excelsior-formats 0.0.9
Sign up to get free protection for your applications and to get access to all the features.
- data/Rakefile +26 -0
- data/ext/excelsior_reader/excelsior_reader.c +547 -0
- data/ext/excelsior_reader/excelsior_reader.rl +155 -0
- data/ext/excelsior_reader/extconf.rb +5 -0
- data/lib/excelsior.rb +7 -0
- metadata +67 -0
data/Rakefile
ADDED
@@ -0,0 +1,26 @@
|
|
1
|
+
require 'rake'
|
2
|
+
require 'rake/testtask'
|
3
|
+
|
4
|
+
PKG_FILES = %w(Rakefile) +
|
5
|
+
Dir.glob("{lib}/**/*") +
|
6
|
+
Dir.glob("ext/**/*.{c,rb,rl}") +
|
7
|
+
%w[ext/excelsior_reader/excelsior_reader.c] # needed because they are generated later
|
8
|
+
|
9
|
+
gem_spec = Gem::Specification.new do |gem_spec|
|
10
|
+
gem_spec.name = 'excelsior'
|
11
|
+
gem_spec.version = '0.0.9'
|
12
|
+
gem_spec.summary = 'A Ruby gem that uses C bindings to read CSV files superfast. I\'m totally serial!'
|
13
|
+
gem_spec.description = 'A Ruby gem that uses C bindings to read CSV files superfast. I\'m totally serial!'
|
14
|
+
gem_spec.email = 'matt@toastyapps.com'
|
15
|
+
gem_spec.homepage = 'http://github.com/toastyapps/excelsior'
|
16
|
+
gem_spec.authors = ["Matthew Mongeau"]
|
17
|
+
gem_spec.files = PKG_FILES
|
18
|
+
gem_spec.extensions = FileList["ext/**/extconf.rb"].to_a
|
19
|
+
end
|
20
|
+
|
21
|
+
desc "Generate a gemspec file"
|
22
|
+
task :gemspec do
|
23
|
+
File.open("#{gem_spec.name}.gemspec", "w") do |f|
|
24
|
+
f.write gem_spec.to_yaml
|
25
|
+
end
|
26
|
+
end
|
@@ -0,0 +1,547 @@
|
|
1
|
+
|
2
|
+
#line 1 "excelsior_reader.rl"
|
3
|
+
#include <ruby.h>
|
4
|
+
|
5
|
+
static ID s_read;
|
6
|
+
|
7
|
+
int has_found = 0;
|
8
|
+
#define BUFSIZE 16384
|
9
|
+
|
10
|
+
|
11
|
+
#line 67 "excelsior_reader.rl"
|
12
|
+
|
13
|
+
|
14
|
+
|
15
|
+
#line 16 "excelsior_reader.c"
|
16
|
+
static const char _excelsior_scan_actions[] = {
|
17
|
+
0, 1, 0, 1, 2, 1, 4, 1,
|
18
|
+
5, 1, 6, 1, 7, 1, 11, 1,
|
19
|
+
12, 1, 13, 1, 14, 1, 15, 1,
|
20
|
+
19, 1, 20, 1, 21, 1, 22, 1,
|
21
|
+
23, 1, 27, 1, 28, 1, 29, 1,
|
22
|
+
30, 1, 31, 1, 35, 1, 36, 1,
|
23
|
+
37, 1, 38, 1, 39, 2, 0, 1,
|
24
|
+
2, 3, 8, 2, 3, 9, 2, 3,
|
25
|
+
10, 2, 3, 16, 2, 3, 17, 2,
|
26
|
+
3, 18, 2, 3, 24, 2, 3, 25,
|
27
|
+
2, 3, 26, 2, 3, 32, 2, 3,
|
28
|
+
33, 2, 3, 34
|
29
|
+
};
|
30
|
+
|
31
|
+
static const char _excelsior_scan_key_offsets[] = {
|
32
|
+
0, 0, 1, 2, 3, 4, 8, 15,
|
33
|
+
19, 20, 21, 28, 32, 33, 34, 41,
|
34
|
+
45, 46, 47, 54, 58, 59
|
35
|
+
};
|
36
|
+
|
37
|
+
static const char _excelsior_scan_trans_keys[] = {
|
38
|
+
34, 34, 34, 34, 9, 44, 59, 124,
|
39
|
+
10, 13, 32, 34, 44, 9, 12, 10,
|
40
|
+
13, 34, 44, 10, 34, 9, 10, 13,
|
41
|
+
32, 34, 11, 12, 13, 34, 9, 10,
|
42
|
+
10, 34, 10, 13, 32, 34, 124, 9,
|
43
|
+
12, 10, 13, 34, 124, 10, 34, 10,
|
44
|
+
13, 32, 34, 59, 9, 12, 10, 13,
|
45
|
+
34, 59, 10, 34, 0
|
46
|
+
};
|
47
|
+
|
48
|
+
static const char _excelsior_scan_single_lengths[] = {
|
49
|
+
0, 1, 1, 1, 1, 4, 5, 4,
|
50
|
+
1, 1, 5, 2, 1, 1, 5, 4,
|
51
|
+
1, 1, 5, 4, 1, 1
|
52
|
+
};
|
53
|
+
|
54
|
+
static const char _excelsior_scan_range_lengths[] = {
|
55
|
+
0, 0, 0, 0, 0, 0, 1, 0,
|
56
|
+
0, 0, 1, 1, 0, 0, 1, 0,
|
57
|
+
0, 0, 1, 0, 0, 0
|
58
|
+
};
|
59
|
+
|
60
|
+
static const char _excelsior_scan_index_offsets[] = {
|
61
|
+
0, 0, 2, 4, 6, 8, 13, 20,
|
62
|
+
25, 27, 29, 36, 40, 42, 44, 51,
|
63
|
+
56, 58, 60, 67, 72, 74
|
64
|
+
};
|
65
|
+
|
66
|
+
static const char _excelsior_scan_trans_targs[] = {
|
67
|
+
9, 1, 13, 2, 17, 3, 21, 4,
|
68
|
+
5, 5, 5, 5, 0, 6, 8, 7,
|
69
|
+
1, 6, 7, 7, 6, 6, 6, 6,
|
70
|
+
7, 6, 6, 1, 6, 10, 10, 12,
|
71
|
+
11, 2, 11, 11, 10, 10, 10, 11,
|
72
|
+
10, 10, 2, 10, 14, 16, 15, 3,
|
73
|
+
14, 15, 15, 14, 14, 14, 14, 15,
|
74
|
+
14, 14, 3, 14, 18, 20, 19, 4,
|
75
|
+
18, 19, 19, 18, 18, 18, 18, 19,
|
76
|
+
18, 18, 4, 18, 6, 10, 14, 18,
|
77
|
+
6, 6, 6, 10, 10, 10, 14, 14,
|
78
|
+
14, 18, 18, 18, 0
|
79
|
+
};
|
80
|
+
|
81
|
+
static const char _excelsior_scan_trans_actions[] = {
|
82
|
+
62, 0, 71, 0, 80, 0, 89, 0,
|
83
|
+
7, 5, 11, 9, 0, 13, 0, 56,
|
84
|
+
0, 15, 56, 59, 21, 21, 21, 21,
|
85
|
+
59, 13, 17, 0, 19, 25, 23, 0,
|
86
|
+
65, 0, 65, 68, 31, 31, 31, 68,
|
87
|
+
23, 27, 0, 29, 33, 0, 74, 0,
|
88
|
+
35, 74, 77, 41, 41, 41, 41, 77,
|
89
|
+
33, 37, 0, 39, 43, 0, 83, 0,
|
90
|
+
45, 83, 86, 51, 51, 51, 51, 86,
|
91
|
+
43, 47, 0, 49, 21, 31, 41, 51,
|
92
|
+
21, 17, 19, 31, 27, 29, 41, 37,
|
93
|
+
39, 51, 47, 49, 0
|
94
|
+
};
|
95
|
+
|
96
|
+
static const char _excelsior_scan_to_state_actions[] = {
|
97
|
+
0, 0, 0, 0, 0, 1, 53, 0,
|
98
|
+
0, 0, 53, 0, 0, 0, 53, 0,
|
99
|
+
0, 0, 53, 0, 0, 0
|
100
|
+
};
|
101
|
+
|
102
|
+
static const char _excelsior_scan_from_state_actions[] = {
|
103
|
+
0, 0, 0, 0, 0, 3, 3, 0,
|
104
|
+
0, 0, 3, 0, 0, 0, 3, 0,
|
105
|
+
0, 0, 3, 0, 0, 0
|
106
|
+
};
|
107
|
+
|
108
|
+
static const char _excelsior_scan_eof_trans[] = {
|
109
|
+
0, 81, 84, 87, 90, 0, 0, 81,
|
110
|
+
82, 83, 0, 84, 85, 86, 0, 87,
|
111
|
+
88, 89, 0, 90, 91, 92
|
112
|
+
};
|
113
|
+
|
114
|
+
static const int excelsior_scan_start = 5;
|
115
|
+
static const int excelsior_scan_error = 0;
|
116
|
+
|
117
|
+
static const int excelsior_scan_en_main = 5;
|
118
|
+
static const int excelsior_scan_en_csv = 6;
|
119
|
+
static const int excelsior_scan_en_tsv = 10;
|
120
|
+
static const int excelsior_scan_en_psv = 14;
|
121
|
+
static const int excelsior_scan_en_ssv = 18;
|
122
|
+
|
123
|
+
|
124
|
+
#line 70 "excelsior_reader.rl"
|
125
|
+
|
126
|
+
|
127
|
+
VALUE e_rows(int argc, VALUE *argv, VALUE self) {
|
128
|
+
|
129
|
+
int cs, act, have = 0, nread = 0, curline = 1, text = 0;
|
130
|
+
char *ts = 0, *te = 0, *buf = NULL, *eof = NULL;
|
131
|
+
char *p, *pe;
|
132
|
+
int buffer_size = BUFSIZE;
|
133
|
+
|
134
|
+
has_found = 0;
|
135
|
+
VALUE arr;
|
136
|
+
VALUE io;
|
137
|
+
VALUE format;
|
138
|
+
int is_io = 0;
|
139
|
+
int done = 0;
|
140
|
+
int first_run = 1;
|
141
|
+
arr = rb_ary_new();
|
142
|
+
rb_scan_args(argc, argv, "11", &io, &format);
|
143
|
+
if(NIL_P(format)) format = rb_str_new2(",");
|
144
|
+
|
145
|
+
|
146
|
+
#line 147 "excelsior_reader.c"
|
147
|
+
{
|
148
|
+
cs = excelsior_scan_start;
|
149
|
+
ts = 0;
|
150
|
+
te = 0;
|
151
|
+
act = 0;
|
152
|
+
}
|
153
|
+
|
154
|
+
#line 91 "excelsior_reader.rl"
|
155
|
+
|
156
|
+
is_io = rb_respond_to(io, s_read);
|
157
|
+
buf = (char *) malloc(buffer_size); //ALLOC_N(char, buffer_size); <= This caused problems
|
158
|
+
|
159
|
+
while(!done) {
|
160
|
+
|
161
|
+
int len, space = buffer_size - have;
|
162
|
+
VALUE str;
|
163
|
+
p = buf + have;
|
164
|
+
|
165
|
+
if(is_io) {
|
166
|
+
str = rb_funcall(io, s_read, 1, INT2FIX(space));
|
167
|
+
if(first_run) {
|
168
|
+
str = rb_str_append(format, str);
|
169
|
+
first_run = 0;
|
170
|
+
}
|
171
|
+
len = RSTRING_LEN(str);
|
172
|
+
p = memcpy(p, StringValuePtr(str), len);
|
173
|
+
} else {
|
174
|
+
// Going to assume it's a string and already in memory
|
175
|
+
//str = io;
|
176
|
+
io = rb_str_append(format, io);
|
177
|
+
p = RSTRING_PTR(io);
|
178
|
+
len = RSTRING_LEN(io);
|
179
|
+
pe = p + len;
|
180
|
+
eof = pe;
|
181
|
+
done = 1;
|
182
|
+
}
|
183
|
+
if(len < space) {
|
184
|
+
done = 1;
|
185
|
+
//p[len++] = 0; can't seem to get it to work with this
|
186
|
+
pe = p + len;
|
187
|
+
eof = pe;
|
188
|
+
} else {
|
189
|
+
pe = p + len;
|
190
|
+
}
|
191
|
+
|
192
|
+
|
193
|
+
#line 194 "excelsior_reader.c"
|
194
|
+
{
|
195
|
+
int _klen;
|
196
|
+
unsigned int _trans;
|
197
|
+
const char *_acts;
|
198
|
+
unsigned int _nacts;
|
199
|
+
const char *_keys;
|
200
|
+
|
201
|
+
if ( p == pe )
|
202
|
+
goto _test_eof;
|
203
|
+
if ( cs == 0 )
|
204
|
+
goto _out;
|
205
|
+
_resume:
|
206
|
+
_acts = _excelsior_scan_actions + _excelsior_scan_from_state_actions[cs];
|
207
|
+
_nacts = (unsigned int) *_acts++;
|
208
|
+
while ( _nacts-- > 0 ) {
|
209
|
+
switch ( *_acts++ ) {
|
210
|
+
case 2:
|
211
|
+
#line 1 "excelsior_reader.rl"
|
212
|
+
{ts = p;}
|
213
|
+
break;
|
214
|
+
#line 215 "excelsior_reader.c"
|
215
|
+
}
|
216
|
+
}
|
217
|
+
|
218
|
+
_keys = _excelsior_scan_trans_keys + _excelsior_scan_key_offsets[cs];
|
219
|
+
_trans = _excelsior_scan_index_offsets[cs];
|
220
|
+
|
221
|
+
_klen = _excelsior_scan_single_lengths[cs];
|
222
|
+
if ( _klen > 0 ) {
|
223
|
+
const char *_lower = _keys;
|
224
|
+
const char *_mid;
|
225
|
+
const char *_upper = _keys + _klen - 1;
|
226
|
+
while (1) {
|
227
|
+
if ( _upper < _lower )
|
228
|
+
break;
|
229
|
+
|
230
|
+
_mid = _lower + ((_upper-_lower) >> 1);
|
231
|
+
if ( (*p) < *_mid )
|
232
|
+
_upper = _mid - 1;
|
233
|
+
else if ( (*p) > *_mid )
|
234
|
+
_lower = _mid + 1;
|
235
|
+
else {
|
236
|
+
_trans += (_mid - _keys);
|
237
|
+
goto _match;
|
238
|
+
}
|
239
|
+
}
|
240
|
+
_keys += _klen;
|
241
|
+
_trans += _klen;
|
242
|
+
}
|
243
|
+
|
244
|
+
_klen = _excelsior_scan_range_lengths[cs];
|
245
|
+
if ( _klen > 0 ) {
|
246
|
+
const char *_lower = _keys;
|
247
|
+
const char *_mid;
|
248
|
+
const char *_upper = _keys + (_klen<<1) - 2;
|
249
|
+
while (1) {
|
250
|
+
if ( _upper < _lower )
|
251
|
+
break;
|
252
|
+
|
253
|
+
_mid = _lower + (((_upper-_lower) >> 1) & ~1);
|
254
|
+
if ( (*p) < _mid[0] )
|
255
|
+
_upper = _mid - 2;
|
256
|
+
else if ( (*p) > _mid[1] )
|
257
|
+
_lower = _mid + 2;
|
258
|
+
else {
|
259
|
+
_trans += ((_mid - _keys)>>1);
|
260
|
+
goto _match;
|
261
|
+
}
|
262
|
+
}
|
263
|
+
_trans += _klen;
|
264
|
+
}
|
265
|
+
|
266
|
+
_match:
|
267
|
+
_eof_trans:
|
268
|
+
cs = _excelsior_scan_trans_targs[_trans];
|
269
|
+
|
270
|
+
if ( _excelsior_scan_trans_actions[_trans] == 0 )
|
271
|
+
goto _again;
|
272
|
+
|
273
|
+
_acts = _excelsior_scan_actions + _excelsior_scan_trans_actions[_trans];
|
274
|
+
_nacts = (unsigned int) *_acts++;
|
275
|
+
while ( _nacts-- > 0 )
|
276
|
+
{
|
277
|
+
switch ( *_acts++ )
|
278
|
+
{
|
279
|
+
case 3:
|
280
|
+
#line 1 "excelsior_reader.rl"
|
281
|
+
{te = p+1;}
|
282
|
+
break;
|
283
|
+
case 4:
|
284
|
+
#line 32 "excelsior_reader.rl"
|
285
|
+
{te = p+1;{ {cs = 6; goto _again;} }}
|
286
|
+
break;
|
287
|
+
case 5:
|
288
|
+
#line 33 "excelsior_reader.rl"
|
289
|
+
{te = p+1;{ {cs = 10; goto _again;} }}
|
290
|
+
break;
|
291
|
+
case 6:
|
292
|
+
#line 34 "excelsior_reader.rl"
|
293
|
+
{te = p+1;{ {cs = 14; goto _again;} }}
|
294
|
+
break;
|
295
|
+
case 7:
|
296
|
+
#line 35 "excelsior_reader.rl"
|
297
|
+
{te = p+1;{ {cs = 18; goto _again;} }}
|
298
|
+
break;
|
299
|
+
case 8:
|
300
|
+
#line 40 "excelsior_reader.rl"
|
301
|
+
{act = 6;}
|
302
|
+
break;
|
303
|
+
case 9:
|
304
|
+
#line 41 "excelsior_reader.rl"
|
305
|
+
{act = 7;}
|
306
|
+
break;
|
307
|
+
case 10:
|
308
|
+
#line 42 "excelsior_reader.rl"
|
309
|
+
{act = 8;}
|
310
|
+
break;
|
311
|
+
case 11:
|
312
|
+
#line 39 "excelsior_reader.rl"
|
313
|
+
{te = p+1;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
314
|
+
break;
|
315
|
+
case 12:
|
316
|
+
#line 43 "excelsior_reader.rl"
|
317
|
+
{te = p+1;{ if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;}}
|
318
|
+
break;
|
319
|
+
case 13:
|
320
|
+
#line 39 "excelsior_reader.rl"
|
321
|
+
{te = p;p--;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
322
|
+
break;
|
323
|
+
case 14:
|
324
|
+
#line 42 "excelsior_reader.rl"
|
325
|
+
{te = p;p--;{ rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}}
|
326
|
+
break;
|
327
|
+
case 15:
|
328
|
+
#line 1 "excelsior_reader.rl"
|
329
|
+
{ switch( act ) {
|
330
|
+
case 0:
|
331
|
+
{{cs = 0; goto _again;}}
|
332
|
+
break;
|
333
|
+
case 7:
|
334
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;}
|
335
|
+
break;
|
336
|
+
case 8:
|
337
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}
|
338
|
+
break;
|
339
|
+
default:
|
340
|
+
{{p = ((te))-1;}}
|
341
|
+
break;
|
342
|
+
}
|
343
|
+
}
|
344
|
+
break;
|
345
|
+
case 16:
|
346
|
+
#line 47 "excelsior_reader.rl"
|
347
|
+
{act = 11;}
|
348
|
+
break;
|
349
|
+
case 17:
|
350
|
+
#line 48 "excelsior_reader.rl"
|
351
|
+
{act = 12;}
|
352
|
+
break;
|
353
|
+
case 18:
|
354
|
+
#line 49 "excelsior_reader.rl"
|
355
|
+
{act = 13;}
|
356
|
+
break;
|
357
|
+
case 19:
|
358
|
+
#line 46 "excelsior_reader.rl"
|
359
|
+
{te = p+1;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
360
|
+
break;
|
361
|
+
case 20:
|
362
|
+
#line 47 "excelsior_reader.rl"
|
363
|
+
{te = p+1;}
|
364
|
+
break;
|
365
|
+
case 21:
|
366
|
+
#line 46 "excelsior_reader.rl"
|
367
|
+
{te = p;p--;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
368
|
+
break;
|
369
|
+
case 22:
|
370
|
+
#line 49 "excelsior_reader.rl"
|
371
|
+
{te = p;p--;{ rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}}
|
372
|
+
break;
|
373
|
+
case 23:
|
374
|
+
#line 1 "excelsior_reader.rl"
|
375
|
+
{ switch( act ) {
|
376
|
+
case 0:
|
377
|
+
{{cs = 0; goto _again;}}
|
378
|
+
break;
|
379
|
+
case 12:
|
380
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;}
|
381
|
+
break;
|
382
|
+
case 13:
|
383
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}
|
384
|
+
break;
|
385
|
+
default:
|
386
|
+
{{p = ((te))-1;}}
|
387
|
+
break;
|
388
|
+
}
|
389
|
+
}
|
390
|
+
break;
|
391
|
+
case 24:
|
392
|
+
#line 54 "excelsior_reader.rl"
|
393
|
+
{act = 16;}
|
394
|
+
break;
|
395
|
+
case 25:
|
396
|
+
#line 55 "excelsior_reader.rl"
|
397
|
+
{act = 17;}
|
398
|
+
break;
|
399
|
+
case 26:
|
400
|
+
#line 56 "excelsior_reader.rl"
|
401
|
+
{act = 18;}
|
402
|
+
break;
|
403
|
+
case 27:
|
404
|
+
#line 53 "excelsior_reader.rl"
|
405
|
+
{te = p+1;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
406
|
+
break;
|
407
|
+
case 28:
|
408
|
+
#line 57 "excelsior_reader.rl"
|
409
|
+
{te = p+1;{ if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;}}
|
410
|
+
break;
|
411
|
+
case 29:
|
412
|
+
#line 53 "excelsior_reader.rl"
|
413
|
+
{te = p;p--;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
414
|
+
break;
|
415
|
+
case 30:
|
416
|
+
#line 56 "excelsior_reader.rl"
|
417
|
+
{te = p;p--;{ rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}}
|
418
|
+
break;
|
419
|
+
case 31:
|
420
|
+
#line 1 "excelsior_reader.rl"
|
421
|
+
{ switch( act ) {
|
422
|
+
case 0:
|
423
|
+
{{cs = 0; goto _again;}}
|
424
|
+
break;
|
425
|
+
case 17:
|
426
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;}
|
427
|
+
break;
|
428
|
+
case 18:
|
429
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}
|
430
|
+
break;
|
431
|
+
default:
|
432
|
+
{{p = ((te))-1;}}
|
433
|
+
break;
|
434
|
+
}
|
435
|
+
}
|
436
|
+
break;
|
437
|
+
case 32:
|
438
|
+
#line 61 "excelsior_reader.rl"
|
439
|
+
{act = 21;}
|
440
|
+
break;
|
441
|
+
case 33:
|
442
|
+
#line 62 "excelsior_reader.rl"
|
443
|
+
{act = 22;}
|
444
|
+
break;
|
445
|
+
case 34:
|
446
|
+
#line 63 "excelsior_reader.rl"
|
447
|
+
{act = 23;}
|
448
|
+
break;
|
449
|
+
case 35:
|
450
|
+
#line 60 "excelsior_reader.rl"
|
451
|
+
{te = p+1;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
452
|
+
break;
|
453
|
+
case 36:
|
454
|
+
#line 64 "excelsior_reader.rl"
|
455
|
+
{te = p+1;{ if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;}}
|
456
|
+
break;
|
457
|
+
case 37:
|
458
|
+
#line 60 "excelsior_reader.rl"
|
459
|
+
{te = p;p--;{ if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; }}
|
460
|
+
break;
|
461
|
+
case 38:
|
462
|
+
#line 63 "excelsior_reader.rl"
|
463
|
+
{te = p;p--;{ rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}}
|
464
|
+
break;
|
465
|
+
case 39:
|
466
|
+
#line 1 "excelsior_reader.rl"
|
467
|
+
{ switch( act ) {
|
468
|
+
case 0:
|
469
|
+
{{cs = 0; goto _again;}}
|
470
|
+
break;
|
471
|
+
case 22:
|
472
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;}
|
473
|
+
break;
|
474
|
+
case 23:
|
475
|
+
{{p = ((te))-1;} rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;}
|
476
|
+
break;
|
477
|
+
default:
|
478
|
+
{{p = ((te))-1;}}
|
479
|
+
break;
|
480
|
+
}
|
481
|
+
}
|
482
|
+
break;
|
483
|
+
#line 484 "excelsior_reader.c"
|
484
|
+
}
|
485
|
+
}
|
486
|
+
|
487
|
+
_again:
|
488
|
+
_acts = _excelsior_scan_actions + _excelsior_scan_to_state_actions[cs];
|
489
|
+
_nacts = (unsigned int) *_acts++;
|
490
|
+
while ( _nacts-- > 0 ) {
|
491
|
+
switch ( *_acts++ ) {
|
492
|
+
case 0:
|
493
|
+
#line 1 "excelsior_reader.rl"
|
494
|
+
{ts = 0;}
|
495
|
+
break;
|
496
|
+
case 1:
|
497
|
+
#line 1 "excelsior_reader.rl"
|
498
|
+
{act = 0;}
|
499
|
+
break;
|
500
|
+
#line 501 "excelsior_reader.c"
|
501
|
+
}
|
502
|
+
}
|
503
|
+
|
504
|
+
if ( cs == 0 )
|
505
|
+
goto _out;
|
506
|
+
if ( ++p != pe )
|
507
|
+
goto _resume;
|
508
|
+
_test_eof: {}
|
509
|
+
if ( p == eof )
|
510
|
+
{
|
511
|
+
if ( _excelsior_scan_eof_trans[cs] > 0 ) {
|
512
|
+
_trans = _excelsior_scan_eof_trans[cs] - 1;
|
513
|
+
goto _eof_trans;
|
514
|
+
}
|
515
|
+
}
|
516
|
+
|
517
|
+
_out: {}
|
518
|
+
}
|
519
|
+
|
520
|
+
#line 129 "excelsior_reader.rl"
|
521
|
+
|
522
|
+
|
523
|
+
if(ts != 0) { // we are not at the end
|
524
|
+
have = pe - ts; //so copy stuff back in
|
525
|
+
memmove(buf, ts, have);
|
526
|
+
te = buf + (te - ts);
|
527
|
+
ts = buf;
|
528
|
+
}
|
529
|
+
|
530
|
+
}
|
531
|
+
|
532
|
+
if(RARRAY_LEN(arr) > 0) { // have a last array to yield
|
533
|
+
rb_yield(arr);
|
534
|
+
}
|
535
|
+
|
536
|
+
return Qnil;
|
537
|
+
}
|
538
|
+
|
539
|
+
VALUE mExcelsior;
|
540
|
+
VALUE cReader;
|
541
|
+
|
542
|
+
void Init_excelsior_reader() {
|
543
|
+
s_read = rb_intern("read");
|
544
|
+
mExcelsior = rb_define_module("Excelsior");
|
545
|
+
cReader = rb_define_class_under(mExcelsior, "Reader", rb_cObject);
|
546
|
+
rb_define_singleton_method(cReader, "rows", e_rows, -1);
|
547
|
+
}
|
@@ -0,0 +1,155 @@
|
|
1
|
+
#include <ruby.h>
|
2
|
+
|
3
|
+
static ID s_read;
|
4
|
+
|
5
|
+
int has_found = 0;
|
6
|
+
#define BUFSIZE 16384
|
7
|
+
|
8
|
+
%%{
|
9
|
+
machine excelsior_scan;
|
10
|
+
|
11
|
+
csv_delimiter = ',';
|
12
|
+
tsv_delimiter = '\t';
|
13
|
+
psv_delimiter = '|';
|
14
|
+
ssv_delimiter = ';';
|
15
|
+
|
16
|
+
|
17
|
+
newline = "\r"? "\n" | "\r";
|
18
|
+
string_character = any - '"';
|
19
|
+
string = '"' (string_character | '""')* '"' ;
|
20
|
+
|
21
|
+
csv_letter = string_character - csv_delimiter - newline;
|
22
|
+
tsv_letter = string_character - tsv_delimiter - newline;
|
23
|
+
psv_letter = string_character - psv_delimiter - newline;
|
24
|
+
ssv_letter = string_character - ssv_delimiter - newline;
|
25
|
+
|
26
|
+
csv_value = csv_letter+;
|
27
|
+
tsv_value = tsv_letter+;
|
28
|
+
psv_value = psv_letter+;
|
29
|
+
ssv_value = ssv_letter+;
|
30
|
+
|
31
|
+
main := |*
|
32
|
+
csv_delimiter { fgoto csv; };
|
33
|
+
tsv_delimiter { fgoto tsv; };
|
34
|
+
psv_delimiter { fgoto psv; };
|
35
|
+
ssv_delimiter { fgoto ssv; };
|
36
|
+
*|;
|
37
|
+
|
38
|
+
csv := |*
|
39
|
+
newline { if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; };
|
40
|
+
space;
|
41
|
+
csv_value { rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;};
|
42
|
+
string { rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;};
|
43
|
+
csv_delimiter { if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;};
|
44
|
+
*|;
|
45
|
+
tsv := |*
|
46
|
+
newline { if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; };
|
47
|
+
space;
|
48
|
+
tsv_value { rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;};
|
49
|
+
string { rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;};
|
50
|
+
tsv_delimiter { if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;};
|
51
|
+
*|;
|
52
|
+
psv := |*
|
53
|
+
newline { if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; };
|
54
|
+
space;
|
55
|
+
psv_value { rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;};
|
56
|
+
string { rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;};
|
57
|
+
psv_delimiter { if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;};
|
58
|
+
*|;
|
59
|
+
ssv := |*
|
60
|
+
newline { if(has_found ==0) rb_ary_push(arr, Qnil); rb_yield(arr); arr = rb_ary_new(); has_found = 0; };
|
61
|
+
space;
|
62
|
+
ssv_value { rb_ary_push(arr, rb_str_new(ts, te - ts)); has_found = 1;};
|
63
|
+
string { rb_ary_push(arr, rb_str_new(ts + 1, te - ts - 2)); has_found = 1;};
|
64
|
+
ssv_delimiter { if(has_found == 0) rb_ary_push(arr, Qnil); has_found = 0;};
|
65
|
+
*|;
|
66
|
+
|
67
|
+
}%%
|
68
|
+
|
69
|
+
%% write data nofinal;
|
70
|
+
|
71
|
+
|
72
|
+
VALUE e_rows(int argc, VALUE *argv, VALUE self) {
|
73
|
+
|
74
|
+
int cs, act, have = 0, nread = 0, curline = 1, text = 0;
|
75
|
+
char *ts = 0, *te = 0, *buf = NULL, *eof = NULL;
|
76
|
+
char *p, *pe;
|
77
|
+
int buffer_size = BUFSIZE;
|
78
|
+
|
79
|
+
has_found = 0;
|
80
|
+
VALUE arr;
|
81
|
+
VALUE io;
|
82
|
+
VALUE format;
|
83
|
+
int is_io = 0;
|
84
|
+
int done = 0;
|
85
|
+
int first_run = 1;
|
86
|
+
arr = rb_ary_new();
|
87
|
+
rb_scan_args(argc, argv, "11", &io, &format);
|
88
|
+
if(NIL_P(format)) format = rb_str_new2(",");
|
89
|
+
|
90
|
+
%% write init;
|
91
|
+
|
92
|
+
is_io = rb_respond_to(io, s_read);
|
93
|
+
buf = (char *) malloc(buffer_size); //ALLOC_N(char, buffer_size); <= This caused problems
|
94
|
+
|
95
|
+
while(!done) {
|
96
|
+
|
97
|
+
int len, space = buffer_size - have;
|
98
|
+
VALUE str;
|
99
|
+
p = buf + have;
|
100
|
+
|
101
|
+
if(is_io) {
|
102
|
+
str = rb_funcall(io, s_read, 1, INT2FIX(space));
|
103
|
+
if(first_run) {
|
104
|
+
str = rb_str_append(format, str);
|
105
|
+
first_run = 0;
|
106
|
+
}
|
107
|
+
len = RSTRING_LEN(str);
|
108
|
+
p = memcpy(p, StringValuePtr(str), len);
|
109
|
+
} else {
|
110
|
+
// Going to assume it's a string and already in memory
|
111
|
+
//str = io;
|
112
|
+
io = rb_str_append(format, io);
|
113
|
+
p = RSTRING_PTR(io);
|
114
|
+
len = RSTRING_LEN(io);
|
115
|
+
pe = p + len;
|
116
|
+
eof = pe;
|
117
|
+
done = 1;
|
118
|
+
}
|
119
|
+
if(len < space) {
|
120
|
+
done = 1;
|
121
|
+
//p[len++] = 0; can't seem to get it to work with this
|
122
|
+
pe = p + len;
|
123
|
+
eof = pe;
|
124
|
+
} else {
|
125
|
+
pe = p + len;
|
126
|
+
}
|
127
|
+
|
128
|
+
%% write exec;
|
129
|
+
|
130
|
+
|
131
|
+
if(ts != 0) { // we are not at the end
|
132
|
+
have = pe - ts; //so copy stuff back in
|
133
|
+
memmove(buf, ts, have);
|
134
|
+
te = buf + (te - ts);
|
135
|
+
ts = buf;
|
136
|
+
}
|
137
|
+
|
138
|
+
}
|
139
|
+
|
140
|
+
if(RARRAY_LEN(arr) > 0) { // have a last array to yield
|
141
|
+
rb_yield(arr);
|
142
|
+
}
|
143
|
+
|
144
|
+
return Qnil;
|
145
|
+
}
|
146
|
+
|
147
|
+
VALUE mExcelsior;
|
148
|
+
VALUE cReader;
|
149
|
+
|
150
|
+
void Init_excelsior_reader() {
|
151
|
+
s_read = rb_intern("read");
|
152
|
+
mExcelsior = rb_define_module("Excelsior");
|
153
|
+
cReader = rb_define_class_under(mExcelsior, "Reader", rb_cObject);
|
154
|
+
rb_define_singleton_method(cReader, "rows", e_rows, -1);
|
155
|
+
}
|
data/lib/excelsior.rb
ADDED
metadata
ADDED
@@ -0,0 +1,67 @@
|
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
|
+
name: excelsior-formats
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
segments:
|
5
|
+
- 0
|
6
|
+
- 0
|
7
|
+
- 9
|
8
|
+
version: 0.0.9
|
9
|
+
platform: ruby
|
10
|
+
authors:
|
11
|
+
- Matthew Mongeau
|
12
|
+
autorequire:
|
13
|
+
bindir: bin
|
14
|
+
cert_chain: []
|
15
|
+
|
16
|
+
date: 2009-12-10 06:00:00 +01:00
|
17
|
+
default_executable:
|
18
|
+
dependencies: []
|
19
|
+
|
20
|
+
description: A Ruby gem that uses C bindings to read CSV files superfast. I'm totally serial!
|
21
|
+
email: matt@toastyapps.com
|
22
|
+
executables: []
|
23
|
+
|
24
|
+
extensions:
|
25
|
+
- ext/excelsior_reader/extconf.rb
|
26
|
+
extra_rdoc_files: []
|
27
|
+
|
28
|
+
files:
|
29
|
+
- Rakefile
|
30
|
+
- lib/excelsior.rb
|
31
|
+
- ext/excelsior_reader/excelsior_reader.c
|
32
|
+
- ext/excelsior_reader/extconf.rb
|
33
|
+
- ext/excelsior_reader/excelsior_reader.rl
|
34
|
+
has_rdoc: true
|
35
|
+
homepage: http://github.com/toastyapps/excelsior
|
36
|
+
licenses: []
|
37
|
+
|
38
|
+
post_install_message:
|
39
|
+
rdoc_options: []
|
40
|
+
|
41
|
+
require_paths:
|
42
|
+
- lib
|
43
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
44
|
+
requirements:
|
45
|
+
- - ">="
|
46
|
+
- !ruby/object:Gem::Version
|
47
|
+
segments:
|
48
|
+
- 0
|
49
|
+
version: "0"
|
50
|
+
version:
|
51
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
52
|
+
requirements:
|
53
|
+
- - ">="
|
54
|
+
- !ruby/object:Gem::Version
|
55
|
+
segments:
|
56
|
+
- 0
|
57
|
+
version: "0"
|
58
|
+
version:
|
59
|
+
requirements: []
|
60
|
+
|
61
|
+
rubyforge_project:
|
62
|
+
rubygems_version: 1.3.6
|
63
|
+
signing_key:
|
64
|
+
specification_version: 3
|
65
|
+
summary: A Ruby gem that uses C bindings to read CSV files superfast. I'm totally serial!
|
66
|
+
test_files: []
|
67
|
+
|