[Bug 4479] New: Bounds error in rich edit EM_FINDTEXT

[Wine Bugs]

http://bugs.winehq.org/show_bug.cgi?id=4479

Summary: Bounds error in rich edit EM_FINDTEXT
           Product: Wine
           Version: CVS
          Platform: All
        OS/Version: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: wine-richedit
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]
                CC: [e-mail]

Bug 4144 mentions a problem in finding text where the second occurance of 'blah'
in 'blahblah' would not be found. A simpler case is finding 'blah' in the range
(0, 4) in document 'blah', which fails only with Wine.

A rich edit conformance test posted several days ago shows this behavior:
http://www.winehq.org/pipermail/wine-patches/2006-January/023871.html

Variable values during the three find statements (lines 148-153 of the patch) are:
fixme:richedit:ME_FindText paraofs=0 rowofs=0, nstart=0, nlen=4, nmax=8
fixme:richedit:ME_FindText paraofs=0 rowofs=0, nstart=4, nlen=4, nmax=8
fixme:richedit:ME_FindText paraofs=0 rowofs=0, nstart=4, nlen=4, nmax=9

Here's the mention from bug 4144:
--snip--
(One interesting note: A Find done not-whole-word will miss a second
version of the word in a row. Ex.: in blahblah only the first "blah" is
found, while if RichEd20.dll is not loaded the Find is almost entirely
malfunctional. It's possible those issues are partially connected to
some problem with offset position numbers used to start the Find. In
both the control and API the RTB text has a zero-based offset system,
but Wine could be starting text at 1?  )
--snip--