* Added min/max match length to irregexp ast

Each RegExtTree node can now report the min and max size of strings it can match.

[Lasse Reichstein, 2008-12-17 09:48:56]

Review please.
Should we delay computation until we need it instead?

[Erik Corry, 2008-12-17 10:16:03]

LGTM

http://codereview.chromium.org/14190/diff/208/8
File src/ast.h (right):

http://codereview.chromium.org/14190/diff/208/8#newcode1222
Line 1222: virtual int min_match() = 0;
Could we call this min_chars instead?  Seems less ambiguous to me.

http://codereview.chromium.org/14190/diff/208/8#newcode1238
Line 1238: min_match_(alternatives->at(0)->min_match()),
I think these initializations should be moved into the body of the constructor,
since they will be overwritten later.

http://codereview.chromium.org/14190/diff/208/8#newcode1241
Line 1241: RegExpTree* alternative = alternatives->at(i);
We have Min and Max in utils.h for stuff like this.  Perhaps with the help of
those we can get this down to a size where it is not unreasonable to have it in
the .h file.

http://codereview.chromium.org/14190/diff/208/8#newcode1246
Line 1246: if (max_match_ != kInfinity) {
You don't need to special case infinity here.

http://codereview.chromium.org/14190/diff/208/8#newcode1299
Line 1299: 
Reordering stuff like this makes the change hard to read.

http://codereview.chromium.org/14190/diff/208/8#newcode1382
Line 1382: void AddElement(TextElement elm) {
I think TextElement should grow a Length() method.

http://codereview.chromium.org/14190/diff/208/8#newcode1392
Line 1392: UNIMPLEMENTED();
unreachable

http://codereview.chromium.org/14190/diff/208/8#newcode1409
Line 1409: min_match_(min * body->min_match()),
move into body

http://codereview.chromium.org/14190/diff/208/8#newcode1411
Line 1411: if (max > 0 && body->max_match() > kInfinity / max) {
Please add a test with a max of 0.