MPLex problem

Dev Shed Forums Sponsor:

December 3rd, 2011, 08:32 AM

minor28

Contributing User

Join Date: Jan 2004

Location: Sweden

Posts: 259

Time spent in forums: 4 h 1 m 26 sec
Reputation Power: 19

MPLex problem

Could anyone help me with a MPLex problem.

I want to find comment blocks like

Code:

/*++ BUILD Version: 0091     Increment this if a change has global effects
Copyright (c) Microsoft Corporation. All rights reserved.
*/

and #define statements with two or more lines like

Code:

#define FILE_GENERIC_READ	(STANDARD_RIGHTS_READ |\
							FILE_READ_DATA |\
							FILE_READ_ATTRIBUTES |\
							FILE_READ_EA |\
							SYNCHRONIZE)

in c++ headers.

The comment blocks is OK but not #define blocks. This is
so far I have come.

Code:

%COMMENT
%DEFINE

W	[0-9a-zA-Z_]
S	[ \t\v\n\f\r]

%%
"/*"				{ BEGIN(COMMENT); return (int)HToken.COMMENTBLOCK; }
<COMMENT>[^\|*\n]*	{ return (int)HToken.COMMENTBLOCK; }
<COMMENT>"*/"		{ BEGIN(INITIAL); return (int)HToken.COMMENTBLOCK; }

#define.*"\\"			{ BEGIN(DEFINE); return (int)HToken.DEFINE; }
<DEFINE>(.*"\\"[^\n])+	{ return (int)HToken.DEFINE; }
<DEFINE>.*")"			{ BEGIN(INITIAL); return (int)HToken.DEFINE; }

%%

This is the result of #define block:

Code:

#define FILE_GENERIC_READ (STANDARD_RIGHTS_READ |\SYNCHRONIZE)

December 19th, 2011, 05:17 PM

ragax

Turn left at the third duck

Join Date: Dec 2011

Location: Nelson, NZ

Posts: 93

Time spent in forums: 1 Day 24 m 37 sec
Reputation Power: 92

Hi there,

I'll say at the outset that I am not familiar with MPlex, but since no one replied so far I thought I would try to give you a hand with a fairly standard syntax.

This expression works using a number of regex flavors: PCRE, Java, Perl, .NET.

Code:

(?s)#define[^(]*?\([^)]*?\)

It matches the define block in your example.

In other flavors, you could remove the inline flag for "dot-matches new line"---the "(?s)" and turn that on using your local syntax.

Hoping this works for you. Please let me know.

December 20th, 2011, 07:29 AM

minor28

Contributing User

Join Date: Jan 2004

Location: Sweden

Posts: 259

Time spent in forums: 4 h 1 m 26 sec
Reputation Power: 19

Thank you for your answer.

Code:

#define{space}+{W}+{space}+	{ BEGIN(DEFINE); return (int)NodeToken.DEFINE_BLOCK; }
<DEFINE>"("{OR}*		{ return (int)NodeToken.DEFINE_BLOCK; }
<DEFINE>[^\\\r\n]+		{
				if (yytext.EndsWith("\\"))
				{
					return (int)NodeToken.DEFINE_BLOCK;
				}
				else
				{
					BEGIN(INITIAL); 
					return (int)NodeToken.DEFINE_BLOCK;
				}
				}
<DEFINE>{OR}*")"$		{ BEGIN(INITIAL); return (int)NodeToken.DEFINE_BLOCK; }

where {W} is word characters and {OR} also include " " and "|" chars.

This will match #defines with one or more lines.

December 20th, 2011, 01:43 PM

ragax

Turn left at the third duck

Join Date: Dec 2011

Location: Nelson, NZ

Posts: 93

Time spent in forums: 1 Day 24 m 37 sec
Reputation Power: 92

Hi, you're welcome. Sorry, I don't understand your last post.
Are you saying that you want to match the pattern in the new box?

December 20th, 2011, 02:37 PM

minor28

Contributing User

Join Date: Jan 2004

Location: Sweden

Posts: 259

Time spent in forums: 4 h 1 m 26 sec
Reputation Power: 19

My lexer (the new box) will match all #defines for example like these (one or more lines) from a c++ header file.

Code:

#define HSHELL_FLASH	(HSHELL_REDRAW|HSHELL_HIGHBIT)
#define GHND		(GMEM_MOVEABLE | \
			GMEM_ZEROINIT)

My parser will then write these lines to a masm include file

Code:

HSHELL_FLASH equ (HSHELL_REDRAW or HSHELL_HIGHBIT)
GHND equ (GMEM_MOVEABLE or GMEM_ZEROINIT)

The problem is solved.

Developer Shed Advertisers and Affiliates