isomorfeus-ferret 0.17.1 → 0.17.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/ext/isomorfeus_ferret_ext/benchmark.c +9 -20
- data/ext/isomorfeus_ferret_ext/benchmarks_all.h +1 -2
- data/ext/isomorfeus_ferret_ext/bm_hash.c +1 -2
- data/ext/isomorfeus_ferret_ext/bm_store.c +2 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_decode.c +4 -2
- data/ext/isomorfeus_ferret_ext/brotli_enc_encode.c +3 -2
- data/ext/isomorfeus_ferret_ext/frb_analysis.c +4 -5
- data/ext/isomorfeus_ferret_ext/frb_field_info.c +3 -4
- data/ext/isomorfeus_ferret_ext/frb_index.c +118 -160
- data/ext/isomorfeus_ferret_ext/frb_lazy_doc.c +14 -16
- data/ext/isomorfeus_ferret_ext/frb_search.c +31 -23
- data/ext/isomorfeus_ferret_ext/frb_store.c +27 -13
- data/ext/isomorfeus_ferret_ext/frb_utils.c +3 -6
- data/ext/isomorfeus_ferret_ext/frt_analysis.c +39 -46
- data/ext/isomorfeus_ferret_ext/frt_analysis.h +9 -9
- data/ext/isomorfeus_ferret_ext/frt_array.c +11 -22
- data/ext/isomorfeus_ferret_ext/frt_bitvector.h +3 -6
- data/ext/isomorfeus_ferret_ext/frt_doc_field.c +87 -0
- data/ext/isomorfeus_ferret_ext/frt_doc_field.h +26 -0
- data/ext/isomorfeus_ferret_ext/frt_document.c +4 -97
- data/ext/isomorfeus_ferret_ext/frt_document.h +2 -27
- data/ext/isomorfeus_ferret_ext/frt_except.c +50 -6
- data/ext/isomorfeus_ferret_ext/frt_except.h +3 -2
- data/ext/isomorfeus_ferret_ext/frt_field_index.c +13 -32
- data/ext/isomorfeus_ferret_ext/frt_field_index.h +0 -6
- data/ext/isomorfeus_ferret_ext/frt_field_info.c +69 -0
- data/ext/isomorfeus_ferret_ext/frt_field_info.h +49 -0
- data/ext/isomorfeus_ferret_ext/frt_field_infos.c +196 -0
- data/ext/isomorfeus_ferret_ext/frt_field_infos.h +35 -0
- data/ext/isomorfeus_ferret_ext/frt_global.c +10 -4
- data/ext/isomorfeus_ferret_ext/frt_global.h +11 -15
- data/ext/isomorfeus_ferret_ext/frt_hash.c +8 -8
- data/ext/isomorfeus_ferret_ext/frt_hash.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_hashset.c +20 -40
- data/ext/isomorfeus_ferret_ext/frt_hashset.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_helper.c +7 -15
- data/ext/isomorfeus_ferret_ext/frt_in_stream.c +482 -0
- data/ext/isomorfeus_ferret_ext/frt_in_stream.h +241 -0
- data/ext/isomorfeus_ferret_ext/frt_ind.c +20 -49
- data/ext/isomorfeus_ferret_ext/frt_ind.h +0 -1
- data/ext/isomorfeus_ferret_ext/frt_index.c +296 -1857
- data/ext/isomorfeus_ferret_ext/frt_index.h +2 -145
- data/ext/isomorfeus_ferret_ext/frt_lang.c +5 -10
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc.c +29 -0
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc.h +19 -0
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc_field.c +93 -0
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc_field.h +33 -0
- data/ext/isomorfeus_ferret_ext/frt_mdbx_store.c +102 -70
- data/ext/isomorfeus_ferret_ext/frt_mempool.c +8 -16
- data/ext/isomorfeus_ferret_ext/frt_multimapper.c +23 -46
- data/ext/isomorfeus_ferret_ext/frt_multimapper.h +4 -8
- data/ext/isomorfeus_ferret_ext/frt_out_stream.c +334 -0
- data/ext/isomorfeus_ferret_ext/frt_out_stream.h +198 -0
- data/ext/isomorfeus_ferret_ext/frt_posh.c +6 -819
- data/ext/isomorfeus_ferret_ext/frt_posh.h +0 -57
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.c +11 -22
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +85 -171
- data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +8 -16
- data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +1 -2
- data/ext/isomorfeus_ferret_ext/frt_q_parser.c +49 -98
- data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +52 -104
- data/ext/isomorfeus_ferret_ext/frt_q_range.c +6 -12
- data/ext/isomorfeus_ferret_ext/frt_q_span.c +113 -226
- data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +1 -2
- data/ext/isomorfeus_ferret_ext/frt_ram_store.c +134 -85
- data/ext/isomorfeus_ferret_ext/frt_ram_store.h +12 -0
- data/ext/isomorfeus_ferret_ext/frt_search.c +82 -164
- data/ext/isomorfeus_ferret_ext/frt_similarity.c +11 -22
- data/ext/isomorfeus_ferret_ext/frt_similarity.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_store.c +13 -536
- data/ext/isomorfeus_ferret_ext/frt_store.h +90 -495
- data/ext/isomorfeus_ferret_ext/frt_stream.h +18 -0
- data/ext/isomorfeus_ferret_ext/frt_term_vectors.c +8 -16
- data/ext/isomorfeus_ferret_ext/frt_win32.h +5 -10
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +12 -11
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +11 -13
- data/ext/isomorfeus_ferret_ext/lz4.c +422 -195
- data/ext/isomorfeus_ferret_ext/lz4.h +114 -46
- data/ext/isomorfeus_ferret_ext/lz4frame.c +421 -242
- data/ext/isomorfeus_ferret_ext/lz4frame.h +122 -53
- data/ext/isomorfeus_ferret_ext/lz4hc.c +127 -111
- data/ext/isomorfeus_ferret_ext/lz4hc.h +14 -14
- data/ext/isomorfeus_ferret_ext/lz4xxhash.h +1 -1
- data/ext/isomorfeus_ferret_ext/mdbx.c +3762 -2526
- data/ext/isomorfeus_ferret_ext/mdbx.h +115 -70
- data/ext/isomorfeus_ferret_ext/test.c +40 -87
- data/ext/isomorfeus_ferret_ext/test.h +3 -6
- data/ext/isomorfeus_ferret_ext/test_1710.c +11 -13
- data/ext/isomorfeus_ferret_ext/test_analysis.c +32 -64
- data/ext/isomorfeus_ferret_ext/test_array.c +6 -12
- data/ext/isomorfeus_ferret_ext/test_bitvector.c +12 -24
- data/ext/isomorfeus_ferret_ext/test_document.c +23 -33
- data/ext/isomorfeus_ferret_ext/test_except.c +10 -21
- data/ext/isomorfeus_ferret_ext/test_fields.c +62 -68
- data/ext/isomorfeus_ferret_ext/test_file_deleter.c +15 -24
- data/ext/isomorfeus_ferret_ext/test_filter.c +17 -27
- data/ext/isomorfeus_ferret_ext/test_global.c +14 -29
- data/ext/isomorfeus_ferret_ext/test_hash.c +19 -38
- data/ext/isomorfeus_ferret_ext/test_hashset.c +8 -16
- data/ext/isomorfeus_ferret_ext/test_helper.c +4 -8
- data/ext/isomorfeus_ferret_ext/test_highlighter.c +16 -28
- data/ext/isomorfeus_ferret_ext/test_index.c +277 -495
- data/ext/isomorfeus_ferret_ext/test_lang.c +7 -14
- data/ext/isomorfeus_ferret_ext/test_mdbx_store.c +2 -5
- data/ext/isomorfeus_ferret_ext/test_mempool.c +5 -10
- data/ext/isomorfeus_ferret_ext/test_multimapper.c +3 -6
- data/ext/isomorfeus_ferret_ext/test_priorityqueue.c +9 -18
- data/ext/isomorfeus_ferret_ext/test_q_const_score.c +4 -6
- data/ext/isomorfeus_ferret_ext/test_q_filtered.c +3 -4
- data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +9 -15
- data/ext/isomorfeus_ferret_ext/test_q_parser.c +8 -16
- data/ext/isomorfeus_ferret_ext/test_q_span.c +19 -35
- data/ext/isomorfeus_ferret_ext/test_ram_store.c +14 -13
- data/ext/isomorfeus_ferret_ext/test_search.c +60 -109
- data/ext/isomorfeus_ferret_ext/test_segments.c +8 -13
- data/ext/isomorfeus_ferret_ext/test_similarity.c +2 -4
- data/ext/isomorfeus_ferret_ext/test_sort.c +14 -24
- data/ext/isomorfeus_ferret_ext/test_store.c +96 -115
- data/ext/isomorfeus_ferret_ext/test_term.c +9 -15
- data/ext/isomorfeus_ferret_ext/test_term_vectors.c +9 -14
- data/ext/isomorfeus_ferret_ext/test_test.c +4 -8
- data/ext/isomorfeus_ferret_ext/test_threading.c +15 -30
- data/ext/isomorfeus_ferret_ext/testhelper.c +11 -21
- data/ext/isomorfeus_ferret_ext/testhelper.h +1 -1
- data/ext/isomorfeus_ferret_ext/tests_all.h +1 -2
- data/lib/isomorfeus/ferret/index/index.rb +1 -12
- data/lib/isomorfeus/ferret/version.rb +1 -1
- metadata +43 -4
@@ -0,0 +1,482 @@
|
|
1
|
+
/* prevent warning: #warning Please include winsock2.h before windows.h [-Wcpp] */
|
2
|
+
#ifdef _WIN32
|
3
|
+
# include <winsock2.h>
|
4
|
+
#endif
|
5
|
+
|
6
|
+
#include "brotli_decode.h"
|
7
|
+
#include "brotli_encode.h"
|
8
|
+
#include "bzlib.h"
|
9
|
+
#include "lz4frame.h"
|
10
|
+
#include "frt_except.h"
|
11
|
+
#include "frt_in_stream.h"
|
12
|
+
|
13
|
+
/**
|
14
|
+
* Create a newly allocated and initialized InStream
|
15
|
+
*
|
16
|
+
* @return a newly allocated and initialized InStream
|
17
|
+
*/
|
18
|
+
FrtInStream *frt_is_new(void) {
|
19
|
+
FrtInStream *is = FRT_ALLOC(FrtInStream);
|
20
|
+
is->folder_name = NULL;
|
21
|
+
is->f = FRT_ALLOC_AND_ZERO(FrtInStreamFile);
|
22
|
+
is->f->ref_cnt = 1;
|
23
|
+
is->buf.start = 0;
|
24
|
+
is->buf.pos = 0;
|
25
|
+
is->buf.len = 0;
|
26
|
+
is->ref_cnt = 1;
|
27
|
+
return is;
|
28
|
+
}
|
29
|
+
|
30
|
+
/**
|
31
|
+
* Refill the InStream's buffer from the store source (filesystem or memory).
|
32
|
+
*
|
33
|
+
* @param is the InStream to refill
|
34
|
+
* @raise FRT_IO_ERROR if there is a error reading from the filesystem
|
35
|
+
* @raise FRT_EOF_ERROR if there is an attempt to read past the end of the file
|
36
|
+
*/
|
37
|
+
static void frt_is_refill(FrtInStream *is) {
|
38
|
+
frt_off_t start = is->buf.start + is->buf.pos;
|
39
|
+
frt_off_t last = start + FRT_BUFFER_SIZE;
|
40
|
+
frt_off_t flen = is->m->length_i(is);
|
41
|
+
|
42
|
+
if (last > flen) { /* don't read past EOF */
|
43
|
+
last = flen;
|
44
|
+
}
|
45
|
+
|
46
|
+
is->buf.len = last - start;
|
47
|
+
if (is->buf.len <= 0) {
|
48
|
+
FRT_RAISE(FRT_EOF_ERROR, "current pos = %"FRT_OFF_T_PFX"d, "
|
49
|
+
"file length = %"FRT_OFF_T_PFX"d", start, flen);
|
50
|
+
}
|
51
|
+
|
52
|
+
is->m->read_i(is, is->buf.buf, is->buf.len);
|
53
|
+
|
54
|
+
is->buf.start = start;
|
55
|
+
is->buf.pos = 0;
|
56
|
+
}
|
57
|
+
|
58
|
+
/**
|
59
|
+
* Unsafe alternative to frt_is_read_byte. Only use this method when you know
|
60
|
+
* there is no chance that you will read past the end of the InStream's
|
61
|
+
* buffer.
|
62
|
+
*/
|
63
|
+
#define read_byte(is) is->buf.buf[is->buf.pos++]
|
64
|
+
|
65
|
+
/**
|
66
|
+
* Read a singly byte (unsigned char) from the InStream +is+.
|
67
|
+
*
|
68
|
+
* @param is the Instream to read from
|
69
|
+
* @return a single unsigned char read from the InStream +is+
|
70
|
+
* @raise FRT_IO_ERROR if there is a error reading from the filesystem
|
71
|
+
* @raise FRT_EOF_ERROR if there is an attempt to read past the end of the file
|
72
|
+
*/
|
73
|
+
frt_uchar frt_is_read_byte(FrtInStream *is) {
|
74
|
+
if (is->buf.pos >= is->buf.len) {
|
75
|
+
frt_is_refill(is);
|
76
|
+
}
|
77
|
+
return read_byte(is);
|
78
|
+
}
|
79
|
+
|
80
|
+
off_t frt_is_pos(FrtInStream *is) {
|
81
|
+
return is->buf.start + is->buf.pos;
|
82
|
+
}
|
83
|
+
|
84
|
+
frt_uchar *frt_is_read_bytes(FrtInStream *is, frt_uchar *buf, int len) {
|
85
|
+
int i;
|
86
|
+
frt_off_t start;
|
87
|
+
|
88
|
+
if ((is->buf.pos + len) < is->buf.len) {
|
89
|
+
for (i = 0; i < len; i++) {
|
90
|
+
buf[i] = read_byte(is);
|
91
|
+
}
|
92
|
+
} else { /* read all-at-once */
|
93
|
+
start = frt_is_pos(is);
|
94
|
+
is->m->seek_i(is, start);
|
95
|
+
is->m->read_i(is, buf, len);
|
96
|
+
|
97
|
+
is->buf.start = start + len; /* adjust stream variables */
|
98
|
+
is->buf.pos = 0;
|
99
|
+
is->buf.len = 0; /* trigger refill on read */
|
100
|
+
}
|
101
|
+
return buf;
|
102
|
+
}
|
103
|
+
|
104
|
+
void frt_is_seek(FrtInStream *is, frt_off_t pos) {
|
105
|
+
if (pos >= is->buf.start && pos < (is->buf.start + is->buf.len)) {
|
106
|
+
is->buf.pos = pos - is->buf.start; /* seek within buffer */
|
107
|
+
} else {
|
108
|
+
is->buf.start = pos;
|
109
|
+
is->buf.pos = 0;
|
110
|
+
is->buf.len = 0; /* trigger refill() on read() */
|
111
|
+
is->m->seek_i(is, pos);
|
112
|
+
}
|
113
|
+
}
|
114
|
+
|
115
|
+
void frt_is_close(FrtInStream *is) {
|
116
|
+
if (is->ref_cnt == 0) {
|
117
|
+
FRT_RAISE(FRT_STATE_ERROR, "is ref_cnt to low\n");
|
118
|
+
}
|
119
|
+
|
120
|
+
if (FRT_DEREF(is) == 0) {
|
121
|
+
if (FRT_DEREF(is->f) == 0) {
|
122
|
+
is->m->close_i(is);
|
123
|
+
free(is->f);
|
124
|
+
}
|
125
|
+
if (is->folder_name) {
|
126
|
+
free((void *)is->folder_name);
|
127
|
+
}
|
128
|
+
free(is);
|
129
|
+
}
|
130
|
+
}
|
131
|
+
|
132
|
+
FrtInStream *frt_is_clone(FrtInStream *is) {
|
133
|
+
if (!(is->f))
|
134
|
+
return NULL;
|
135
|
+
FrtInStream *new_is = FRT_ALLOC(FrtInStream);
|
136
|
+
memcpy(new_is, is, sizeof(FrtInStream));
|
137
|
+
if (is->folder_name) {
|
138
|
+
new_is->folder_name = frt_estrdup(is->folder_name);
|
139
|
+
}
|
140
|
+
new_is->ref_cnt = 1;
|
141
|
+
FRT_REF(new_is->f);
|
142
|
+
return new_is;
|
143
|
+
}
|
144
|
+
|
145
|
+
frt_i32 frt_is_read_i32(FrtInStream *is) {
|
146
|
+
return ((frt_i32)frt_is_read_byte(is) << 24) |
|
147
|
+
((frt_i32)frt_is_read_byte(is) << 16) |
|
148
|
+
((frt_i32)frt_is_read_byte(is) << 8) |
|
149
|
+
((frt_i32)frt_is_read_byte(is));
|
150
|
+
}
|
151
|
+
|
152
|
+
|
153
|
+
frt_i64 frt_is_read_i64(FrtInStream *is) {
|
154
|
+
return ((frt_i64)frt_is_read_byte(is) << 56) |
|
155
|
+
((frt_i64)frt_is_read_byte(is) << 48) |
|
156
|
+
((frt_i64)frt_is_read_byte(is) << 40) |
|
157
|
+
((frt_i64)frt_is_read_byte(is) << 32) |
|
158
|
+
((frt_i64)frt_is_read_byte(is) << 24) |
|
159
|
+
((frt_i64)frt_is_read_byte(is) << 16) |
|
160
|
+
((frt_i64)frt_is_read_byte(is) << 8) |
|
161
|
+
((frt_i64)frt_is_read_byte(is));
|
162
|
+
}
|
163
|
+
|
164
|
+
|
165
|
+
frt_u32 frt_is_read_u32(FrtInStream *is) {
|
166
|
+
return ((frt_u32)frt_is_read_byte(is) << 24) |
|
167
|
+
((frt_u32)frt_is_read_byte(is) << 16) |
|
168
|
+
((frt_u32)frt_is_read_byte(is) << 8) |
|
169
|
+
((frt_u32)frt_is_read_byte(is));
|
170
|
+
}
|
171
|
+
|
172
|
+
frt_u64 frt_is_read_u64(FrtInStream *is) {
|
173
|
+
return ((frt_u64)frt_is_read_byte(is) << 56) |
|
174
|
+
((frt_u64)frt_is_read_byte(is) << 48) |
|
175
|
+
((frt_u64)frt_is_read_byte(is) << 40) |
|
176
|
+
((frt_u64)frt_is_read_byte(is) << 32) |
|
177
|
+
((frt_u64)frt_is_read_byte(is) << 24) |
|
178
|
+
((frt_u64)frt_is_read_byte(is) << 16) |
|
179
|
+
((frt_u64)frt_is_read_byte(is) << 8) |
|
180
|
+
((frt_u64)frt_is_read_byte(is));
|
181
|
+
}
|
182
|
+
|
183
|
+
|
184
|
+
/* optimized to use unchecked read_byte if there is definitely space */
|
185
|
+
unsigned int frt_is_read_vint(FrtInStream *is) {
|
186
|
+
register unsigned int res, b;
|
187
|
+
register int shift = 7;
|
188
|
+
|
189
|
+
if (is->buf.pos > (is->buf.len - VINT_MAX_LEN)) {
|
190
|
+
b = frt_is_read_byte(is);
|
191
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
192
|
+
|
193
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
194
|
+
b = frt_is_read_byte(is);
|
195
|
+
res |= (b & 0x7F) << shift;
|
196
|
+
shift += 7;
|
197
|
+
}
|
198
|
+
} else { /* unchecked optimization */
|
199
|
+
b = read_byte(is);
|
200
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
201
|
+
|
202
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
203
|
+
b = read_byte(is);
|
204
|
+
res |= (b & 0x7F) << shift;
|
205
|
+
shift += 7;
|
206
|
+
}
|
207
|
+
}
|
208
|
+
|
209
|
+
return res;
|
210
|
+
}
|
211
|
+
|
212
|
+
|
213
|
+
/* optimized to use unchecked read_byte if there is definitely space */
|
214
|
+
off_t frt_is_read_voff_t(FrtInStream *is) {
|
215
|
+
register frt_off_t res, b;
|
216
|
+
register int shift = 7;
|
217
|
+
|
218
|
+
if (is->buf.pos > (is->buf.len - VINT_MAX_LEN)) {
|
219
|
+
b = frt_is_read_byte(is);
|
220
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
221
|
+
|
222
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
223
|
+
b = frt_is_read_byte(is);
|
224
|
+
res |= (b & 0x7F) << shift;
|
225
|
+
shift += 7;
|
226
|
+
}
|
227
|
+
} else { /* unchecked optimization */
|
228
|
+
b = read_byte(is);
|
229
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
230
|
+
|
231
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
232
|
+
b = read_byte(is);
|
233
|
+
res |= (b & 0x7F) << shift;
|
234
|
+
shift += 7;
|
235
|
+
}
|
236
|
+
}
|
237
|
+
|
238
|
+
return res;
|
239
|
+
}
|
240
|
+
|
241
|
+
/* optimized to use unchecked read_byte if there is definitely space */
|
242
|
+
frt_u64 frt_is_read_vll(FrtInStream *is) {
|
243
|
+
register frt_u64 res, b;
|
244
|
+
register int shift = 7;
|
245
|
+
|
246
|
+
if (is->buf.pos > (is->buf.len - VINT_MAX_LEN)) {
|
247
|
+
b = frt_is_read_byte(is);
|
248
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
249
|
+
|
250
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
251
|
+
b = frt_is_read_byte(is);
|
252
|
+
res |= (b & 0x7F) << shift;
|
253
|
+
shift += 7;
|
254
|
+
}
|
255
|
+
} else { /* unchecked optimization */
|
256
|
+
b = read_byte(is);
|
257
|
+
res = b & 0x7F; /* 0x7F = 0b01111111 */
|
258
|
+
|
259
|
+
while ((b & 0x80) != 0) { /* 0x80 = 0b10000000 */
|
260
|
+
b = read_byte(is);
|
261
|
+
res |= (b & 0x7F) << shift;
|
262
|
+
shift += 7;
|
263
|
+
}
|
264
|
+
}
|
265
|
+
|
266
|
+
return res;
|
267
|
+
}
|
268
|
+
|
269
|
+
void frt_is_skip_vints(FrtInStream *is, register int cnt) {
|
270
|
+
for (; cnt > 0; cnt--) {
|
271
|
+
while ((frt_is_read_byte(is) & 0x80) != 0) {
|
272
|
+
}
|
273
|
+
}
|
274
|
+
}
|
275
|
+
|
276
|
+
char *frt_is_read_string(FrtInStream *is) {
|
277
|
+
register int length = (int) frt_is_read_vint(is);
|
278
|
+
char *str = FRT_ALLOC_N(char, length + 1);
|
279
|
+
str[length] = '\0';
|
280
|
+
|
281
|
+
if (is->buf.pos > (is->buf.len - length)) {
|
282
|
+
register int i;
|
283
|
+
for (i = 0; i < length; i++) {
|
284
|
+
str[i] = frt_is_read_byte(is);
|
285
|
+
}
|
286
|
+
} else { /* unchecked optimization */
|
287
|
+
memcpy(str, is->buf.buf + is->buf.pos, length);
|
288
|
+
is->buf.pos += length;
|
289
|
+
}
|
290
|
+
|
291
|
+
return str;
|
292
|
+
}
|
293
|
+
|
294
|
+
char *frt_is_read_string_safe(FrtInStream *is) {
|
295
|
+
register int length = (int) frt_is_read_vint(is);
|
296
|
+
char *str = FRT_ALLOC_N(char, length + 1);
|
297
|
+
str[length] = '\0';
|
298
|
+
|
299
|
+
FRT_TRY
|
300
|
+
if (is->buf.pos > (is->buf.len - length)) {
|
301
|
+
register int i;
|
302
|
+
for (i = 0; i < length; i++) {
|
303
|
+
str[i] = frt_is_read_byte(is);
|
304
|
+
}
|
305
|
+
} else { /* unchecked optimization */
|
306
|
+
memcpy(str, is->buf.buf + is->buf.pos, length);
|
307
|
+
is->buf.pos += length;
|
308
|
+
}
|
309
|
+
FRT_XCATCHALL
|
310
|
+
free(str);
|
311
|
+
FRT_XENDTRY
|
312
|
+
|
313
|
+
return str;
|
314
|
+
}
|
315
|
+
|
316
|
+
static char *frt_is_read_brotli_compressed_bytes(FrtInStream *is, int compressed_len, int *len) {
|
317
|
+
int buf_out_idx = 0;
|
318
|
+
int read_len;
|
319
|
+
frt_uchar buf_in[FRT_COMPRESSION_BUFFER_SIZE];
|
320
|
+
const frt_uchar *next_in;
|
321
|
+
size_t available_in;
|
322
|
+
frt_uchar *buf_out = NULL;
|
323
|
+
frt_uchar *next_out;
|
324
|
+
size_t available_out;
|
325
|
+
|
326
|
+
BrotliDecoderState *b_state = BrotliDecoderCreateInstance(NULL, NULL, NULL);
|
327
|
+
BrotliDecoderResult b_result = BROTLI_DECODER_RESULT_NEEDS_MORE_INPUT;
|
328
|
+
if (!b_state) { frt_comp_raise(); return NULL; }
|
329
|
+
|
330
|
+
do {
|
331
|
+
read_len = (compressed_len > FRT_COMPRESSION_BUFFER_SIZE) ? FRT_COMPRESSION_BUFFER_SIZE : compressed_len;
|
332
|
+
frt_is_read_bytes(is, buf_in, read_len);
|
333
|
+
compressed_len -= read_len;
|
334
|
+
available_in = read_len;
|
335
|
+
next_in = buf_in;
|
336
|
+
available_out = FRT_COMPRESSION_BUFFER_SIZE;
|
337
|
+
do {
|
338
|
+
FRT_REALLOC_N(buf_out, frt_uchar, buf_out_idx + FRT_COMPRESSION_BUFFER_SIZE);
|
339
|
+
next_out = buf_out + buf_out_idx;
|
340
|
+
b_result = BrotliDecoderDecompressStream(b_state,
|
341
|
+
&available_in, &next_in,
|
342
|
+
&available_out, &next_out, NULL);
|
343
|
+
if (b_result == BROTLI_DECODER_RESULT_ERROR) { frt_comp_raise(); return NULL; }
|
344
|
+
buf_out_idx += FRT_COMPRESSION_BUFFER_SIZE - available_out;
|
345
|
+
} while (b_result == BROTLI_DECODER_RESULT_NEEDS_MORE_OUTPUT);
|
346
|
+
} while (b_result != BROTLI_DECODER_RESULT_SUCCESS && compressed_len > 0);
|
347
|
+
|
348
|
+
BrotliDecoderDestroyInstance(b_state);
|
349
|
+
|
350
|
+
FRT_REALLOC_N(buf_out, frt_uchar, buf_out_idx + 1);
|
351
|
+
buf_out[buf_out_idx] = '\0';
|
352
|
+
*len = buf_out_idx;
|
353
|
+
return (char *)buf_out;
|
354
|
+
}
|
355
|
+
|
356
|
+
static char *frt_is_read_bz2_compressed_bytes(FrtInStream *is, int compressed_len, int *len) {
|
357
|
+
int buf_out_idx = 0, ret, read_len;
|
358
|
+
char *buf_out = NULL;
|
359
|
+
char buf_in[FRT_COMPRESSION_BUFFER_SIZE];
|
360
|
+
bz_stream zstrm;
|
361
|
+
zstrm.bzalloc = NULL;
|
362
|
+
zstrm.bzfree = NULL;
|
363
|
+
zstrm.opaque = NULL;
|
364
|
+
zstrm.next_in = NULL;
|
365
|
+
zstrm.avail_in = 0;
|
366
|
+
if ((ret = BZ2_bzDecompressInit(&zstrm, 0, 0)) != BZ_OK) frt_zraise(ret);
|
367
|
+
|
368
|
+
do {
|
369
|
+
read_len = (compressed_len > FRT_COMPRESSION_BUFFER_SIZE) ? FRT_COMPRESSION_BUFFER_SIZE : compressed_len;
|
370
|
+
frt_is_read_bytes(is, (frt_uchar *)buf_in, read_len);
|
371
|
+
compressed_len -= read_len;
|
372
|
+
zstrm.avail_in = read_len;
|
373
|
+
zstrm.next_in = buf_in;
|
374
|
+
zstrm.avail_out = FRT_COMPRESSION_BUFFER_SIZE;
|
375
|
+
|
376
|
+
do {
|
377
|
+
REALLOC_N(buf_out, char, buf_out_idx + FRT_COMPRESSION_BUFFER_SIZE);
|
378
|
+
zstrm.next_out = buf_out + buf_out_idx;
|
379
|
+
ret = BZ2_bzDecompress(&zstrm);
|
380
|
+
assert(ret != BZ_SEQUENCE_ERROR); /* state not clobbered */
|
381
|
+
if (ret != BZ_OK && ret != BZ_STREAM_END) {
|
382
|
+
(void)BZ2_bzDecompressEnd(&zstrm);
|
383
|
+
frt_zraise(ret);
|
384
|
+
}
|
385
|
+
buf_out_idx += FRT_COMPRESSION_BUFFER_SIZE - zstrm.avail_out;
|
386
|
+
} while (zstrm.avail_out == 0);
|
387
|
+
} while (ret != BZ_STREAM_END && compressed_len != 0);
|
388
|
+
|
389
|
+
(void)BZ2_bzDecompressEnd(&zstrm);
|
390
|
+
|
391
|
+
FRT_REALLOC_N(buf_out, char, buf_out_idx + 1);
|
392
|
+
buf_out[buf_out_idx] = '\0';
|
393
|
+
|
394
|
+
*len = buf_out_idx;
|
395
|
+
return (char *)buf_out;
|
396
|
+
}
|
397
|
+
|
398
|
+
static char *frt_is_read_lz4_compressed_bytes(FrtInStream *is, int compressed_len, int *length) {
|
399
|
+
frt_uchar buf_in[FRT_COMPRESSION_BUFFER_SIZE];
|
400
|
+
char *buf_out = NULL;
|
401
|
+
int dc_length = 0;
|
402
|
+
LZ4F_dctx *dctx;
|
403
|
+
LZ4F_frameInfo_t frame_info;
|
404
|
+
LZ4F_errorCode_t dctx_status = LZ4F_createDecompressionContext(&dctx, LZ4F_VERSION);
|
405
|
+
if (LZ4F_isError(dctx_status)) { *length = -1; return NULL; }
|
406
|
+
|
407
|
+
/* header and buffer */
|
408
|
+
int read_length = (compressed_len > FRT_COMPRESSION_BUFFER_SIZE) ? FRT_COMPRESSION_BUFFER_SIZE : compressed_len;
|
409
|
+
frt_is_read_bytes(is, buf_in, read_length);
|
410
|
+
compressed_len -= read_length;
|
411
|
+
|
412
|
+
size_t consumed_size = read_length;
|
413
|
+
size_t res = LZ4F_getFrameInfo(dctx, &frame_info, buf_in, &consumed_size);
|
414
|
+
if (LZ4F_isError(res)) { *length = -1; return NULL; }
|
415
|
+
size_t buf_out_length;
|
416
|
+
switch(frame_info.blockSizeID) {
|
417
|
+
case LZ4F_default:
|
418
|
+
case LZ4F_max64KB:
|
419
|
+
buf_out_length = 1 << 16;
|
420
|
+
break;
|
421
|
+
case LZ4F_max256KB:
|
422
|
+
buf_out_length = 1 << 18;
|
423
|
+
break;
|
424
|
+
case LZ4F_max1MB:
|
425
|
+
buf_out_length = 1 << 20;
|
426
|
+
break;
|
427
|
+
case LZ4F_max4MB:
|
428
|
+
buf_out_length = 1 << 22;
|
429
|
+
break;
|
430
|
+
default:
|
431
|
+
buf_out_length = 0;
|
432
|
+
}
|
433
|
+
|
434
|
+
res = 1;
|
435
|
+
int first_chunk = 1;
|
436
|
+
|
437
|
+
/* decompress data */
|
438
|
+
while (res != 0) {
|
439
|
+
if (!first_chunk) {
|
440
|
+
read_length = (compressed_len > FRT_COMPRESSION_BUFFER_SIZE) ? FRT_COMPRESSION_BUFFER_SIZE : compressed_len;
|
441
|
+
frt_is_read_bytes(is, buf_in, read_length);
|
442
|
+
compressed_len -= read_length;
|
443
|
+
consumed_size = 0;
|
444
|
+
}
|
445
|
+
first_chunk = 0;
|
446
|
+
|
447
|
+
char *src = (char *)(buf_in + consumed_size);
|
448
|
+
char *src_end = (char *)buf_in + read_length;
|
449
|
+
|
450
|
+
while (src < src_end && res != 0){
|
451
|
+
size_t dest_length = buf_out_length;
|
452
|
+
size_t consumed_size = read_length;
|
453
|
+
FRT_REALLOC_N(buf_out, char, dc_length + buf_out_length);
|
454
|
+
res = LZ4F_decompress(dctx, buf_out + dc_length, &dest_length, src, &consumed_size, NULL);
|
455
|
+
if (LZ4F_isError(res)) { *length = -1; return NULL; }
|
456
|
+
dc_length += dest_length;
|
457
|
+
src = src + consumed_size;
|
458
|
+
}
|
459
|
+
}
|
460
|
+
|
461
|
+
/* finish up */
|
462
|
+
LZ4F_freeDecompressionContext(dctx);
|
463
|
+
|
464
|
+
FRT_REALLOC_N(buf_out, char, dc_length + 1);
|
465
|
+
buf_out[dc_length] = '\0';
|
466
|
+
|
467
|
+
*length = dc_length;
|
468
|
+
return buf_out;
|
469
|
+
}
|
470
|
+
|
471
|
+
char *frt_is_read_compressed_bytes(FrtInStream *is, int compressed_len, int *len, FrtCompressionType compression_type) {
|
472
|
+
switch (compression_type) {
|
473
|
+
case FRT_COMPRESSION_BROTLI:
|
474
|
+
return frt_is_read_brotli_compressed_bytes(is, compressed_len, len);
|
475
|
+
case FRT_COMPRESSION_BZ2:
|
476
|
+
return frt_is_read_bz2_compressed_bytes(is, compressed_len, len);
|
477
|
+
case FRT_COMPRESSION_LZ4:
|
478
|
+
return frt_is_read_lz4_compressed_bytes(is, compressed_len, len);
|
479
|
+
default:
|
480
|
+
return NULL;
|
481
|
+
}
|
482
|
+
}
|