418099 - RegExp classrange (case-insensitive) does not follow ECMA spec.

Status: RESOLVED DUPLICATE of bug 416933

(Core :: JavaScript Engine, defect)

Description

[mozilla]

ECMA-spec 15.10.2.16 (page 142) gives an example for classranges when case-insensitive is activated:
/[E-f]/i should match the symbols [,\,],^,_ and `.
FF2.0.0.12 and FF3.0beta3 do not match these symbols. (neither does Rhino 1.6rel7 btw).

Comment 1

[timeless]

fwiw, MSIE seems to get this right.

I'd suggest the test be:
/[E-c]/i.test(...) instead of /[E-f]/i.

/[E-c]/i currently yields: Error: invalid range in character class

Comment 2

[timeless]

/[E-e]/i is also interesting

Comment 3

[Brian Crowder]

The spec is internally inconsistent here, I think.  For example:

The internal helper function CharacterRange takes two CharSet parameters A and B and performs the following:
1. If A does not contain exactly one character or B does not contain exactly one character then throw a SyntaxError exception.
2. Let a be the one character in CharSet A.
3. Let b be the one character in CharSet B.
4. Let i be the code point value of character a.
5. Let j be the code point value of character b.
6. If I > j then throw a SyntaxError exception.
7. Return the set containing all characters numbered i through j, inclusive.

In other words, the example that it later provides: /[E-f]/i should yield a syntax error (according to item #6 above, assuming the sudden change in capitalization is an accident and not referring to some other variable "I").

Comment 4

[Brian Crowder]

I think the pseudocode here should be considered more authoritative than the example.  This means that we're accepting bogus ranges that we shouldn't be accepting.  I'll check it out in a bit...

Comment 5

[mozilla]

The spec seems fine to me: code point of E is 69 and of f=102. So Classrange returns 69-102. The case-sensitivity (or insensitivity) comes into play elsewhere: 15.10.2.8 (3rd production):
Atom::CharacterClass:
1. Evaluate CharacterClass to obtain a CharSet A and a boolean invert.
2. Call CharacterSetMatcher(A, invert) and return its Matcher result.

It's in the CharacterSetMatcher that the chars are then canonicalized (on the next page):
CharacterSetMatcher step 6/8:
If there [does not exist/exists] a member a of set A such that Canonicalize(a) == cc, then return [true/failure].

So each char of the CharSet needs to be uppercased and then compared to the uppercased input-char. (semantically at least).

Comment 6

[Brian Crowder]

Re: comment 5 -- ooops, thanks, my bad.