6.2 VBScript Regular Expressions

Note

This section is RD-VBAL's own. MS-VBAL has no section for the VBScript RegExp 5.5 library, and there is no MS-VBAL §6.2.1.

MS-VBA programs use the regular expressions of Microsoft VBScript Regular Expressions 5.5, either through a reference to it or with CreateObject("VBScript.RegExp"). RD-VBA implements the library itself, as classes of the standard library: New RegExp needs no reference, and CreateObject("VBScript.RegExp") creates the same object on any platform, without asking the machine for a server (RD-VBAL §6.1.2.8 Interaction). What follows is what MS-VBA does, recorded in Excel, and what RD-VBA does with it.

Class	SDK interface	What it is
RegExp	IStdRegExpClass	A pattern and its options: Pattern (""), Global, IgnoreCase, Multiline (all False for a new object); Test, Execute, Replace. It has no default member.
MatchCollection	IStdMatchCollectionClass	What Execute found: Count, and Item, its default member, from 0, declared Object as MS-VBA's type library declares it. A For Each enumerates it.
Match	IStdMatchClass	One match: Value, its default member; FirstIndex, from 0; Length; SubMatches.
SubMatches	IStdSubMatchesClass	What each group matched: Count, and Item, its default member, from 0 - a String, or Empty for a group that took no part in the match. A For Each enumerates it.

Only RegExp can be created. The matches are what was found when Execute was called: changing the pattern afterwards changes none of them. A position that is not one (m(5) of a collection of one match) is error 5.

Patterns

A pattern is written in VBScript's syntax (that of JScript, which ECMAScript 3 describes), and means what it means in MS-VBA:

Construct	What it matches
.	Any character but a line feed (a carriage return is one).
\d, \w, \s and their negations \D, \W, \S	An ASCII digit; an ASCII letter, digit or _; a space, tab, line feed, vertical tab, form feed or carriage return. é is no word character, and a no-break space is no space.
\b, \B	A boundary between a word character and another character, as \w has them; anywhere that is not one.
^, $	The start of the text and its very end (a final line feed is not skipped). With Multiline, also after and before a carriage return, a line feed, U+2028 or U+2029, each on its own: a vbCrLf has an empty line between its two characters.
[...], [^...]	One of the characters given, or none of them. A ] first in the set is one of its characters; a - first or last is one too; \b is a backspace.
*, +, ?, {n}, {n,}, {n,m}, each with a ? to make it lazy	Repetition. A brace that is not one of these is a brace (a{,2}, a{2).
(...), (?:...), (?=...), (?!...)	A group, numbered by its opening parenthesis; a group that captures nothing; lookahead. A group inside a repetition is what it matched last.
\1 to \99	A backreference, when the pattern has that many groups: with fewer, \10 is group 1 followed by 0, and \2 is an octal escape. A backreference to a group that has not matched yet fails.
\n, \r, \t, \v, \f, \cX, \xhh, \uhhhh, \0 and octal escapes	The character. \x4 and \u12, which give too few digits, are x4 and u12; \c before anything but a letter matches nothing.
\ before any other character	That character: \p is p, \Z is Z, \k is k.

With IgnoreCase, a letter matches another when the two are the same in lower case: é and É, i and İ. The Kelvin, Ohm and Angstrom signs are not the letters they look like.

What is not a pattern is reported when it is first used - by Test, Execute or Replace - never when it is assigned. MS-VBA has no description of its own for these errors: it says "Application-defined or object-defined error".

Error	What is wrong	Example
5017	Syntax error in regular expression.	a{3,2}, (?<=a), (?i), (?<name>a), a ) that closes nothing, a final \
5018	Unexpected quantifier.	*a, a**, a?+, ^*, \b+
5019	Expected ] in regular expression.	[a, []a
5020	Expected ) in regular expression.	(a, (?:a
5021	Invalid range in character set.	[z-a], [a-\d]

Replacements

In what Replace writes for a match, $1 to $99 are what a group matched ($10 is group 10 when there is one, and group 1 followed by 0 when there is not), $& is the match, $` and $' the text before and after it, and $$ a dollar sign. Anything else, $0 and \n included, is written as it is. A replacement that is not text is written as text - a Boolean as -1 or 0, Empty as nothing - and Null is error 13.

Differences

• A member of these classes called or assigned late-bound (Dim m As Object) is called as MS-VBA calls it through IDispatch, whose errors are not those of an early-bound call: Null for a String is 13, a read-only property assigned 450 (RD-VBAL §6.0 Late-Bound Calls).
• 🚧 A match is not bounded in time: a pattern that backtracks without end does here what it does in MS-VBA.

---

⏮️ RD-VBAL §6.1.3.3 Global Class