/* scrub.c
 * A poor attempt at a program to strip comments and whitespace from C programs.
 * February 2009 - PUBLIC DOMAIN
 */
/* TODO: remove unneeded spaces between operators */

#include <ctype.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#define MAX_LINE_LENGTH 4194304
#define WHITESPACE "\t\v\r\n "
#define SPACE_OR_TAB "\t "

/** state varibles */
static int comments, quotes;

/** configuration variables */
static int strip_leading_whitespace_fl=1;
static int compact_all_whitespace_fl=1;

static void uncomment(char *s, const char *filename, unsigned line_no) {
	char *t;
	int backslash=0;
	for(;*s;s++) {
		if(comments) {
			/*
			fprintf(stderr, "%s:%u:Still in comment mode\n", filename, line_no);
			*/
			goto check_comment;
		} else if(!quotes) {
			if(*s=='/'&&s[1]=='/') {
				*s=0;
				return;
			} else if(*s=='/'&&s[1]=='*') {
check_comment:
				t=strstr(s, "*/");
				if(!t) { /* multiline comment */
					*s=0;
					comments=1;
					return;
				} else {
					/* replace comment with a single whitespace */
					fprintf(stderr, "%s:%u:partial line comment: '%.*s'\n", filename, line_no, t-s+2, s);
					t+=2;
					memmove(s+1, t, strlen(t)+1);
					comments=0;
					*s=' ';
				}
			} else if(*s=='"') {
				quotes='"';
			} else if(*s=='\'') {
				quotes='\'';
			}
		} else if(quotes && backslash) {
			backslash=0;
		} else if(quotes=='"') {
			/* TODO: handle \ at end of line inside of strings */
			if(*s=='\\') {
				fprintf(stderr, "%s:%u:Backslash in string literal\n", filename, line_no);
				backslash=1;
			} else if(!backslash&&*s=='"') {
				quotes=0;
			} else {
				backslash=0;
			}
		} else if(quotes=='\'') {
			/* TODO: handle \ at end of line inside of quotes */
			if(*s=='\\') {
				fprintf(stderr, "%s:%u:Backslash in character literal\n", filename, line_no);
				backslash=1;
			} else if(!backslash&&*s=='\'') {
				quotes=0;
			} else {
				backslash=0;
			}
		} else {
			abort();
		}
	}
}

static void strip_trailing_whitespace(char *s, const char *filename, unsigned line_no) {
	char last;
	size_t len;

	/* ignore if we are quoting */
	if(quotes) return;

	len=strlen(s);
	if(len<=0) return; /* ignore empty lines */
	len--; /* start at last character */

	last=s[len--]; /* save last character */

	/* find space */
	while(len>0 && isspace(s[len])) {
		len--;
	}

	/* truncate the string using last character and null */
	s[len+1]=last;
	s[len+2]=0;
}

static void quote_process(int *backslash, int *quote_mode, char ch, char quote_ch) {
	if(*backslash) {
		/* already in backslash mode, exit it */
		*backslash=0;
	} else if(ch=='\\') {
		/* enter backslash */
		*backslash=1;
	} else if(ch==quote_ch) {
		*quote_mode=0; /* leaving quotes mode */
	}
}

static void compact_spaces(char *s, const char *filename, unsigned line_no) {
	char *t;
	int backslash=0, whitespace=0, quote_mode=quotes;

	for(t=s;*t;t++) {
		if(whitespace && !isspace(*t)) { /* leave whitespace mode */
			*s++=whitespace; /* emit the held whitespace */
			whitespace=0;
		}

		/* do not modify whitespace between quotes */
		if(quote_mode=='\"') {
			quote_process(&backslash, &quote_mode, *t, '\"');
		} else if(quote_mode=='\'') {
			quote_process(&backslash, &quote_mode, *t, '\'');
		} else if(isspace(*t)) {
			/* use the first whitespace and ignore the remaining */
			if(whitespace) {
				fprintf(stderr, "%s:%u:skipping extra spaces.\n", filename, line_no);

				/* found a newline, hold that instead */
				if(*t=='\n')
					whitespace='\n';

				continue; /* ignore further whitespace */
			}
			/* else hold the first whitespace */
			whitespace=*t;
			continue;
		} else if (*t=='\"'||*t=='\'') {
			quote_mode=*t; /* enter quote_mode mode */
		}
		/* else: normal stream of characters being copied */

		/* copy */
		*s++=*t;
	}

	if(whitespace) {
		*s++=whitespace; /* emit the held whitespace */
	}

	*s=0;
}

void do_file(const char *filename, FILE *in, FILE *out) {
	char *line;
	unsigned line_no; /** current line number */
	size_t start;

	/* create a very large buffer */
	line=malloc(MAX_LINE_LENGTH);
	if(!line) {
		perror(filename);
		exit(EXIT_FAILURE);
	}

	line_no=0;
	while(fgets(line, MAX_LINE_LENGTH, in)) {
		line_no++;

		/* loop through leading whitespace */
		if(strip_leading_whitespace_fl) {
			start=strspn(line, SPACE_OR_TAB);
		} else {
			start=0;
		}

		uncomment(line+start, filename, line_no);

		/* go to next line if inside a comment */
		if(comments) continue;

		if(compact_all_whitespace_fl) {
			compact_spaces(line+start, filename, line_no);
		}

		strip_trailing_whitespace(line+start, filename, line_no);

		/* check that we are manipulating the lines correctly
		if(line[start]==' ') {
			fprintf(stderr, "%s:%u:space found at start of line but shouldn't be there.\n", filename, line_no);
			fprintf(stderr, "\"%s\"\n", line+start);
			abort();
		}
		*/
		if(!line[start]||line[start]=='\n') continue; /* swallow empty lines */

		fputs(line+start, out);
	}
	free(line);
}

int main(int argc, char **argv) {
	int i;
	if(argc<1) {
		do_file("<stdin>", stdin, stdout);
	} else for(i=1;i<argc;i++) {
		FILE *f;
		f=fopen(argv[i], "r");
		if(!f) {
			perror(argv[i]);
			return EXIT_FAILURE;
		}
		do_file(argv[i], f, stdout);
		fclose(f);
	}
	return 0;
}