path: root/usr.sbin/ctm/mkCTM/mkctm.c

/* Still missing:
 *
 * Damage counter
 * Change counter
 * Time stamp
 * prefix
 * cmd-line args
 * %100 deltas
 * delta and Add are different. delta -> Equ.
 *
 * mkctm 
 *	-B regex	Bogus
 *	-I regex	Ignore
 *	-D int		Damage
 *	-v		increase verbosity
 *	-l str		control logging.
 *	name		cvs-cur
 *	prefix		src/secure
 *	dir1		"Soll"
 *	dir2		"Ist"
 *
 */

#include <sys/types.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <sys/wait.h>
#include <dirent.h>
#include <regex.h>
#include <stdio.h>
#include <fcntl.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <md5.h>
#include <err.h>
#include <signal.h>

#define DEFAULT_IGNORE	"/CVS$|/\\.#"
#define DEFAULT_BOGUS	"\\.core$"
regex_t reg_ignore,  reg_bogus;
int	flag_ignore, flag_bogus;

u_long s1_ignored,	s2_ignored;
u_long s1_bogus,	s2_bogus;
u_long s1_wrong,	s2_wrong;
u_long s_same_dirs,	s_same_files,	s_same_bytes;
u_long 			s_files_chg,	s_bytes_add,	s_bytes_del;
u_long s_new_dirs,	s_new_files,	s_new_bytes;
u_long s_del_dirs,	s_del_files,	s_del_bytes;
u_long            	s_chg_files,	s_chg_bytes;
u_long 			s_edit_files,	s_edit_bytes,	s_edit_saves;
u_long 			s_sub_files,	s_sub_bytes;

void
print_stat(FILE *fd, char *pre)
{
    fprintf(fd,"%sAvoided:\n",pre);
    fprintf(fd,"%s  ignore:  %5lu old   %5lu new\n",
	    pre, s1_ignored, s2_ignored);
    fprintf(fd,"%s  bogus:   %5lu old   %5lu new\n",
	    pre, s1_bogus, s2_bogus);
    fprintf(fd,"%s  wrong:   %5lu old   %5lu new\n", 
	    pre, s1_wrong, s2_wrong);
    fprintf(fd,"%sDelta:\n",pre);
    fprintf(fd,"%s  new:     %5lu dirs  %5lu files  %9lu plus\n", 
	    pre, s_new_dirs, s_new_files, s_new_bytes);
    fprintf(fd,"%s  del:     %5lu dirs  %5lu files                   %9lu minus\n", 
	    pre, s_del_dirs, s_del_files, s_del_bytes);
    fprintf(fd,"%s  chg:                 %5lu files  %9lu plus   %9lu minus\n",
	    pre, s_files_chg, s_bytes_add, s_bytes_del);
    fprintf(fd,"%s  same:    %5lu dirs  %5lu files  %9lu bytes\n", 
	    pre, s_same_dirs, s_same_files, s_same_bytes);
    fprintf(fd,"%sMethod:\n",pre);
    fprintf(fd,"%s  edit:                %5lu files  %9lu bytes  %9lu saved\n", 
	    pre, s_edit_files, s_edit_bytes, s_edit_saves);
    fprintf(fd,"%s  sub:                 %5lu files  %9lu bytes\n", 
	    pre, s_sub_files, s_sub_bytes);
}

void
stat_info(int foo)
{
	signal(SIGINFO,stat_info);
	print_stat(stderr,"INFO: ");
}

void DoDir(const char *dir1, const char *dir2, const char *name);

static struct stat st;
static __inline struct stat *
StatFile(char *name)
{
	if (lstat(name,&st) < 0) 
		err(1,"Couldn't stat %s\n",name);
	return &st;
}

int
dirselect(struct dirent *de)
{
	if (!strcmp(de->d_name,"."))	return 0;
	if (!strcmp(de->d_name,".."))	return 0;
	return 1;
}

void
name_stat(const char *pfx, const char *dir, const char *name, struct dirent *de)
{
	char *buf = alloca(strlen(dir) + strlen(name) + 
		strlen(de->d_name) + 3);
	struct stat *st;

	strcpy(buf,dir); 
		strcat(buf,"/"); strcat(buf,name);
		strcat(buf,"/"); strcat(buf,de->d_name);
	st = StatFile(buf);
	printf("%s %s%s %lu %lu %o", 
	    pfx, name, de->d_name, 
	    st->st_uid, st->st_gid, st->st_mode & ~S_IFMT);
}

void
Equ(const char *dir1, const char *dir2, const char *name, struct dirent *de)
{
	if (de->d_type == DT_DIR) {
		char *p = alloca(strlen(name)+strlen(de->d_name)+2);

		strcpy(p,name);  strcat(p,de->d_name); strcat(p, "/");
		DoDir(dir1,dir2,p);
		s_same_dirs++;
	} else {
		char *buf1 = alloca(strlen(dir1) + strlen(name) + 
			strlen(de->d_name) + 3);
		char *buf2 = alloca(strlen(dir2) + strlen(name) + 
			strlen(de->d_name) + 3);
		char *m1,md5_1[33],*m2, md5_2[33];
		void *p1,*p2;
		int fd1,fd2;
		struct stat s1,s2;

		strcpy(buf1,dir1); 
			strcat(buf1,"/"); strcat(buf1,name);
			strcat(buf1,"/"); strcat(buf1,de->d_name);
		fd1 = open(buf1,O_RDONLY);
		if(fd1 < 0) { perror(buf1); exit(3); }
		fstat(fd1,&s1);
		strcpy(buf2,dir2); 
			strcat(buf2,"/"); strcat(buf2,name);
			strcat(buf2,"/"); strcat(buf2,de->d_name);
		fd2 = open(buf2,O_RDONLY);
		if(fd2 < 0) { perror(buf2); exit(3); }
		fstat(fd2,&s2);
#if 1
		if (s1.st_size == s2.st_size) {
			s_same_files++;
			s_same_bytes += s1.st_size;
			close(fd1);
			close(fd2);
			goto finish;
		}
#endif
		p1=mmap(0,s1.st_size,PROT_READ,MAP_PRIVATE,fd1,0);
		if ((int)p1 == -1) { perror(buf1); exit(3); }
		close(fd1);

		p2=mmap(0,s2.st_size,PROT_READ,MAP_PRIVATE,fd2,0);
		if ((int)p2 == -1) { perror(buf2); exit(3); }
		close(fd2);

		/* If identical, we're done. */
		if((s1.st_size == s2.st_size) && !memcmp(p1,p2,s1.st_size)) {
			s_same_files++;
			s_same_bytes += s1.st_size;
			goto finish;
		}

		s_files_chg++;
		if (s1.st_size > s2.st_size)
			s_bytes_del += (s1.st_size - s2.st_size);
		else
			s_bytes_add += (s2.st_size - s1.st_size);

		m1 = MD5Data(p1, s1.st_size, md5_1);
		m2 = MD5Data(p2, s2.st_size, md5_2);
		
		/* Just a curiosity... */
		if(!strcmp(m1,m2)) {
			if (s1.st_size != s2.st_size) 
				fprintf(stderr,
		"Notice: MD5 same for files of diffent size:\n\t%s\n\t%s\n",
					buf1,buf2);
			goto finish;
		}

		{
			u_long l = s1.st_size + s2.st_size;
			u_char *ob = alloca(l);
			Dissect(p1,p1+s1.st_size,p2,p2+s2.st_size,ob,&l);
			if (l && l < s2.st_size) {
				name_stat("CTMFB",dir2,name,de);
				printf(" %s %s %d\n",m1,m2,(unsigned)s1.st_size);
				s_edit_files++;
				s_edit_bytes += l;
				s_edit_saves += (s2.st_size - l);
				fwrite(ob,1,l,stdout);
				putchar('\n');
			} else {
				name_stat("CTMFS",dir2,name,de);
				printf(" %s %s %u\n",m1,m2,(unsigned)s2.st_size);
				fwrite(p2,1,s2.st_size,stdout);
				s_sub_files++;
				s_sub_bytes += s2.st_size;
			}
		}
	    finish:
		munmap(p1,s1.st_size);
		munmap(p2,s2.st_size);
	}
}

void
Add(const char *dir1, const char *dir2, const char *name, struct dirent *de)
{
	if (de->d_type == DT_DIR) {
		char *p = alloca(strlen(name)+strlen(de->d_name)+2);
		strcpy(p,name);  strcat(p,de->d_name); strcat(p, "/");
		name_stat("CTMDM",dir2,name,de);
		putchar('\n');
		s_new_dirs++;
		DoDir(dir1,dir2,p);
	} else if (de->d_type == DT_REG) {
		char *buf2 = alloca(strlen(dir2) + strlen(name) + 
			strlen(de->d_name) + 3);
		char *m2, md5_2[33];
		u_char *p1;
		struct stat st;
		int fd1;

		strcpy(buf2,dir2); 
			strcat(buf2,"/"); strcat(buf2,name);
			strcat(buf2,"/"); strcat(buf2,de->d_name);
		fd1 = open(buf2,O_RDONLY);
		if (fd1 < 0) {perror(buf2); exit (3); }
		fstat(fd1,&st);
		p1=mmap(0,st.st_size,PROT_READ,MAP_PRIVATE,fd1,0);
		if ((int)p1 == -1) { perror(buf2); exit(3); }
		close(fd1);
		m2 = MD5Data(p1, st.st_size, md5_2);
		name_stat("CTMFM",dir2,name,de);
		printf(" %s %u\n",m2,(unsigned)st.st_size);
		fwrite(p1,1,st.st_size,stdout);
		putchar('\n');
		munmap(p1,st.st_size);
		s_new_files++;
		s_new_bytes += st.st_size;
	}
}

void
Del (const char *dir1, const char *dir2, const char *name, struct dirent *de)
{
	if (de->d_type == DT_DIR) {
		char *p = alloca(strlen(name)+strlen(de->d_name)+2);
		strcpy(p,name);  strcat(p,de->d_name); strcat(p, "/");
		DoDir(dir1,dir2,p);
		printf("CTMDR %s%s\n",name,de->d_name);
		s_del_dirs++;
	} else if (de->d_type == DT_REG) {
		char *buf1 = alloca(strlen(dir1) + strlen(name) + 
			strlen(de->d_name) + 3);
		char *m1, md5_1[33];
		strcpy(buf1,dir1); 
			strcat(buf1,"/"); strcat(buf1,name);
			strcat(buf1,"/"); strcat(buf1,de->d_name);
		m1 = MD5File(buf1, md5_1);
		printf("CTMFR %s%s %s\n",name,de->d_name,m1);
		s_del_files++;
		s_del_bytes += StatFile(buf1)->st_size;
	}
}

void
GetNext(int *i, int *n, struct dirent **nl, const char *dir, const char *name, u_long *ignored, u_long *bogus, u_long *wrong)
{
	char buf[BUFSIZ];

	for (;;) {
		for (;;) {
			(*i)++;
			if (*i >= *n)
				return;
			*buf = 0;
			if (*dir != '/')
				strcat(buf,"/");
			strcat(buf,dir); 
			if (buf[strlen(buf)-1] != '/')
				strcat(buf,"/"); 
			strcat(buf,name);
			if (buf[strlen(buf)-1] != '/')
				strcat(buf,"/"); 
			fprintf(stderr,">%d<%s>\n",strlen(nl[*i]->d_name),nl[*i]->d_name);
			fflush(stderr);
			strcat(buf,nl[*i]->d_name);
			if (flag_ignore && !regexec(&reg_ignore,buf,0,0,0))
				(*ignored)++;
			else if (flag_bogus && !regexec(&reg_bogus,buf,0,0,0))
				(*bogus)++;
			else
				break;
			free(nl[*i]); nl[*i] = 0;
		}
		/* If the filesystem didn't tell us, find type */
		if (nl[*i]->d_type == DT_UNKNOWN) 
			nl[*i]->d_type = IFTODT(StatFile(buf)->st_mode);
		if (nl[*i]->d_type == DT_REG || nl[*i]->d_type == DT_DIR)
			break;
		(*wrong)++;
		free(nl[*i]); nl[*i] = 0;
	}
}

void
DoDir(const char *dir1, const char *dir2, const char *name)
{
	int i1,i2,n1,n2,i;
	struct dirent **nl1,**nl2;
	char *buf1 = alloca(strlen(dir1) + strlen(name) + 4);
	char *buf2 = alloca(strlen(dir2) + strlen(name) + 4);

	strcpy(buf1,dir1); strcat(buf1,"/"); strcat(buf1,name);
	strcpy(buf2,dir2); strcat(buf2,"/"); strcat(buf2,name);
	n1 = scandir(buf1, &nl1, dirselect, alphasort);
	n2 = scandir(buf2, &nl2, dirselect, alphasort);
	i1 = i2 = -1;
	GetNext(&i1, &n1, nl1, dir1, name, &s1_ignored, &s1_bogus, &s1_wrong);
	GetNext(&i2, &n2, nl2, dir2, name, &s2_ignored, &s2_bogus, &s2_wrong);
	for (;i1 < n1 || i2 < n2;) {

		/* Get next item from list 1 */
		if (i1 < n1 && !nl1[i1]) 
			GetNext(&i1, &n1, nl1, dir1, name, 
				&s1_ignored, &s1_bogus, &s1_wrong);

		/* Get next item from list 2 */
		if (i2 < n2 && !nl2[i2]) 
			GetNext(&i2, &n2, nl2, dir2, name, 
				&s2_ignored, &s2_bogus, &s2_wrong);

		if (i1 >= n1 && i2 >= n2) {
			/* Done */
			break;
		} else if (i1 >= n1 && i2 < n2) {
			/* end of list 1, add anything left on list 2 */
			Add(dir1,dir2,name,nl2[i2]);
			free(nl2[i2]); nl2[i2] = 0;
		} else if (i1 < n1 && i2 >= n2) {
			/* end of list 2, delete anything left on list 1 */
			Del(dir1,dir2,name,nl1[i1]);
			free(nl1[i1]); nl1[i1] = 0;
		} else if (!(i = strcmp(nl1[i1]->d_name, nl2[i2]->d_name))) {
			/* Identical names */
			if (nl1[i1]->d_type == nl2[i2]->d_type) {
				/* same type */
				Equ(dir1,dir2,name,nl1[i1]);
			} else {
				/* different types */
				Del(dir1,dir2,name,nl1[i1]);
				Add(dir1,dir2,name,nl2[i2]);
			}
			free(nl1[i1]); nl1[i1] = 0;
			free(nl2[i2]); nl2[i2] = 0;
		} else if (i < 0) {
			/* Something extra in list 1, delete it */
			Del(dir1,dir2,name,nl1[i1]);
			free(nl1[i1]); nl1[i1] = 0;
		} else {
			/* Something extra in list 2, add it */
			Add(dir1,dir2,name,nl2[i2]);
			free(nl2[i2]); nl2[i2] = 0;
		}
	}
	if (n1 >= 0)
		free(nl1);
	if (n2 >= 0)
		free(nl2);
}

int
main(int argc, char **argv)
{
	int i;
	extern char *optarg;
	extern int optind;

	if (regcomp(&reg_bogus,DEFAULT_BOGUS, REG_EXTENDED | REG_NEWLINE))
		/* XXX use regerror to explain it */
		err(1,"Default regular expression argument to -B is botched");
	flag_bogus = 1;

	if (regcomp(&reg_ignore,DEFAULT_IGNORE, REG_EXTENDED | REG_NEWLINE))
		/* XXX use regerror to explain it */
		err(1,"Default regular expression argument to -I is botched");
	flag_ignore = 1;

	while ((i = getopt(argc,argv,"I:B:")) != EOF)
		switch (i) {
		case 'I':
			if (flag_ignore)
				regfree(&reg_ignore);
			flag_ignore = 0;
			if (!*optarg)
				break;
			if (regcomp(&reg_ignore,optarg,
			    REG_EXTENDED | REG_NEWLINE))
				/* XXX use regerror to explain it */
				err(1,"Regular expression argument to -I is botched");
			flag_ignore = 1;
			break;
		case 'B':
			if (flag_bogus)
				regfree(&reg_bogus);
			flag_bogus = 0;
			if (!*optarg)
				break;
			if (regcomp(&reg_bogus,optarg,
			    REG_EXTENDED | REG_NEWLINE))
				/* XXX use regerror to explain it */
				err(1,"Regular expression argument to -B is botched");
			flag_bogus = 1;
			break;
		case '?':
		default:
			fprintf(stderr,"Usage:\n\t%s: %s\n", argv[0],
"[-I ignore_re] [-B bogus_re]");
			return (1);
		}
	argc -= optind;
	argv += optind;

	setbuf(stdout,0);

	signal(SIGINFO,stat_info);
	printf("CTM_BEGIN 2.0 tst 0 950326022230Z .\n");
	DoDir(argv[0],argv[1],"");
	printf("CTM_END ");
	print_stat(stderr,"");
	exit(0);
}