15 // There's no 'ps' on win32 anyway, and Py_GetArgcArgv() isn't available.
16 static void unpythonize_argv(void) { }
18 #else // not __WIN32__
20 // For some reason this isn't declared in Python.h
21 extern void Py_GetArgcArgv(int *argc, char ***argv);
23 static void unpythonize_argv(void)
28 Py_GetArgcArgv(&argc, &argv);
30 for (i = 0; i < argc-1; i++)
32 if (argv[i] + strlen(argv[i]) + 1 != argv[i+1])
34 // The argv block doesn't work the way we expected; it's unsafe
40 arge = argv[argc-1] + strlen(argv[argc-1]) + 1;
42 if (strstr(argv[0], "python") && argv[1] == argv[0] + strlen(argv[0]) + 1)
46 p = strrchr(argv[1], '/');
52 memmove(argv[0], p, len);
53 memset(arge - diff, 0, diff);
54 for (i = 0; i < argc; i++)
55 argv[i] = argv[i+1] ? argv[i+1]-diff : NULL;
60 #endif // not __WIN32__
63 static PyObject *selftest(PyObject *self, PyObject *args)
65 if (!PyArg_ParseTuple(args, ""))
68 return Py_BuildValue("i", !bupsplit_selftest());
72 static PyObject *blobbits(PyObject *self, PyObject *args)
74 if (!PyArg_ParseTuple(args, ""))
76 return Py_BuildValue("i", BUP_BLOBBITS);
80 static PyObject *splitbuf(PyObject *self, PyObject *args)
82 unsigned char *buf = NULL;
83 int len = 0, out = 0, bits = -1;
85 if (!PyArg_ParseTuple(args, "t#", &buf, &len))
87 out = bupsplit_find_ofs(buf, len, &bits);
88 return Py_BuildValue("ii", out, bits);
92 static PyObject *bitmatch(PyObject *self, PyObject *args)
94 unsigned char *buf1 = NULL, *buf2 = NULL;
95 int len1 = 0, len2 = 0;
98 if (!PyArg_ParseTuple(args, "t#t#", &buf1, &len1, &buf2, &len2))
102 for (byte = 0; byte < len1 && byte < len2; byte++)
104 int b1 = buf1[byte], b2 = buf2[byte];
107 for (bit = 0; bit < 8; bit++)
108 if ( (b1 & (0x80 >> bit)) != (b2 & (0x80 >> bit)) )
114 return Py_BuildValue("i", byte*8 + bit);
118 static PyObject *firstword(PyObject *self, PyObject *args)
120 unsigned char *buf = NULL;
124 if (!PyArg_ParseTuple(args, "t#", &buf, &len))
130 v = ntohl(*(uint32_t *)buf);
131 return PyLong_FromUnsignedLong(v);
141 static void to_bloom_address_bitmask4(const bits40_t *buf,
142 const int nbits, uint64_t *v, unsigned char *bitmask)
147 mask = (1<<nbits) - 1;
148 raw = (((uint64_t)ntohl(buf->high)) << 8) | buf->low;
149 bit = (raw >> (37-nbits)) & 0x7;
150 *v = (raw >> (40-nbits)) & mask;
154 static void to_bloom_address_bitmask5(const uint32_t *buf,
155 const int nbits, uint32_t *v, unsigned char *bitmask)
160 mask = (1<<nbits) - 1;
162 bit = (raw >> (29-nbits)) & 0x7;
163 *v = (raw >> (32-nbits)) & mask;
168 #define BLOOM_SET_BIT(name, address, itype, otype) \
169 static void name(unsigned char *bloom, const void *buf, const int nbits)\
171 unsigned char bitmask;\
173 address((itype *)buf, nbits, &v, &bitmask);\
174 bloom[16+v] |= bitmask;\
176 BLOOM_SET_BIT(bloom_set_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
177 BLOOM_SET_BIT(bloom_set_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
180 #define BLOOM_GET_BIT(name, address, itype, otype) \
181 static int name(const unsigned char *bloom, const void *buf, const int nbits)\
183 unsigned char bitmask;\
185 address((itype *)buf, nbits, &v, &bitmask);\
186 return bloom[16+v] & bitmask;\
188 BLOOM_GET_BIT(bloom_get_bit4, to_bloom_address_bitmask4, bits40_t, uint64_t)
189 BLOOM_GET_BIT(bloom_get_bit5, to_bloom_address_bitmask5, uint32_t, uint32_t)
192 static PyObject *bloom_add(PyObject *self, PyObject *args)
194 unsigned char *sha = NULL, *bloom = NULL;
196 int len = 0, blen = 0, nbits = 0, k = 0;
198 if (!PyArg_ParseTuple(args, "w#s#ii", &bloom, &blen, &sha, &len, &nbits, &k))
201 if (blen < 16+(1<<nbits) || len % 20 != 0)
208 for (end = sha + len; sha < end; sha += 20/k)
209 bloom_set_bit5(bloom, sha, nbits);
215 for (end = sha + len; sha < end; sha += 20/k)
216 bloom_set_bit4(bloom, sha, nbits);
222 return Py_BuildValue("i", len/20);
225 static PyObject *bloom_contains(PyObject *self, PyObject *args)
227 unsigned char *sha = NULL, *bloom = NULL;
228 int len = 0, blen = 0, nbits = 0, k = 0;
232 if (!PyArg_ParseTuple(args, "t#s#ii", &bloom, &blen, &sha, &len, &nbits, &k))
242 for (steps = 1, end = sha + 20; sha < end; sha += 20/k, steps++)
243 if (!bloom_get_bit5(bloom, sha, nbits))
244 return Py_BuildValue("Oi", Py_None, steps);
250 for (steps = 1, end = sha + 20; sha < end; sha += 20/k, steps++)
251 if (!bloom_get_bit4(bloom, sha, nbits))
252 return Py_BuildValue("Oi", Py_None, steps);
257 return Py_BuildValue("Oi", Py_True, k);
261 static uint32_t _extract_bits(unsigned char *buf, int nbits)
265 mask = (1<<nbits) - 1;
266 v = ntohl(*(uint32_t *)buf);
267 v = (v >> (32-nbits)) & mask;
270 static PyObject *extract_bits(PyObject *self, PyObject *args)
272 unsigned char *buf = NULL;
273 int len = 0, nbits = 0;
275 if (!PyArg_ParseTuple(args, "t#i", &buf, &len, &nbits))
281 return PyLong_FromUnsignedLong(_extract_bits(buf, nbits));
286 unsigned char bytes[20];
298 static int _cmp_sha(const struct sha *sha1, const struct sha *sha2)
301 for (i = 0; i < 20; i++)
302 if (sha1->bytes[i] != sha2->bytes[i])
303 return sha1->bytes[i] - sha2->bytes[i];
308 static void _fix_idx_order(struct idx **idxs, int *last_i)
311 int low, mid, high, c = 0;
314 if (idxs[*last_i]->cur >= idxs[*last_i]->end)
316 idxs[*last_i] = NULL;
329 mid = (low + high) / 2;
330 c = _cmp_sha(idx->cur, idxs[mid]->cur);
342 memmove(&idxs[mid+1], &idxs[mid], (*last_i-mid)*sizeof(struct idx *));
347 static uint32_t _get_idx_i(struct idx *idx)
349 if (idx->cur_name == NULL)
350 return idx->name_base;
351 return ntohl(*idx->cur_name) + idx->name_base;
355 static PyObject *merge_into(PyObject *self, PyObject *args)
357 PyObject *ilist = NULL;
358 unsigned char *fmap = NULL;
359 struct sha *sha_ptr, *last = NULL;
360 uint32_t *table_ptr, *name_ptr;
361 struct idx **idxs = NULL;
362 int flen = 0, bits = 0, i;
363 uint32_t total, count, prefix;
367 if (!PyArg_ParseTuple(args, "w#iIO", &fmap, &flen, &bits, &total, &ilist))
370 num_i = PyList_Size(ilist);
371 idxs = (struct idx **)PyMem_Malloc(num_i * sizeof(struct idx *));
373 for (i = 0; i < num_i; i++)
375 long len, sha_ofs, name_map_ofs;
376 idxs[i] = (struct idx *)PyMem_Malloc(sizeof(struct idx));
377 PyObject *itup = PyList_GetItem(ilist, i);
378 if (!PyArg_ParseTuple(itup, "t#llli", &idxs[i]->map, &idxs[i]->bytes,
379 &len, &sha_ofs, &name_map_ofs, &idxs[i]->name_base))
381 idxs[i]->cur = (struct sha *)&idxs[i]->map[sha_ofs];
382 idxs[i]->end = &idxs[i]->cur[len];
384 idxs[i]->cur_name = (uint32_t *)&idxs[i]->map[name_map_ofs];
386 idxs[i]->cur_name = NULL;
388 table_ptr = (uint32_t *)&fmap[12];
389 sha_ptr = (struct sha *)&table_ptr[1<<bits];
390 name_ptr = (uint32_t *)&sha_ptr[total];
399 if (count % 102424 == 0 && istty)
400 fprintf(stderr, "midx: writing %.2f%% (%d/%d)\r",
401 count*100.0/total, count, total);
403 new_prefix = _extract_bits((unsigned char *)idx->cur, bits);
404 while (prefix < new_prefix)
405 table_ptr[prefix++] = htonl(count);
406 if (last == NULL || _cmp_sha(last, idx->cur) != 0)
408 memcpy(sha_ptr++, idx->cur, 20);
409 *name_ptr++ = htonl(_get_idx_i(idx));
413 if (idx->cur_name != NULL)
415 _fix_idx_order(idxs, &last_i);
418 table_ptr[prefix] = htonl(count);
421 return PyLong_FromUnsignedLong(count);
424 // This function should techniclly be macro'd out if it's going to be used
425 // more than ocasionally. As of this writing, it'll actually never be called
426 // in real world bup scenarios (because our packs are < MAX_INT bytes).
427 static uint64_t htonll(uint64_t value)
429 static const int endian_test = 42;
431 if (*(char *)&endian_test == endian_test) // LSB-MSB
432 return ((uint64_t)htonl(value & 0xFFFFFFFF) << 32) | htonl(value >> 32);
433 return value; // already in network byte order MSB-LSB
436 static PyObject *write_idx(PyObject *self, PyObject *args)
438 PyObject *pf = NULL, *idx = NULL;
441 unsigned char *fmap = NULL;
445 int i, j, ofs64_count;
446 uint32_t *fan_ptr, *crc_ptr, *ofs_ptr;
449 if (!PyArg_ParseTuple(args, "Ow#OI", &pf, &fmap, &flen, &idx, &total))
452 fan_ptr = (uint32_t *)&fmap[8];
453 sha_ptr = (struct sha *)&fan_ptr[256];
454 crc_ptr = (uint32_t *)&sha_ptr[total];
455 ofs_ptr = (uint32_t *)&crc_ptr[total];
456 f = PyFile_AsFile(pf);
460 for (i = 0; i < 256; ++i)
463 part = PyList_GET_ITEM(idx, i);
465 plen = PyList_GET_SIZE(part);
467 *fan_ptr++ = htonl(count);
468 for (j = 0; j < plen; ++j)
470 unsigned char *sha = NULL;
474 if (!PyArg_ParseTuple(PyList_GET_ITEM(part, j), "t#IK",
475 &sha, &sha_len, &crc, &ofs))
479 memcpy(sha_ptr++, sha, 20);
480 *crc_ptr++ = htonl(crc);
481 if (ofs > 0x7fffffff)
483 uint64_t nofs = htonll(ofs);
484 if (fwrite(&nofs, 8, 1, f) != 1)
485 return PyErr_SetFromErrno(PyExc_OSError);
486 ofs = 0x80000000 | ofs64_count++;
488 *ofs_ptr++ = htonl((uint32_t)ofs);
491 return PyLong_FromUnsignedLong(count);
495 // I would have made this a lower-level function that just fills in a buffer
496 // with random values, and then written those values from python. But that's
497 // about 20% slower in my tests, and since we typically generate random
498 // numbers for benchmarking other parts of bup, any slowness in generating
499 // random bytes will make our benchmarks inaccurate. Plus nobody wants
500 // pseudorandom bytes much except for this anyway.
501 static PyObject *write_random(PyObject *self, PyObject *args)
503 uint32_t buf[1024/4];
504 int fd = -1, seed = 0, verbose = 0;
506 long long len = 0, kbytes = 0, written = 0;
508 if (!PyArg_ParseTuple(args, "iLii", &fd, &len, &seed, &verbose))
513 for (kbytes = 0; kbytes < len/1024; kbytes++)
516 for (i = 0; i < sizeof(buf)/sizeof(buf[0]); i++)
518 ret = write(fd, buf, sizeof(buf));
522 if (ret < (int)sizeof(buf))
524 if (verbose && kbytes/1024 > 0 && !(kbytes%1024))
525 fprintf(stderr, "Random: %lld Mbytes\r", kbytes/1024);
528 // handle non-multiples of 1024
532 for (i = 0; i < sizeof(buf)/sizeof(buf[0]); i++)
534 ret = write(fd, buf, len % 1024);
541 fprintf(stderr, "Random: %lld Mbytes, done.\n", kbytes/1024);
542 return Py_BuildValue("L", written);
546 static PyObject *random_sha(PyObject *self, PyObject *args)
548 static int seeded = 0;
549 uint32_t shabuf[20/4];
554 assert(sizeof(shabuf) == 20);
559 if (!PyArg_ParseTuple(args, ""))
562 memset(shabuf, 0, sizeof(shabuf));
563 for (i=0; i < 20/4; i++)
564 shabuf[i] = random();
565 return Py_BuildValue("s#", shabuf, 20);
569 static PyObject *open_noatime(PyObject *self, PyObject *args)
571 char *filename = NULL;
572 int attrs, attrs_noatime, fd;
573 if (!PyArg_ParseTuple(args, "s", &filename))
580 attrs |= O_LARGEFILE;
582 attrs_noatime = attrs;
584 attrs_noatime |= O_NOATIME;
586 fd = open(filename, attrs_noatime);
587 if (fd < 0 && errno == EPERM)
589 // older Linux kernels would return EPERM if you used O_NOATIME
590 // and weren't the file's owner. This pointless restriction was
591 // relaxed eventually, but we have to handle it anyway.
592 // (VERY old kernels didn't recognized O_NOATIME, but they would
593 // just harmlessly ignore it, so this branch won't trigger)
594 fd = open(filename, attrs);
597 return PyErr_SetFromErrnoWithFilename(PyExc_IOError, filename);
598 return Py_BuildValue("i", fd);
602 static PyObject *fadvise_done(PyObject *self, PyObject *args)
606 if (!PyArg_ParseTuple(args, "iL", &fd, &ofs))
608 #ifdef POSIX_FADV_DONTNEED
609 posix_fadvise(fd, 0, ofs, POSIX_FADV_DONTNEED);
611 return Py_BuildValue("");
615 static PyMethodDef faster_methods[] = {
616 { "selftest", selftest, METH_VARARGS,
617 "Check that the rolling checksum rolls correctly (for unit tests)." },
618 { "blobbits", blobbits, METH_VARARGS,
619 "Return the number of bits in the rolling checksum." },
620 { "splitbuf", splitbuf, METH_VARARGS,
621 "Split a list of strings based on a rolling checksum." },
622 { "bitmatch", bitmatch, METH_VARARGS,
623 "Count the number of matching prefix bits between two strings." },
624 { "firstword", firstword, METH_VARARGS,
625 "Return an int corresponding to the first 32 bits of buf." },
626 { "bloom_contains", bloom_contains, METH_VARARGS,
627 "Check if a bloom filter of 2^nbits bytes contains an object" },
628 { "bloom_add", bloom_add, METH_VARARGS,
629 "Add an object to a bloom filter of 2^nbits bytes" },
630 { "extract_bits", extract_bits, METH_VARARGS,
631 "Take the first 'nbits' bits from 'buf' and return them as an int." },
632 { "merge_into", merge_into, METH_VARARGS,
633 "Merges a bunch of idx and midx files into a single midx." },
634 { "write_idx", write_idx, METH_VARARGS,
635 "Write a PackIdxV2 file from an idx list of lists of tuples" },
636 { "write_random", write_random, METH_VARARGS,
637 "Write random bytes to the given file descriptor" },
638 { "random_sha", random_sha, METH_VARARGS,
639 "Return a random 20-byte string" },
640 { "open_noatime", open_noatime, METH_VARARGS,
641 "open() the given filename for read with O_NOATIME if possible" },
642 { "fadvise_done", fadvise_done, METH_VARARGS,
643 "Inform the kernel that we're finished with earlier parts of a file" },
644 { NULL, NULL, 0, NULL }, // sentinel
647 PyMODINIT_FUNC init_helpers(void)
649 Py_InitModule("_helpers", faster_methods);
650 istty = isatty(2) || getenv("BUP_FORCE_TTY");