void rf (const char *name) { struct stat st, st2, st3; const size_t namelen = strlen(name); nobjects++; if (lstat (name, &st)) return; if (st.st_dev != dev && !force) { if (dev) { fprintf(stderr, "%s is on different filesystem than the rest.\nUse -f option to override.\n", name); doexit(6); } dev = st.st_dev; } if (S_ISDIR (st.st_mode)) { d * dp = malloc(add3(sizeof(d), namelen, 1)); if (!dp) { fprintf(stderr, "\nOut of memory 3\n"); doexit(3); } memcpy(dp->name, name, namelen + 1); dp->next = dirs; dirs = dp; } else if (S_ISREG (st.st_mode)) { int fd, i; f * fp, * fp2; h * hp; const char *n1, *n2; int cksumsize = sizeof(buf); unsigned int cksum; time_t mtime = content_only ? 0 : st.st_mtime; unsigned int hsh = hash (st.st_size, mtime); off_t fsize; nregfiles++; if (verbose > 1) fprintf(stderr, " %s", name); fd = open (name, O_RDONLY); if (fd < 0) return; if (st.st_size < sizeof(buf)) { cksumsize = st.st_size; memset (((char *)buf) + cksumsize, 0, (sizeof(buf) - cksumsize) % sizeof(buf[0])); } if (read (fd, buf, cksumsize) != cksumsize) { close(fd); if (verbose > 1 && namelen <= NAMELEN) fprintf(stderr, "\r%*s\r", (int)(namelen + 2), ""); return; } cksumsize = (cksumsize + sizeof(buf[0]) - 1) / sizeof(buf[0]); for (i = 0, cksum = 0; i < cksumsize; i++) { if (cksum + buf[i] < cksum) cksum += buf[i] + 1; else cksum += buf[i]; } for (hp = hps[hsh]; hp; hp = hp->next) if (hp->size == st.st_size && hp->mtime == mtime) break; if (!hp) { hp = malloc(sizeof(h)); if (!hp) { fprintf(stderr, "\nOut of memory 1\n"); doexit(1); } hp->size = st.st_size; hp->mtime = mtime; hp->chain = NULL; hp->next = hps[hsh]; hps[hsh] = hp; } for (fp = hp->chain; fp; fp = fp->next) if (fp->cksum == cksum) break; for (fp2 = fp; fp2 && fp2->cksum == cksum; fp2 = fp2->next) if (fp2->ino == st.st_ino && fp2->dev == st.st_dev) { close(fd); if (verbose > 1 && namelen <= NAMELEN) fprintf(stderr, "\r%*s\r", (int)(namelen + 2), ""); return; } for (fp2 = fp; fp2 && fp2->cksum == cksum; fp2 = fp2->next) if (!lstat (fp2->name, &st2) && S_ISREG (st2.st_mode) && !stcmp (&st, &st2, content_only) && st2.st_ino != st.st_ino && st2.st_dev == st.st_dev) { int fd2 = open (fp2->name, O_RDONLY); if (fd2 < 0) continue; if (fstat (fd2, &st2) || !S_ISREG (st2.st_mode) || st2.st_size == 0) { close (fd2); continue; } ncomp++; lseek(fd, 0, SEEK_SET); for (fsize = st.st_size; fsize > 0; fsize -= NIOBUF) { off_t rsize = fsize >= NIOBUF ? NIOBUF : fsize; if (read (fd, iobuf1, rsize) != rsize || read (fd2, iobuf2, rsize) != rsize) { close(fd); close(fd2); fprintf(stderr, "\nReading error\n"); return; } if (memcmp (iobuf1, iobuf2, rsize)) break; } close(fd2); if (fsize > 0) continue; if (lstat (name, &st3)) { fprintf(stderr, "\nCould not stat %s again\n", name); close(fd); return; } st3.st_atime = st.st_atime; if (stcmp (&st, &st3, 0)) { fprintf(stderr, "\nFile %s changed underneath us\n", name); close(fd); return; } n1 = fp2->name; n2 = name; if (!no_link) { const char *suffix = ".$$$___cleanit___$$$"; const size_t suffixlen = strlen(suffix); size_t n2len = strlen(n2); dynstr nam2 = {NULL, 0}; growstr(&nam2, add2(n2len, suffixlen)); memcpy(nam2.buf, n2, n2len); memcpy(&nam2.buf[n2len], suffix, suffixlen + 1); if (rename (n2, nam2.buf)) { fprintf(stderr, "\nFailed to rename %s to %s\n", n2, nam2.buf); free(nam2.buf); continue; } if (link (n1, n2)) { fprintf(stderr, "\nFailed to hardlink %s to %s\n", n1, n2); if (rename (nam2.buf, n2)) { fprintf(stderr, "\nBad bad - failed to rename back %s to %s\n", nam2.buf, n2); } close(fd); free(nam2.buf); return; } unlink (nam2.buf); free(nam2.buf); } nlinks++; if (st3.st_nlink > 1) { /* We actually did not save anything this time, since the link second argument had some other links as well. */ if (verbose > 1) fprintf(stderr, "\r%*s\r%s %s to %s\n", (int)(((namelen > NAMELEN) ? 0 : namelen) + 2), "", (no_link ? "Would link" : "Linked"), n1, n2); } else { nsaved+=((st.st_size+4095)/4096)*4096; if (verbose > 1) fprintf(stderr, "\r%*s\r%s %s to %s, %s %ld\n", (int)(((namelen > NAMELEN) ? 0 : namelen) + 2), "", (no_link ? "Would link" : "Linked"), n1, n2, (no_link ? "would save" : "saved"), st.st_size); } close(fd); return; } fp2 = malloc(add3(sizeof(f), namelen, 1)); if (!fp2) { fprintf(stderr, "\nOut of memory 2\n"); doexit(2); } close(fd); fp2->ino = st.st_ino; fp2->dev = st.st_dev; fp2->cksum = cksum; memcpy(fp2->name, name, namelen + 1); if (fp) { fp2->next = fp->next; fp->next = fp2; } else { fp2->next = hp->chain; hp->chain = fp2; } if (verbose > 1 && namelen <= NAMELEN) fprintf(stderr, "\r%*s\r", (int)(namelen + 2), ""); return; } }
void rf (char *name) { struct stat st, st2, st3; nobjects++; if (lstat (name, &st)) return; if (S_ISDIR (st.st_mode)) { d * dp = malloc(sizeof(d) + 1 + strlen (name)); if (!dp) { fprintf(stderr, "\nOut of memory 3\n"); doexit(3); } strcpy (dp->name, name); dp->next = dirs; dirs = dp; } else if (S_ISREG (st.st_mode)) { int fd, i; f * fp, * fp2; h * hp; char *p = NULL, *q; char *n1, *n2; int cksumsize = sizeof(buf); unsigned int cksum; time_t mtime = content_only ? 0 : st.st_mtime; unsigned int hsh = hash (st.st_size, mtime); nregfiles++; if (verbose > 1) fprintf(stderr, " %s", name); fd = open (name, O_RDONLY); if (fd < 0) return; if (st.st_size < sizeof(buf)) { cksumsize = st.st_size; memset (((char *)buf) + cksumsize, 0, (sizeof(buf) - cksumsize) % sizeof(buf[0])); } if (read (fd, buf, cksumsize) != cksumsize) { close(fd); if (verbose > 1) fprintf(stderr, "\r%*s\r", (int)strlen(name)+2, ""); return; } cksumsize = (cksumsize + sizeof(buf[0]) - 1) / sizeof(buf[0]); for (i = 0, cksum = 0; i < cksumsize; i++) { if (cksum + buf[i] < cksum) cksum += buf[i] + 1; else cksum += buf[i]; } for (hp = hps[hsh]; hp; hp = hp->next) if (hp->size == st.st_size && hp->mtime == mtime) break; if (!hp) { hp = malloc(sizeof(h)); if (!hp) { fprintf(stderr, "\nOut of memory 1\n"); doexit(1); } hp->size = st.st_size; hp->mtime = mtime; hp->chain = NULL; hp->next = hps[hsh]; hps[hsh] = hp; } for (fp = hp->chain; fp; fp = fp->next) if (fp->cksum == cksum) break; for (fp2 = fp; fp2 && fp2->cksum == cksum; fp2 = fp2->next) if (fp2->ino == st.st_ino && fp2->dev == st.st_dev) { close(fd); if (verbose > 1) fprintf(stderr, "\r%*s\r", (int)strlen(name)+2, ""); return; } if (fp && st.st_size > 0) { p = mmap (NULL, st.st_size, PROT_READ, MAP_SHARED, fd, 0); nmmap++; if (p == (void *)-1) { close(fd); fprintf(stderr, "\nFailed to mmap %s\n", name); return; } } for (fp2 = fp; fp2 && fp2->cksum == cksum; fp2 = fp2->next) if (!lstat (fp2->name, &st2) && S_ISREG (st2.st_mode) && !stcmp (&st, &st2, content_only) && st2.st_ino != st.st_ino && st2.st_dev == st.st_dev) { int fd2 = open (fp2->name, O_RDONLY); if (fd2 < 0) continue; if (fstat (fd2, &st2) || !S_ISREG (st2.st_mode) || st2.st_size == 0) { close (fd2); continue; } ncomp++; q = mmap (NULL, st.st_size, PROT_READ, MAP_SHARED, fd2, 0); if (q == (void *)-1) { close(fd2); fprintf(stderr, "\nFailed to mmap %s\n", fp2->name); continue; } if (memcmp (p, q, st.st_size)) { munmap (q, st.st_size); close(fd2); continue; } munmap (q, st.st_size); close(fd2); if (lstat (name, &st3)) { fprintf(stderr, "\nCould not stat %s again\n", name); munmap (p, st.st_size); close(fd); return; } st3.st_atime = st.st_atime; if (stcmp (&st, &st3, 0)) { fprintf(stderr, "\nFile %s changed underneath us\n", name); munmap (p, st.st_size); close(fd); return; } n1 = fp2->name; n2 = name; if (!no_link) { strcpy (stpcpy (nambuf2, n2), ".$$$___cleanit___$$$"); if (rename (n2, nambuf2)) { fprintf(stderr, "\nFailed to rename %s to %s\n", n2, nambuf2); continue; } if (link (n1, n2)) { fprintf(stderr, "\nFailed to hardlink %s to %s\n", n1, n2); if (rename (nambuf2, n2)) { fprintf(stderr, "\nBad bad - failed to rename back %s to %s\n", nambuf2, n2); } munmap (p, st.st_size); close(fd); return; } unlink (nambuf2); } nlinks++; if (st3.st_nlink > 1) { /* We actually did not save anything this time, since the link second argument had some other links as well. */ if (verbose > 1) fprintf(stderr, "\r%*s\r%s %s to %s\n", (int)strlen(name)+2, "", (no_link ? "Would link" : "Linked"), n1, n2); } else { nsaved+=((st.st_size+4095)/4096)*4096; if (verbose > 1) fprintf(stderr, "\r%*s\r%s %s to %s, %s %ld\n", (int)strlen(name)+2, "", (no_link ? "Would link" : "Linked"), n1, n2, (no_link ? "would save" : "saved"), st.st_size); } munmap (p, st.st_size); close(fd); return; } if (fp) munmap (p, st.st_size); fp2 = malloc(sizeof(f) + 1 + strlen (name)); if (!fp2) { fprintf(stderr, "\nOut of memory 2\n"); doexit(2); } close(fd); fp2->ino = st.st_ino; fp2->dev = st.st_dev; fp2->cksum = cksum; strcpy(fp2->name, name); if (fp) { fp2->next = fp->next; fp->next = fp2; } else { fp2->next = hp->chain; hp->chain = fp2; } if (verbose > 1) fprintf(stderr, "\r%*s\r", (int)strlen(name)+2, ""); return; } }