text file cleanup: unix->dos line endings, missing newline at EOF, and trailing whitespace

dropcr.c

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

char *find_nonws(char *ptr, char *end);

int trailing_ws, fixed_trailing;

int main(int argc, char **argv) {

	int i, j, count = 0;

	for(i=1; i<argc; i++) {
		if(argv[i][0] == '-') {
			if(argv[i][2] == 0) {
				switch(argv[i][1]) {
				case 'w':
					trailing_ws = 1;
					break;

				case 'h':
				default:
					goto usage;
				}
			} else {
				goto usage;
			}
		} else {
			count++;
		}
	}

	if(!count) {
usage:	printf("Usage %s [options] <file> [<file 2> ... <file N>]\n", argv[0]);
		printf("Options:\n");
		printf("  -w: clean trailing whitespace\n");
		printf("  -h: print usage and exit\n");
		return 1;
	}


	for(j=1; j<argc; j++) {
		FILE *fp;
		char *memfile, *ptr, *end;
		long filesize;
		int lines = 0;

		if(argv[j][0] == '-') continue;

		if(!(fp = fopen(argv[j], "rb"))) {
			fprintf(stderr, "Could not open file %s for reading\n", argv[j]);
			continue;
		}

		fseek(fp, 0, SEEK_END);
		filesize = ftell(fp);

		ptr = memfile = malloc(filesize);
		if(!memfile) {
			fprintf(stderr, "failed to allocate %ld bytes for: %s\n", filesize, argv[j]);
			fclose(fp);
			continue;
		}

		fseek(fp, 0, SEEK_SET);
		for(i=0; i<filesize; i++) {
			fread(ptr++, 1, 1, fp);
		}
		fclose(fp);

		if(!(fp = fopen(argv[j], "wb"))) {
			fprintf(stderr, "Could not open file %s for writing\n", argv[j]);
			free(memfile);
			continue;
		}

		ptr = memfile;
		end = memfile + filesize;
		while(ptr < end) {

			switch(*ptr) {
			case 13:
				if(ptr[1] != 10) {
					fputc(10, fp);
				}
				ptr++;
				lines++;
				break;

			case 10:
				fputc(*ptr++, fp);
				break;

			default:
				if(trailing_ws && isspace(*ptr)) {
					char *ws_end = find_nonws(ptr + 1, end);
					if(ws_end > ptr && (*ws_end == 13 || *ws_end == 10)) {
						ptr = ws_end;
						fixed_trailing++;
					} else {
						fwrite(ptr, 1, ws_end - ptr, fp);
						ptr = ws_end;
					}
				} else {
					fputc(*ptr++, fp);
				}
			}
		}

		printf("%s - fixed %d line endings", argv[j], lines);
		if(fixed_trailing) {
			printf(", %d trailing w/s", fixed_trailing);
		}

		if(memfile[filesize - 1] != '\n') {
			fputc('\n', fp);
			puts(", and newline @ eof");
		} else {
			putchar('\n');
		}

		fclose(fp);
		free(memfile);
		fixed_trailing = 0;
	}

	return 0;
}

char *find_nonws(char *ptr, char *end)
{
	while(ptr < end && (*ptr == ' ' || *ptr == '\t' || *ptr == '\v')) ptr++;
	return ptr;
}