session(), type, input_params));
OX(type.set_length_semantics(type_ctx.get_session()->get_actual_nls_length_semantics()));
} else {
type.set_varchar();
type.set_length_semantics(types[0].is_varchar_or_char() ? types[0].get_length_semantics() : default_length_semantics);
const common::ObLengthSemantics default_length_semantics = (OB_NOT_NULL(type_ctx.get_session())
? type_ctx.get_session()->get_actual_nls_length_semantics()
: common::LS_BYTE);
if (types[0].is_lob()) {
type.set_type(types[0].get_type());
} else {
type.set_clob();
type.set_length_semantics(types[0].is_varchar_or_char() ? types[0].get_length_semantics() : default_length_semantics);
}
//建表列的最大长度
type.set_length(max_allowed_packet);
ret = aggregate_charsets_for_string_result(type, types, 1, type_ctx.get_coll_type());
}
//建表列的最大长度
type.set_length(OB_MAX_COLUMN_NAMES_LENGTH);
ret = aggregate_charsets_for_string_result(type, types, 1, type_ctx.get_coll_type());
}
if (OB_SUCC(ret)) {
switch (param_num) {//because of regexp engine need utf8 code, here need reset calc collation type and can implicit cast.
case 6:
bool is_case_sensitive = ObCharset::is_bin_sort(types[0].get_calc_collation_type());
bool need_utf8 = false;
switch (param_num) {
case 6/*match type*/:
types[5].set_calc_type(ObVarcharType);
case 5:
if (lib::is_oracle_mode()) {
types[4].set_calc_type(ObNumberType);
} else {
type_ctx.set_cast_mode(type_ctx.get_cast_mode() | CM_STRING_INTEGER_TRUNC);
types[4].set_calc_type(ObIntType);
}
case 4:
if (lib::is_oracle_mode()) {
types[3].set_calc_type(ObNumberType);
} else {
type_ctx.set_cast_mode(type_ctx.get_cast_mode() | CM_STRING_INTEGER_TRUNC);
types[3].set_calc_type(ObIntType);
}
case 3:
types[5].set_calc_collation_type(CS_TYPE_UTF8MB4_BIN);
types[5].set_calc_collation_level(CS_LEVEL_IMPLICIT);
case 5/*occurence*/:
types[4].set_calc_type(lib::is_oracle_mode() ? ObNumberType : ObIntType);
types[4].set_scale(lib::is_oracle_mode() ? NUMBER_SCALE_UNKNOWN_YET : DEFAULT_SCALE_FOR_INTEGER);
types[4].set_precision(lib::is_oracle_mode() ? PRECISION_UNKNOWN_YET :
ObAccuracy::DDL_DEFAULT_ACCURACY[ObIntType].precision_);
case 4/*position*/:
types[3].set_calc_type(lib::is_oracle_mode() ? ObNumberType : ObIntType);
types[3].set_scale(lib::is_oracle_mode() ? NUMBER_SCALE_UNKNOWN_YET : DEFAULT_SCALE_FOR_INTEGER);
types[3].set_precision(lib::is_oracle_mode() ? PRECISION_UNKNOWN_YET :
ObAccuracy::DDL_DEFAULT_ACCURACY[ObIntType].precision_);
type_ctx.set_cast_mode(type_ctx.get_cast_mode() | CM_STRING_INTEGER_TRUNC);
case 3/*replace string*/:
types[2].set_calc_type(ObVarcharType);
if (ObCharset::is_bin_sort(types[2].get_calc_collation_type())) {
types[2].set_calc_collation_type(CS_TYPE_UTF8MB4_BIN);
types[2].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF8MB4_BIN : CS_TYPE_UTF8MB4_GENERAL_CI);
types[2].set_calc_collation_level(CS_LEVEL_IMPLICIT);
need_utf8 = false;
if (OB_FAIL(ret)) {
} else if (OB_FAIL(ObExprRegexContext::check_need_utf8(raw_expr->get_param_expr(2), need_utf8))) {
LOG_WARN("fail to check need utf8", K(ret));
} else if (need_utf8) {
types[2].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF8MB4_BIN : CS_TYPE_UTF8MB4_GENERAL_CI);
} else {
types[2].set_calc_collation_type(CS_TYPE_UTF8MB4_GENERAL_CI);
types[2].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI);
}
types[2].set_calc_collation_level(type.get_collation_level());
case 2:
case 2/*pattern and text*/:
types[1].set_calc_type(ObVarcharType);
if (ObCharset::is_bin_sort(types[1].get_calc_collation_type())) {
types[1].set_calc_collation_type(CS_TYPE_UTF8MB4_BIN);
} else {
types[1].set_calc_collation_type(CS_TYPE_UTF8MB4_GENERAL_CI);
}
types[1].set_calc_collation_level(type.get_collation_level());
types[1].set_calc_collation_level(CS_LEVEL_IMPLICIT);
if (!types[0].is_clob()) {
types[0].set_calc_type(ObVarcharType);
if (ObCharset::is_bin_sort(types[0].get_calc_collation_type())) {
types[0].set_calc_collation_type(CS_TYPE_UTF8MB4_BIN);
} else {
types[0].set_calc_collation_type(CS_TYPE_UTF8MB4_GENERAL_CI);
}
types[0].set_calc_collation_level(type.get_collation_level());
}
types[0].set_calc_collation_level(CS_LEVEL_IMPLICIT);
need_utf8 = false;
if (OB_FAIL(ret)) {
} else if (OB_FAIL(ObExprRegexContext::check_need_utf8(raw_expr->get_param_expr(1), need_utf8))) {
LOG_WARN("fail to check need utf8", K(ret));
} else if (need_utf8) {
types[1].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF8MB4_BIN : CS_TYPE_UTF8MB4_GENERAL_CI);
} else {
types[1].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI);
}
need_utf8 = false;
if (OB_FAIL(ret)) {
} else if (OB_FAIL(ObExprRegexContext::check_need_utf8(raw_expr->get_param_expr(0), need_utf8))) {
LOG_WARN("fail to check need utf8", K(ret));
} else if (need_utf8) {
types[0].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF8MB4_BIN : CS_TYPE_UTF8MB4_GENERAL_CI);
} else {
types[0].set_calc_collation_type(is_case_sensitive ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI);
}
default:
// already check before
@ -136,108 +159,6 @@ int ObExprRegexpReplace::calc_result_typeN(ObExprResType &type,
return ret;
}
int ObExprRegexpReplace::calc(common::ObString &ret_str,
const ObString &text,
const ObString &pattern,
const ObString &replacement_string,
int64_t position,
int64_t occurrence,
const ObCollationType cs_type,
const ObString &match_param,
int null_argument_idx,
bool reusable,
ObExprRegexContext* regexp_ptr,
ObExprStringBuf &string_buf,
ObIAllocator &exec_ctx_alloc)
{
int ret = OB_SUCCESS;
//begin with '^'
bool from_begin = false;
//end with '$'
bool from_end = false;
ret_str.reset();
ObExprRegexContext *regexp_replace_ctx = regexp_ptr;
ObSEArray<uint32_t, 4> begin_locations(common::ObModIds::OB_SQL_EXPR_REPLACE,
common::OB_MALLOC_NORMAL_BLOCK_SIZE);
int flags = 0, multi_flag = 0;
if (OB_ISNULL(regexp_replace_ctx)) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("regexp ptr is null", K(ret));
} else if (OB_FAIL(ObExprRegexpCount::get_regexp_flags(cs_type, match_param,
flags, multi_flag))) {
LOG_WARN("fail to get regexp flags", K(ret), K(match_param));
} else if (OB_FAIL(regexp_replace_ctx->init(pattern, flags,
reusable ? exec_ctx_alloc : string_buf,
reusable))) {
LOG_WARN("fail to init regexp", K(flags));
}
if (OB_SUCC(ret)) {
if (-1 != null_argument_idx) {
//此处是为了更好的和oracle行为一致
if ((null_argument_idx == 0 || (lib::is_oracle_mode() && null_argument_idx == 1)) &&
text.length() != 0) {
ret_str = text;
} else {
ret_str = ObString(); // NULL
}
} else if (position <= 0 || occurrence < 0) {
if (lib::is_oracle_mode()) {
ret = OB_ERR_ARGUMENT_OUT_OF_RANGE;
} else {
ret = OB_INVALID_ARGUMENT;
}
LOG_WARN("position or occurrence is invalid", K(ret), K(position), K(occurrence));
} else {
if (OB_NOT_NULL(pattern)) {
const char* pat = pattern.ptr();
if ('^' == pat[0]) {
from_begin = true;
} else if ('$' == pat[pattern.length() - 1]) {
from_end = true;
}
}
if (OB_FAIL(begin_locations.push_back(position - 1))) {
LOG_WARN("begin_locations push_back error", K(ret));
} else if (from_begin && 1 != position && !multi_flag) {
ret_str = text;
} else {
ObSEArray<size_t, 4> byte_num;
ObSEArray<size_t, 4> byte_offsets;
if (OB_FAIL(ObExprUtil::get_mb_str_info(text,
ObCharset::get_default_collation_oracle(CHARSET_UTF8MB4),
byte_num,
byte_offsets))) {
LOG_WARN("failed to get mb str info", K(ret));
} else if (multi_flag) {
if (from_begin && 1 != position) {
begin_locations.pop_back();
}
const char *text_ptr = text.ptr();
for (int64_t idx = position; OB_SUCC(ret) && idx < byte_offsets.count() - 1; ++idx) {
if (OB_UNLIKELY(byte_offsets.at(idx) >= text.length())) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("get unexpected error", K(byte_offsets.at(idx)), K(text.length()), K(ret));
} else if (text_ptr[byte_offsets.at(idx)] == '\n') {
if (OB_FAIL(begin_locations.push_back(idx + 1))) {
LOG_WARN("begin_locations push_back error", K(ret));
}
}
}
}
if (OB_SUCC(ret)) {
if (OB_FAIL(regexp_replace_ctx->replace_substr(text, replacement_string,
occurrence, string_buf,
byte_offsets, ret_str, from_begin,
from_end, begin_locations))) {
LOG_WARN("fail to replace substr", K(ret), K(text), K(replacement_string));
}
}
}
}
}
return ret;
}
int ObExprRegexpReplace::cg_expr(ObExprCGCtx &op_cg_ctx, const ObRawExpr &raw_expr, ObExpr &rt_expr) const
{
int ret = OB_SUCCESS;
@ -270,103 +191,166 @@ int ObExprRegexpReplace::eval_regexp_replace(
ObDatum *to = NULL;
ObDatum *position = NULL;
ObDatum *occurrence = NULL;
ObDatum *flags = NULL;
if (OB_FAIL(expr.eval_param_value(ctx, text, pattern, to, position, occurrence, flags))) {
ObDatum *match_type = NULL;
ObString res_replace;
bool need_convert = false;
ObCollationType res_coll_type = CS_TYPE_INVALID;
if (OB_FAIL(expr.eval_param_value(ctx, text, pattern, to, position, occurrence, match_type))) {
LOG_WARN("evaluate parameters failed", K(ret));
} else if (lib::is_oracle_mode() && pattern->is_null()) {
expr_datum.set_datum(*text);
if (text->is_null()) {
expr_datum.set_null();
} else {
res_replace = text->get_string();
res_coll_type = expr.args_[0]->datum_meta_.cs_type_;
need_convert = true;
}
} else if (expr.args_[0]->datum_meta_.is_clob() && (0 == text->get_string().length())) {
expr_datum.set_datum(*text);
} else {
// %to and %flags may be null
bool is_flag_null = (NULL != flags && flags->is_null());
const bool null_result = (text->is_null()
|| (NULL != position && position->is_null())
|| (NULL != occurrence && occurrence->is_null())
|| (lib::is_mysql_mode() && NULL != pattern && pattern->is_null())
|| (lib::is_mysql_mode() && NULL != to && to->is_null())
|| (lib::is_mysql_mode() && is_flag_null));
// same with calc_resultN
int null_idx = -1;
for (int64_t i = 0; i < expr.arg_cnt_; i++) {
if (expr.locate_param_datum(ctx, i).is_null() && (i != 5 && i != 2)) {
null_idx = null_idx > i ? null_idx : i;
}
} else if (OB_UNLIKELY(expr.arg_cnt_ < 2 ||
(expr.args_[0]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_GENERAL_CI &&
expr.args_[0]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_BIN &&
expr.args_[0]->datum_meta_.cs_type_ != CS_TYPE_UTF16_GENERAL_CI &&
expr.args_[0]->datum_meta_.cs_type_ != CS_TYPE_UTF16_BIN) ||
(expr.args_[1]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_GENERAL_CI &&
expr.args_[1]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_BIN &&
expr.args_[1]->datum_meta_.cs_type_ != CS_TYPE_UTF16_GENERAL_CI &&
expr.args_[1]->datum_meta_.cs_type_ != CS_TYPE_UTF16_BIN))) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("get unexpected error", K(ret), K(expr));
} else if (OB_UNLIKELY(expr.arg_cnt_ > 2 &&
expr.args_[2]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_GENERAL_CI &&
expr.args_[2]->datum_meta_.cs_type_ != CS_TYPE_UTF8MB4_BIN &&
expr.args_[2]->datum_meta_.cs_type_ != CS_TYPE_UTF16_GENERAL_CI &&
expr.args_[2]->datum_meta_.cs_type_ != CS_TYPE_UTF16_BIN)) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("get unexpected error", K(ret), K(expr));
} else if (lib::is_mysql_mode() && !pattern->is_null() && pattern->get_string().empty()) {
if (NULL == match_type || !match_type->is_null()) {
ret = OB_ERR_REGEXP_ERROR;
LOG_WARN("empty regex expression", K(ret));
} else {
expr_datum.set_null();
}
} else {
int64_t pos = 1;
int64_t occur = 0;
bool null_result = (position != NULL && position->is_null()) ||
(occurrence != NULL && occurrence->is_null()) ||
(lib::is_mysql_mode() && match_type != NULL && match_type->is_null());
if (OB_FAIL(ObExprUtil::get_int_param_val(position, pos))
|| OB_FAIL(ObExprUtil::get_int_param_val(occurrence, occur))) {
LOG_WARN("get integer parameter value failed", K(ret));
} else if (pos < 0 || occur < 0) {
} else if (!null_result && (pos <= 0 || occur < 0)) {
ret = OB_INVALID_ARGUMENT;
LOG_WARN("regexp_replace position or occurrence is invalid", K(ret), K(pos), K(occur));
}
ObString to_str = (NULL != to && !to->is_null()) ? to->get_string() : ObString();
ObString match_param = (NULL != flags && !flags->is_null())
? flags->get_string()
: ObString();
ObString res;
ObEvalCtx::TempAllocGuard alloc_guard(ctx);
ObIAllocator &alloc = alloc_guard.get_allocator();
ObExprRegexContext local_regex_ctx;
ObExprRegexContext *regexp_ctx = &local_regex_ctx;
const bool reusable = (0 != expr.extra_) && ObExpr::INVALID_EXP_CTX_ID != expr.expr_ctx_id_;
if (OB_SUCC(ret) && reusable) {
if (NULL == (regexp_ctx = static_cast<ObExprRegexContext *>(
ctx.exec_ctx_.get_expr_op_ctx(expr.expr_ctx_id_)))) {
if (OB_FAIL(ctx.exec_ctx_.create_expr_op_ctx(expr.expr_ctx_id_, regexp_ctx))) {
LOG_WARN("create expr regex context failed", K(ret), K(expr));
} else if (OB_ISNULL(regexp_ctx)) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("NULL context returned", K(ret));
LOG_USER_ERROR(OB_INVALID_ARGUMENT, "use position or occurrence in regexp_replace");
} else {
ObString to_str = (NULL != to && !to->is_null()) ? to->get_string() : ObString();
ObString match_param = (NULL != match_type && !match_type->is_null()) ? match_type->get_string() : ObString();
ObEvalCtx::TempAllocGuard alloc_guard(ctx);
ObIAllocator &tmp_alloc = alloc_guard.get_allocator();
ObExprRegexContext local_regex_ctx;
ObExprRegexContext *regexp_ctx = &local_regex_ctx;
const bool reusable = (0 != expr.extra_) && ObExpr::INVALID_EXP_CTX_ID != expr.expr_ctx_id_;
uint32_t flags = 0;
bool is_case_sensitive = ObCharset::is_bin_sort(expr.args_[0]->datum_meta_.cs_type_);
if (reusable) {
if (NULL == (regexp_ctx = static_cast<ObExprRegexContext *>(
ctx.exec_ctx_.get_expr_op_ctx(expr.expr_ctx_id_)))) {
if (OB_FAIL(ctx.exec_ctx_.create_expr_op_ctx(expr.expr_ctx_id_, regexp_ctx))) {
LOG_WARN("create expr regex context failed", K(ret), K(expr));
} else if (OB_ISNULL(regexp_ctx)) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("NULL context returned", K(ret));
}
}
}
}
if (OB_FAIL(ret)) {
} else if (lib::is_mysql_mode() && !pattern->is_null() &&
pattern->get_string().empty() && !is_flag_null) {//compatible mysql
ret = OB_ERR_REGEXP_ERROR;
LOG_WARN("empty regex expression", K(ret));
} else if (OB_FAIL(calc(res, text->get_string(), pattern->get_string(), to_str, pos, occur,
expr.datum_meta_.cs_type_, match_param,
null_idx, reusable, regexp_ctx, alloc,
ctx.exec_ctx_.get_allocator()))) {
LOG_WARN("calc failed", K(ret));
} else if (null_result) {
expr_datum.set_null();
} else {
if (res.empty() && lib::is_oracle_mode()) {
expr_datum.set_null();
} else {
ObExprStrResAlloc out_alloc(expr, ctx);
ObString out;
if (OB_FAIL(ObExprUtil::convert_string_collation(res,
expr.args_[0]->datum_meta_.cs_type_,
out,
expr.datum_meta_.cs_type_,
out_alloc))) {
LOG_WARN("convert charset failed", K(ret));
} else {
if (out.ptr() == res.ptr()) {
// res is allocated in temporary allocator, deep copy here.
char *mem = expr.get_str_res_mem(ctx, res.length());
if (OB_ISNULL(mem)) {
ret = OB_ALLOCATE_MEMORY_FAILED;
LOG_WARN("allocate memory failed", K(ret));
} else {
MEMCPY(mem, res.ptr(), res.length());
expr_datum.set_string(mem, res.length());
}
if (OB_FAIL(ret)) {
} else if (OB_FAIL(ObExprRegexContext::get_regexp_flags(match_param, is_case_sensitive, flags))) {
LOG_WARN("fail to get regexp flags", K(ret), K(match_param));
} else if (!pattern->is_null() && !null_result &&
OB_FAIL(regexp_ctx->init(reusable ? ctx.exec_ctx_.get_allocator() : tmp_alloc,
ctx.exec_ctx_.get_my_session(),
pattern->get_string(), flags, reusable,
expr.args_[1]->datum_meta_.cs_type_))) {
LOG_WARN("fail to init regexp", K(pattern), K(flags), K(ret));
} else if (text->is_null() ||
pattern->is_null() ||
null_result ||
(NULL != position && position->is_null()) ||
(NULL != occurrence && occurrence->is_null()) ||
(lib::is_mysql_mode() && NULL != pattern && pattern->is_null()) ||
(lib::is_mysql_mode() && NULL != to && to->is_null()) ||
(lib::is_mysql_mode() && NULL != match_type && match_type->is_null())) {
if (lib::is_oracle_mode() && !text->is_null() && !text->get_string().empty()) {
if ((NULL != position && position->is_null()) ||
(NULL != occurrence && occurrence->is_null())) {
expr_datum.set_null();
} else {
expr_datum.set_string(out.ptr(), out.length());
res_replace = text->get_string();
res_coll_type = expr.args_[0]->datum_meta_.cs_type_;
need_convert = true;
}
} else {
expr_datum.set_null();
}
} else {
ObString text_utf16;
ObString to_utf16;
if (expr.args_[0]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_BIN ||
expr.args_[0]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_GENERAL_CI) {
res_coll_type = ObCharset::is_bin_sort(expr.args_[0]->datum_meta_.cs_type_) ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI;
if (OB_FAIL(ObExprUtil::convert_string_collation(text->get_string(), expr.args_[0]->datum_meta_.cs_type_, text_utf16,
res_coll_type, tmp_alloc))) {
LOG_WARN("convert charset failed", K(ret));
}
} else {
res_coll_type = expr.args_[0]->datum_meta_.cs_type_;
text_utf16 = text->get_string();
}
if (OB_FAIL(ret)) {
} else if (expr.arg_cnt_ > 2 && (expr.args_[2]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_BIN ||
expr.args_[2]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_GENERAL_CI)) {
if (OB_FAIL(ObExprUtil::convert_string_collation(to_str, expr.args_[2]->datum_meta_.cs_type_, to_utf16,
ObCharset::is_bin_sort(expr.args_[2]->datum_meta_.cs_type_) ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI,
tmp_alloc))) {
LOG_WARN("convert charset failed", K(ret));
}
} else {
to_utf16 = to_str;
}
if (OB_FAIL(ret)) {
} else if (OB_FAIL(regexp_ctx->replace(tmp_alloc, text_utf16, to_utf16, pos - 1,
occur, res_replace))) {
LOG_WARN("failed to regexp replace str", K(ret));
} else if (res_replace.empty() && lib::is_oracle_mode()) {
expr_datum.set_null();
} else {
need_convert = true;
}
}
}
}
if (OB_SUCC(ret) && need_convert) {
ObExprStrResAlloc out_alloc(expr, ctx);
ObString out;
if (OB_FAIL(ObExprUtil::convert_string_collation(res_replace, res_coll_type, out, expr.datum_meta_.cs_type_, out_alloc))) {
LOG_WARN("convert charset failed", K(ret));
} else if (out.ptr() == res_replace.ptr()) {
// res_replace is allocated in temporary allocator, deep copy here.
char *mem = expr.get_str_res_mem(ctx, res_replace.length());
if (OB_ISNULL(mem)) {
ret = OB_ALLOCATE_MEMORY_FAILED;
LOG_WARN("allocate memory failed", K(ret));
} else {
MEMCPY(mem, res_replace.ptr(), res_replace.length());
expr_datum.set_string(mem, res_replace.length());
}
} else {
expr_datum.set_string(out.ptr(), out.length());
}
}
return ret;
}