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

257
src/sql/engine/expr/ob_expr_regexp_like.cpp
View File

@ -43,6 +43,7 @@ int ObExprRegexpLike::calc_result_typeN(ObExprResType &type,
{
UNUSED(type_ctx);
int ret = OB_SUCCESS;
ObRawExpr * raw_expr = type_ctx.get_raw_expr();
CK(NULL != type_ctx.get_raw_expr());
if (OB_FAIL(ret)) {
} else if (OB_UNLIKELY(param_num < 2 || param_num > 3)) {
@ -55,163 +56,49 @@ int ObExprRegexpLike::calc_result_typeN(ObExprResType &type,
LOG_WARN("the parameter is not castable", K(ret), K(i));
}
}
if (lib::is_oracle_mode()) {//set default value
type.set_calc_type(ObVarcharType);
type.set_calc_collation_type(ObCharset::get_default_collation_oracle(CHARSET_UTF8MB4));
type.set_calc_collation_level(CS_LEVEL_IMPLICIT);
} else {
ObObjMeta type_metas[2] = {types[0], types[1]};
if (OB_FAIL(aggregate_charsets_for_comparison(
type.get_calc_meta(), type_metas, 2, type_ctx.get_coll_type()))) {
LOG_WARN("fail to aggregate charsets for comparison", K(ret), K(types[0]), K(types[1]));
}
}
if (OB_SUCC(ret)) {//because of regexp engine need utf8 code, here need reset calc collation type and can implicit cast.
switch (param_num) {
case 3:
types[2].set_calc_type(ObVarcharType);
case 2:
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);
}
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);
}
}
default:
// already check before
break;
}
}
if (OB_SUCC(ret)) {
if (param_num == 3) {/*match type*/
types[2].set_calc_type(ObVarcharType);
types[2].set_calc_collation_type(CS_TYPE_UTF8MB4_BIN);
types[2].set_calc_collation_level(CS_LEVEL_IMPLICIT);
}
//we set the calc collation type to utf8 and convert it to utf16 in excution stage, because the ICU regexp engine is used uft16,
//we need convert it the need collation in advance, and no need to think about in regexp.
//lob TODO,jiangxiu.wt
bool is_case_sensitive = ObCharset::is_bin_sort(types[0].get_calc_collation_type());
bool need_utf8 = false;
types[1].set_calc_type(ObVarcharType);
types[1].set_calc_collation_level(CS_LEVEL_IMPLICIT);
if (!types[0].is_clob()) {
types[0].set_calc_type(ObVarcharType);
}
types[0].set_calc_collation_level(CS_LEVEL_IMPLICIT);
type.set_tinyint();
type.set_precision(DEFAULT_PRECISION_FOR_BOOL);
type.set_scale(DEFAULT_SCALE_FOR_INTEGER);
ObExprOperator::calc_result_flag2(type, types[0], types[1]);
}
}
return ret;
}
int ObExprRegexpLike::regexp_like(bool &match,
const ObString &text,
const ObString &pattern,
int64_t position,
int64_t occurrence,
const ObCollationType calc_cs_type,
const ObString &match_param,
bool has_null_argument,
bool reusable,
ObExprRegexContext* regexp_ptr,
ObExprStringBuf &string_buf,
ObIAllocator &exec_ctx_alloc)
{
int ret = OB_SUCCESS;
bool sub = false;
//begin with '^'
bool from_begin = false;
//end with '$'
bool from_end = false;
ObExprRegexContext *regexp_like_ctx = regexp_ptr;
ObSEArray<uint32_t, 4> begin_locations(common::ObModIds::OB_SQL_EXPR_REPLACE,
common::OB_MALLOC_NORMAL_BLOCK_SIZE);
int flags = 0, multi_flag = 0;
if (OB_ISNULL(regexp_like_ctx)) {
ret = OB_ERR_UNEXPECTED;
LOG_WARN("regexp ptr is null", K(ret));
} else if (OB_FAIL(ObExprRegexpCount::get_regexp_flags(calc_cs_type, match_param,
flags, multi_flag))) {
LOG_WARN("fail to get regexp flags", K(ret), K(match_param));
} else if (OB_FAIL(regexp_like_ctx->init(pattern, flags,
reusable ? exec_ctx_alloc : string_buf,
reusable))) {
LOG_WARN("fail to init regexp", K(pattern), K(flags));
}
if (OB_SUCC(ret)) {
//为了更好的和oracle兼容
if (has_null_argument) {
match = false;
} else {
if (!pattern.empty()) {
const char* pat = pattern.ptr();
if ('^' == pat[0]) {
from_begin = true;
} else if ('$' == 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) {
match = sub;
need_utf8 = false;
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 {
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_like_ctx->like(text, occurrence, sub, string_buf,
from_begin, from_end, begin_locations))) {
LOG_WARN("fail to replace substr", K(ret), K(text));
} else {
match = sub;
}
}
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);
}
}
}
return ret;
}
int ObExprRegexpLike::calc(common::ObObj &result,
const ObString &text,
const ObString &pattern,
int64_t position,
int64_t occurrence,
const ObString &match_param,
bool has_null,
ObExprRegexContext* regexp_ptr,
ObExprStringBuf &string_buf,
ObIAllocator &exec_ctx_alloc) const
{
int ret = OB_SUCCESS;
bool match = false;
if (OB_FAIL(regexp_like(match, text, pattern, position, occurrence,
result_type_.get_calc_collation_type(),
match_param, has_null, false, regexp_ptr, string_buf, exec_ctx_alloc))) {
LOG_WARN("regexp like failed", K(ret));
} else {
result.set_int(result_type_.get_type(), static_cast<int64_t>(match));
}
return ret;
}
int ObExprRegexpLike::cg_expr(ObExprCGCtx &op_cg_ctx, const ObRawExpr &raw_expr, ObExpr &rt_expr) const
{
int ret = OB_SUCCESS;
@ -241,26 +128,40 @@ int ObExprRegexpLike::eval_regexp_like(
int ret = OB_SUCCESS;
ObDatum *text = NULL;
ObDatum *pattern = NULL;
ObDatum *flags = NULL;
if (OB_FAIL(expr.eval_param_value(ctx, text, pattern, flags))) {
ObDatum *match_type = NULL;
if (OB_FAIL(expr.eval_param_value(ctx, text, pattern, match_type))) {
LOG_WARN("evaluate parameters failed", K(ret));
} 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 (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 {
// use default flags if NULL == flags and flags->is_null()
bool is_flag_null = (NULL != flags && flags->is_null());
const bool null_result = (text->is_null() || pattern->is_null() ||
(lib::is_mysql_mode() && is_flag_null));
ObString match_param = (NULL != flags && !flags->is_null())
? flags->get_string()
: ObString();
const int64_t pos = 1;
const int64_t occurrence = 1;
ObString match_param = (NULL != match_type && !match_type->is_null()) ? match_type->get_string() : ObString();
ObEvalCtx::TempAllocGuard alloc_guard(ctx);
ObIAllocator &alloc = alloc_guard.get_allocator();
bool match = false;
ObIAllocator &tmp_alloc = alloc_guard.get_allocator();
ObExprRegexContext local_regex_ctx;
ObExprRegexContext *regexp_ctx = &local_regex_ctx;
uint32_t flags = 0;
int64_t start_pos = 1;
bool match = false;
bool is_case_sensitive = ObCharset::is_bin_sort(expr.args_[0]->datum_meta_.cs_type_);
const bool reusable = (0 != expr.extra_) && ObExpr::INVALID_EXP_CTX_ID != expr.expr_ctx_id_;
if (OB_SUCC(ret) && reusable) {
ObString text_utf16;
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))) {
@ -272,19 +173,35 @@ int ObExprRegexpLike::eval_regexp_like(
}
}
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(regexp_like(match, text->get_string(), pattern->get_string(), pos,
occurrence, expr.args_[0]->datum_meta_.cs_type_, match_param,
null_result, reusable, regexp_ctx, alloc,
ctx.exec_ctx_.get_allocator()))) {
LOG_WARN("do regexp like failed", K(ret));
} else if (null_result) {
} 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() &&
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));
//need pre check the pattern valid, and then set result.
} else if (text->is_null() || pattern->is_null() ||
(lib::is_mysql_mode() && NULL != match_type && match_type->is_null())) {
expr_datum.set_null();
} else {
expr_datum.set_int32(match);
if (expr.args_[0]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_BIN ||
expr.args_[0]->datum_meta_.cs_type_ == CS_TYPE_UTF8MB4_GENERAL_CI) {
if (OB_FAIL(ObExprUtil::convert_string_collation(text->get_string(), expr.args_[0]->datum_meta_.cs_type_, text_utf16,
ObCharset::is_bin_sort(expr.args_[0]->datum_meta_.cs_type_) ? CS_TYPE_UTF16_BIN : CS_TYPE_UTF16_GENERAL_CI,
tmp_alloc))) {
LOG_WARN("convert charset failed", K(ret));
}
} else {
text_utf16 = text->get_string();
}
if (OB_FAIL(ret)) {
} else if (OB_FAIL(regexp_ctx->match(tmp_alloc, text_utf16, start_pos - 1, match))) {
LOG_WARN("fail to match", K(ret), K(text));
} else {
expr_datum.set_int32(match);
}
}
}
return ret;