js-stream-sas7bdat 0.1.0 → 0.1.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/binding.gyp +58 -0
- package/dist/class/datasetSas7BDat.d.ts.map +1 -1
- package/dist/class/datasetSas7BDat.js +4 -3
- package/dist/class/datasetSas7BDat.js.map +1 -1
- package/dist/tsconfig.tsbuildinfo +1 -1
- package/package.json +16 -7
- package/prebuilds/darwin-arm64/js-stream-sas7bdat.node +0 -0
- package/prebuilds/darwin-x64/js-stream-sas7bdat.node +0 -0
- package/prebuilds/linux-arm64/js-stream-sas7bdat.node +0 -0
- package/prebuilds/linux-x64/js-stream-sas7bdat.node +0 -0
- package/prebuilds/win32-ia32/js-stream-sas7bdat.node +0 -0
- package/prebuilds/win32-x64/js-stream-sas7bdat.node +0 -0
- package/src/binding/ReadStat/LICENSE +19 -0
- package/src/binding/ReadStat/README.md +483 -0
- package/src/binding/ReadStat/src/CKHashTable.c +309 -0
- package/src/binding/ReadStat/src/CKHashTable.h +37 -0
- package/src/binding/ReadStat/src/readstat.h +627 -0
- package/src/binding/ReadStat/src/readstat_bits.c +69 -0
- package/src/binding/ReadStat/src/readstat_bits.h +20 -0
- package/src/binding/ReadStat/src/readstat_convert.c +36 -0
- package/src/binding/ReadStat/src/readstat_convert.h +2 -0
- package/src/binding/ReadStat/src/readstat_error.c +126 -0
- package/src/binding/ReadStat/src/readstat_iconv.h +15 -0
- package/src/binding/ReadStat/src/readstat_io_unistd.c +147 -0
- package/src/binding/ReadStat/src/readstat_io_unistd.h +11 -0
- package/src/binding/ReadStat/src/readstat_malloc.c +34 -0
- package/src/binding/ReadStat/src/readstat_malloc.h +4 -0
- package/src/binding/ReadStat/src/readstat_metadata.c +53 -0
- package/src/binding/ReadStat/src/readstat_parser.c +121 -0
- package/src/binding/ReadStat/src/readstat_strings.h +6 -0
- package/src/binding/ReadStat/src/readstat_value.c +178 -0
- package/src/binding/ReadStat/src/readstat_variable.c +123 -0
- package/src/binding/ReadStat/src/readstat_writer.c +677 -0
- package/src/binding/ReadStat/src/readstat_writer.h +21 -0
- package/src/binding/ReadStat/src/sas/ieee.c +420 -0
- package/src/binding/ReadStat/src/sas/ieee.h +6 -0
- package/src/binding/ReadStat/src/sas/readstat_sas.c +528 -0
- package/src/binding/ReadStat/src/sas/readstat_sas.h +131 -0
- package/src/binding/ReadStat/src/sas/readstat_sas7bcat_read.c +515 -0
- package/src/binding/ReadStat/src/sas/readstat_sas7bcat_write.c +218 -0
- package/src/binding/ReadStat/src/sas/readstat_sas7bdat_read.c +1304 -0
- package/src/binding/ReadStat/src/sas/readstat_sas7bdat_write.c +812 -0
- package/src/binding/ReadStat/src/sas/readstat_sas_rle.c +286 -0
- package/src/binding/ReadStat/src/sas/readstat_sas_rle.h +8 -0
- package/src/binding/ReadStat/src/sas/readstat_xport.c +28 -0
- package/src/binding/ReadStat/src/sas/readstat_xport.h +47 -0
- package/src/binding/ReadStat/src/sas/readstat_xport_parse_format.c +265 -0
- package/src/binding/ReadStat/src/sas/readstat_xport_parse_format.h +4 -0
- package/src/binding/ReadStat/src/sas/readstat_xport_parse_format.rl +68 -0
- package/src/binding/ReadStat/src/sas/readstat_xport_read.c +777 -0
- package/src/binding/ReadStat/src/sas/readstat_xport_write.c +561 -0
- package/src/binding/readstat_binding.cc +393 -0
|
@@ -0,0 +1,286 @@
|
|
|
1
|
+
|
|
2
|
+
#include <sys/types.h>
|
|
3
|
+
#include <string.h>
|
|
4
|
+
#include <stdio.h>
|
|
5
|
+
|
|
6
|
+
#if defined(_MSC_VER)
|
|
7
|
+
#include <BaseTsd.h>
|
|
8
|
+
typedef SSIZE_T ssize_t;
|
|
9
|
+
#endif
|
|
10
|
+
|
|
11
|
+
#include "readstat_sas_rle.h"
|
|
12
|
+
|
|
13
|
+
#define SAS_RLE_COMMAND_COPY64 0
|
|
14
|
+
#define SAS_RLE_COMMAND_COPY64_PLUS_4096 1
|
|
15
|
+
#define SAS_RLE_COMMAND_COPY96 2
|
|
16
|
+
#define SAS_RLE_COMMAND_INSERT_BYTE18 4
|
|
17
|
+
#define SAS_RLE_COMMAND_INSERT_AT17 5
|
|
18
|
+
#define SAS_RLE_COMMAND_INSERT_BLANK17 6
|
|
19
|
+
#define SAS_RLE_COMMAND_INSERT_ZERO17 7
|
|
20
|
+
#define SAS_RLE_COMMAND_COPY1 8
|
|
21
|
+
#define SAS_RLE_COMMAND_COPY17 9
|
|
22
|
+
#define SAS_RLE_COMMAND_COPY33 10
|
|
23
|
+
#define SAS_RLE_COMMAND_COPY49 11
|
|
24
|
+
#define SAS_RLE_COMMAND_INSERT_BYTE3 12
|
|
25
|
+
#define SAS_RLE_COMMAND_INSERT_AT2 13
|
|
26
|
+
#define SAS_RLE_COMMAND_INSERT_BLANK2 14
|
|
27
|
+
#define SAS_RLE_COMMAND_INSERT_ZERO2 15
|
|
28
|
+
|
|
29
|
+
#define MAX_INSERT_RUN 4112 // 4095 + 17
|
|
30
|
+
#define MAX_COPY_RUN 4159 // 4095 + 64
|
|
31
|
+
|
|
32
|
+
static size_t command_lengths[16] = {
|
|
33
|
+
[SAS_RLE_COMMAND_COPY64] = 1,
|
|
34
|
+
[SAS_RLE_COMMAND_COPY64_PLUS_4096] = 1,
|
|
35
|
+
[SAS_RLE_COMMAND_INSERT_BYTE18] = 2,
|
|
36
|
+
[SAS_RLE_COMMAND_INSERT_AT17] = 1,
|
|
37
|
+
[SAS_RLE_COMMAND_INSERT_BLANK17] = 1,
|
|
38
|
+
[SAS_RLE_COMMAND_INSERT_ZERO17] = 1,
|
|
39
|
+
[SAS_RLE_COMMAND_INSERT_BYTE3] = 1
|
|
40
|
+
};
|
|
41
|
+
|
|
42
|
+
ssize_t sas_rle_decompressed_len(const void *input_buf, size_t input_len) {
|
|
43
|
+
return sas_rle_decompress(NULL, 0, input_buf, input_len);
|
|
44
|
+
}
|
|
45
|
+
|
|
46
|
+
ssize_t sas_rle_decompress(void *output_buf, size_t output_len,
|
|
47
|
+
const void *input_buf, size_t input_len) {
|
|
48
|
+
unsigned char *buffer = (unsigned char *)output_buf;
|
|
49
|
+
unsigned char *output = buffer;
|
|
50
|
+
size_t output_written = 0;
|
|
51
|
+
|
|
52
|
+
const unsigned char *input = (const unsigned char *)input_buf;
|
|
53
|
+
|
|
54
|
+
while (input < (const unsigned char *)input_buf + input_len) {
|
|
55
|
+
unsigned char control = *input++;
|
|
56
|
+
unsigned char command = (control & 0xF0) >> 4;
|
|
57
|
+
unsigned char length = (control & 0x0F);
|
|
58
|
+
int copy_len = 0;
|
|
59
|
+
int insert_len = 0;
|
|
60
|
+
unsigned char insert_byte = '\0';
|
|
61
|
+
if (input + command_lengths[command] > (const unsigned char *)input_buf + input_len) {
|
|
62
|
+
return -1;
|
|
63
|
+
}
|
|
64
|
+
switch (command) {
|
|
65
|
+
case SAS_RLE_COMMAND_COPY64:
|
|
66
|
+
copy_len = (*input++) + 64 + length * 256;
|
|
67
|
+
break;
|
|
68
|
+
case SAS_RLE_COMMAND_COPY64_PLUS_4096:
|
|
69
|
+
copy_len = (*input++) + 64 + length * 256 + 4096;
|
|
70
|
+
break;
|
|
71
|
+
case SAS_RLE_COMMAND_COPY96: copy_len = length + 96; break;
|
|
72
|
+
case SAS_RLE_COMMAND_INSERT_BYTE18:
|
|
73
|
+
insert_len = (*input++) + 18 + length * 256;
|
|
74
|
+
insert_byte = *input++;
|
|
75
|
+
break;
|
|
76
|
+
case SAS_RLE_COMMAND_INSERT_AT17:
|
|
77
|
+
insert_len = (*input++) + 17 + length * 256;
|
|
78
|
+
insert_byte = '@';
|
|
79
|
+
break;
|
|
80
|
+
case SAS_RLE_COMMAND_INSERT_BLANK17:
|
|
81
|
+
insert_len = (*input++) + 17 + length * 256;
|
|
82
|
+
insert_byte = ' ';
|
|
83
|
+
break;
|
|
84
|
+
case SAS_RLE_COMMAND_INSERT_ZERO17:
|
|
85
|
+
insert_len = (*input++) + 17 + length * 256;
|
|
86
|
+
insert_byte = '\0';
|
|
87
|
+
break;
|
|
88
|
+
case SAS_RLE_COMMAND_COPY1: copy_len = length + 1; break;
|
|
89
|
+
case SAS_RLE_COMMAND_COPY17: copy_len = length + 17; break;
|
|
90
|
+
case SAS_RLE_COMMAND_COPY33: copy_len = length + 33; break;
|
|
91
|
+
case SAS_RLE_COMMAND_COPY49: copy_len = length + 49; break;
|
|
92
|
+
case SAS_RLE_COMMAND_INSERT_BYTE3:
|
|
93
|
+
insert_byte = *input++;
|
|
94
|
+
insert_len = length + 3;
|
|
95
|
+
break;
|
|
96
|
+
case SAS_RLE_COMMAND_INSERT_AT2:
|
|
97
|
+
insert_byte = '@';
|
|
98
|
+
insert_len = length + 2;
|
|
99
|
+
break;
|
|
100
|
+
case SAS_RLE_COMMAND_INSERT_BLANK2:
|
|
101
|
+
insert_byte = ' ';
|
|
102
|
+
insert_len = length + 2;
|
|
103
|
+
break;
|
|
104
|
+
case SAS_RLE_COMMAND_INSERT_ZERO2:
|
|
105
|
+
insert_byte = '\0';
|
|
106
|
+
insert_len = length + 2;
|
|
107
|
+
break;
|
|
108
|
+
default:
|
|
109
|
+
/* error out here? */
|
|
110
|
+
break;
|
|
111
|
+
}
|
|
112
|
+
if (copy_len) {
|
|
113
|
+
if (output_written + copy_len > output_len) {
|
|
114
|
+
return -1;
|
|
115
|
+
}
|
|
116
|
+
if (input + copy_len > (const unsigned char *)input_buf + input_len) {
|
|
117
|
+
return -1;
|
|
118
|
+
}
|
|
119
|
+
if (output) {
|
|
120
|
+
memcpy(&output[output_written], input, copy_len);
|
|
121
|
+
}
|
|
122
|
+
input += copy_len;
|
|
123
|
+
output_written += copy_len;
|
|
124
|
+
}
|
|
125
|
+
if (insert_len) {
|
|
126
|
+
if (output_written + insert_len > output_len) {
|
|
127
|
+
return -1;
|
|
128
|
+
}
|
|
129
|
+
if (output) {
|
|
130
|
+
memset(&output[output_written], insert_byte, insert_len);
|
|
131
|
+
}
|
|
132
|
+
output_written += insert_len;
|
|
133
|
+
}
|
|
134
|
+
}
|
|
135
|
+
|
|
136
|
+
return output_written;
|
|
137
|
+
}
|
|
138
|
+
|
|
139
|
+
static size_t sas_rle_measure_copy_run(size_t copy_run) {
|
|
140
|
+
size_t len = 0;
|
|
141
|
+
while (copy_run >= MAX_COPY_RUN) {
|
|
142
|
+
len += 2 + MAX_COPY_RUN;
|
|
143
|
+
copy_run -= MAX_COPY_RUN;
|
|
144
|
+
}
|
|
145
|
+
return len + (copy_run > 64) + (copy_run > 0) + copy_run;
|
|
146
|
+
}
|
|
147
|
+
|
|
148
|
+
static size_t sas_rle_copy_run(unsigned char *output_buf, size_t offset,
|
|
149
|
+
const unsigned char *copy, size_t copy_run) {
|
|
150
|
+
unsigned char *out = output_buf + offset;
|
|
151
|
+
if (output_buf == NULL)
|
|
152
|
+
return sas_rle_measure_copy_run(copy_run);
|
|
153
|
+
|
|
154
|
+
while (copy_run >= MAX_COPY_RUN) {
|
|
155
|
+
*out++ = (SAS_RLE_COMMAND_COPY64 << 4) + 0x0F;
|
|
156
|
+
*out++ = 0xFF;
|
|
157
|
+
memcpy(out, copy, MAX_COPY_RUN);
|
|
158
|
+
out += MAX_COPY_RUN;
|
|
159
|
+
copy += MAX_COPY_RUN;
|
|
160
|
+
copy_run -= MAX_COPY_RUN;
|
|
161
|
+
}
|
|
162
|
+
|
|
163
|
+
if (copy_run > 64) {
|
|
164
|
+
int length = (copy_run - 64) / 256;
|
|
165
|
+
unsigned char rem = (copy_run - 64) % 256;
|
|
166
|
+
*out++ = (SAS_RLE_COMMAND_COPY64 << 4) + (length & 0x0F);
|
|
167
|
+
*out++ = rem;
|
|
168
|
+
} else if (copy_run >= 49) {
|
|
169
|
+
*out++ = (SAS_RLE_COMMAND_COPY49 << 4) + (copy_run - 49);
|
|
170
|
+
} else if (copy_run >= 33) {
|
|
171
|
+
*out++ = (SAS_RLE_COMMAND_COPY33 << 4) + (copy_run - 33);
|
|
172
|
+
} else if (copy_run >= 17) {
|
|
173
|
+
*out++ = (SAS_RLE_COMMAND_COPY17 << 4) + (copy_run - 17);
|
|
174
|
+
} else if (copy_run >= 1) {
|
|
175
|
+
*out++ = (SAS_RLE_COMMAND_COPY1 << 4) + (copy_run - 1);
|
|
176
|
+
}
|
|
177
|
+
memcpy(out, copy, copy_run);
|
|
178
|
+
out += copy_run;
|
|
179
|
+
return out - (output_buf + offset);
|
|
180
|
+
}
|
|
181
|
+
|
|
182
|
+
static int sas_rle_is_special_byte(unsigned char last_byte) {
|
|
183
|
+
return (last_byte == '@' || last_byte == ' ' || last_byte == '\0');
|
|
184
|
+
}
|
|
185
|
+
|
|
186
|
+
static size_t sas_rle_measure_insert_run(unsigned char last_byte, size_t insert_run) {
|
|
187
|
+
if (sas_rle_is_special_byte(last_byte))
|
|
188
|
+
return insert_run > 17 ? 2 : 1;
|
|
189
|
+
|
|
190
|
+
return insert_run > 18 ? 3 : 2;
|
|
191
|
+
}
|
|
192
|
+
|
|
193
|
+
static size_t sas_rle_insert_run(unsigned char *output_buf, size_t offset, unsigned char last_byte, size_t insert_run) {
|
|
194
|
+
unsigned char *out = output_buf + offset;
|
|
195
|
+
if (output_buf == NULL)
|
|
196
|
+
return sas_rle_measure_insert_run(last_byte, insert_run);
|
|
197
|
+
|
|
198
|
+
if (sas_rle_is_special_byte(last_byte)) {
|
|
199
|
+
if (insert_run > 17) {
|
|
200
|
+
int length = (insert_run - 17) / 256;
|
|
201
|
+
unsigned char rem = (insert_run - 17) % 256;
|
|
202
|
+
if (last_byte == '@') {
|
|
203
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_AT17 << 4) + (length & 0x0F);
|
|
204
|
+
} else if (last_byte == ' ') {
|
|
205
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_BLANK17 << 4) + (length & 0x0F);
|
|
206
|
+
} else if (last_byte == '\0') {
|
|
207
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_ZERO17 << 4) + (length & 0x0F);
|
|
208
|
+
}
|
|
209
|
+
*out++ = rem;
|
|
210
|
+
} else if (insert_run >= 2) {
|
|
211
|
+
if (last_byte == '@') {
|
|
212
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_AT2 << 4) + (insert_run - 2);
|
|
213
|
+
} else if (last_byte == ' ') {
|
|
214
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_BLANK2 << 4) + (insert_run - 2);
|
|
215
|
+
} else if (last_byte == '\0') {
|
|
216
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_ZERO2 << 4) + (insert_run - 2);
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
} else if (insert_run > 18) {
|
|
220
|
+
int length = (insert_run - 18) / 256;
|
|
221
|
+
unsigned char rem = (insert_run - 18) % 256;
|
|
222
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_BYTE18 << 4) + (length & 0x0F);
|
|
223
|
+
*out++ = rem;
|
|
224
|
+
*out++ = last_byte;
|
|
225
|
+
} else if (insert_run >= 3) {
|
|
226
|
+
*out++ = (SAS_RLE_COMMAND_INSERT_BYTE3 << 4) + (insert_run - 3);
|
|
227
|
+
*out++ = last_byte;
|
|
228
|
+
}
|
|
229
|
+
return out - (output_buf + offset);
|
|
230
|
+
}
|
|
231
|
+
|
|
232
|
+
static int sas_rle_is_insert_run(unsigned char last_byte, size_t insert_run) {
|
|
233
|
+
if (sas_rle_is_special_byte(last_byte))
|
|
234
|
+
return (insert_run > 1);
|
|
235
|
+
|
|
236
|
+
return (insert_run > 2);
|
|
237
|
+
}
|
|
238
|
+
|
|
239
|
+
ssize_t sas_rle_compressed_len(const void *bytes, size_t len) {
|
|
240
|
+
return sas_rle_compress(NULL, 0, bytes, len);
|
|
241
|
+
}
|
|
242
|
+
|
|
243
|
+
ssize_t sas_rle_compress(void *output_buf, size_t output_len,
|
|
244
|
+
const void *input_buf, size_t input_len) {
|
|
245
|
+
/* TODO bounds check */
|
|
246
|
+
const unsigned char *p = (const unsigned char *)input_buf;
|
|
247
|
+
const unsigned char *pe = p + input_len;
|
|
248
|
+
const unsigned char *copy = p;
|
|
249
|
+
|
|
250
|
+
unsigned char *out = (unsigned char *)output_buf;
|
|
251
|
+
|
|
252
|
+
size_t insert_run = 0;
|
|
253
|
+
size_t copy_run = 0;
|
|
254
|
+
size_t out_written = 0;
|
|
255
|
+
unsigned char last_byte = 0;
|
|
256
|
+
|
|
257
|
+
while (p < pe) {
|
|
258
|
+
unsigned char c = *p;
|
|
259
|
+
if (insert_run == 0) {
|
|
260
|
+
insert_run = 1;
|
|
261
|
+
} else if (c == last_byte && insert_run < MAX_INSERT_RUN) {
|
|
262
|
+
insert_run++;
|
|
263
|
+
} else {
|
|
264
|
+
if (sas_rle_is_insert_run(last_byte, insert_run)) {
|
|
265
|
+
out_written += sas_rle_copy_run(out, out_written, copy, copy_run);
|
|
266
|
+
out_written += sas_rle_insert_run(out, out_written, last_byte, insert_run);
|
|
267
|
+
copy_run = 0;
|
|
268
|
+
copy = p;
|
|
269
|
+
} else {
|
|
270
|
+
copy_run += insert_run;
|
|
271
|
+
}
|
|
272
|
+
insert_run = 1;
|
|
273
|
+
}
|
|
274
|
+
last_byte = c;
|
|
275
|
+
p++;
|
|
276
|
+
}
|
|
277
|
+
|
|
278
|
+
if (sas_rle_is_insert_run(last_byte, insert_run)) {
|
|
279
|
+
out_written += sas_rle_copy_run(out, out_written, copy, copy_run);
|
|
280
|
+
out_written += sas_rle_insert_run(out, out_written, last_byte, insert_run);
|
|
281
|
+
} else {
|
|
282
|
+
out_written += sas_rle_copy_run(out, out_written, copy, copy_run + insert_run);
|
|
283
|
+
}
|
|
284
|
+
|
|
285
|
+
return out_written;
|
|
286
|
+
}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
|
|
2
|
+
ssize_t sas_rle_decompress(void *output_buf, size_t output_len,
|
|
3
|
+
const void *input_buf, size_t input_len);
|
|
4
|
+
ssize_t sas_rle_compress(void *output_buf, size_t output_len,
|
|
5
|
+
const void *input_buf, size_t input_len);
|
|
6
|
+
|
|
7
|
+
ssize_t sas_rle_decompressed_len(const void *input_buf, size_t input_len);
|
|
8
|
+
ssize_t sas_rle_compressed_len(const void *bytes, size_t len);
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
#include <stdint.h>
|
|
2
|
+
|
|
3
|
+
#include "readstat_xport.h"
|
|
4
|
+
#include "../readstat_bits.h"
|
|
5
|
+
|
|
6
|
+
char _xport_months[12][4] = {
|
|
7
|
+
"JAN", "FEB", "MAR", "APR", "MAY", "JUN",
|
|
8
|
+
"JUL", "AUG", "SEP", "OCT", "NOV", "DEC" };
|
|
9
|
+
|
|
10
|
+
void xport_namestr_bswap(xport_namestr_t *namestr) {
|
|
11
|
+
if (!machine_is_little_endian())
|
|
12
|
+
return;
|
|
13
|
+
|
|
14
|
+
namestr->ntype = byteswap2(namestr->ntype);
|
|
15
|
+
namestr->nhfun = byteswap2(namestr->nhfun);
|
|
16
|
+
namestr->nlng = byteswap2(namestr->nlng);
|
|
17
|
+
namestr->nvar0 = byteswap2(namestr->nvar0);
|
|
18
|
+
|
|
19
|
+
namestr->nfl = byteswap2(namestr->nfl);
|
|
20
|
+
namestr->nfd = byteswap2(namestr->nfd);
|
|
21
|
+
namestr->nfj = byteswap2(namestr->nfj);
|
|
22
|
+
|
|
23
|
+
namestr->nifl = byteswap2(namestr->nifl);
|
|
24
|
+
namestr->nifd = byteswap2(namestr->nifd);
|
|
25
|
+
namestr->npos = byteswap4(namestr->npos);
|
|
26
|
+
|
|
27
|
+
namestr->labeln = byteswap2(namestr->labeln);
|
|
28
|
+
}
|
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
|
|
2
|
+
typedef struct xport_header_record_s {
|
|
3
|
+
char name[9];
|
|
4
|
+
|
|
5
|
+
int num1;
|
|
6
|
+
int num2;
|
|
7
|
+
int num3;
|
|
8
|
+
int num4;
|
|
9
|
+
int num5;
|
|
10
|
+
int num6;
|
|
11
|
+
} xport_header_record_t;
|
|
12
|
+
|
|
13
|
+
extern char _xport_months[12][4];
|
|
14
|
+
|
|
15
|
+
#pragma pack(push, 1)
|
|
16
|
+
typedef struct xport_namestr_s {
|
|
17
|
+
uint16_t ntype;
|
|
18
|
+
uint16_t nhfun;
|
|
19
|
+
uint16_t nlng;
|
|
20
|
+
uint16_t nvar0;
|
|
21
|
+
char nname[8];
|
|
22
|
+
char nlabel[40];
|
|
23
|
+
char nform[8];
|
|
24
|
+
uint16_t nfl;
|
|
25
|
+
uint16_t nfd;
|
|
26
|
+
uint16_t nfj;
|
|
27
|
+
char nfill[2];
|
|
28
|
+
char niform[8];
|
|
29
|
+
uint16_t nifl;
|
|
30
|
+
uint16_t nifd;
|
|
31
|
+
uint32_t npos;
|
|
32
|
+
char longname[32];
|
|
33
|
+
uint16_t labeln;
|
|
34
|
+
char rest[18];
|
|
35
|
+
} xport_namestr_t;
|
|
36
|
+
#pragma pack(pop)
|
|
37
|
+
|
|
38
|
+
typedef struct xport_format_s {
|
|
39
|
+
char name[32];
|
|
40
|
+
int width;
|
|
41
|
+
int decimals;
|
|
42
|
+
} xport_format_t;
|
|
43
|
+
|
|
44
|
+
#define XPORT_MIN_DOUBLE_SIZE 3
|
|
45
|
+
#define XPORT_MAX_DOUBLE_SIZE 8
|
|
46
|
+
|
|
47
|
+
void xport_namestr_bswap(xport_namestr_t *namestr);
|
|
@@ -0,0 +1,265 @@
|
|
|
1
|
+
|
|
2
|
+
#line 1 "src/sas/readstat_xport_parse_format.rl"
|
|
3
|
+
|
|
4
|
+
#include "../readstat.h"
|
|
5
|
+
#include "readstat_xport.h"
|
|
6
|
+
#include "readstat_xport_parse_format.h"
|
|
7
|
+
|
|
8
|
+
|
|
9
|
+
#line 10 "src/sas/readstat_xport_parse_format.c"
|
|
10
|
+
static const char _xport_format_parse_actions[] = {
|
|
11
|
+
0, 1, 0, 1, 1, 1, 3, 1,
|
|
12
|
+
4, 2, 2, 0, 3, 1, 2, 0
|
|
13
|
+
|
|
14
|
+
};
|
|
15
|
+
|
|
16
|
+
static const char _xport_format_parse_key_offsets[] = {
|
|
17
|
+
0, 0, 7, 14, 23, 31, 31, 34,
|
|
18
|
+
42, 50, 58, 60, 62, 65, 73, 81
|
|
19
|
+
};
|
|
20
|
+
|
|
21
|
+
static const char _xport_format_parse_trans_keys[] = {
|
|
22
|
+
95, 48, 57, 65, 90, 97, 122, 95,
|
|
23
|
+
48, 57, 65, 90, 97, 122, 36, 46,
|
|
24
|
+
95, 48, 57, 65, 90, 97, 122, 46,
|
|
25
|
+
95, 48, 57, 65, 90, 97, 122, 46,
|
|
26
|
+
48, 57, 46, 95, 48, 57, 65, 90,
|
|
27
|
+
97, 122, 46, 95, 48, 57, 65, 90,
|
|
28
|
+
97, 122, 46, 95, 48, 57, 65, 90,
|
|
29
|
+
97, 122, 48, 57, 48, 57, 46, 48,
|
|
30
|
+
57, 46, 95, 48, 57, 65, 90, 97,
|
|
31
|
+
122, 46, 95, 48, 57, 65, 90, 97,
|
|
32
|
+
122, 46, 95, 48, 57, 65, 90, 97,
|
|
33
|
+
122, 0
|
|
34
|
+
};
|
|
35
|
+
|
|
36
|
+
static const char _xport_format_parse_single_lengths[] = {
|
|
37
|
+
0, 1, 1, 3, 2, 0, 1, 2,
|
|
38
|
+
2, 2, 0, 0, 1, 2, 2, 2
|
|
39
|
+
};
|
|
40
|
+
|
|
41
|
+
static const char _xport_format_parse_range_lengths[] = {
|
|
42
|
+
0, 3, 3, 3, 3, 0, 1, 3,
|
|
43
|
+
3, 3, 1, 1, 1, 3, 3, 3
|
|
44
|
+
};
|
|
45
|
+
|
|
46
|
+
static const char _xport_format_parse_index_offsets[] = {
|
|
47
|
+
0, 0, 5, 10, 17, 23, 24, 27,
|
|
48
|
+
33, 39, 45, 47, 49, 52, 58, 64
|
|
49
|
+
};
|
|
50
|
+
|
|
51
|
+
static const char _xport_format_parse_indicies[] = {
|
|
52
|
+
2, 0, 2, 2, 1, 4, 3, 4,
|
|
53
|
+
4, 1, 5, 6, 8, 7, 8, 8,
|
|
54
|
+
1, 9, 11, 10, 11, 11, 1, 1,
|
|
55
|
+
12, 13, 1, 9, 0, 14, 0, 0,
|
|
56
|
+
1, 12, 2, 15, 2, 2, 1, 9,
|
|
57
|
+
2, 14, 2, 2, 1, 16, 1, 17,
|
|
58
|
+
1, 18, 19, 1, 6, 3, 20, 3,
|
|
59
|
+
3, 1, 18, 4, 21, 4, 4, 1,
|
|
60
|
+
6, 4, 20, 4, 4, 1, 0
|
|
61
|
+
};
|
|
62
|
+
|
|
63
|
+
static const char _xport_format_parse_trans_targs[] = {
|
|
64
|
+
1, 0, 9, 2, 15, 4, 10, 12,
|
|
65
|
+
13, 5, 6, 7, 5, 6, 8, 8,
|
|
66
|
+
11, 11, 10, 12, 14, 14
|
|
67
|
+
};
|
|
68
|
+
|
|
69
|
+
static const char _xport_format_parse_trans_actions[] = {
|
|
70
|
+
0, 0, 0, 0, 0, 0, 3, 12,
|
|
71
|
+
0, 3, 12, 0, 5, 1, 12, 1,
|
|
72
|
+
9, 1, 5, 1, 12, 1
|
|
73
|
+
};
|
|
74
|
+
|
|
75
|
+
static const char _xport_format_parse_eof_actions[] = {
|
|
76
|
+
0, 0, 0, 3, 3, 0, 5, 3,
|
|
77
|
+
5, 3, 0, 7, 5, 3, 5, 3
|
|
78
|
+
};
|
|
79
|
+
|
|
80
|
+
static const int xport_format_parse_start = 3;
|
|
81
|
+
|
|
82
|
+
static const int xport_format_parse_en_main = 3;
|
|
83
|
+
|
|
84
|
+
|
|
85
|
+
#line 9 "src/sas/readstat_xport_parse_format.rl"
|
|
86
|
+
|
|
87
|
+
|
|
88
|
+
readstat_error_t xport_parse_format(const char *data, size_t len, xport_format_t *fmt,
|
|
89
|
+
readstat_error_handler error_handler, void *user_ctx) {
|
|
90
|
+
|
|
91
|
+
fmt->name[0] = '\0';
|
|
92
|
+
fmt->width = 0;
|
|
93
|
+
fmt->decimals = 0;
|
|
94
|
+
|
|
95
|
+
readstat_error_t retval = READSTAT_OK;
|
|
96
|
+
const char *p = data;
|
|
97
|
+
const char *pe = p + len;
|
|
98
|
+
const char *eof = pe;
|
|
99
|
+
|
|
100
|
+
int cs;
|
|
101
|
+
unsigned int temp_val = 0;
|
|
102
|
+
size_t parsed_len = 0;
|
|
103
|
+
|
|
104
|
+
|
|
105
|
+
#line 106 "src/sas/readstat_xport_parse_format.c"
|
|
106
|
+
{
|
|
107
|
+
cs = xport_format_parse_start;
|
|
108
|
+
}
|
|
109
|
+
|
|
110
|
+
#line 111 "src/sas/readstat_xport_parse_format.c"
|
|
111
|
+
{
|
|
112
|
+
int _klen;
|
|
113
|
+
unsigned int _trans;
|
|
114
|
+
const char *_acts;
|
|
115
|
+
unsigned int _nacts;
|
|
116
|
+
const char *_keys;
|
|
117
|
+
|
|
118
|
+
if ( p == pe )
|
|
119
|
+
goto _test_eof;
|
|
120
|
+
if ( cs == 0 )
|
|
121
|
+
goto _out;
|
|
122
|
+
_resume:
|
|
123
|
+
_keys = _xport_format_parse_trans_keys + _xport_format_parse_key_offsets[cs];
|
|
124
|
+
_trans = _xport_format_parse_index_offsets[cs];
|
|
125
|
+
|
|
126
|
+
_klen = _xport_format_parse_single_lengths[cs];
|
|
127
|
+
if ( _klen > 0 ) {
|
|
128
|
+
const char *_lower = _keys;
|
|
129
|
+
const char *_mid;
|
|
130
|
+
const char *_upper = _keys + _klen - 1;
|
|
131
|
+
while (1) {
|
|
132
|
+
if ( _upper < _lower )
|
|
133
|
+
break;
|
|
134
|
+
|
|
135
|
+
_mid = _lower + ((_upper-_lower) >> 1);
|
|
136
|
+
if ( (*p) < *_mid )
|
|
137
|
+
_upper = _mid - 1;
|
|
138
|
+
else if ( (*p) > *_mid )
|
|
139
|
+
_lower = _mid + 1;
|
|
140
|
+
else {
|
|
141
|
+
_trans += (unsigned int)(_mid - _keys);
|
|
142
|
+
goto _match;
|
|
143
|
+
}
|
|
144
|
+
}
|
|
145
|
+
_keys += _klen;
|
|
146
|
+
_trans += _klen;
|
|
147
|
+
}
|
|
148
|
+
|
|
149
|
+
_klen = _xport_format_parse_range_lengths[cs];
|
|
150
|
+
if ( _klen > 0 ) {
|
|
151
|
+
const char *_lower = _keys;
|
|
152
|
+
const char *_mid;
|
|
153
|
+
const char *_upper = _keys + (_klen<<1) - 2;
|
|
154
|
+
while (1) {
|
|
155
|
+
if ( _upper < _lower )
|
|
156
|
+
break;
|
|
157
|
+
|
|
158
|
+
_mid = _lower + (((_upper-_lower) >> 1) & ~1);
|
|
159
|
+
if ( (*p) < _mid[0] )
|
|
160
|
+
_upper = _mid - 2;
|
|
161
|
+
else if ( (*p) > _mid[1] )
|
|
162
|
+
_lower = _mid + 2;
|
|
163
|
+
else {
|
|
164
|
+
_trans += (unsigned int)((_mid - _keys)>>1);
|
|
165
|
+
goto _match;
|
|
166
|
+
}
|
|
167
|
+
}
|
|
168
|
+
_trans += _klen;
|
|
169
|
+
}
|
|
170
|
+
|
|
171
|
+
_match:
|
|
172
|
+
_trans = _xport_format_parse_indicies[_trans];
|
|
173
|
+
cs = _xport_format_parse_trans_targs[_trans];
|
|
174
|
+
|
|
175
|
+
if ( _xport_format_parse_trans_actions[_trans] == 0 )
|
|
176
|
+
goto _again;
|
|
177
|
+
|
|
178
|
+
_acts = _xport_format_parse_actions + _xport_format_parse_trans_actions[_trans];
|
|
179
|
+
_nacts = (unsigned int) *_acts++;
|
|
180
|
+
while ( _nacts-- > 0 )
|
|
181
|
+
{
|
|
182
|
+
switch ( *_acts++ )
|
|
183
|
+
{
|
|
184
|
+
case 0:
|
|
185
|
+
#line 28 "src/sas/readstat_xport_parse_format.rl"
|
|
186
|
+
{
|
|
187
|
+
temp_val = 10 * temp_val + ((*p) - '0');
|
|
188
|
+
}
|
|
189
|
+
break;
|
|
190
|
+
case 1:
|
|
191
|
+
#line 32 "src/sas/readstat_xport_parse_format.rl"
|
|
192
|
+
{
|
|
193
|
+
parsed_len = p - data;
|
|
194
|
+
if (parsed_len < sizeof(fmt->name)) {
|
|
195
|
+
memcpy(fmt->name, data, parsed_len);
|
|
196
|
+
fmt->name[parsed_len] = '\0';
|
|
197
|
+
}
|
|
198
|
+
}
|
|
199
|
+
break;
|
|
200
|
+
case 2:
|
|
201
|
+
#line 40 "src/sas/readstat_xport_parse_format.rl"
|
|
202
|
+
{ temp_val = 0; }
|
|
203
|
+
break;
|
|
204
|
+
case 3:
|
|
205
|
+
#line 47 "src/sas/readstat_xport_parse_format.rl"
|
|
206
|
+
{ fmt->width = temp_val; }
|
|
207
|
+
break;
|
|
208
|
+
#line 209 "src/sas/readstat_xport_parse_format.c"
|
|
209
|
+
}
|
|
210
|
+
}
|
|
211
|
+
|
|
212
|
+
_again:
|
|
213
|
+
if ( cs == 0 )
|
|
214
|
+
goto _out;
|
|
215
|
+
if ( ++p != pe )
|
|
216
|
+
goto _resume;
|
|
217
|
+
_test_eof: {}
|
|
218
|
+
if ( p == eof )
|
|
219
|
+
{
|
|
220
|
+
const char *__acts = _xport_format_parse_actions + _xport_format_parse_eof_actions[cs];
|
|
221
|
+
unsigned int __nacts = (unsigned int) *__acts++;
|
|
222
|
+
while ( __nacts-- > 0 ) {
|
|
223
|
+
switch ( *__acts++ ) {
|
|
224
|
+
case 1:
|
|
225
|
+
#line 32 "src/sas/readstat_xport_parse_format.rl"
|
|
226
|
+
{
|
|
227
|
+
parsed_len = p - data;
|
|
228
|
+
if (parsed_len < sizeof(fmt->name)) {
|
|
229
|
+
memcpy(fmt->name, data, parsed_len);
|
|
230
|
+
fmt->name[parsed_len] = '\0';
|
|
231
|
+
}
|
|
232
|
+
}
|
|
233
|
+
break;
|
|
234
|
+
case 3:
|
|
235
|
+
#line 47 "src/sas/readstat_xport_parse_format.rl"
|
|
236
|
+
{ fmt->width = temp_val; }
|
|
237
|
+
break;
|
|
238
|
+
case 4:
|
|
239
|
+
#line 48 "src/sas/readstat_xport_parse_format.rl"
|
|
240
|
+
{ fmt->decimals = temp_val; }
|
|
241
|
+
break;
|
|
242
|
+
#line 243 "src/sas/readstat_xport_parse_format.c"
|
|
243
|
+
}
|
|
244
|
+
}
|
|
245
|
+
}
|
|
246
|
+
|
|
247
|
+
_out: {}
|
|
248
|
+
}
|
|
249
|
+
|
|
250
|
+
#line 54 "src/sas/readstat_xport_parse_format.rl"
|
|
251
|
+
|
|
252
|
+
|
|
253
|
+
if (cs < 3|| p != pe || parsed_len + 1 > sizeof(fmt->name)) {
|
|
254
|
+
char error_buf[1024];
|
|
255
|
+
if (error_handler) {
|
|
256
|
+
snprintf(error_buf, sizeof(error_buf), "Invalid format string (length=%d): %.*s", (int)len, (int)len, data);
|
|
257
|
+
error_handler(error_buf, user_ctx);
|
|
258
|
+
}
|
|
259
|
+
retval = READSTAT_ERROR_BAD_FORMAT_STRING;
|
|
260
|
+
}
|
|
261
|
+
|
|
262
|
+
(void)xport_format_parse_en_main;
|
|
263
|
+
|
|
264
|
+
return retval;
|
|
265
|
+
}
|
|
@@ -0,0 +1,68 @@
|
|
|
1
|
+
|
|
2
|
+
#include "../readstat.h"
|
|
3
|
+
#include "readstat_xport.h"
|
|
4
|
+
#include "readstat_xport_parse_format.h"
|
|
5
|
+
|
|
6
|
+
%%{
|
|
7
|
+
machine xport_format_parse;
|
|
8
|
+
write data nofinal noerror;
|
|
9
|
+
}%%
|
|
10
|
+
|
|
11
|
+
readstat_error_t xport_parse_format(const char *data, size_t len, xport_format_t *fmt,
|
|
12
|
+
readstat_error_handler error_handler, void *user_ctx) {
|
|
13
|
+
|
|
14
|
+
fmt->name[0] = '\0';
|
|
15
|
+
fmt->width = 0;
|
|
16
|
+
fmt->decimals = 0;
|
|
17
|
+
|
|
18
|
+
readstat_error_t retval = READSTAT_OK;
|
|
19
|
+
const char *p = data;
|
|
20
|
+
const char *pe = p + len;
|
|
21
|
+
const char *eof = pe;
|
|
22
|
+
|
|
23
|
+
int cs;
|
|
24
|
+
unsigned int temp_val = 0;
|
|
25
|
+
size_t parsed_len = 0;
|
|
26
|
+
|
|
27
|
+
%%{
|
|
28
|
+
action incr_val {
|
|
29
|
+
temp_val = 10 * temp_val + (fc - '0');
|
|
30
|
+
}
|
|
31
|
+
|
|
32
|
+
action write_name {
|
|
33
|
+
parsed_len = p - data;
|
|
34
|
+
if (parsed_len < sizeof(fmt->name)) {
|
|
35
|
+
memcpy(fmt->name, data, parsed_len);
|
|
36
|
+
fmt->name[parsed_len] = '\0';
|
|
37
|
+
}
|
|
38
|
+
}
|
|
39
|
+
|
|
40
|
+
integer = [0-9]+ >{ temp_val = 0; } $incr_val;
|
|
41
|
+
|
|
42
|
+
name = [A-Za-z_] ([A-Za-z0-9_]+ [A-Za-z_a])?;
|
|
43
|
+
|
|
44
|
+
char_name = '$' name? %write_name;
|
|
45
|
+
dbl_name = name? %write_name;
|
|
46
|
+
|
|
47
|
+
width = integer %{ fmt->width = temp_val; };
|
|
48
|
+
decimals = integer %{ fmt->decimals = temp_val; };
|
|
49
|
+
|
|
50
|
+
main := (char_name width? '.'?) | (dbl_name? (width? ('.' decimals?)?)?);
|
|
51
|
+
|
|
52
|
+
write init;
|
|
53
|
+
write exec;
|
|
54
|
+
}%%
|
|
55
|
+
|
|
56
|
+
if (cs < %%{ write first_final; }%%|| p != pe || parsed_len + 1 > sizeof(fmt->name)) {
|
|
57
|
+
char error_buf[1024];
|
|
58
|
+
if (error_handler) {
|
|
59
|
+
snprintf(error_buf, sizeof(error_buf), "Invalid format string (length=%d): %.*s", (int)len, (int)len, data);
|
|
60
|
+
error_handler(error_buf, user_ctx);
|
|
61
|
+
}
|
|
62
|
+
retval = READSTAT_ERROR_BAD_FORMAT_STRING;
|
|
63
|
+
}
|
|
64
|
+
|
|
65
|
+
(void)xport_format_parse_en_main;
|
|
66
|
+
|
|
67
|
+
return retval;
|
|
68
|
+
}
|