int
Scan_File(char *inpathname, Index *ind, Pathstore *store, int discardDups)
{
// Save the pathname in the store
char *pathname = Pathstore_path(store, inpathname,discardDups);
if (pathname == NULL) {
numdups++;
DPRINTF('s',("Scan_Pathname discard dup (%s)\n", inpathname));
return 0;
}
numfiles++;
DPRINTF('s', ("Scan_Pathname(%s)\n", pathname));
int fd = Fileops_open(pathname);
if (fd < 0) {
fprintf(stderr, "Can't open pathname %s\n", pathname);
return -1;
}
int size = 0;
int inum = Fileops_getinumber(fd, &size);
printf("Size: %d\n", size);
if(inum < 0) {Fileops_close(fd); return 0; }
unsigned char buf[DISKIMG_SECTOR_SIZE];
int blockNo = Fileops_getblockno(fd);
int bytesMoved = Fileops_getblock(inum, blockNo, buf);
int ch = Fileops_getchar(fd, inum, size, buf, &bytesMoved);
numchars++;
while (!(ch < 0)) { // Process words until we reach the end of the file
while (!isalpha(ch)) { // Skip any leading non-alpha characters
ch = Fileops_getchar(fd, inum, size, buf, &bytesMoved);
if (ch < 0) {Fileops_close(fd); return 0; }
numchars++;
}
// Found a word - record it in the index.
int offset = Fileops_tell(fd);
int pos = 0;
// read off the word until we hit the end of the word buffer
// or the end of the file or we hit an non-alpah characters
char word[MAX_WORD_SIZE+1];
while ((pos < MAX_WORD_SIZE) && !(ch < 0) && isalpha(ch)) {
word[pos++] = ch;
ch = Fileops_getchar(fd, inum, size, buf, &bytesMoved);
numchars++;
}
numwords++;
word[pos] = 0; // terminate string
bool ok = Index_StoreEntry(ind, word, pathname, offset);
assert(ok);
}
Fileops_close(fd);
return 0;
}
int Scan_TreeAndIndex(char *pathname, Index *ind, Pathstore *store,int discardDups) {
const uint32_t MAXPATH = 1024;
if (Fileops_isfile(pathname)) {
return Scan_File(pathname, ind, store, discardDups);
}
// Not a file must be directory, process all entries in the directory
if (strlen(pathname) > MAXPATH-16) {
fprintf(stderr, "Too deep of directories %s\n", pathname);
return -1;
}
numdirs++;
int dirfd = Fileops_open(pathname);
if (dirfd < 0) {
fprintf(stderr, "Can't open pathname %s\n", pathname);
return -1;
}
if (pathname[1] == 0) {
// pathame == "/"
pathname++; // Delete extra / character
}
int ret;
while (1) {
struct direntv6 dirent;
ret = Fileops_read(dirfd, (char *)&dirent, sizeof(struct direntv6));
if (ret == 0) {
/* Done with directory */
break;
}
if (ret != sizeof(struct direntv6)) {
fprintf(stderr, "Error reading directory %s\n", pathname);
ret = -1;
break;
}
numdirents++;
char *n = dirent.d_name;
if (n[0] == '.') {
if ((n[1] == 0) || ((n[1] == '.') && (n[2] == 0))) {
/* Skip over "." and ".." */
continue;
}
}
char nextpath[MAXPATH];
sprintf(nextpath, "%s/%s",pathname, n);
Scan_TreeAndIndex(nextpath, ind, store, discardDups);
}
Fileops_close(dirfd);
return ret;
}
/*
* Do the two pathnames refer to a file with the same contents.
*/
static int
IsSameFile(Pathstore *store, char *pathname1, char *pathname2)
{
char chksum1[CHKSUMFILE_SIZE],
chksum2[CHKSUMFILE_SIZE];
struct unixfilesystem *fs = (struct unixfilesystem *) (store->fshandle);
numcompares++;
if (strcmp(pathname1, pathname2) == 0) {
return 1; // Same pathname must be same file.
}
/* Compute the chksumfile of each file to see if they are the same */
int err = chksumfile_bypathname(fs, pathname1, chksum1);
if (err < 0) {
fprintf(stderr,"Can't checksum path %s\n", pathname1);
return 0;
}
err = chksumfile_bypathname(fs, pathname2, chksum2);
if (err < 0) {
fprintf(stderr,"Can't checksum path %s\n", pathname2);
return 0;
}
if (chksumfile_compare(chksum1, chksum2) == 0) {
numdiffchecksum++;
return 0; // Checksum mismatch, not the same file
}
/* Checksums match, do a content comparison */
int fd1 = Fileops_open(pathname1);
if (fd1 < 0) {
fprintf(stderr, "Can't open path %s\n", pathname1);
return 0;
}
int fd2 = Fileops_open(pathname2);
if (fd2 < 0) {
Fileops_close(fd1);
fprintf(stderr, "Can't open path %s\n", pathname2);
return 0;
}
int ch1, ch2;
do {
ch1 = Fileops_getchar(fd1);
ch2 = Fileops_getchar(fd2);
if (ch1 != ch2) {
break; // Mismatch - exit loop with ch1 != ch2
}
} while (ch1 != -1);
// if files match then ch1 == ch2 == -1
Fileops_close(fd1);
Fileops_close(fd2);
if (ch1 == ch2) {
numsamefiles++;
} else {
numdifferentfiles++;
}
return ch1 == ch2;
}
