CMake regex match header names in include declarations

Question

I'm attempting to get the header names of all the #include declarations in my robot.cpp, i.e., if I have an example.cpp with two declarations:

example.cpp

#include "MyLib.h"
#include <iosream>

//some code

I want to see in my output something like this:

INCLUDE_NAME MyLib.h
INCLUDE_NAME iosream

I have these robot.cpp and CMakeLists.txt files (they are simplified):

CMakeLists.txt

...
   file(STRINGS ${SRC_FILE} SRC_CONTENT) #copy the file content

   foreach(SRC_LINE ${SRC_CONTENT})
     if("${SRC_LINE }" MATCHES "^ *#include *[<\"](.*)[>\"]")
        message("INCLUDE_NAME ${CMAKE_MATCH_1}")
...

robot.cpp

#include "ArduinoRobot.h"
#include "EasyTransfer2.h"

void RobotControl::pointTo(int angle){
    int target=angle;
    uint8_t speed=80;
    target=target%360;
    if(target<0){
        target+=360;
    }
    int direction=angle;
    while(1){
        if(direction>0){
            motorsWrite(speed,-speed);//right
            delay(10);
        }else{
            motorsWrite(-speed,speed);//left
            delay(10);
        }
        int currentAngle=compassRead();
        int diff=target-currentAngle;
        if(diff<-180) 
            diff += 360;
        else if(diff> 180) 
            diff -= 360;
        direction=-diff;

        if(abs(diff)<5){
            motorsWrite(0,0);
            return;
        }
    }
}

Then, if I execute in a console my CMakeList.txt, I get this output:

...
    INCLUDE_NAME ArduinoRobot.h"#include "EasyTransfer2.h"void RobotControl::pointTo(int angle){     int target=angle;       uint8_t speed=80;       target=target%360;      if(target<0){           target+=360; }       int direction=angle;    while(1){               if(direction>0){                        motorsWrite(speed,-speed);//right                       delay(10);              }else{                  motorsWrite(-speed,speed);//left                     delay(10);              }               int currentAngle=compassRead();         int diff=target-currentAngle;           if(diff<-180)                   diff += 360;            else if(diff    
...

However, if I change all the symbols > to <, the result is ok, and I get the following output:

...
INCLUDE_NAME ArduinoRobot.h
INCLUDE_NAME EasyTransfer2.h
...

I guess it is because the regex, but I don't know how fix it.

Does anyone?

Edward · Accepted Answer · 2014-04-07 15:40:30Z

I believe there are two problems. The first one, I strongly suspect, is line endings in your source code. Your code would have worked as you'd had it originally except that the code you're parsing is probably downloaded from somewhere and that file likely has different line endings. This can happen, for example, if the original was developed under Linux and you're operating under Windows. That needs to be fixed first.

Secondly, what you really want is a regex that doesn't accept just any character in the middle, but only characters that don't match " or >. Here's what I tested with just now:

CMAKE_MINIMUM_REQUIRED(VERSION 2.8 FATAL_ERROR)
PROJECT(NONSENSE NONE)
file(STRINGS "silly.txt" SRC_CONTENT)
foreach(SRC_LINE ${SRC_CONTENT})
    if("${SRC_LINE}" MATCHES "^[ \t]*#[ \t]*include[ \t]*[<\"]([^>\"]*)[>\"]")
        message("INCLUDE_NAME ${CMAKE_MATCH_1}")
    endif()
ENDFOREACH()

The contents of silly.txt:

#  include "foo.h"
#include    "bar.h"

#include "ArduinoRobot.h"#include "EasyTransfer2.h"int main()
{
    printf("This is not a header.\n");
    return 0;
}

    #include "surprise.h"

Note that there is a tab character in the middle of bar.h line and at the begining of the surprise.h line. Weird, but still accepted by preprocessors is also the variation shown on the foo.h line in which there are spaces or tabs between the # and the directive. The regex shown handles all of these.

This the the output of that:

INCLUDE_NAME foo
INCLUDE_NAME bar
INCLUDE_NAME foo.h
INCLUDE_NAME bar.h
INCLUDE_NAME ArduinoRobot.h
INCLUDE_NAME surprise.h

Thanks a lot @Edward, it works!! :) I just tested my code and it goes smoothly

Barth · Accepted Answer · 2014-04-07 15:04:01Z

0

CMake regexes are greedy, thus the output.

Maybe try to match less things in between the apostrophes :

"^ *#include *[<\"](\w+)[>\"]"

I am wondering if CMake accepts explicit non-greedy syntax such as

"^ *#include *[<\"]\\(.*)\\[>\"]"

(not tested!)

edited Apr 7, 2014 at 15:04

answered Apr 7, 2014 at 14:58

Barth

15.9k22 gold badges74 silver badges108 bronze badges

1 Comment

fenix688 Over a year ago

Thanks so much @Barth but it doesn't work :( I know CMake regexes are greedy but I didn't find anything to fix it.

Collectives™ on Stack Overflow

CMake regex match header names in include declarations

2 Answers 2

1 Comment

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related