fix replace expr charset mistaken code bug

This commit is contained in:
SevenJ-swj
2023-07-11 10:42:33 +00:00
committed by ob-robot
parent c5c92dfe7c
commit fa010efcd4
5 changed files with 51 additions and 57 deletions

View File

@ -3141,10 +3141,10 @@ int ObCharsetUtils::init(ObIAllocator &allocator)
return ret;
}
bool ObStringScanner::next_character(ObString &encoding, int32_t &wchar, int &ret)
bool ObStringScanner::next_character(ObString &encoding_value, int32_t &unicode_value, int &ret)
{
bool has_next = false;
ret = next_character(encoding, wchar);
ret = next_character(encoding_value, unicode_value);
if (OB_ITER_END == ret) {
has_next = false;
@ -3158,20 +3158,26 @@ bool ObStringScanner::next_character(ObString &encoding, int32_t &wchar, int &re
return has_next;
}
int ObStringScanner::next_character(ObString &encoding, int32_t &wchar)
int ObStringScanner::next_character(ObString &encoding_value, int32_t &unicode_value)
{
int ret = OB_SUCCESS;
int32_t length = 0;
ObString &str = const_cast<ObString &>(str_);
ObString &str = str_;
if (str.empty()) {
ret = OB_ITER_END;
} else if (OB_FAIL(ObCharset::mb_wc(collation_type_, str.ptr(), str.length(), length, wchar))) {
ret = OB_ERR_INCORRECT_STRING_VALUE;
LOG_WARN("fail to call mb_wc", K(ret), KPHEX(str.ptr(), str.length()));
} else {
encoding.assign_ptr(str.ptr(), length);
} else if (OB_FAIL(ObCharset::mb_wc(collation_type_, str.ptr(), str.length(), length, unicode_value))) {
if (!!(IGNORE_INVALID_CHARACTER & flags_)) {
ret = OB_SUCCESS;
length = 1;
} else {
ret = OB_ERR_INCORRECT_STRING_VALUE;
LOG_WARN("fail to call mb_wc", K(ret), KPHEX(str.ptr(), str.length()));
}
}
if (OB_SUCC(ret)) {
encoding_value.assign_ptr(str.ptr(), length);
LOG_DEBUG("next_character", K(ret), KPHEX(str.ptr(), str.length()));
str += length;
}

View File

@ -608,15 +608,22 @@ private:
class ObStringScanner
{
public:
ObStringScanner(const ObString &str, common::ObCollationType collation_type)
: str_(str), collation_type_(collation_type)
enum {
IGNORE_INVALID_CHARACTER = 1<<0,
};
ObStringScanner(const ObString &str, common::ObCollationType collation_type, uint64_t flags = 0)
: origin_str_(str), str_(str), collation_type_(collation_type), flags_(flags)
{}
int next_character(ObString &encoding, int32_t &wchar);
bool next_character(ObString &encoding, int32_t &wchar, int &ret);
int next_character(ObString &encoding_value, int32_t &unicode_value);
bool next_character(ObString &encoding_value, int32_t &unicode_value, int &ret);
ObString get_remain_str() { return str_; }
void forward_bytes(int64_t n) { str_ += n; }
TO_STRING_KV(K_(str), K_(collation_type));
private:
const ObString &str_;
const ObString &origin_str_;
ObString str_;
common::ObCollationType collation_type_;
uint64_t flags_;
};
class ObCharSetString