r195910 - InclusionRewriter: Avoid duplicated BOM check

[Alp Toker]

On 03/12/2013 20:01, Lubos Lunak wrote:
> On Tuesday 03 of December 2013, Reid Kleckner wrote:
>> This seems like it has the side effect of trimming leading whitespace from
>> the rewritten file, but that's probably fine.
>   It's not. The main file does not start with any line markers, so it is
> possible that lines in warnings/errors will be wrong. Please fix this or
> revert the change to the previous one that works.

Hi Lubos,

I don't see the problem here. Can you explain, preferably with a test case?

Alp.


>
>> On Wed, Nov 27, 2013 at 11:21 PM, Alp Toker <alp at nuanti.com> wrote:
>>> Author: alp
>>> Date: Thu Nov 28 01:21:44 2013
>>> New Revision: 195910
>>>
>>> URL: http://llvm.org/viewvc/llvm-project?rev=195910&view=rev
>>> Log:
>>> InclusionRewriter: Avoid duplicated BOM check
>>>
>>> The lexer already knows its position in the file, so use that instead of
>>> guessing it might be 3.
>>>
>>> Modified:
>>>      cfe/trunk/lib/Rewrite/Frontend/InclusionRewriter.cpp
>>>
>>> Modified: cfe/trunk/lib/Rewrite/Frontend/InclusionRewriter.cpp
>>> URL:
>>> http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Rewrite/Frontend/Inclus
>>> ionRewriter.cpp?rev=195910&r1=195909&r2=195910&view=diff
>>>
>>> =========================================================================
>>> ===== --- cfe/trunk/lib/Rewrite/Frontend/InclusionRewriter.cpp (original)
>>> +++ cfe/trunk/lib/Rewrite/Frontend/InclusionRewriter.cpp Thu Nov 28
>>> 01:21:44 2013
>>> @@ -363,15 +363,11 @@ bool InclusionRewriter::Process(FileID F
>>>     if (SM.getFileIDSize(FileId) == 0)
>>>       return false;
>>>
>>> -  // The next byte to be copied from the source file
>>> -  unsigned NextToWrite = 0;
>>> +  // The next byte to be copied from the source file, which may be
>>> non-zero if
>>> +  // the lexer handled a BOM.
>>> +  unsigned NextToWrite = SM.getFileOffset(RawLex.getSourceLocation());
>>>     int Line = 1; // The current input file line number.
>>>
>>> -  // Ignore UTF-8 BOM, otherwise it'd end up somewhere else than the
>>> start -  // of the resulting file.
>>> -  if (FromFile.getBuffer().startswith("\xEF\xBB\xBF"))
>>> -    NextToWrite = 3;
>>> -
>>>     Token RawToken;
>>>     RawLex.LexFromRawLexer(RawToken);

-- 
http://www.nuanti.com
the browser experts