2 Copyright (c) 2012 Frank Lahm <franklahm@gmail.com>
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
17 #endif /* HAVE_CONFIG_H */
27 #include <atalk/errchk.h>
28 #include <atalk/util.h>
29 #include <atalk/logger.h>
30 #include <atalk/talloc.h>
31 #include <atalk/dalloc.h>
32 #include <atalk/byteorder.h>
33 #include <atalk/netatalk_conf.h>
34 #include <atalk/volume.h>
36 #include "spotlight.h"
38 /**************************************************************************************************
39 * RPC data marshalling and unmarshalling
40 **************************************************************************************************/
42 /* Spotlight epoch is UNIX epoch minus SPOTLIGHT_TIME_DELTA */
43 #define SPOTLIGHT_TIME_DELTA INT64_C(280878921600U)
45 #define SQ_TYPE_NULL 0x0000
46 #define SQ_TYPE_COMPLEX 0x0200
47 #define SQ_TYPE_INT64 0x8400
48 #define SQ_TYPE_BOOL 0x0100
49 #define SQ_TYPE_FLOAT 0x8500
50 #define SQ_TYPE_DATA 0x0700
51 #define SQ_TYPE_CNIDS 0x8700
52 #define SQ_TYPE_UUID 0x0e00
53 #define SQ_TYPE_DATE 0x8600
54 #define SQ_TYPE_TOC 0x8800
56 #define SQ_CPX_TYPE_ARRAY 0x0a00
57 #define SQ_CPX_TYPE_STRING 0x0c00
58 #define SQ_CPX_TYPE_UTF16_STRING 0x1c00
59 #define SQ_CPX_TYPE_DICT 0x0d00
60 #define SQ_CPX_TYPE_CNIDS 0x1a00
61 #define SQ_CPX_TYPE_FILEMETA 0x1b00
63 #define SUBQ_SAFETY_LIM 20
65 /* Forward declarations */
66 static int sl_pack_loop(DALLOC_CTX *query, char *buf, int offset, char *toc_buf, int *toc_idx);
67 static int sl_unpack_loop(DALLOC_CTX *query, const char *buf, int offset, uint count, const uint toc_offset, const uint encoding);
70 * Returns the UTF-16 string encoding, by checking the 2-byte byte order mark.
71 * If there is no byte order mark, -1 is returned.
73 static uint spotlight_get_utf16_string_encoding(const char *buf, int offset, int query_length, uint encoding) {
76 /* check for byte order mark */
77 utf16_encoding = SL_ENC_BIG_ENDIAN;
78 if (query_length >= 2) {
79 uint16_t byte_order_mark;
80 if (encoding == SL_ENC_LITTLE_ENDIAN)
81 byte_order_mark = SVAL(buf, offset);
83 byte_order_mark = RSVAL(buf, offset);
85 if (byte_order_mark == 0xFFFE) {
86 utf16_encoding = SL_ENC_BIG_ENDIAN | SL_ENC_UTF_16;
88 else if (byte_order_mark == 0xFEFF) {
89 utf16_encoding = SL_ENC_LITTLE_ENDIAN | SL_ENC_UTF_16;
93 return utf16_encoding;
96 /**************************************************************************************************
97 * marshalling functions
98 **************************************************************************************************/
100 #define SL_OFFSET_DELTA 16
102 static uint64_t sl_pack_tag(uint16_t type, uint16_t size_or_count, uint32_t val)
104 uint64_t tag = ((uint64_t)val << 32) | ((uint64_t)type << 16) | size_or_count;
108 static int sl_pack_float(double d, char *buf, int offset)
115 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_FLOAT, 2, 1));
116 SLVAL(buf, offset + 8, ieee_fp_union.w);
118 return offset + 2 * sizeof(uint64_t);
121 static int sl_pack_uint64(uint64_t u, char *buf, int offset)
123 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_INT64, 2, 1));
124 SLVAL(buf, offset + 8, u);
126 return offset + 2 * sizeof(uint64_t);
129 static int sl_pack_bool(sl_bool_t bl, char *buf, int offset)
131 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_BOOL, 1, bl ? 1 : 0));
133 return offset + sizeof(uint64_t);
136 static int sl_pack_nil(char *buf, int offset)
138 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_NULL, 1, 1));
140 return offset + sizeof(uint64_t);
143 static int sl_pack_date(sl_time_t t, char *buf, int offset)
147 data = (t.tv_sec + SPOTLIGHT_TIME_DELTA) << 24;
149 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_DATE, 2, 1));
150 SLVAL(buf, offset + 8, data);
152 return offset + 2 * sizeof(uint64_t);
155 static int sl_pack_uuid(sl_uuid_t *uuid, char *buf, int offset)
157 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_UUID, 3, 1));
158 memcpy(buf + offset + 8, uuid, 16);
160 return offset + sizeof(uint64_t) + 16;
163 static int sl_pack_CNID(sl_cnids_t *cnids, char *buf, int offset, char *toc_buf, int *toc_idx)
165 int len = 0, off = 0;
166 int cnid_count = talloc_array_length(cnids->ca_cnids);
168 SLVAL(toc_buf, *toc_idx * 8, sl_pack_tag(SQ_CPX_TYPE_CNIDS, (offset + SL_OFFSET_DELTA) / 8, cnid_count));
169 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_COMPLEX, 1, *toc_idx + 1));
173 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_CNIDS, 2 + cnid_count, 8 /* unknown meaning, but always 8 */));
176 if (cnid_count > 0) {
177 SLVAL(buf, offset, sl_pack_tag(0x0add, cnid_count, cnids->ca_context));
180 for (int i = 0; i < cnid_count; i++) {
181 SLVAL(buf, offset, cnids->ca_cnids->dd_talloc_array[i]);
189 static int sl_pack_array(sl_array_t *array, char *buf, int offset, char *toc_buf, int *toc_idx)
191 int count = talloc_array_length(array->dd_talloc_array);
192 int octets = (offset + SL_OFFSET_DELTA) / 8;
194 LOG(log_maxdebug, logtype_sl, "sl_pack_array: count: %d, offset:%d, octets: %d", count, offset, octets);
196 SLVAL(toc_buf, *toc_idx * 8, sl_pack_tag(SQ_CPX_TYPE_ARRAY, octets, count));
197 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_COMPLEX, 1, *toc_idx + 1));
201 offset = sl_pack_loop(array, buf, offset, toc_buf, toc_idx);
206 static int sl_pack_dict(sl_array_t *dict, char *buf, int offset, char *toc_buf, int *toc_idx)
208 SLVAL(toc_buf, *toc_idx * 8, sl_pack_tag(SQ_CPX_TYPE_DICT, (offset + SL_OFFSET_DELTA) / 8, talloc_array_length(dict->dd_talloc_array)));
209 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_COMPLEX, 1, *toc_idx + 1));
213 offset = sl_pack_loop(dict, buf, offset, toc_buf, toc_idx);
218 static int sl_pack_string(char **string, char *buf, int offset, char *toc_buf, int *toc_idx)
220 int len, octets, used_in_last_octet;
223 octets = (len / 8) + (len & 7 ? 1 : 0);
224 used_in_last_octet = 8 - (octets * 8 - len);
226 LOG(log_maxdebug, logtype_sl, "sl_pack_string(\"%s\"): len: %d, octets: %d, used_in_last_octet: %d",
227 s, len, octets, used_in_last_octet);
229 SLVAL(toc_buf, *toc_idx * 8, sl_pack_tag(SQ_CPX_TYPE_STRING, (offset + SL_OFFSET_DELTA) / 8, used_in_last_octet));
230 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_COMPLEX, 1, *toc_idx + 1));
234 SLVAL(buf, offset, sl_pack_tag(SQ_TYPE_DATA, octets + 1, used_in_last_octet));
237 memset(buf + offset, 0, octets * 8);
238 strncpy(buf + offset, s, len);
239 offset += octets * 8;
244 static int sl_pack_loop(DALLOC_CTX *query, char *buf, int offset, char *toc_buf, int *toc_idx)
248 for (int n = 0; n < talloc_array_length(query->dd_talloc_array); n++) {
250 type = talloc_get_name(query->dd_talloc_array[n]);
252 if (STRCMP(type, ==, "sl_array_t")) {
253 offset = sl_pack_array(query->dd_talloc_array[n], buf, offset, toc_buf, toc_idx);
254 } else if (STRCMP(type, ==, "sl_dict_t")) {
255 offset = sl_pack_dict(query->dd_talloc_array[n], buf, offset, toc_buf, toc_idx);
256 } else if (STRCMP(type, ==, "uint64_t")) {
258 memcpy(&i, query->dd_talloc_array[n], sizeof(uint64_t));
259 offset = sl_pack_uint64(i, buf, offset);
260 } else if (STRCMP(type, ==, "char *")) {
261 offset = sl_pack_string(query->dd_talloc_array[n], buf, offset, toc_buf, toc_idx);
262 } else if (STRCMP(type, ==, "sl_bool_t")) {
264 memcpy(&bl, query->dd_talloc_array[n], sizeof(sl_bool_t));
265 offset = sl_pack_bool(bl, buf, offset);
266 } else if (STRCMP(type, ==, "double")) {
268 memcpy(&d, query->dd_talloc_array[n], sizeof(double));
269 offset = sl_pack_float(d, buf, offset);
270 } else if (STRCMP(type, ==, "sl_nil_t")) {
271 offset = sl_pack_nil(buf, offset);
272 } else if (STRCMP(type, ==, "sl_time_t")) {
274 memcpy(&t, query->dd_talloc_array[n], sizeof(sl_time_t));
275 offset = sl_pack_date(t, buf, offset);
276 } else if (STRCMP(type, ==, "sl_uuid_t")) {
277 offset = sl_pack_uuid(query->dd_talloc_array[n], buf, offset);
278 } else if (STRCMP(type, ==, "sl_cnids_t")) {
279 offset = sl_pack_CNID(query->dd_talloc_array[n], buf, offset, toc_buf, toc_idx);
286 /**************************************************************************************************
287 * unmarshalling functions
288 **************************************************************************************************/
290 static uint64_t sl_unpack_uint64(const char *buf, int offset, uint encoding)
292 if (encoding == SL_ENC_LITTLE_ENDIAN)
293 return LVAL(buf, offset);
295 return RLVAL(buf, offset);
298 static int sl_unpack_ints(DALLOC_CTX *query, const char *buf, int offset, uint encoding)
301 uint64_t query_data64;
303 query_data64 = sl_unpack_uint64(buf, offset, encoding);
304 count = query_data64 >> 32;
308 while (i++ < count) {
309 query_data64 = sl_unpack_uint64(buf, offset, encoding);
310 dalloc_add(query, &query_data64, uint64_t);
317 static int sl_unpack_date(DALLOC_CTX *query, const char *buf, int offset, uint encoding)
320 uint64_t query_data64;
323 query_data64 = sl_unpack_uint64(buf, offset, encoding);
324 count = query_data64 >> 32;
328 while (i++ < count) {
329 query_data64 = sl_unpack_uint64(buf, offset, encoding) >> 24;
330 t.tv_sec = query_data64 - SPOTLIGHT_TIME_DELTA;
332 dalloc_add(query, &t, sl_time_t);
339 static int sl_unpack_uuid(DALLOC_CTX *query, const char *buf, int offset, uint encoding)
342 uint64_t query_data64;
344 query_data64 = sl_unpack_uint64(buf, offset, encoding);
345 count = query_data64 >> 32;
349 while (i++ < count) {
350 memcpy(uuid.sl_uuid, buf + offset, 16);
351 dalloc_add(query, &uuid, sl_uuid_t);
358 static int sl_unpack_floats(DALLOC_CTX *query, const char *buf, int offset, uint encoding)
361 uint64_t query_data64;
368 query_data64 = sl_unpack_uint64(buf, offset, encoding);
369 count = query_data64 >> 32;
373 while (i++ < count) {
374 if (encoding == SL_ENC_LITTLE_ENDIAN) {
375 #ifdef WORDS_BIGENDIAN
376 ieee_fp_union.w[0] = IVAL(buf, offset + 4);
377 ieee_fp_union.w[1] = IVAL(buf, offset);
379 ieee_fp_union.w[0] = IVAL(buf, offset);
380 ieee_fp_union.w[1] = IVAL(buf, offset + 4);
383 #ifdef WORDS_BIGENDIAN
384 ieee_fp_union.w[0] = RIVAL(buf, offset);
385 ieee_fp_union.w[1] = RIVAL(buf, offset + 4);
387 ieee_fp_union.w[0] = RIVAL(buf, offset + 4);
388 ieee_fp_union.w[1] = RIVAL(buf, offset);
391 dalloc_add(query, &ieee_fp_union.d, double);
398 static int sl_unpack_CNID(DALLOC_CTX *query, const char *buf, int offset, int length, uint encoding)
402 uint64_t query_data64;
405 EC_NULL( cnids.ca_cnids = talloc_zero(query, DALLOC_CTX) );
408 /* that's permitted, it's an empty array */
411 query_data64 = sl_unpack_uint64(buf, offset, encoding);
412 count = query_data64 & 0xffff;
414 cnids.ca_unkn1 = (query_data64 & 0xffff0000) >> 16;
415 cnids.ca_context = query_data64 >> 32;
420 query_data64 = sl_unpack_uint64(buf, offset, encoding);
421 dalloc_add(cnids.ca_cnids, &query_data64, uint64_t);
425 dalloc_add(query, &cnids, sl_cnids_t);
431 static const char *spotlight_get_qtype_string(uint64_t query_type)
433 switch (query_type) {
436 case SQ_TYPE_COMPLEX:
453 static const char *spotlight_get_cpx_qtype_string(uint64_t cpx_query_type)
455 switch (cpx_query_type) {
456 case SQ_CPX_TYPE_ARRAY:
458 case SQ_CPX_TYPE_STRING:
460 case SQ_CPX_TYPE_UTF16_STRING:
461 return "utf-16 string";
462 case SQ_CPX_TYPE_DICT:
464 case SQ_CPX_TYPE_CNIDS:
466 case SQ_CPX_TYPE_FILEMETA:
473 static int sl_unpack_cpx(DALLOC_CTX *query,
477 uint cpx_query_count,
478 const uint toc_offset,
483 int roffset = offset;
484 uint64_t query_data64;
485 uint unicode_encoding;
488 int qlen, used_in_last_block, slen;
489 sl_array_t *sl_arrary;
492 switch (cpx_query_type) {
493 case SQ_CPX_TYPE_ARRAY:
494 sl_arrary = talloc_zero(query, sl_array_t);
495 EC_NEG1_LOG( roffset = sl_unpack_loop(sl_arrary, buf, offset, cpx_query_count, toc_offset, encoding) );
496 dalloc_add(query, sl_arrary, sl_array_t);
499 case SQ_CPX_TYPE_DICT:
500 sl_dict = talloc_zero(query, sl_dict_t);
501 EC_NEG1_LOG( roffset = sl_unpack_loop(sl_dict, buf, offset, cpx_query_count, toc_offset, encoding) );
502 dalloc_add(query, sl_dict, sl_dict_t);
505 case SQ_CPX_TYPE_STRING:
506 case SQ_CPX_TYPE_UTF16_STRING:
507 query_data64 = sl_unpack_uint64(buf, offset, encoding);
508 qlen = (query_data64 & 0xffff) * 8;
509 used_in_last_block = query_data64 >> 32;
510 slen = qlen - 8 + used_in_last_block;
512 if (cpx_query_type == SQ_CPX_TYPE_STRING) {
513 p = talloc_strndup(query, buf + offset + 8, slen);
515 unicode_encoding = spotlight_get_utf16_string_encoding(buf, offset + 8, slen, encoding);
516 mark_exists = (unicode_encoding & SL_ENC_UTF_16);
517 unicode_encoding &= ~SL_ENC_UTF_16;
518 EC_NEG1( convert_string_allocate(CH_UCS2, CH_UTF8, buf + offset + (mark_exists ? 18 : 16), slen, &p) );
521 dalloc_add(query, &p, char *);
525 case SQ_CPX_TYPE_FILEMETA:
526 query_data64 = sl_unpack_uint64(buf, offset, encoding);
527 qlen = (query_data64 & 0xffff) * 8;
529 EC_FAIL_LOG("SQ_CPX_TYPE_FILEMETA: query_length <= 8: %d", qlen);
531 EC_NEG1_LOG( sl_unpack(query, buf + offset + 8) );
536 case SQ_CPX_TYPE_CNIDS:
537 query_data64 = sl_unpack_uint64(buf, offset, encoding);
538 qlen = (query_data64 & 0xffff) * 8;
539 EC_NEG1_LOG( sl_unpack_CNID(query, buf, offset + 8, qlen, encoding) );
553 static int sl_unpack_loop(DALLOC_CTX *query,
557 const uint toc_offset,
561 int i, toc_index, query_length;
563 uint64_t query_data64, query_type;
564 uint cpx_query_type, cpx_query_count;
568 while (count > 0 && (offset < toc_offset)) {
569 query_data64 = sl_unpack_uint64(buf, offset, encoding);
570 query_length = (query_data64 & 0xffff) * 8;
571 query_type = (query_data64 & 0xffff0000) >> 16;
572 if (query_length == 0)
575 switch (query_type) {
576 case SQ_TYPE_COMPLEX:
577 toc_index = (query_data64 >> 32) - 1;
578 query_data64 = sl_unpack_uint64(buf, toc_offset + toc_index * 8, encoding);
579 cpx_query_type = (query_data64 & 0xffff0000) >> 16;
580 cpx_query_count = query_data64 >> 32;
582 EC_NEG1_LOG( offset = sl_unpack_cpx(query, buf, offset + 8, cpx_query_type, cpx_query_count, toc_offset, encoding));
586 subcount = query_data64 >> 32;
590 for (i = 0; i < subcount; i++)
591 dalloc_add(query, &nil, sl_nil_t);
592 offset += query_length;
596 b = query_data64 >> 32;
597 dalloc_add(query, &b, sl_bool_t);
598 offset += query_length;
602 EC_NEG1_LOG( subcount = sl_unpack_ints(query, buf, offset, encoding) );
603 offset += query_length;
607 EC_NEG1_LOG( subcount = sl_unpack_uuid(query, buf, offset, encoding) );
608 offset += query_length;
612 EC_NEG1_LOG( subcount = sl_unpack_floats(query, buf, offset, encoding) );
613 offset += query_length;
617 EC_NEG1_LOG( subcount = sl_unpack_date(query, buf, offset, encoding) );
618 offset += query_length;
633 /**************************************************************************************************
634 * Global functions for packing und unpacking
635 **************************************************************************************************/
637 #define MAX_SLQ_DAT 65000
638 #define MAX_SLQ_TOC 2048
640 int sl_pack(DALLOC_CTX *query, char *buf)
643 char toc_buf[MAX_SLQ_TOC];
647 memcpy(buf, "432130dm", 8);
648 EC_NEG1_LOG( len = sl_pack_loop(query, buf + 16, 0, toc_buf + 8, &toc_index) );
649 SIVAL(buf, 8, len / 8 + 1 + toc_index + 1);
650 SIVAL(buf, 12, len / 8 + 1);
652 SLVAL(toc_buf, 0, sl_pack_tag(SQ_TYPE_TOC, toc_index + 1, 0));
653 memcpy(buf + 16 + len, toc_buf, (toc_index + 1 ) * 8);
655 len += 16 + (toc_index + 1 ) * 8;
663 int sl_unpack(DALLOC_CTX *query, const char *buf)
666 int encoding, i, toc_entries;
667 uint64_t toc_offset, tquerylen, toc_entry;
669 if (strncmp(buf, "md031234", 8) == 0)
670 encoding = SL_ENC_BIG_ENDIAN;
672 encoding = SL_ENC_LITTLE_ENDIAN;
676 toc_offset = ((sl_unpack_uint64(buf, 0, encoding) >> 32) - 1 ) * 8;
677 if (toc_offset < 0 || (toc_offset > 65000)) {
683 toc_entries = (int)(sl_unpack_uint64(buf, toc_offset, encoding) & 0xffff);
685 EC_NEG1( sl_unpack_loop(query, buf, 0, 1, toc_offset + 8, encoding) );