line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
#ifndef SRL_READER_DECOMPRESS_H_ |
2
|
|
|
|
|
|
|
#define SRL_READER_DECOMPRESS_H_ |
3
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
#include "srl_inline.h" |
5
|
|
|
|
|
|
|
#include "srl_common.h" |
6
|
|
|
|
|
|
|
#include "srl_reader.h" |
7
|
|
|
|
|
|
|
#include "srl_reader_error.h" |
8
|
|
|
|
|
|
|
#include "srl_reader_varint.h" |
9
|
|
|
|
|
|
|
#include "srl_protocol.h" |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
#if defined(HAVE_CSNAPPY) |
12
|
|
|
|
|
|
|
#include |
13
|
|
|
|
|
|
|
#else |
14
|
|
|
|
|
|
|
/* PLEASE READ!!! |
15
|
|
|
|
|
|
|
* Since the decompression code was migrated to this header file |
16
|
|
|
|
|
|
|
* and it's not a good practise to include C files into headers (actually, doing so |
17
|
|
|
|
|
|
|
* cases 'duplicate symbol' errors in Sereal::Path, expected huh?) and I didn't |
18
|
|
|
|
|
|
|
* found right way how to compile snappy's implementation files and link them |
19
|
|
|
|
|
|
|
* (main problem is that snappy files in a subdirectory) I'm only including |
20
|
|
|
|
|
|
|
* header file here and a dev should include C file in his code. |
21
|
|
|
|
|
|
|
* |
22
|
|
|
|
|
|
|
* It should be something like this: |
23
|
|
|
|
|
|
|
* #if !defined(HAVE_CSNAPPY) |
24
|
|
|
|
|
|
|
* # include "snappy/csnappy_decompress.c" |
25
|
|
|
|
|
|
|
* #endif |
26
|
|
|
|
|
|
|
*/ |
27
|
|
|
|
|
|
|
#include "snappy/csnappy.h" |
28
|
|
|
|
|
|
|
#endif |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
#if defined(HAVE_ZSTD) |
31
|
|
|
|
|
|
|
#include |
32
|
|
|
|
|
|
|
#else |
33
|
|
|
|
|
|
|
#include "zstd/zstd.h" |
34
|
|
|
|
|
|
|
#endif |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
#if defined(HAVE_MINIZ) |
37
|
|
|
|
|
|
|
#include |
38
|
|
|
|
|
|
|
#else |
39
|
|
|
|
|
|
|
#include "miniz.h" |
40
|
|
|
|
|
|
|
#endif |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
/* Creates a new buffer of size header_len + body_len + 1 and swaps it into place |
43
|
|
|
|
|
|
|
* of the current reader's buffer. Sets reader position to right after the |
44
|
|
|
|
|
|
|
* header and makes the reader state internally consistent. The buffer is |
45
|
|
|
|
|
|
|
* owned by a mortal SV which is returned. */ |
46
|
|
|
|
|
|
|
/* TODO reuse the buffer */ |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
SRL_STATIC_INLINE SV * |
49
|
88001
|
|
|
|
|
|
srl_realloc_empty_buffer(pTHX_ srl_reader_buffer_t *buf, |
50
|
|
|
|
|
|
|
const STRLEN header_len, |
51
|
|
|
|
|
|
|
const STRLEN body_len) |
52
|
|
|
|
|
|
|
{ |
53
|
|
|
|
|
|
|
SV *b_sv; |
54
|
|
|
|
|
|
|
srl_reader_char_ptr b; |
55
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
/* Let perl clean this up. Yes, it's not the most efficient thing |
57
|
|
|
|
|
|
|
* ever, but it's just one mortal per full decompression, so not |
58
|
|
|
|
|
|
|
* a bottle-neck. */ |
59
|
88001
|
|
|
|
|
|
b_sv = sv_2mortal( newSV(header_len + body_len + 1 )); |
60
|
88001
|
|
|
|
|
|
b = (srl_reader_char_ptr) SvPVX(b_sv); |
61
|
|
|
|
|
|
|
|
62
|
88001
|
|
|
|
|
|
buf->start = b; |
63
|
88001
|
|
|
|
|
|
buf->pos = b + header_len; |
64
|
88001
|
|
|
|
|
|
buf->end = buf->pos + body_len; |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
/*SRL_RDR_UPDATE_BODY_POS(buf); // XXX caller *MUST* call by himself/herself */ |
67
|
88001
|
|
|
|
|
|
return b_sv; |
68
|
|
|
|
|
|
|
} |
69
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
/* Decompress a Snappy-compressed document body and put the resulting document |
71
|
|
|
|
|
|
|
* body back in the place of the old compressed blob. The function internaly |
72
|
|
|
|
|
|
|
* creates temporary buffer which is owned by mortal SV. If the caller is |
73
|
|
|
|
|
|
|
* interested in keeping the buffer around for longer time, it should pass |
74
|
|
|
|
|
|
|
* buf_owner parameter and unmortalize it. |
75
|
|
|
|
|
|
|
* The caller *MUST* call SRL_RDR_UPDATE_BODY_POS right after existing from this function. */ |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
SRL_STATIC_INLINE UV |
78
|
58322
|
|
|
|
|
|
srl_decompress_body_snappy(pTHX_ srl_reader_buffer_t *buf, U8 encoding_flags, SV** buf_owner) |
79
|
|
|
|
|
|
|
{ |
80
|
|
|
|
|
|
|
SV *buf_sv; |
81
|
|
|
|
|
|
|
int header_len; |
82
|
|
|
|
|
|
|
int decompress_ok; |
83
|
|
|
|
|
|
|
uint32_t dest_len; |
84
|
|
|
|
|
|
|
UV bytes_consumed; |
85
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
srl_reader_char_ptr old_pos; |
87
|
58322
|
|
|
|
|
|
const STRLEN sereal_header_len = (STRLEN) SRL_RDR_POS_OFS(buf); |
88
|
58322
|
|
|
|
|
|
const STRLEN compressed_packet_len = |
89
|
|
|
|
|
|
|
encoding_flags == SRL_PROTOCOL_ENCODING_SNAPPY_INCREMENTAL |
90
|
|
|
|
|
|
|
? (STRLEN) srl_read_varint_uv_length(aTHX_ buf, " while reading compressed packet size") |
91
|
58322
|
100
|
|
|
|
|
: (STRLEN) SRL_RDR_SPACE_LEFT(buf); |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
/* All bufl's above here, or we break C89 compilers */ |
94
|
58322
|
|
|
|
|
|
old_pos = buf->pos; |
95
|
58322
|
|
|
|
|
|
bytes_consumed = compressed_packet_len + SRL_RDR_POS_OFS(buf); |
96
|
58322
|
|
|
|
|
|
header_len = csnappy_get_uncompressed_length((char *)buf->pos, |
97
|
|
|
|
|
|
|
compressed_packet_len, |
98
|
|
|
|
|
|
|
&dest_len); |
99
|
|
|
|
|
|
|
|
100
|
58322
|
50
|
|
|
|
|
if (header_len == CSNAPPY_E_HEADER_BAD) |
101
|
0
|
|
|
|
|
|
SRL_RDR_ERROR(buf, "Invalid Snappy header in Snappy-compressed Sereal packet"); |
102
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
/* Allocate output buffer and swap it into place within the bufoder. */ |
104
|
58322
|
|
|
|
|
|
buf_sv = srl_realloc_empty_buffer(aTHX_ buf, sereal_header_len, dest_len); |
105
|
58322
|
50
|
|
|
|
|
if (buf_owner) *buf_owner = buf_sv; |
106
|
|
|
|
|
|
|
|
107
|
58322
|
|
|
|
|
|
decompress_ok = csnappy_decompress_noheader((char *)(old_pos + header_len), |
108
|
|
|
|
|
|
|
compressed_packet_len - header_len, |
109
|
58322
|
|
|
|
|
|
(char *)buf->pos, |
110
|
|
|
|
|
|
|
&dest_len); |
111
|
|
|
|
|
|
|
|
112
|
58322
|
50
|
|
|
|
|
if (expect_false( decompress_ok != 0 )) { |
113
|
0
|
|
|
|
|
|
SRL_RDR_ERRORf1(buf, "Snappy decompression of Sereal packet payload failed with error %i!", |
114
|
|
|
|
|
|
|
decompress_ok); |
115
|
|
|
|
|
|
|
} |
116
|
|
|
|
|
|
|
|
117
|
58322
|
|
|
|
|
|
return bytes_consumed; |
118
|
|
|
|
|
|
|
} |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
/* Decompress a zlib-compressed document body and put the resulting |
121
|
|
|
|
|
|
|
* document body back in the place of the old compressed blob. The function |
122
|
|
|
|
|
|
|
* internaly creates temporary buffer which is owned by mortal SV. If the |
123
|
|
|
|
|
|
|
* caller is interested in keeping the buffer around for longer time, it should |
124
|
|
|
|
|
|
|
* pass buf_owner parameter and unmortalize it. |
125
|
|
|
|
|
|
|
* The caller *MUST* call SRL_RDR_UPDATE_BODY_POS right after existing from this function. */ |
126
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
SRL_STATIC_INLINE UV |
128
|
20143
|
|
|
|
|
|
srl_decompress_body_zlib(pTHX_ srl_reader_buffer_t *buf, SV** buf_owner) |
129
|
|
|
|
|
|
|
{ |
130
|
|
|
|
|
|
|
SV *buf_sv; |
131
|
|
|
|
|
|
|
mz_ulong tmp; |
132
|
|
|
|
|
|
|
int decompress_ok; |
133
|
|
|
|
|
|
|
UV bytes_consumed; |
134
|
|
|
|
|
|
|
srl_reader_char_ptr old_pos; |
135
|
20143
|
|
|
|
|
|
const STRLEN sereal_header_len = (STRLEN)SRL_RDR_POS_OFS(buf); |
136
|
20143
|
|
|
|
|
|
const STRLEN uncompressed_packet_len = (STRLEN)srl_read_varint_uv(aTHX_ buf); |
137
|
20143
|
|
|
|
|
|
const STRLEN compressed_packet_len = |
138
|
|
|
|
|
|
|
(STRLEN)srl_read_varint_uv_length(aTHX_ buf, " while reading compressed packet size"); |
139
|
|
|
|
|
|
|
|
140
|
20143
|
50
|
|
|
|
|
SRL_RDR_ASSERT_SPACE(buf, compressed_packet_len, " while reading compressed packet"); |
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
/* All decl's above here, or we break C89 compilers */ |
143
|
20143
|
|
|
|
|
|
old_pos = buf->pos; |
144
|
20143
|
|
|
|
|
|
bytes_consumed = compressed_packet_len + SRL_RDR_POS_OFS(buf); |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
/* Allocate output buffer and swap it into place within the decoder. */ |
147
|
20143
|
|
|
|
|
|
buf_sv = srl_realloc_empty_buffer(aTHX_ buf, sereal_header_len, uncompressed_packet_len); |
148
|
20143
|
50
|
|
|
|
|
if (buf_owner) *buf_owner = buf_sv; |
149
|
|
|
|
|
|
|
|
150
|
20143
|
|
|
|
|
|
tmp = uncompressed_packet_len; |
151
|
20143
|
|
|
|
|
|
decompress_ok = mz_uncompress((unsigned char *)buf->pos, |
152
|
|
|
|
|
|
|
&tmp, old_pos, compressed_packet_len); |
153
|
|
|
|
|
|
|
|
154
|
20143
|
50
|
|
|
|
|
if (expect_false( decompress_ok != Z_OK )) { |
155
|
0
|
|
|
|
|
|
SRL_RDR_ERRORf1(buf, "ZLIB decompression of Sereal packet payload failed with error %i!", decompress_ok); |
156
|
|
|
|
|
|
|
} |
157
|
|
|
|
|
|
|
|
158
|
20143
|
|
|
|
|
|
return bytes_consumed; |
159
|
|
|
|
|
|
|
} |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
/* Decompress a zstd-compressed document body and put the resulting document |
162
|
|
|
|
|
|
|
* body back in the place of the old compressed blob. The function internaly |
163
|
|
|
|
|
|
|
* creates temporary buffer which is owned by mortal SV. If the caller is |
164
|
|
|
|
|
|
|
* interested in keeping the buffer around for longer time, it should pass |
165
|
|
|
|
|
|
|
* buf_owner parameter and unmortalize it. The caller *MUST* call |
166
|
|
|
|
|
|
|
* SRL_RDR_UPDATE_BODY_POS right after existing from this function. */ |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
SRL_STATIC_INLINE UV |
169
|
9536
|
|
|
|
|
|
srl_decompress_body_zstd(pTHX_ srl_reader_buffer_t *buf, SV** buf_owner) |
170
|
|
|
|
|
|
|
{ |
171
|
|
|
|
|
|
|
SV *buf_sv; |
172
|
|
|
|
|
|
|
UV bytes_consumed; |
173
|
|
|
|
|
|
|
size_t decompress_code; |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
srl_reader_char_ptr old_pos; |
176
|
|
|
|
|
|
|
unsigned long long uncompressed_packet_len; |
177
|
9536
|
|
|
|
|
|
const STRLEN sereal_header_len = (STRLEN) SRL_RDR_POS_OFS(buf); |
178
|
9536
|
|
|
|
|
|
const STRLEN compressed_packet_len = (STRLEN) srl_read_varint_uv_length(aTHX_ buf, |
179
|
|
|
|
|
|
|
" while reading compressed packet size"); |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
/* All bufl's above here, or we break C89 compilers */ |
182
|
9536
|
|
|
|
|
|
old_pos = buf->pos; |
183
|
9536
|
|
|
|
|
|
bytes_consumed = compressed_packet_len + SRL_RDR_POS_OFS(buf); |
184
|
|
|
|
|
|
|
|
185
|
9536
|
|
|
|
|
|
uncompressed_packet_len = ZSTD_getDecompressedSize((const void *)buf->pos, (size_t) compressed_packet_len); |
186
|
9536
|
50
|
|
|
|
|
if (expect_false(uncompressed_packet_len == 0)) |
187
|
0
|
|
|
|
|
|
SRL_RDR_ERROR(buf, "Invalid zstd packet with unknown uncompressed size"); |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
/* Allocate output buffer and swap it into place within the decoder. */ |
190
|
9536
|
|
|
|
|
|
buf_sv = srl_realloc_empty_buffer(aTHX_ buf, sereal_header_len, (STRLEN) uncompressed_packet_len); |
191
|
9536
|
50
|
|
|
|
|
if (buf_owner) *buf_owner = buf_sv; |
192
|
|
|
|
|
|
|
|
193
|
9536
|
|
|
|
|
|
decompress_code = ZSTD_decompress((void *)buf->pos, (size_t) uncompressed_packet_len, |
194
|
|
|
|
|
|
|
(void *)old_pos, (size_t) compressed_packet_len); |
195
|
|
|
|
|
|
|
|
196
|
9536
|
50
|
|
|
|
|
if (expect_false( ZSTD_isError(decompress_code) )) { |
197
|
0
|
|
|
|
|
|
SRL_RDR_ERRORf1(buf, "Zstd decompression of Sereal packet payload failed with error %s!", |
198
|
|
|
|
|
|
|
ZSTD_getErrorName(decompress_code)); |
199
|
|
|
|
|
|
|
} |
200
|
|
|
|
|
|
|
|
201
|
9536
|
|
|
|
|
|
return bytes_consumed; |
202
|
|
|
|
|
|
|
} |
203
|
|
|
|
|
|
|
|
204
|
|
|
|
|
|
|
#endif |