pg 0.18.0 → 1.1.4

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.
Files changed (80) hide show
  1. checksums.yaml +5 -5
  2. checksums.yaml.gz.sig +0 -0
  3. data/BSDL +2 -2
  4. data/ChangeLog +1221 -4
  5. data/History.rdoc +200 -0
  6. data/Manifest.txt +5 -18
  7. data/README-Windows.rdoc +15 -26
  8. data/README.rdoc +27 -10
  9. data/Rakefile +33 -24
  10. data/Rakefile.cross +57 -39
  11. data/ext/errorcodes.def +37 -0
  12. data/ext/errorcodes.rb +1 -1
  13. data/ext/errorcodes.txt +16 -1
  14. data/ext/extconf.rb +29 -35
  15. data/ext/gvl_wrappers.c +4 -0
  16. data/ext/gvl_wrappers.h +27 -39
  17. data/ext/pg.c +27 -53
  18. data/ext/pg.h +66 -83
  19. data/ext/pg_binary_decoder.c +75 -6
  20. data/ext/pg_binary_encoder.c +14 -12
  21. data/ext/pg_coder.c +83 -13
  22. data/ext/pg_connection.c +627 -351
  23. data/ext/pg_copy_coder.c +44 -9
  24. data/ext/pg_result.c +364 -134
  25. data/ext/pg_text_decoder.c +605 -46
  26. data/ext/pg_text_encoder.c +95 -76
  27. data/ext/pg_tuple.c +541 -0
  28. data/ext/pg_type_map.c +20 -13
  29. data/ext/pg_type_map_by_column.c +7 -7
  30. data/ext/pg_type_map_by_mri_type.c +2 -2
  31. data/ext/pg_type_map_in_ruby.c +4 -7
  32. data/ext/util.c +7 -7
  33. data/ext/util.h +3 -3
  34. data/lib/pg/basic_type_mapping.rb +105 -45
  35. data/lib/pg/binary_decoder.rb +22 -0
  36. data/lib/pg/coder.rb +1 -1
  37. data/lib/pg/connection.rb +109 -39
  38. data/lib/pg/constants.rb +1 -1
  39. data/lib/pg/exceptions.rb +1 -1
  40. data/lib/pg/result.rb +11 -6
  41. data/lib/pg/text_decoder.rb +25 -20
  42. data/lib/pg/text_encoder.rb +43 -1
  43. data/lib/pg/tuple.rb +30 -0
  44. data/lib/pg/type_map_by_column.rb +1 -1
  45. data/lib/pg.rb +21 -11
  46. data/spec/helpers.rb +50 -25
  47. data/spec/pg/basic_type_mapping_spec.rb +287 -30
  48. data/spec/pg/connection_spec.rb +695 -282
  49. data/spec/pg/connection_sync_spec.rb +41 -0
  50. data/spec/pg/result_spec.rb +59 -17
  51. data/spec/pg/tuple_spec.rb +280 -0
  52. data/spec/pg/type_map_by_class_spec.rb +3 -3
  53. data/spec/pg/type_map_by_column_spec.rb +1 -1
  54. data/spec/pg/type_map_by_mri_type_spec.rb +2 -2
  55. data/spec/pg/type_map_by_oid_spec.rb +1 -1
  56. data/spec/pg/type_map_in_ruby_spec.rb +1 -1
  57. data/spec/pg/type_map_spec.rb +1 -1
  58. data/spec/pg/type_spec.rb +319 -35
  59. data/spec/pg_spec.rb +2 -2
  60. data.tar.gz.sig +0 -0
  61. metadata +68 -68
  62. metadata.gz.sig +0 -0
  63. data/sample/array_insert.rb +0 -20
  64. data/sample/async_api.rb +0 -106
  65. data/sample/async_copyto.rb +0 -39
  66. data/sample/async_mixed.rb +0 -56
  67. data/sample/check_conn.rb +0 -21
  68. data/sample/copyfrom.rb +0 -81
  69. data/sample/copyto.rb +0 -19
  70. data/sample/cursor.rb +0 -21
  71. data/sample/disk_usage_report.rb +0 -186
  72. data/sample/issue-119.rb +0 -94
  73. data/sample/losample.rb +0 -69
  74. data/sample/minimal-testcase.rb +0 -17
  75. data/sample/notify_wait.rb +0 -72
  76. data/sample/pg_statistics.rb +0 -294
  77. data/sample/replication_monitor.rb +0 -231
  78. data/sample/test_binary_values.rb +0 -33
  79. data/sample/wal_shipper.rb +0 -434
  80. data/sample/warehouse_partitions.rb +0 -320
@@ -1,6 +1,6 @@
1
1
  /*
2
2
  * pg_text_decoder.c - PG::TextDecoder module
3
- * $Id: pg_text_decoder.c,v b7f9daeeba29 2014/11/21 19:53:47 lars $
3
+ * $Id: pg_text_decoder.c,v cee615e0ea2c 2018/07/30 05:27:05 lars $
4
4
  *
5
5
  */
6
6
 
@@ -12,13 +12,13 @@
12
12
  * assigns a decoder function.
13
13
  *
14
14
  * Signature of all type cast decoders is:
15
- * VALUE decoder_function(t_pg_coder *this, char *val, int len, int tuple, int field, int enc_idx)
15
+ * VALUE decoder_function(t_pg_coder *this, const char *val, int len, int tuple, int field, int enc_idx)
16
16
  *
17
17
  * Params:
18
18
  * this - The data part of the coder object that belongs to the decoder function.
19
- * val, len - The text or binary data to decode. The caller ensures, that the data is
20
- * zero terminated ( that is val[len] = 0 ). The memory should be used read
21
- * only by the callee.
19
+ * val, len - The text or binary data to decode.
20
+ * The caller ensures, that text data (format=0) is zero terminated so that val[len]=0.
21
+ * The memory should be used read-only by the callee.
22
22
  * tuple - Row of the value within the result set.
23
23
  * field - Column of the value within the result set.
24
24
  * enc_idx - Index of the Encoding that any output String should get assigned.
@@ -28,13 +28,39 @@
28
28
  *
29
29
  */
30
30
 
31
+ #include "ruby/version.h"
31
32
  #include "pg.h"
32
33
  #include "util.h"
34
+ #ifdef HAVE_INTTYPES_H
33
35
  #include <inttypes.h>
36
+ #endif
37
+ #include <ctype.h>
38
+ #include <time.h>
39
+ #if !defined(_WIN32)
40
+ #include <arpa/inet.h>
41
+ #include <sys/socket.h>
42
+ #endif
43
+ #include <string.h>
34
44
 
35
45
  VALUE rb_mPG_TextDecoder;
36
46
  static ID s_id_decode;
37
-
47
+ static ID s_id_Rational;
48
+ static ID s_id_new;
49
+ static ID s_id_utc;
50
+ static ID s_id_getlocal;
51
+ static ID s_id_BigDecimal;
52
+
53
+ static VALUE s_IPAddr;
54
+ static VALUE s_vmasks4;
55
+ static VALUE s_vmasks6;
56
+ static VALUE s_nan, s_pos_inf, s_neg_inf;
57
+ static int use_ipaddr_alloc;
58
+ static ID s_id_lshift;
59
+ static ID s_id_add;
60
+ static ID s_id_mask;
61
+ static ID s_ivar_family;
62
+ static ID s_ivar_addr;
63
+ static ID s_ivar_mask_addr;
38
64
 
39
65
  /*
40
66
  * Document-class: PG::TextDecoder::Boolean < PG::SimpleDecoder
@@ -44,7 +70,7 @@ static ID s_id_decode;
44
70
  *
45
71
  */
46
72
  static VALUE
47
- pg_text_dec_boolean(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
73
+ pg_text_dec_boolean(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
48
74
  {
49
75
  if (len < 1) {
50
76
  rb_raise( rb_eTypeError, "wrong data for text boolean converter in tuple %d field %d", tuple, field);
@@ -61,7 +87,7 @@ pg_text_dec_boolean(t_pg_coder *conv, char *val, int len, int tuple, int field,
61
87
  *
62
88
  */
63
89
  VALUE
64
- pg_text_dec_string(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
90
+ pg_text_dec_string(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
65
91
  {
66
92
  VALUE ret = rb_tainted_str_new( val, len );
67
93
  PG_ENCODING_SET_NOCHECK( ret, enc_idx );
@@ -76,7 +102,7 @@ pg_text_dec_string(t_pg_coder *conv, char *val, int len, int tuple, int field, i
76
102
  *
77
103
  */
78
104
  static VALUE
79
- pg_text_dec_integer(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
105
+ pg_text_dec_integer(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
80
106
  {
81
107
  long i;
82
108
  int max_len;
@@ -101,7 +127,7 @@ pg_text_dec_integer(t_pg_coder *conv, char *val, int len, int tuple, int field,
101
127
  * conn.exec("select generate_series(1,1000000)").values }
102
128
  * end
103
129
  */
104
- char *val_pos = val;
130
+ const char *val_pos = val;
105
131
  char digit = *val_pos;
106
132
  int neg;
107
133
  int error = 0;
@@ -132,6 +158,19 @@ pg_text_dec_integer(t_pg_coder *conv, char *val, int len, int tuple, int field,
132
158
  return rb_cstr2inum(val, 10);
133
159
  }
134
160
 
161
+ /*
162
+ * Document-class: PG::TextDecoder::Numeric < PG::SimpleDecoder
163
+ *
164
+ * This is a decoder class for conversion of PostgreSQL numeric types
165
+ * to Ruby BigDecimal objects.
166
+ *
167
+ */
168
+ static VALUE
169
+ pg_text_dec_numeric(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
170
+ {
171
+ return rb_funcall(rb_cObject, s_id_BigDecimal, 1, rb_str_new(val, len));
172
+ }
173
+
135
174
  /*
136
175
  * Document-class: PG::TextDecoder::Float < PG::SimpleDecoder
137
176
  *
@@ -140,9 +179,32 @@ pg_text_dec_integer(t_pg_coder *conv, char *val, int len, int tuple, int field,
140
179
  *
141
180
  */
142
181
  static VALUE
143
- pg_text_dec_float(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
182
+ pg_text_dec_float(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
144
183
  {
145
- return rb_float_new(strtod(val, NULL));
184
+ switch(*val) {
185
+ case 'N':
186
+ return s_nan;
187
+ case 'I':
188
+ return s_pos_inf;
189
+ case '-':
190
+ if (val[1] == 'I') {
191
+ return s_neg_inf;
192
+ } else {
193
+ return rb_float_new(rb_cstr_to_dbl(val, Qfalse));
194
+ }
195
+ default:
196
+ return rb_float_new(rb_cstr_to_dbl(val, Qfalse));
197
+ }
198
+ }
199
+
200
+ struct pg_blob_initialization {
201
+ char *blob_string;
202
+ size_t length;
203
+ };
204
+
205
+ static VALUE pg_create_blob(VALUE v) {
206
+ struct pg_blob_initialization *bi = (struct pg_blob_initialization *)v;
207
+ return rb_tainted_str_new(bi->blob_string, bi->length);
146
208
  }
147
209
 
148
210
  /*
@@ -153,18 +215,54 @@ pg_text_dec_float(t_pg_coder *conv, char *val, int len, int tuple, int field, in
153
215
  *
154
216
  */
155
217
  static VALUE
156
- pg_text_dec_bytea(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
218
+ pg_text_dec_bytea(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
157
219
  {
158
- unsigned char *to;
159
- size_t to_len;
160
- VALUE ret;
220
+ struct pg_blob_initialization bi;
161
221
 
162
- to = PQunescapeBytea( (unsigned char *)val, &to_len);
222
+ bi.blob_string = (char *)PQunescapeBytea((unsigned char*)val, &bi.length);
223
+ if (bi.blob_string == NULL) {
224
+ rb_raise(rb_eNoMemError, "PQunescapeBytea failure: probably not enough memory");
225
+ }
226
+ return rb_ensure(pg_create_blob, (VALUE)&bi, (VALUE(*)())PQfreemem, (VALUE)bi.blob_string);
227
+ }
163
228
 
164
- ret = rb_tainted_str_new((char*)to, to_len);
165
- PQfreemem(to);
229
+ /*
230
+ * array_isspace() --- a non-locale-dependent isspace()
231
+ *
232
+ * We used to use isspace() for parsing array values, but that has
233
+ * undesirable results: an array value might be silently interpreted
234
+ * differently depending on the locale setting. Now we just hard-wire
235
+ * the traditional ASCII definition of isspace().
236
+ */
237
+ static int
238
+ array_isspace(char ch)
239
+ {
240
+ if (ch == ' ' ||
241
+ ch == '\t' ||
242
+ ch == '\n' ||
243
+ ch == '\r' ||
244
+ ch == '\v' ||
245
+ ch == '\f')
246
+ return 1;
247
+ return 0;
248
+ }
166
249
 
167
- return ret;
250
+ static int
251
+ array_isdim(char ch)
252
+ {
253
+ if ( (ch >= '0' && ch <= '9') ||
254
+ (ch == '-') ||
255
+ (ch == '+') ||
256
+ (ch == ':') )
257
+ return 1;
258
+ return 0;
259
+ }
260
+
261
+ static void
262
+ array_parser_error(t_pg_composite_coder *this, const char *text){
263
+ if( (this->comp.flags & PG_CODER_FORMAT_ERROR_MASK) == PG_CODER_FORMAT_ERROR_TO_RAISE ){
264
+ rb_raise( rb_eTypeError, "%s", text );
265
+ }
168
266
  }
169
267
 
170
268
  /*
@@ -172,7 +270,7 @@ pg_text_dec_bytea(t_pg_coder *conv, char *val, int len, int tuple, int field, in
172
270
  * https://github.com/dockyard/pg_array_parser
173
271
  */
174
272
  static VALUE
175
- read_array(t_pg_composite_coder *this, int *index, char *c_pg_array_string, int array_string_length, char *word, int enc_idx, int tuple, int field, t_pg_coder_dec_func dec_func)
273
+ read_array_without_dim(t_pg_composite_coder *this, int *index, const char *c_pg_array_string, int array_string_length, char *word, int enc_idx, int tuple, int field, t_pg_coder_dec_func dec_func)
176
274
  {
177
275
  /* Return value: array */
178
276
  VALUE array;
@@ -196,7 +294,7 @@ read_array(t_pg_composite_coder *this, int *index, char *c_pg_array_string, int
196
294
 
197
295
  /* Special case the empty array, so it doesn't need to be handled manually inside
198
296
  * the loop. */
199
- if(((*index) < array_string_length) && c_pg_array_string[(*index)] == '}')
297
+ if(((*index) < array_string_length) && c_pg_array_string[*index] == '}')
200
298
  {
201
299
  return array;
202
300
  }
@@ -236,10 +334,17 @@ read_array(t_pg_composite_coder *this, int *index, char *c_pg_array_string, int
236
334
  }
237
335
  else if(c == '{')
238
336
  {
337
+ VALUE subarray;
239
338
  (*index)++;
240
- rb_ary_push(array, read_array(this, index, c_pg_array_string, array_string_length, word, enc_idx, tuple, field, dec_func));
339
+ subarray = read_array_without_dim(this, index, c_pg_array_string, array_string_length, word, enc_idx, tuple, field, dec_func);
340
+ rb_ary_push(array, subarray);
241
341
  escapeNext = 1;
242
342
  }
343
+ else if(c == 0)
344
+ {
345
+ array_parser_error( this, "premature end of the array string" );
346
+ return array;
347
+ }
243
348
  else
244
349
  {
245
350
  word[word_index] = c;
@@ -266,34 +371,116 @@ read_array(t_pg_composite_coder *this, int *index, char *c_pg_array_string, int
266
371
  }
267
372
  }
268
373
 
374
+ array_parser_error( this, "premature end of the array string" );
269
375
  return array;
270
376
  }
271
377
 
272
378
  /*
273
379
  * Document-class: PG::TextDecoder::Array < PG::CompositeDecoder
274
380
  *
275
- * This is the decoder class for PostgreSQL array types.
381
+ * This is a decoder class for PostgreSQL array types.
382
+ *
383
+ * It returns an Array with possibly an arbitrary number of sub-Arrays.
384
+ * All values are decoded according to the #elements_type accessor.
385
+ * Sub-arrays are decoded recursively.
276
386
  *
277
- * All values are decoded according to the #elements_type
278
- * accessor. Sub-arrays are decoded recursively.
387
+ * This decoder simply ignores any dimension decorations preceding the array values.
388
+ * It returns all array values as regular ruby Array with a zero based index, regardless of the index given in the dimension decoration.
389
+ *
390
+ * An array decoder which respects dimension decorations is waiting to be implemented.
279
391
  *
280
392
  */
281
393
  static VALUE
282
- pg_text_dec_array(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
394
+ pg_text_dec_array(t_pg_coder *conv, const char *c_pg_array_string, int array_string_length, int tuple, int field, int enc_idx)
283
395
  {
396
+ int index = 0;
397
+ int ndim = 0;
398
+ VALUE ret;
284
399
  t_pg_composite_coder *this = (t_pg_composite_coder *)conv;
285
- t_pg_coder_dec_func dec_func = pg_coder_dec_func(this->elem, 0);
286
- /* create a buffer of the same length, as that will be the worst case */
287
- char *word = xmalloc(len + 1);
288
- int index = 1;
289
-
290
- VALUE return_value = read_array(this, &index, val, len, word, enc_idx, tuple, field, dec_func);
291
- free(word);
292
- return return_value;
400
+
401
+ /*
402
+ * If the input string starts with dimension info, read and use that.
403
+ * Otherwise, we require the input to be in curly-brace style, and we
404
+ * prescan the input to determine dimensions.
405
+ *
406
+ * Dimension info takes the form of one or more [n] or [m:n] items. The
407
+ * outer loop iterates once per dimension item.
408
+ */
409
+ for (;;)
410
+ {
411
+ /*
412
+ * Note: we currently allow whitespace between, but not within,
413
+ * dimension items.
414
+ */
415
+ while (array_isspace(c_pg_array_string[index]))
416
+ index++;
417
+ if (c_pg_array_string[index] != '[')
418
+ break; /* no more dimension items */
419
+ index++;
420
+
421
+ while (array_isdim(c_pg_array_string[index]))
422
+ index++;
423
+
424
+ if (c_pg_array_string[index] != ']'){
425
+ array_parser_error( this, "missing \"]\" in array dimensions");
426
+ break;
427
+ }
428
+ index++;
429
+
430
+ ndim++;
431
+ }
432
+
433
+ if (ndim == 0)
434
+ {
435
+ /* No array dimensions */
436
+ }
437
+ else
438
+ {
439
+ /* If array dimensions are given, expect '=' operator */
440
+ if (c_pg_array_string[index] != '=') {
441
+ array_parser_error( this, "missing assignment operator");
442
+ index-=2; /* jump back to before "]" so that we don't break behavior to pg < 1.1 */
443
+ }
444
+ index++;
445
+
446
+ while (array_isspace(c_pg_array_string[index]))
447
+ index++;
448
+ }
449
+
450
+ if (c_pg_array_string[index] != '{')
451
+ array_parser_error( this, "array value must start with \"{\" or dimension information");
452
+ index++;
453
+
454
+ if ( index < array_string_length && c_pg_array_string[index] == '}' ) {
455
+ /* avoid buffer allocation for empty array */
456
+ ret = rb_ary_new();
457
+ } else {
458
+ t_pg_coder_dec_func dec_func = pg_coder_dec_func(this->elem, 0);
459
+ /* create a buffer of the same length, as that will be the worst case */
460
+ VALUE buf = rb_str_new(NULL, array_string_length);
461
+ char *word = RSTRING_PTR(buf);
462
+
463
+ ret = read_array_without_dim(this, &index, c_pg_array_string, array_string_length, word, enc_idx, tuple, field, dec_func);
464
+
465
+ RB_GC_GUARD(buf);
466
+ }
467
+
468
+ if (c_pg_array_string[index] != '}' )
469
+ array_parser_error( this, "array value must end with \"}\"");
470
+ index++;
471
+
472
+ /* only whitespace is allowed after the closing brace */
473
+ for(;index < array_string_length; ++index)
474
+ {
475
+ if (!array_isspace(c_pg_array_string[index]))
476
+ array_parser_error( this, "malformed array literal: Junk after closing right brace.");
477
+ }
478
+
479
+ return ret;
293
480
  }
294
481
 
295
482
  /*
296
- * Document-class: PG::TextDecoder::Identifier < PG::CompositeDecoder
483
+ * Document-class: PG::TextDecoder::Identifier < PG::SimpleDecoder
297
484
  *
298
485
  * This is the decoder class for PostgreSQL identifiers.
299
486
  *
@@ -303,18 +490,15 @@ pg_text_dec_array(t_pg_coder *conv, char *val, int len, int tuple, int field, in
303
490
  *
304
491
  */
305
492
  static VALUE
306
- pg_text_dec_identifier(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
493
+ pg_text_dec_identifier(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
307
494
  {
308
- t_pg_composite_coder *this = (t_pg_composite_coder *)conv;
309
- t_pg_coder_dec_func dec_func = pg_coder_dec_func(this->elem, 0);
310
-
311
495
  /* Return value: array */
312
496
  VALUE array;
313
497
  VALUE elem;
314
498
  int word_index = 0;
315
499
  int index;
316
500
  /* Use a buffer of the same length, as that will be the worst case */
317
- char word[len + 1];
501
+ PG_VARIABLE_LENGTH_ARRAY(char, word, len + 1, NAMEDATALEN)
318
502
 
319
503
  /* The current character in the input string. */
320
504
  char c;
@@ -331,7 +515,7 @@ pg_text_dec_identifier(t_pg_coder *conv, char *val, int len, int tuple, int fiel
331
515
  if(c == '.' && openQuote < 2 ) {
332
516
  word[word_index] = 0;
333
517
 
334
- elem = dec_func(conv, word, word_index, tuple, field, enc_idx);
518
+ elem = pg_text_dec_string(conv, word, word_index, tuple, field, enc_idx);
335
519
  rb_ary_push(array, elem);
336
520
 
337
521
  openQuote = 0;
@@ -353,7 +537,7 @@ pg_text_dec_identifier(t_pg_coder *conv, char *val, int len, int tuple, int fiel
353
537
  }
354
538
 
355
539
  word[word_index] = 0;
356
- elem = dec_func(conv, word, word_index, tuple, field, enc_idx);
540
+ elem = pg_text_dec_string(conv, word, word_index, tuple, field, enc_idx);
357
541
  rb_ary_push(array, elem);
358
542
 
359
543
  return array;
@@ -368,7 +552,7 @@ pg_text_dec_identifier(t_pg_coder *conv, char *val, int len, int tuple, int fiel
368
552
  *
369
553
  */
370
554
  static VALUE
371
- pg_text_dec_from_base64(t_pg_coder *conv, char *val, int len, int tuple, int field, int enc_idx)
555
+ pg_text_dec_from_base64(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
372
556
  {
373
557
  t_pg_composite_coder *this = (t_pg_composite_coder *)conv;
374
558
  t_pg_coder_dec_func dec_func = pg_coder_dec_func(this->elem, this->comp.format);
@@ -393,10 +577,379 @@ pg_text_dec_from_base64(t_pg_coder *conv, char *val, int len, int tuple, int fie
393
577
  return out_value;
394
578
  }
395
579
 
580
+ static inline int char_to_digit(char c)
581
+ {
582
+ return c - '0';
583
+ }
584
+
585
+ static int str2_to_int(const char *str)
586
+ {
587
+ return char_to_digit(str[0]) * 10
588
+ + char_to_digit(str[1]);
589
+ }
590
+
591
+ static int parse_year(const char **str) {
592
+ int year = 0;
593
+ int i;
594
+ const char * p = *str;
595
+
596
+ for(i = 0; isdigit(*p) && i < 7; i++, p++) {
597
+ year = 10 * year + char_to_digit(*p);
598
+ }
599
+
600
+ *str = p;
601
+ return year;
602
+ }
603
+
604
+ #define TZ_NEG 1
605
+ #define TZ_POS 2
606
+
607
+ /*
608
+ * Document-class: PG::TextDecoder::Timestamp < PG::SimpleDecoder
609
+ *
610
+ * This is a decoder class for conversion of PostgreSQL text timestamps
611
+ * to Ruby Time objects.
612
+ *
613
+ * The following flags can be used to specify timezone interpretation:
614
+ * * +PG::Coder::TIMESTAMP_DB_UTC+ : Interpret timestamp as UTC time (default)
615
+ * * +PG::Coder::TIMESTAMP_DB_LOCAL+ : Interpret timestamp as local time
616
+ * * +PG::Coder::TIMESTAMP_APP_UTC+ : Return timestamp as UTC time (default)
617
+ * * +PG::Coder::TIMESTAMP_APP_LOCAL+ : Return timestamp as local time
618
+ *
619
+ * Example:
620
+ * deco = PG::TextDecoder::Timestamp.new(flags: PG::Coder::TIMESTAMP_DB_UTC | PG::Coder::TIMESTAMP_APP_LOCAL)
621
+ * deco.decode("2000-01-01 00:00:00") # => 2000-01-01 01:00:00 +0100
622
+ */
623
+ static VALUE pg_text_dec_timestamp(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
624
+ {
625
+ const char *str = val;
626
+ int year, mon, day;
627
+ int hour, min, sec;
628
+ int nsec = 0;
629
+ int tz_given = 0;
630
+ int tz_hour = 0;
631
+ int tz_min = 0;
632
+ int tz_sec = 0;
633
+
634
+ year = parse_year(&str);
635
+
636
+ if ( year > 0
637
+ && str[0] == '-' && isdigit(str[1]) && isdigit(str[2])
638
+ && str[3] == '-' && isdigit(str[4]) && isdigit(str[5])
639
+ && str[6] == ' ' && isdigit(str[7]) && isdigit(str[8])
640
+ && str[9] == ':' && isdigit(str[10]) && isdigit(str[11])
641
+ && str[12] == ':' && isdigit(str[13]) && isdigit(str[14])
642
+ ) {
643
+
644
+ mon = str2_to_int(str+1); str += 3;
645
+ day = str2_to_int(str+1); str += 3;
646
+ hour = str2_to_int(str+1); str += 3;
647
+ min = str2_to_int(str+1); str += 3;
648
+ sec = str2_to_int(str+1); str += 3;
649
+
650
+ if (str[0] == '.' && isdigit(str[1])) {
651
+ /* nano second part, up to 9 digits */
652
+ static const int coef[9] = {
653
+ 100000000, 10000000, 1000000,
654
+ 100000, 10000, 1000, 100, 10, 1
655
+ };
656
+ int i;
657
+
658
+ str++;
659
+ for (i = 0; i < 9 && isdigit(*str); i++)
660
+ {
661
+ nsec += coef[i] * char_to_digit(*str++);
662
+ }
663
+ /* consume digits smaller than nsec */
664
+ while(isdigit(*str)) str++;
665
+ }
666
+
667
+ if ((str[0] == '+' || str[0] == '-') && isdigit(str[1]) && isdigit(str[2])) {
668
+ tz_given = str[0] == '-' ? TZ_NEG : TZ_POS;
669
+ tz_hour = str2_to_int(str+1); str += 3;
670
+
671
+ if (str[0] == ':' && isdigit(str[1]) && isdigit(str[2]))
672
+ {
673
+ tz_min = str2_to_int(str+1); str += 3;
674
+ }
675
+ if (str[0] == ':' && isdigit(str[1]) && isdigit(str[2]))
676
+ {
677
+ tz_sec = str2_to_int(str+1); str += 3;
678
+ }
679
+ }
680
+
681
+ if (str[0] == ' ' && str[1] == 'B' && str[2] == 'C') {
682
+ year = -year + 1;
683
+ str += 3;
684
+ }
685
+
686
+ if (*str == '\0') { /* must have consumed all the string */
687
+ VALUE sec_value;
688
+ VALUE gmt_offset_value;
689
+ VALUE res;
690
+
691
+ #if (RUBY_API_VERSION_MAJOR > 2 || (RUBY_API_VERSION_MAJOR == 2 && RUBY_API_VERSION_MINOR >= 3)) && defined(HAVE_TIMEGM)
692
+ /* Fast path for time conversion */
693
+ struct tm tm;
694
+ struct timespec ts;
695
+ tm.tm_year = year - 1900;
696
+ tm.tm_mon = mon - 1;
697
+ tm.tm_mday = day;
698
+ tm.tm_hour = hour;
699
+ tm.tm_min = min;
700
+ tm.tm_sec = sec;
701
+ tm.tm_isdst = -1;
702
+
703
+ if (tz_given) {
704
+ /* with timezone */
705
+ time_t time = timegm(&tm);
706
+ if (time != -1){
707
+ int gmt_offset;
708
+
709
+ gmt_offset = tz_hour * 3600 + tz_min * 60 + tz_sec;
710
+ if (tz_given == TZ_NEG)
711
+ {
712
+ gmt_offset = - gmt_offset;
713
+ }
714
+ ts.tv_sec = time - gmt_offset;
715
+ ts.tv_nsec = nsec;
716
+ return rb_time_timespec_new(&ts, gmt_offset);
717
+ }
718
+ } else {
719
+ /* without timezone */
720
+ time_t time;
721
+
722
+ if( conv->flags & PG_CODER_TIMESTAMP_DB_LOCAL ) {
723
+ time = mktime(&tm);
724
+ } else {
725
+ time = timegm(&tm);
726
+ }
727
+ if (time != -1){
728
+ ts.tv_sec = time;
729
+ ts.tv_nsec = nsec;
730
+ return rb_time_timespec_new(&ts, conv->flags & PG_CODER_TIMESTAMP_APP_LOCAL ? INT_MAX : INT_MAX-1);
731
+ }
732
+ }
733
+ /* Some libc implementations fail to convert certain values,
734
+ * so that we fall through to the slow path.
735
+ */
736
+ #endif
737
+ if (nsec) {
738
+ int sec_numerator = sec * 1000000 + nsec / 1000;
739
+ int sec_denominator = 1000000;
740
+ sec_value = rb_funcall(Qnil, s_id_Rational, 2,
741
+ INT2NUM(sec_numerator), INT2NUM(sec_denominator));
742
+ } else {
743
+ sec_value = INT2NUM(sec);
744
+ }
745
+
746
+ if (tz_given) {
747
+ /* with timezone */
748
+ int gmt_offset;
749
+
750
+ gmt_offset = tz_hour * 3600 + tz_min * 60 + tz_sec;
751
+ if (tz_given == TZ_NEG)
752
+ {
753
+ gmt_offset = - gmt_offset;
754
+ }
755
+ gmt_offset_value = INT2NUM(gmt_offset);
756
+ } else {
757
+ /* without timezone */
758
+ gmt_offset_value = conv->flags & PG_CODER_TIMESTAMP_DB_LOCAL ? Qnil : INT2NUM(0);
759
+ }
760
+
761
+ res = rb_funcall(rb_cTime, s_id_new, 7,
762
+ INT2NUM(year),
763
+ INT2NUM(mon),
764
+ INT2NUM(day),
765
+ INT2NUM(hour),
766
+ INT2NUM(min),
767
+ sec_value,
768
+ gmt_offset_value);
769
+
770
+ if (tz_given) {
771
+ /* with timezone */
772
+ return res;
773
+ } else {
774
+ /* without timezone */
775
+ if( (conv->flags & PG_CODER_TIMESTAMP_DB_LOCAL) && (conv->flags & PG_CODER_TIMESTAMP_APP_LOCAL) ) {
776
+ return res;
777
+ } else if( conv->flags & PG_CODER_TIMESTAMP_APP_LOCAL ) {
778
+ return rb_funcall(res, s_id_getlocal, 0);
779
+ } else {
780
+ return rb_funcall(res, s_id_utc, 0);
781
+ }
782
+ }
783
+ }
784
+ }
785
+
786
+ /* fall through to string conversion */
787
+ return pg_text_dec_string(conv, val, len, tuple, field, enc_idx);
788
+ }
789
+
790
+ /*
791
+ * Document-class: PG::TextDecoder::Inet < PG::SimpleDecoder
792
+ *
793
+ * This is a decoder class for conversion of PostgreSQL inet type
794
+ * to Ruby IPAddr values.
795
+ *
796
+ */
797
+ static VALUE
798
+ pg_text_dec_inet(t_pg_coder *conv, const char *val, int len, int tuple, int field, int enc_idx)
799
+ {
800
+ VALUE ip;
801
+ #if defined(_WIN32)
802
+ ip = rb_str_new(val, len);
803
+ ip = rb_class_new_instance(1, &ip, s_IPAddr);
804
+ #else
805
+ VALUE ip_int;
806
+ VALUE vmasks;
807
+ char dst[16];
808
+ char buf[64];
809
+ int af = strchr(val, '.') ? AF_INET : AF_INET6;
810
+ int mask = -1;
811
+
812
+ if (len >= 64) {
813
+ rb_raise(rb_eTypeError, "too long data for text inet converter in tuple %d field %d", tuple, field);
814
+ }
815
+
816
+ if (len >= 4) {
817
+ if (val[len-2] == '/') {
818
+ mask = val[len-1] - '0';
819
+ memcpy(buf, val, len-2);
820
+ buf[len-2] = '\0';
821
+ val = buf;
822
+ } else if (val[len-3] == '/') {
823
+ mask = (val[len-2]- '0')*10 + val[len-1] - '0';
824
+ memcpy(buf, val, len-3);
825
+ buf[len-3] = '\0';
826
+ val = buf;
827
+ } else if (val[len-4] == '/') {
828
+ mask = (val[len-3]- '0')*100 + (val[len-2]- '0')*10 + val[len-1] - '0';
829
+ memcpy(buf, val, len-4);
830
+ buf[len-4] = '\0';
831
+ val = buf;
832
+ }
833
+ }
834
+
835
+ if (1 != inet_pton(af, val, dst)) {
836
+ rb_raise(rb_eTypeError, "wrong data for text inet converter in tuple %d field %d val", tuple, field);
837
+ }
838
+
839
+ if (af == AF_INET) {
840
+ unsigned int ip_int_native;
841
+
842
+ if (mask == -1) {
843
+ mask = 32;
844
+ } else if (mask < 0 || mask > 32) {
845
+ rb_raise(rb_eTypeError, "invalid mask for IPv4: %d", mask);
846
+ }
847
+ vmasks = s_vmasks4;
848
+
849
+ ip_int_native = read_nbo32(dst);
850
+
851
+ /* Work around broken IPAddr behavior of convering portion
852
+ of address after netmask to 0 */
853
+ switch (mask) {
854
+ case 0:
855
+ ip_int_native = 0;
856
+ break;
857
+ case 32:
858
+ /* nothing to do */
859
+ break;
860
+ default:
861
+ ip_int_native &= ~((1UL<<(32-mask))-1);
862
+ break;
863
+ }
864
+
865
+ ip_int = UINT2NUM(ip_int_native);
866
+ } else {
867
+ unsigned long long * dstllp = (unsigned long long *)dst;
868
+ unsigned long long ip_int_native1;
869
+ unsigned long long ip_int_native2;
870
+
871
+ if (mask == -1) {
872
+ mask = 128;
873
+ } else if (mask < 0 || mask > 128) {
874
+ rb_raise(rb_eTypeError, "invalid mask for IPv6: %d", mask);
875
+ }
876
+ vmasks = s_vmasks6;
877
+
878
+ ip_int_native1 = read_nbo64(dstllp);
879
+ dstllp++;
880
+ ip_int_native2 = read_nbo64(dstllp);
881
+
882
+ if (mask == 128) {
883
+ /* nothing to do */
884
+ } else if (mask == 64) {
885
+ ip_int_native2 = 0;
886
+ } else if (mask == 0) {
887
+ ip_int_native1 = 0;
888
+ ip_int_native2 = 0;
889
+ } else if (mask < 64) {
890
+ ip_int_native1 &= ~((1ULL<<(64-mask))-1);
891
+ ip_int_native2 = 0;
892
+ } else {
893
+ ip_int_native2 &= ~((1ULL<<(128-mask))-1);
894
+ }
895
+
896
+ /* 4 Bignum allocations */
897
+ ip_int = ULL2NUM(ip_int_native1);
898
+ ip_int = rb_funcall(ip_int, s_id_lshift, 1, INT2NUM(64));
899
+ ip_int = rb_funcall(ip_int, s_id_add, 1, ULL2NUM(ip_int_native2));
900
+ }
901
+
902
+ if (use_ipaddr_alloc) {
903
+ ip = rb_obj_alloc(s_IPAddr);
904
+ rb_ivar_set(ip, s_ivar_family, INT2NUM(af));
905
+ rb_ivar_set(ip, s_ivar_addr, ip_int);
906
+ rb_ivar_set(ip, s_ivar_mask_addr, RARRAY_AREF(vmasks, mask));
907
+ } else {
908
+ VALUE ip_args[2];
909
+ ip_args[0] = ip_int;
910
+ ip_args[1] = INT2NUM(af);
911
+ ip = rb_class_new_instance(2, ip_args, s_IPAddr);
912
+ ip = rb_funcall(ip, s_id_mask, 1, INT2NUM(mask));
913
+ }
914
+
915
+ #endif
916
+ return ip;
917
+ }
918
+
396
919
  void
397
920
  init_pg_text_decoder()
398
921
  {
922
+ rb_require("ipaddr");
923
+ s_IPAddr = rb_funcall(rb_cObject, rb_intern("const_get"), 1, rb_str_new2("IPAddr"));
924
+ rb_global_variable(&s_IPAddr);
925
+ s_ivar_family = rb_intern("@family");
926
+ s_ivar_addr = rb_intern("@addr");
927
+ s_ivar_mask_addr = rb_intern("@mask_addr");
928
+ s_id_lshift = rb_intern("<<");
929
+ s_id_add = rb_intern("+");
930
+ s_id_mask = rb_intern("mask");
931
+
932
+ use_ipaddr_alloc = RTEST(rb_eval_string("IPAddr.new.instance_variables.sort == [:@addr, :@family, :@mask_addr]"));
933
+
934
+ s_vmasks4 = rb_eval_string("a = [0]*33; a[0] = 0; a[32] = 0xffffffff; 31.downto(1){|i| a[i] = a[i+1] - (1 << (31 - i))}; a.freeze");
935
+ rb_global_variable(&s_vmasks4);
936
+ s_vmasks6 = rb_eval_string("a = [0]*129; a[0] = 0; a[128] = 0xffffffffffffffffffffffffffffffff; 127.downto(1){|i| a[i] = a[i+1] - (1 << (127 - i))}; a.freeze");
937
+ rb_global_variable(&s_vmasks6);
938
+
399
939
  s_id_decode = rb_intern("decode");
940
+ s_id_Rational = rb_intern("Rational");
941
+ s_id_new = rb_intern("new");
942
+ s_id_utc = rb_intern("utc");
943
+ s_id_getlocal = rb_intern("getlocal");
944
+
945
+ rb_require("bigdecimal");
946
+ s_id_BigDecimal = rb_intern("BigDecimal");
947
+ s_nan = rb_eval_string("0.0/0.0");
948
+ rb_global_variable(&s_nan);
949
+ s_pos_inf = rb_eval_string("1.0/0.0");
950
+ rb_global_variable(&s_pos_inf);
951
+ s_neg_inf = rb_eval_string("-1.0/0.0");
952
+ rb_global_variable(&s_neg_inf);
400
953
 
401
954
  /* This module encapsulates all decoder classes with text input format */
402
955
  rb_mPG_TextDecoder = rb_define_module_under( rb_mPG, "TextDecoder" );
@@ -408,15 +961,21 @@ init_pg_text_decoder()
408
961
  pg_define_coder( "Integer", pg_text_dec_integer, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
409
962
  /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Float", rb_cPG_SimpleDecoder ); */
410
963
  pg_define_coder( "Float", pg_text_dec_float, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
964
+ /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "BigDecimal", rb_cPG_SimpleDecoder ); */
965
+ pg_define_coder( "Numeric", pg_text_dec_numeric, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
411
966
  /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "String", rb_cPG_SimpleDecoder ); */
412
967
  pg_define_coder( "String", pg_text_dec_string, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
413
968
  /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Bytea", rb_cPG_SimpleDecoder ); */
414
969
  pg_define_coder( "Bytea", pg_text_dec_bytea, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
970
+ /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Identifier", rb_cPG_SimpleDecoder ); */
971
+ pg_define_coder( "Identifier", pg_text_dec_identifier, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder );
972
+ /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Timestamp", rb_cPG_SimpleDecoder ); */
973
+ pg_define_coder( "Timestamp", pg_text_dec_timestamp, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder);
974
+ /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Inet", rb_cPG_SimpleDecoder ); */
975
+ pg_define_coder( "Inet", pg_text_dec_inet, rb_cPG_SimpleDecoder, rb_mPG_TextDecoder);
415
976
 
416
977
  /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Array", rb_cPG_CompositeDecoder ); */
417
978
  pg_define_coder( "Array", pg_text_dec_array, rb_cPG_CompositeDecoder, rb_mPG_TextDecoder );
418
- /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "Identifier", rb_cPG_CompositeDecoder ); */
419
- pg_define_coder( "Identifier", pg_text_dec_identifier, rb_cPG_CompositeDecoder, rb_mPG_TextDecoder );
420
979
  /* dummy = rb_define_class_under( rb_mPG_TextDecoder, "FromBase64", rb_cPG_CompositeDecoder ); */
421
980
  pg_define_coder( "FromBase64", pg_text_dec_from_base64, rb_cPG_CompositeDecoder, rb_mPG_TextDecoder );
422
981
  }