Command to strip out comments in source files

[cyberfish]

    printf("/* Make sure you deal with this case. */");

Ah, that I didn't think of.

The code is not in source control.

sed 's@//.*@@g'

That takes out the "//" comments, but I need to take out the "/* ... */" ones, too, which I heard is more difficult (multi-lines).

I am thinking about just writing a simple one myself in C++, as it doesn't really need to be fool proof. As long as it works for my code... .

[cyberfish]

My ghetto solution, with a consecutive-blank-lines-stripper -

#include <iostream>
#include <fstream>
#include <string>

#include <cassert>
#include <cctype>

const int max_length = 1024*1024*10; //10 MB

int main(int argc, char **argv) {
	int file_length = 0;
	char *buffer = new char[max_length]; 
	std::string out_data;
	
	std::ifstream in_file(argv[1], std::ios::in | std::ios::binary);
	
	if (!in_file) {
		std::cerr << "Error opening file: " << argv[1] << std::endl;
		return 1;
	}
	
	if (in_file.read(buffer, max_length-1)) { //if the read was successful without hitting an EOF...
		std::cerr << "File too big." << std::endl;
		return 1;
	}
	
	file_length = in_file.gcount();
	
	buffer[file_length] = 0;
	
	std::string in_data(buffer); //make it a string so we have nice functions to play with
	
	//first replace every "\r" by space
	for (int i = 0; i < in_data.size(); ++i) {
		if (in_data[i] == '\r') {
			in_data[i] = ' ';
		}
	}
	
	int state = 0; //0 = normal, 1 = ignore until end of line, 2 = ignore until "*/", 3 = ignore until "#endif"
	
	int skip_next = 0;
	
	for (int i = 0; i < in_data.size(); ++i) {
		
		if (skip_next) {
			--skip_next;
			continue;
		}
		
		switch(state) {
			case 0:
				if ((i + 2 <= in_data.size()) && (in_data.substr(i, 2) == "//")) {
					state = 1;
				} else if ((i + 2 <= in_data.size()) && (in_data.substr(i, 2) == "/*")) {
					state = 2;
				} else if ((i + 5 <= in_data.size()) && (in_data.substr(i, 5) == "#if 0")) {
					state = 3;
				} else {
					out_data.push_back(in_data[i]);
				}
				break;
			case 1:
				if (in_data[i] == '\n') {
					out_data.push_back('\n');
					state = 0;
				}
				break;
			case 2: 
				if ((i + 2 <= in_data.size()) && (in_data.substr(i, 2) == "*/")) {
					out_data.push_back('\n');
					state = 0;
					skip_next = 1;
				}
				break;
			case 3: 
				if ((i + 6 <= in_data.size()) && (in_data.substr(i, 6) == "#endif")) {
					out_data.push_back('\n');
					state = 0;
					skip_next = 5;
				}
				break;
			default:
				assert(false);
		}
		
	}
	
	int n_count = 0;
	for (int i = 0; i < out_data.size(); ++i) {
		if (n_count > 1 && (out_data[i] == '\n' || out_data[i] == '\t' || out_data[i] == ' ')) {
			continue;
		} else {
			if (out_data[i] != '\t' && out_data[i] != '\n' && out_data[i] != ' ') {
				if (n_count > 1) { //get all the junk we threw away back!
					int j = i;
					while (out_data[j] != '\n') {
						--j;
					}
					for (++j; j < i; ++j) {
						std::cout << out_data[j];
					}
				}
				n_count = 0;
			}
			if (out_data[i] == '\n') {
				++n_count;
			}
			std::cout << out_data[i];
		}
	}
}

It's unbreakable until you try to break it .

EDIT: code edited for the consecutive-blank-lines-stripper.

[dwks]

I've never put it to the test, but I have a program called rmccmt on my Linux machine ("remove c/c++ comments").

I do have a program which calculates how much source code consists of comments. I tested it pretty thoroughly, and I could post it here if you wanted, but I suspect you'd be able to figure it out yourself without too much difficulty.

@cyberfish: About your code: consider looking at isspace().

[brewbuck]

@cyberfish: About your code: consider looking at isspace().

isspace() is locale-specific, but the language itself isn't. I think it's better to explicitly check for the characters which are declared by the standard to be "whitespace."

[cyberfish]

I've never put it to the test, but I have a program called rmccmt on my Linux machine ("remove c/c++ comments").

That is strange. I have Ubuntu 8.04 and don't have it. Not even in the APT repository.

*edit* ah nvm, it's in liwc */edit*

*edit2* It messes up indentations, too, for some reason, and doesn't catch "#if 0/#endif" Guess I will just use mine . */edit2*