(PL)SQL - Regular Expressions

I encountered some strange behaviour when trying to remove all characters from a varchar2, except a-z, A-Z and 0-9.
Using regular expressions, you can define that in a number of ways, such as [^A-Za-z0-9] or [^[:alnum:]]

Example 1 works as expected (see the query below):
For A1 the ë (ascii 235) is not included, since it falls outside of the [^A-Za-z0-9] range.
For A2 the ë is included, since it is first converted to a regular e character.

Example 2 is (a dummy version of) the exact same query on another (also 11gR2) database and client. Here, the result for A1 was a bit different:
For A1 the ë was included.
A2 was the same.

Probably the characterset settings of the databases and/or clients were the cause? I have included the info gathered by PL/SQL Developer for both setups.
My favourite online regular expressions tool regex101.com agrees that example 1 is as expected.

A1	A2
Ab3X1	Ab3eX1

Character size: 4 byte(s)

CharSetID: 873

NCharSetID: 2000

Unicode Support: True

NLS_LANG: AMERICAN_AMERICA.WE8MSWIN1252

NLS_CHARACTERSET: AL32UTF8

NLS_NCHAR_CHARACTERSET: AL16UTF16

A1	A2
Ab3ëX1	Ab3eX1

Character size: 1 byte(s)

CharSetID: 31

NCharSetID: 2000

Unicode Support: True

NLS_LANG: DUTCH_THE NETHERLANDS.WE8MSWIN1252

NLS_NCHAR_CHARACTERSET: AL16UTF16

NLS_CHARACTERSET: WE8ISO8859P1

A simple example how to split string like 'ABC:123;XYZ:789' into multiple value pairs.

R	Value 1	Value 2
ABC:123	ABC	123
ABCD:1234	ABCD	1234
WXYZ:6789	WXYZ	6789
XYZ:789	XYZ	789

Creating a simple check for phone numbers was not as easy as I expected.
Getting it to work in regex101.com was easy, but that expression was not handled as expected when using it in the Oracle database.
I did get it to work by changing the expression, but I'm still not sure why the first version didn't work.

Tel	R_1	R_2
088 223 22 33	0	1
+31 (0)88 - 223 22 33	0	1
088-223 22 33	0	1
0882232233	0	1
abc 123	0	0

1 - 5

Region

Identification
	Sequence	10
	Title	Accented Characters Example 1
	Type	Report
Source
	Source Type	Report
	Region Source	select regexp_replace( '$Ab3@ëX 1' , '[^A-Za-z0-9]' ) as a1 , regexp_replace( convert('$Ab3@ëX 1', 'US7ASCII'), '[^A-Za-z0-9]' ) as a2 from dual

Region

Identification
	Sequence	30
	Title	Accented Characters Example 2
	Type	Report
Source
	Source Type	Report
	Region Source	select 'Ab3ëX1' as a1 , 'Ab3eX1' as a2 from dual

Region

Identification
	Sequence	10
	Title	Value Pairs Example
	Type	Report
Source
	Source Type	Report
	Region Source	with t as ( select 'ABC:123;XYZ:789' as col from dual union all select 'ABCD:1234;WXYZ:6789' as col from dual ) , t2 as ( select t0.column_value as r from t cross join table(apex_string.split(t.col, ';')) t0 ) select t2.r , trim( ':' from regexp_substr( t2.r, '^.:' ) ) as value_1 , trim( ':' from regexp_substr( t2.r, ':.$' ) ) as value_2 from t2 order by 1 /* Note that apex_util.string_to_table was only introduced with Apex 5.1 (and apex_string.split in a later version), so for older versions, or a schema without Apex access, you can use xmltable (simply replace the t2 query above with the query below): select extract(column_value, '//text()').getStringVal() as r from t , xmltable( replace( '''' \|\| t.col \|\| '''', ';', ''', ''') ) */

Region

Identification
	Sequence	10
	Title	Phone Numbers Example
	Type	Report
Source
	Source Type	Report
	Region Source	with a as ( select '088 223 22 33' as tel from dual union all select '+31 (0)88 - 223 22 33 ' as tel from dual union all select '088-223 22 33' as tel from dual union all select '0882232233' as tel from dual union all select 'abc 123' as tel from dual ) select a.tel , regexp_instr(a.tel, '^[\d\| \|\-\|+\|(\|)]+$' ) as r_1 -- works in regex101.com, not in Oracle database , regexp_instr(a.tel, '^(\d\| \|-\|\+\|$\|$)+$') as r_2 -- works in both from a

Accented Characters

Accented Characters Example 1

Accented Characters Example 2

Value Pairs

Value Pairs Example

Phone Numbers

Phone Numbers Example

Code

Region

Region

Region

Region