geo_coder 0.1.0

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 (119) hide show
  1. data/Gemfile +12 -0
  2. data/Gemfile.lock +32 -0
  3. data/History.txt +6 -0
  4. data/Makefile +13 -0
  5. data/Manifest.txt +18 -0
  6. data/README.rdoc +197 -0
  7. data/Rakefile +53 -0
  8. data/TODO.txt +8 -0
  9. data/VERSION +1 -0
  10. data/bin/build_indexes +8 -0
  11. data/bin/rebuild_cluster +22 -0
  12. data/bin/rebuild_metaphones +23 -0
  13. data/bin/tiger_import +59 -0
  14. data/demos/demo/app/ext/geocodewrap.rb +84 -0
  15. data/demos/demo/app/views/index.builder +13 -0
  16. data/demos/demo/app/views/index.erb +71 -0
  17. data/demos/demo/config.ru +12 -0
  18. data/demos/demo/config/bootstraps.rb +130 -0
  19. data/demos/demo/config/geoenvironment.rb +25 -0
  20. data/demos/demo/geocoder_helper.rb +12 -0
  21. data/demos/demo/geocom_geocode.rb +10 -0
  22. data/demos/demo/main.rb +3 -0
  23. data/demos/demo/rakefile.rb +17 -0
  24. data/demos/demo/tmp/restart.txt +0 -0
  25. data/demos/simpledemo/views/index.builder +13 -0
  26. data/demos/simpledemo/views/index.erb +69 -0
  27. data/demos/simpledemo/ws.rb +83 -0
  28. data/doc/Makefile +7 -0
  29. data/doc/html4css1.css +279 -0
  30. data/doc/lookup.rst +193 -0
  31. data/doc/parsing.rst +125 -0
  32. data/doc/voidspace.css +147 -0
  33. data/geo_coder.gemspec +172 -0
  34. data/lib/geocoder/us.rb +21 -0
  35. data/lib/geocoder/us/address.rb +290 -0
  36. data/lib/geocoder/us/constants.rb +670 -0
  37. data/lib/geocoder/us/database.rb +745 -0
  38. data/lib/geocoder/us/import.rb +181 -0
  39. data/lib/geocoder/us/import/tiger.rb +13 -0
  40. data/lib/geocoder/us/numbers.rb +58 -0
  41. data/navteq/README +4 -0
  42. data/navteq/convert.sql +37 -0
  43. data/navteq/navteq_import +39 -0
  44. data/navteq/prepare.sql +92 -0
  45. data/sql/cluster.sql +16 -0
  46. data/sql/convert.sql +80 -0
  47. data/sql/create.sql +37 -0
  48. data/sql/index.sql +12 -0
  49. data/sql/place.csv +104944 -0
  50. data/sql/place.sql +104948 -0
  51. data/sql/setup.sql +78 -0
  52. data/src/Makefile +13 -0
  53. data/src/README +14 -0
  54. data/src/liblwgeom/Makefile +75 -0
  55. data/src/liblwgeom/box2d.c +54 -0
  56. data/src/liblwgeom/lex.yy.c +4799 -0
  57. data/src/liblwgeom/liblwgeom.h +1405 -0
  58. data/src/liblwgeom/lwalgorithm.c +946 -0
  59. data/src/liblwgeom/lwalgorithm.h +52 -0
  60. data/src/liblwgeom/lwcircstring.c +759 -0
  61. data/src/liblwgeom/lwcollection.c +541 -0
  62. data/src/liblwgeom/lwcompound.c +118 -0
  63. data/src/liblwgeom/lwcurvepoly.c +86 -0
  64. data/src/liblwgeom/lwgeom.c +886 -0
  65. data/src/liblwgeom/lwgeom_api.c +2201 -0
  66. data/src/liblwgeom/lwgparse.c +1219 -0
  67. data/src/liblwgeom/lwgunparse.c +1054 -0
  68. data/src/liblwgeom/lwline.c +525 -0
  69. data/src/liblwgeom/lwmcurve.c +125 -0
  70. data/src/liblwgeom/lwmline.c +137 -0
  71. data/src/liblwgeom/lwmpoint.c +138 -0
  72. data/src/liblwgeom/lwmpoly.c +141 -0
  73. data/src/liblwgeom/lwmsurface.c +129 -0
  74. data/src/liblwgeom/lwpoint.c +439 -0
  75. data/src/liblwgeom/lwpoly.c +579 -0
  76. data/src/liblwgeom/lwsegmentize.c +1047 -0
  77. data/src/liblwgeom/lwutil.c +369 -0
  78. data/src/liblwgeom/measures.c +861 -0
  79. data/src/liblwgeom/postgis_config.h +93 -0
  80. data/src/liblwgeom/ptarray.c +847 -0
  81. data/src/liblwgeom/vsprintf.c +179 -0
  82. data/src/liblwgeom/wktparse.h +126 -0
  83. data/src/liblwgeom/wktparse.lex +74 -0
  84. data/src/liblwgeom/wktparse.tab.c +2353 -0
  85. data/src/liblwgeom/wktparse.tab.h +145 -0
  86. data/src/liblwgeom/wktparse.y +385 -0
  87. data/src/libsqlite3_geocoder/Makefile +22 -0
  88. data/src/libsqlite3_geocoder/Makefile.nix +15 -0
  89. data/src/libsqlite3_geocoder/Makefile.redhat +15 -0
  90. data/src/libsqlite3_geocoder/extension.c +121 -0
  91. data/src/libsqlite3_geocoder/extension.h +13 -0
  92. data/src/libsqlite3_geocoder/levenshtein.c +42 -0
  93. data/src/libsqlite3_geocoder/metaphon.c +278 -0
  94. data/src/libsqlite3_geocoder/util.c +37 -0
  95. data/src/libsqlite3_geocoder/wkb_compress.c +54 -0
  96. data/src/metaphone/Makefile +7 -0
  97. data/src/metaphone/README +49 -0
  98. data/src/metaphone/extension.c +37 -0
  99. data/src/metaphone/metaphon.c +251 -0
  100. data/src/shp2sqlite/Makefile +37 -0
  101. data/src/shp2sqlite/Makefile.nix +36 -0
  102. data/src/shp2sqlite/Makefile.redhat +35 -0
  103. data/src/shp2sqlite/dbfopen.c +1595 -0
  104. data/src/shp2sqlite/getopt.c +695 -0
  105. data/src/shp2sqlite/getopt.h +127 -0
  106. data/src/shp2sqlite/shapefil.h +500 -0
  107. data/src/shp2sqlite/shp2sqlite.c +1974 -0
  108. data/src/shp2sqlite/shpopen.c +1894 -0
  109. data/tests/address.rb +236 -0
  110. data/tests/benchmark.rb +20 -0
  111. data/tests/constants.rb +57 -0
  112. data/tests/data/address-sample.csv +52 -0
  113. data/tests/data/db-test.csv +57 -0
  114. data/tests/data/locations.csv +4 -0
  115. data/tests/database.rb +137 -0
  116. data/tests/generate.rb +34 -0
  117. data/tests/numbers.rb +46 -0
  118. data/tests/run.rb +11 -0
  119. metadata +237 -0
@@ -0,0 +1,1219 @@
1
+ /*
2
+ * Written by Ralph Mason ralph.mason<at>telogis.com
3
+ *
4
+ * Copyright Telogis 2004
5
+ * www.telogis.com
6
+ *
7
+ */
8
+ #include <string.h>
9
+ #include <stdio.h>
10
+ /* Solaris9 does not provide stdint.h */
11
+ /* #include <stdint.h> */
12
+ #include <inttypes.h>
13
+
14
+ #include "liblwgeom.h"
15
+ #include "wktparse.h"
16
+ #include "wktparse.tab.h"
17
+
18
+ /*
19
+ * To get byte order
20
+ #include <sys/types.h>
21
+ #include <netinet/in.h>
22
+ #include <netinet/in_systm.h>
23
+ #include <netinet/ip.h>
24
+ */
25
+
26
+ void set_zm(char z, char m);
27
+ void close_parser(void);
28
+
29
+ typedef uint32_t int4;
30
+
31
+ typedef struct tag_tuple tuple;
32
+
33
+ struct tag_outputstate{
34
+ uchar* pos;
35
+ };
36
+
37
+ typedef struct tag_outputstate output_state;
38
+ typedef void (*output_func)(tuple* this, output_state* out);
39
+ typedef void (*read_col_func)(const char **f);
40
+
41
+
42
+
43
+ /* Globals */
44
+
45
+ int srid=-1;
46
+
47
+ static int parser_ferror_occured;
48
+ static allocator local_malloc;
49
+ static report_error error_func;
50
+
51
+ struct tag_tuple{
52
+ output_func of;
53
+ union union_tag {
54
+ double points[4];
55
+ int4 pointsi[4];
56
+
57
+ struct struct_tag {
58
+ tuple* stack_next;
59
+ int type;
60
+ int num;
61
+ int size_here;
62
+ } nn;
63
+ } uu;
64
+ struct tag_tuple *next;
65
+ };
66
+
67
+ struct {
68
+ int type;
69
+ int flags;
70
+ int srid;
71
+ int ndims;
72
+ int hasZ;
73
+ int hasM;
74
+ /* create integer version */
75
+ int lwgi;
76
+ /* input is integer (wkb only)*/
77
+ int from_lwgi;
78
+
79
+ int4 alloc_size;
80
+
81
+ /*
82
+ linked list of all tuples
83
+ */
84
+ tuple* first;
85
+ tuple* last;
86
+
87
+ /*
88
+ stack of open geometries
89
+ */
90
+ tuple* stack;
91
+
92
+ } the_geom;
93
+
94
+ tuple* free_list=0;
95
+
96
+
97
+ /*
98
+ * Parser current instance check flags - a bitmap of flags that determine which checks are enabled during the current parse
99
+ * (see liblwgeom.h for the related PARSER_CHECK constants)
100
+ */
101
+ int current_parser_check_flags;
102
+
103
+ /*
104
+ * Parser current instance result structure - the result structure being used for the current parse
105
+ */
106
+ LWGEOM_PARSER_RESULT *current_lwg_parser_result;
107
+
108
+
109
+ /* Parser state flags - these are set automatically by the parser */
110
+ int minpoints;
111
+ int checkclosed;
112
+
113
+ /*
114
+ * This indicates if the number of points in the geometry is required to
115
+ * be odd (one) or even (zero, currently not enforced) or whatever (-one)
116
+ */
117
+ int isodd;
118
+ double *first_point=NULL;
119
+ double *last_point=NULL;
120
+
121
+
122
+ /*
123
+ * Parser error messages
124
+ *
125
+ * IMPORTANT: Make sure the order of these messages matches the PARSER_ERROR constants in liblwgeom.h!
126
+ * The 0th element should always be empty since it is unused (error constants start from -1)
127
+ */
128
+
129
+ const char *parser_error_messages[] = {
130
+ "",
131
+ "geometry requires more points",
132
+ "geometry must have an odd number of points",
133
+ "geometry contains non-closed rings",
134
+ "can not mix dimensionality in a geometry",
135
+ "parse error - invalid geometry",
136
+ "invalid WKB type"
137
+ };
138
+
139
+ /* Macro to return the error message and the current position within WKT */
140
+ #define LWGEOM_WKT_PARSER_ERROR(errcode) \
141
+ do { \
142
+ if (!parser_ferror_occured) { \
143
+ parser_ferror_occured = -1 * errcode; \
144
+ current_lwg_parser_result->message = parser_error_messages[errcode]; \
145
+ current_lwg_parser_result->errlocation = lwg_parse_yylloc.last_column; \
146
+ } \
147
+ } while (0);
148
+
149
+
150
+ /* Macro to return the error message and the current position within WKB
151
+ NOTE: the position is handled automatically by strhex_readbyte */
152
+ #define LWGEOM_WKB_PARSER_ERROR(errcode) \
153
+ do { \
154
+ if (!parser_ferror_occured) { \
155
+ parser_ferror_occured = -1 * errcode; \
156
+ current_lwg_parser_result->message = parser_error_messages[errcode]; \
157
+ } \
158
+ } while (0);
159
+
160
+
161
+ /* External functions */
162
+ extern void init_parser(const char *);
163
+
164
+ /* Prototypes */
165
+ tuple* alloc_tuple(output_func of,size_t size);
166
+ void free_tuple(tuple* to_free);
167
+ void inc_num(void);
168
+ void alloc_stack_tuple(int type,output_func of,size_t size);
169
+ void check_dims(int num);
170
+ void WRITE_DOUBLES(output_state* out,double* points, int cnt);
171
+ #ifdef SHRINK_INTS
172
+ void WRITE_INT4(output_state * out,int4 val);
173
+ #endif
174
+ void empty_stack(tuple* this,output_state* out);
175
+ void alloc_lwgeom(int srid);
176
+ void write_point_2(tuple* this,output_state* out);
177
+ void write_point_3(tuple* this,output_state* out);
178
+ void write_point_4(tuple* this,output_state* out);
179
+ void write_point_2i(tuple* this,output_state* out);
180
+ void write_point_3i(tuple* this,output_state* out);
181
+ void write_point_4i(tuple* this,output_state* out);
182
+ void alloc_point_2d(double x,double y);
183
+ void alloc_point_3d(double x,double y,double z);
184
+ void alloc_point_4d(double x,double y,double z,double m);
185
+ void write_type(tuple* this,output_state* out);
186
+ void write_count(tuple* this,output_state* out);
187
+ void write_type_count(tuple* this,output_state* out);
188
+ void alloc_point(void);
189
+ void alloc_linestring(void);
190
+ void alloc_linestring_closed(void);
191
+ void alloc_circularstring(void);
192
+ void alloc_circularstring_closed(void);
193
+ void alloc_polygon(void);
194
+ void alloc_multipoint(void);
195
+ void alloc_multilinestring(void);
196
+ void alloc_multicurve(void);
197
+ void alloc_multipolygon(void);
198
+ void alloc_multisurface(void);
199
+ void alloc_geomertycollection(void);
200
+ void alloc_counter(void);
201
+ void alloc_empty(void);
202
+ void make_serialized_lwgeom(LWGEOM_PARSER_RESULT *lwg_parser_result);
203
+ uchar strhex_readbyte(const char *in);
204
+ uchar read_wkb_byte(const char **in);
205
+ void read_wkb_bytes(const char **in, uchar* out, int cnt);
206
+ int4 read_wkb_int(const char **in);
207
+ double read_wkb_double(const char **in, int convert_from_int);
208
+ void read_wkb_point(const char **b);
209
+ void read_wkb_polygon(const char **b);
210
+ void read_wkb_linestring(const char **b);
211
+ void read_wkb_circstring(const char **b);
212
+ void read_wkb_ordinate_array(const char **b);
213
+ void read_collection(const char **b, read_col_func f);
214
+ void parse_wkb(const char **b);
215
+ void alloc_wkb(const char *parser);
216
+ int parse_it(LWGEOM_PARSER_RESULT *lwg_parser_result, const char* geometry, int flags, allocator allocfunc, report_error errfunc);
217
+ int parse_lwg(LWGEOM_PARSER_RESULT *lwg_parser_result, const char* geometry, int flags, allocator allocfunc, report_error errfunc);
218
+ int parse_lwgi(LWGEOM_PARSER_RESULT *lwg_parser_result, const char* geometry, int flags, allocator allocfunc, report_error errfunc);
219
+
220
+ void
221
+ set_srid(double d_srid)
222
+ {
223
+ if ( d_srid >= 0 )
224
+ d_srid+=0.1;
225
+ else
226
+ d_srid-=0.1;
227
+
228
+
229
+ srid=(int)(d_srid+0.1);
230
+ }
231
+
232
+ tuple *
233
+ alloc_tuple(output_func of,size_t size)
234
+ {
235
+ tuple* ret = free_list;
236
+
237
+ if ( ! ret ){
238
+ int toalloc = (ALLOC_CHUNKS /sizeof(tuple));
239
+ ret = malloc( toalloc *sizeof(tuple) );
240
+
241
+ free_list = ret;
242
+
243
+ while(--toalloc){
244
+ ret->next = ret+1;
245
+ ret++;
246
+ }
247
+
248
+ ret->next = NULL;
249
+
250
+ return alloc_tuple(of,size);
251
+ }
252
+
253
+ free_list = ret->next;
254
+ ret->of = of;
255
+ ret->next = NULL;
256
+
257
+ if ( the_geom.last ) {
258
+ the_geom.last->next = ret;
259
+ the_geom.last = ret;
260
+ }
261
+ else {
262
+ the_geom.first = the_geom.last = ret;
263
+ }
264
+
265
+ the_geom.alloc_size += size;
266
+ return ret;
267
+ }
268
+
269
+ void
270
+ free_tuple(tuple* to_free)
271
+ {
272
+
273
+ tuple* list_end = to_free;
274
+
275
+ if( !to_free)
276
+ return;
277
+
278
+ while(list_end->next){
279
+ list_end=list_end->next;
280
+ }
281
+
282
+ list_end->next = free_list;
283
+ free_list = to_free;
284
+ }
285
+
286
+ void
287
+ inc_num(void)
288
+ {
289
+ the_geom.stack->uu.nn.num++;
290
+ }
291
+
292
+ /*
293
+ Allocate a 'counting' tuple
294
+ */
295
+ void
296
+ alloc_stack_tuple(int type,output_func of,size_t size)
297
+ {
298
+ tuple* p;
299
+ inc_num();
300
+
301
+ LWDEBUGF(2, "alloc_stack_tuple %d, %d", type, size);
302
+
303
+ p = alloc_tuple(of,size);
304
+ p->uu.nn.stack_next = the_geom.stack;
305
+ p->uu.nn.type = type;
306
+ p->uu.nn.size_here = the_geom.alloc_size;
307
+ p->uu.nn.num = 0;
308
+ the_geom.stack = p;
309
+ }
310
+
311
+ void
312
+ pop(void)
313
+ {
314
+ the_geom.stack = the_geom.stack->uu.nn.stack_next;
315
+ }
316
+
317
+ void
318
+ popc(void)
319
+ {
320
+ /* If the minimum point check has been enabled, perform it */
321
+ if (current_parser_check_flags & PARSER_CHECK_MINPOINTS) {
322
+ if ( the_geom.stack->uu.nn.num < minpoints){
323
+ LWGEOM_WKT_PARSER_ERROR(PARSER_ERROR_MOREPOINTS);
324
+ }
325
+ }
326
+
327
+ /* If the odd number point check has been enabled, perform it */
328
+ if (current_parser_check_flags & PARSER_CHECK_ODD) {
329
+ if(isodd != -1 && the_geom.stack->uu.nn.num % 2 != isodd) {
330
+ LWGEOM_WKT_PARSER_ERROR(PARSER_ERROR_ODDPOINTS);
331
+ }
332
+ }
333
+
334
+ /* If the polygon closure check has been enabled, perform it */
335
+ if (current_parser_check_flags & PARSER_CHECK_CLOSURE) {
336
+ if ( checkclosed && first_point && last_point) {
337
+ if ( memcmp(first_point, last_point,
338
+ sizeof(double)*the_geom.ndims) )
339
+ {
340
+ LWGEOM_WKT_PARSER_ERROR(PARSER_ERROR_UNCLOSED);
341
+ }
342
+ }
343
+ }
344
+
345
+ the_geom.stack = the_geom.stack->uu.nn.stack_next;
346
+ }
347
+
348
+
349
+ void
350
+ check_dims(int num)
351
+ {
352
+ LWDEBUGF(2, "check_dims the_geom.ndims = %d, num = %d", the_geom.ndims, num);
353
+
354
+ if( the_geom.ndims != num){
355
+ if (the_geom.ndims) {
356
+ LWGEOM_WKT_PARSER_ERROR(PARSER_ERROR_MIXDIMS);
357
+ } else {
358
+
359
+ LWDEBUGF(3, "check_dims: setting dim %d", num);
360
+
361
+ the_geom.ndims = num;
362
+ if ( num > 2 ) the_geom.hasZ = 1;
363
+ if ( num > 3 ) the_geom.hasM = 1;
364
+ }
365
+ }
366
+ }
367
+
368
+ #define WRITE_INT4_REAL(x,y) { memcpy(x->pos,&y,4); x->pos+=4;}
369
+ #define WRITE_INT4_REAL_MULTIPLE(x,y,z) { memcpy(x->pos,&y,z*4); x->pos+=(z*4);}
370
+
371
+ /*
372
+ we can shrink ints to one byte if they are less than 127
373
+ by setting the extra bit. Because if the different byte
374
+ ordering possibilities we need to set the lsb on little
375
+ endian to show a packed one and the msb on a big endian
376
+ machine
377
+ */
378
+ #ifdef SHRINK_INTS
379
+ void
380
+ WRITE_INT4(output_state * out,int4 val)
381
+ {
382
+ if ( val <= 0x7f ){
383
+ if ( getMachineEndian() == NDR ){
384
+ val = (val<<1) | 1;
385
+ }
386
+ else{
387
+ val |=0x80;
388
+ }
389
+
390
+ *out->pos++ = (uchar)val;
391
+ the_geom.alloc_size-=3;
392
+ }
393
+ else{
394
+ if ( getMachineEndian() == NDR ){
395
+ val <<=1;
396
+ }
397
+ WRITE_INT4_REAL(out,val);
398
+ }
399
+ }
400
+ #else
401
+ #define WRITE_INT4 WRITE_INT4_REAL
402
+ #endif
403
+
404
+
405
+ void
406
+ WRITE_DOUBLES(output_state* out,double* points, int cnt)
407
+ {
408
+ if ( the_geom.lwgi){
409
+ int4 vals[4];
410
+ int i;
411
+
412
+ for(i=0; i<cnt;i++){
413
+ vals[i] = (int4)(((points[i]+180.0)*0xB60B60)+.5);
414
+ }
415
+ memcpy(out->pos,vals,sizeof(int4)*cnt);
416
+ out->pos+=sizeof(int4)*cnt;
417
+ }
418
+ else{
419
+ memcpy(out->pos,points,sizeof(double)*cnt);
420
+ out->pos+=sizeof(double)*cnt;
421
+ }
422
+
423
+ }
424
+
425
+ void
426
+ empty_stack(tuple *this,output_state* out)
427
+ {
428
+ /* Do nothing but provide an empty base for the geometry stack */
429
+ }
430
+
431
+ void
432
+ alloc_lwgeom(int srid)
433
+ {
434
+ LWDEBUGF(2, "alloc_lwgeom %d", srid);
435
+
436
+ the_geom.srid=srid;
437
+ the_geom.alloc_size=0;
438
+ the_geom.stack=NULL;
439
+ the_geom.ndims=0;
440
+ the_geom.hasZ=0;
441
+ the_geom.hasM=0;
442
+
443
+ /* Free if used already */
444
+ if ( the_geom.first ){
445
+ free_tuple(the_geom.first);
446
+ the_geom.first=the_geom.last=NULL;
447
+ }
448
+
449
+ if ( srid != -1 ){
450
+ the_geom.alloc_size+=4;
451
+ }
452
+
453
+ /* Setup up an empty tuple as the stack base */
454
+ the_geom.stack = alloc_tuple(empty_stack, 0);
455
+ }
456
+
457
+ void
458
+ write_point_2(tuple* this,output_state* out)
459
+ {
460
+ WRITE_DOUBLES(out,this->uu.points,2);
461
+ }
462
+
463
+ void
464
+ write_point_3(tuple* this,output_state* out)
465
+ {
466
+ WRITE_DOUBLES(out,this->uu.points,3);
467
+ }
468
+
469
+ void
470
+ write_point_4(tuple* this,output_state* out)
471
+ {
472
+ WRITE_DOUBLES(out,this->uu.points,4);
473
+ }
474
+
475
+ void
476
+ write_point_2i(tuple* this,output_state* out)
477
+ {
478
+ WRITE_INT4_REAL_MULTIPLE(out,this->uu.points,2);
479
+ }
480
+
481
+ void
482
+ write_point_3i(tuple* this,output_state* out)
483
+ {
484
+ WRITE_INT4_REAL_MULTIPLE(out,this->uu.points,3);
485
+ }
486
+
487
+ void
488
+ write_point_4i(tuple* this,output_state* out)
489
+ {
490
+ WRITE_INT4_REAL_MULTIPLE(out,this->uu.points,4);
491
+ }
492
+
493
+ void
494
+ alloc_point_2d(double x,double y)
495
+ {
496
+ tuple* p = alloc_tuple(write_point_2,the_geom.lwgi?8:16);
497
+ p->uu.points[0] = x;
498
+ p->uu.points[1] = y;
499
+
500
+ LWDEBUGF(2, "alloc_point_2d %f,%f", x, y);
501
+
502
+ /* keep track of point */
503
+ if ( checkclosed ) {
504
+ if ( ! the_geom.stack->uu.nn.num )
505
+ first_point = p->uu.points;
506
+ last_point = p->uu.points;
507
+ }
508
+
509
+ inc_num();
510
+ check_dims(2);
511
+ }
512
+
513
+ void
514
+ alloc_point_3d(double x,double y,double z)
515
+ {
516
+ tuple* p = alloc_tuple(write_point_3,the_geom.lwgi?12:24);
517
+ p->uu.points[0] = x;
518
+ p->uu.points[1] = y;
519
+ p->uu.points[2] = z;
520
+
521
+ LWDEBUGF(2, "alloc_point_3d %f, %f, %f", x, y, z);
522
+
523
+ /* keep track of point */
524
+ if ( checkclosed ) {
525
+ if ( ! the_geom.stack->uu.nn.num )
526
+ first_point = p->uu.points;
527
+ last_point = p->uu.points;
528
+ }
529
+
530
+ inc_num();
531
+ check_dims(3);
532
+ }
533
+
534
+ void
535
+ alloc_point_4d(double x,double y,double z,double m)
536
+ {
537
+ tuple* p = alloc_tuple(write_point_4,the_geom.lwgi?16:32);
538
+ p->uu.points[0] = x;
539
+ p->uu.points[1] = y;
540
+ p->uu.points[2] = z;
541
+ p->uu.points[3] = m;
542
+
543
+ LWDEBUGF(2, "alloc_point_4d %f, %f, %f, %f", x, y, z, m);
544
+
545
+ /* keep track of point */
546
+ if ( checkclosed ) {
547
+ if ( ! the_geom.stack->uu.nn.num )
548
+ first_point = p->uu.points;
549
+ last_point = p->uu.points;
550
+ }
551
+
552
+ inc_num();
553
+ check_dims(4);
554
+ }
555
+
556
+ void
557
+ write_type(tuple* this,output_state* out)
558
+ {
559
+ uchar type=0;
560
+
561
+ /* Empty handler - switch back */
562
+ if ( this->uu.nn.type == 0xff )
563
+ this->uu.nn.type = COLLECTIONTYPE;
564
+
565
+ type |= this->uu.nn.type;
566
+
567
+ if (the_geom.ndims) /* Support empty */
568
+ {
569
+ TYPE_SETZM(type, the_geom.hasZ, the_geom.hasM);
570
+ }
571
+
572
+ if ( the_geom.srid != -1 ){
573
+ type |= 0x40;
574
+ }
575
+
576
+ *(out->pos)=type;
577
+ out->pos++;
578
+
579
+ if ( the_geom.srid != -1 ){
580
+ /* Only the first geometry will have a srid attached */
581
+ WRITE_INT4(out,the_geom.srid);
582
+ the_geom.srid = -1;
583
+ }
584
+ }
585
+
586
+ void
587
+ write_count(tuple* this,output_state* out)
588
+ {
589
+ int num = this->uu.nn.num;
590
+ WRITE_INT4(out,num);
591
+ }
592
+
593
+ void
594
+ write_type_count(tuple* this,output_state* out)
595
+ {
596
+ write_type(this,out);
597
+ write_count(this,out);
598
+ }
599
+
600
+ void
601
+ alloc_point(void)
602
+ {
603
+ LWDEBUG(2, "alloc_point");
604
+
605
+ if( the_geom.lwgi)
606
+ alloc_stack_tuple(POINTTYPEI,write_type,1);
607
+ else
608
+ alloc_stack_tuple(POINTTYPE,write_type,1);
609
+
610
+ minpoints=1;
611
+ checkclosed=0;
612
+ isodd=-1;
613
+ }
614
+
615
+ void
616
+ alloc_linestring(void)
617
+ {
618
+ LWDEBUG(2, "alloc_linestring");
619
+
620
+ if( the_geom.lwgi)
621
+ alloc_stack_tuple(LINETYPEI,write_type,1);
622
+ else
623
+ alloc_stack_tuple(LINETYPE,write_type,1);
624
+
625
+ minpoints=2;
626
+ checkclosed=0;
627
+ isodd=-1;
628
+ }
629
+
630
+ void alloc_linestring_closed(void)
631
+ {
632
+ LWDEBUG(2, "alloc_linestring_closed called.");
633
+
634
+ alloc_linestring();
635
+ checkclosed=1;
636
+ }
637
+
638
+ void
639
+ alloc_circularstring(void)
640
+ {
641
+ LWDEBUG(2, "alloc_circularstring");
642
+
643
+ alloc_stack_tuple(CIRCSTRINGTYPE,write_type,1);
644
+ minpoints=3;
645
+ checkclosed=0;
646
+ isodd=1;
647
+ }
648
+
649
+ void alloc_circularstring_closed(void)
650
+ {
651
+ LWDEBUG(2, "alloc_circularstring_closed");
652
+
653
+ alloc_circularstring();
654
+ checkclosed=1;
655
+ }
656
+
657
+ void
658
+ alloc_polygon(void)
659
+ {
660
+ LWDEBUG(2, "alloc_polygon");
661
+
662
+ if( the_geom.lwgi)
663
+ alloc_stack_tuple(POLYGONTYPEI, write_type,1);
664
+ else
665
+ alloc_stack_tuple(POLYGONTYPE, write_type,1);
666
+
667
+ minpoints=4;
668
+ checkclosed=1;
669
+ isodd=-1;
670
+
671
+ }
672
+
673
+ void
674
+ alloc_curvepolygon(void)
675
+ {
676
+ LWDEBUG(2, "alloc_curvepolygon called.");
677
+
678
+ alloc_stack_tuple(CURVEPOLYTYPE, write_type, 1);
679
+ minpoints=4;
680
+ checkclosed=1;
681
+ isodd=-1;
682
+ }
683
+
684
+ void
685
+ alloc_compoundcurve(void)
686
+ {
687
+ LWDEBUG(2, "alloc_compoundcurve called.");
688
+
689
+ alloc_stack_tuple(COMPOUNDTYPE, write_type, 1);
690
+ }
691
+
692
+ void
693
+ alloc_multipoint(void)
694
+ {
695
+ LWDEBUG(2, "alloc_multipoint");
696
+
697
+ alloc_stack_tuple(MULTIPOINTTYPE,write_type,1);
698
+ }
699
+
700
+ void
701
+ alloc_multilinestring(void)
702
+ {
703
+ LWDEBUG(2, "alloc_multilinestring");
704
+
705
+ alloc_stack_tuple(MULTILINETYPE,write_type,1);
706
+ }
707
+
708
+ void
709
+ alloc_multicurve(void)
710
+ {
711
+ LWDEBUG(2, "alloc_multicurve");
712
+
713
+ alloc_stack_tuple(MULTICURVETYPE,write_type,1);
714
+ }
715
+
716
+ void
717
+ alloc_multipolygon(void)
718
+ {
719
+ LWDEBUG(2, "alloc_multipolygon");
720
+
721
+ alloc_stack_tuple(MULTIPOLYGONTYPE,write_type,1);
722
+ }
723
+
724
+ void
725
+ alloc_multisurface(void)
726
+ {
727
+ LWDEBUG(2, "alloc_multisurface called");
728
+
729
+ alloc_stack_tuple(MULTISURFACETYPE,write_type,1);
730
+ }
731
+
732
+ void
733
+ alloc_geomertycollection(void)
734
+ {
735
+ LWDEBUG(2, "alloc_geometrycollection");
736
+
737
+ alloc_stack_tuple(COLLECTIONTYPE,write_type,1);
738
+ }
739
+
740
+ void
741
+ alloc_counter(void)
742
+ {
743
+ LWDEBUG(2, "alloc_counter");
744
+
745
+ alloc_stack_tuple(0,write_count,4);
746
+ }
747
+
748
+ void
749
+ alloc_empty(void)
750
+ {
751
+ tuple* st = the_geom.stack;
752
+
753
+ LWDEBUG(2, "alloc_empty");
754
+
755
+ /* Find the last geometry */
756
+ while(st->uu.nn.type == 0){
757
+ st =st->uu.nn.stack_next;
758
+ }
759
+
760
+ /* Reclaim memory */
761
+ free_tuple(st->next);
762
+
763
+ /* Put an empty geometry collection on the top of the stack */
764
+ st->next=NULL;
765
+ the_geom.stack=st;
766
+ the_geom.alloc_size=st->uu.nn.size_here;
767
+
768
+ /* Mark as a empty stop */
769
+ if (st->uu.nn.type != 0xFF){
770
+ st->uu.nn.type=0xFF;
771
+ st->of = write_type_count;
772
+ the_geom.alloc_size+=4;
773
+ st->uu.nn.size_here=the_geom.alloc_size;
774
+ }
775
+
776
+ st->uu.nn.num=0;
777
+ }
778
+
779
+ void
780
+ make_serialized_lwgeom(LWGEOM_PARSER_RESULT *lwg_parser_result)
781
+ {
782
+ uchar* out_c;
783
+ output_state out;
784
+ tuple* cur;
785
+
786
+ LWDEBUG(2, "make_serialized_lwgeom");
787
+
788
+ /* Allocate the LWGEOM itself */
789
+ out_c = (uchar*)local_malloc(the_geom.alloc_size);
790
+ out.pos = out_c;
791
+ cur = the_geom.first ;
792
+
793
+ while(cur){
794
+ cur->of(cur,&out);
795
+ cur=cur->next;
796
+ }
797
+
798
+ /* Setup the LWGEOM_PARSER_RESULT structure */
799
+ lwg_parser_result->serialized_lwgeom = out_c;
800
+ lwg_parser_result->size = the_geom.alloc_size;
801
+
802
+ return;
803
+ }
804
+
805
+ void
806
+ lwg_parse_yynotice(char* s)
807
+ {
808
+ lwnotice(s);
809
+ }
810
+
811
+ int
812
+ lwg_parse_yyerror(char* s)
813
+ {
814
+ LWGEOM_WKT_PARSER_ERROR(PARSER_ERROR_INVALIDGEOM);
815
+ return 1;
816
+ }
817
+
818
+ /*
819
+ Table below generated using this ruby.
820
+
821
+ a=(0..0xff).to_a.collect{|x|0xff};('0'..'9').each{|x|a[x[0]]=x[0]-'0'[0]}
822
+ ('a'..'f').each{|x|v=x[0]-'a'[0]+10;a[x[0]]=a[x.upcase[0]]=v}
823
+ puts '{'+a.join(",")+'}'
824
+
825
+ */
826
+ static const uchar to_hex[] = {
827
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
828
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
829
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
830
+ 0,1,2,3,4,5,6,7,8,9,255,255,255,255,255,255,255,10,11,12,13,14,
831
+ 15,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
832
+ 255,255,255,255,255,255,255,255,255,255,255,10,11,12,13,14,15,255,
833
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
834
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
835
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
836
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
837
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
838
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
839
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
840
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
841
+ 255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
842
+ 255,255,255,255,255,255,255,255};
843
+
844
+ uchar
845
+ strhex_readbyte(const char* in)
846
+ {
847
+ if ( *in == 0 ){
848
+ if ( ! parser_ferror_occured){
849
+ LWGEOM_WKB_PARSER_ERROR(PARSER_ERROR_INVALIDGEOM);
850
+ }
851
+ return 0;
852
+ }
853
+
854
+ if (!parser_ferror_occured) {
855
+ lwg_parse_yylloc.last_column++;
856
+ return to_hex[(int)*in]<<4 | to_hex[(int)*(in+1)];
857
+ } else {
858
+ return 0;
859
+ }
860
+ }
861
+
862
+ uchar
863
+ read_wkb_byte(const char** in)
864
+ {
865
+ uchar ret = strhex_readbyte(*in);
866
+ (*in)+=2;
867
+ return ret;
868
+ }
869
+
870
+ int swap_order;
871
+
872
+ void
873
+ read_wkb_bytes(const char** in, uchar* out, int cnt)
874
+ {
875
+ if ( ! swap_order ){
876
+ while(cnt--) *out++ = read_wkb_byte(in);
877
+ }
878
+ else{
879
+ out += (cnt-1);
880
+ while(cnt--) *out-- = read_wkb_byte(in);
881
+ }
882
+ }
883
+
884
+ int4
885
+ read_wkb_int(const char** in)
886
+ {
887
+ int4 ret=0;
888
+ read_wkb_bytes(in,(uchar*)&ret,4);
889
+ return ret;
890
+ }
891
+
892
+ double
893
+ read_wkb_double(const char** in, int convert_from_int)
894
+ {
895
+ double ret=0;
896
+
897
+ if ( ! convert_from_int){
898
+ read_wkb_bytes(in,(uchar*)&ret,8);
899
+ return ret;
900
+ }else{
901
+ ret = read_wkb_int(in);
902
+ ret /= 0xb60b60;
903
+ return ret-180.0;
904
+ }
905
+ }
906
+
907
+ void
908
+ read_wkb_point(const char **b)
909
+ {
910
+ int i;
911
+ tuple* p = NULL;
912
+
913
+
914
+ if(the_geom.lwgi && the_geom.from_lwgi ){
915
+ /*
916
+ * Special case - reading from lwgi to lwgi
917
+ * we don't want to go via doubles in the middle.
918
+ */
919
+ switch(the_geom.ndims){
920
+ case 2: p=alloc_tuple(write_point_2i,8); break;
921
+ case 3: p=alloc_tuple(write_point_3i,12); break;
922
+ case 4: p=alloc_tuple(write_point_4i,16); break;
923
+ }
924
+
925
+ for(i=0;i<the_geom.ndims;i++){
926
+ p->uu.pointsi[i]=read_wkb_int(b);
927
+ }
928
+ }
929
+ else{
930
+ int mul = the_geom.lwgi ? 1 : 2;
931
+
932
+ switch(the_geom.ndims){
933
+ case 2: p=alloc_tuple(write_point_2,8*mul); break;
934
+ case 3: p=alloc_tuple(write_point_3,12*mul); break;
935
+ case 4: p=alloc_tuple(write_point_4,16*mul); break;
936
+ }
937
+
938
+ for(i=0;i<the_geom.ndims;i++){
939
+ p->uu.points[i]=read_wkb_double(b,the_geom.from_lwgi);
940
+ }
941
+ }
942
+
943
+
944
+ /* keep track of point */
945
+ if ( checkclosed ) {
946
+ if ( ! the_geom.stack->uu.nn.num )
947
+ first_point = p->uu.points;
948
+ last_point = p->uu.points;
949
+ }
950
+
951
+ inc_num();
952
+ check_dims(the_geom.ndims);
953
+ }
954
+
955
+ void
956
+ read_wkb_polygon(const char **b)
957
+ {
958
+ /* Stack the number of ORDINATE_ARRAYs (rings) */
959
+ int4 cnt=read_wkb_int(b);
960
+ alloc_counter();
961
+
962
+ /* Read through each ORDINATE_ARRAY in turn */
963
+ while(cnt--){
964
+ if ( parser_ferror_occured ) return;
965
+
966
+ /* Things to check for POLYGON ORDINATE_ARRAYs */
967
+ minpoints=4;
968
+ checkclosed=1;
969
+ isodd=-1;
970
+
971
+ read_wkb_ordinate_array(b);
972
+ }
973
+
974
+ pop();
975
+ }
976
+
977
+ void
978
+ read_wkb_linestring(const char **b)
979
+ {
980
+
981
+ /* Things to check for LINESTRING ORDINATE_ARRAYs */
982
+ minpoints=2;
983
+ checkclosed=0;
984
+ isodd=-1;
985
+
986
+ read_wkb_ordinate_array(b);
987
+ }
988
+
989
+
990
+ void
991
+ read_wkb_circstring(const char **b)
992
+ {
993
+
994
+ /* Things to check for CIRCULARSTRING ORDINATE_ARRAYs */
995
+ minpoints=3;
996
+ checkclosed=0;
997
+ isodd=-1;
998
+
999
+ read_wkb_ordinate_array(b);
1000
+ }
1001
+
1002
+ void
1003
+ read_wkb_ordinate_array(const char **b)
1004
+ {
1005
+ /* Read through a WKB ordinate array */
1006
+ int4 cnt=read_wkb_int(b);
1007
+ alloc_counter();
1008
+
1009
+ while(cnt--){
1010
+ if ( parser_ferror_occured ) return;
1011
+ read_wkb_point(b);
1012
+ }
1013
+
1014
+ /* Perform a check of the ordinate array */
1015
+ popc();
1016
+ }
1017
+
1018
+ void
1019
+ read_collection(const char **b, read_col_func f)
1020
+ {
1021
+ /* Read through a COLLECTION or an EWKB */
1022
+ int4 cnt=read_wkb_int(b);
1023
+ alloc_counter();
1024
+
1025
+ while(cnt--){
1026
+ if ( parser_ferror_occured ) return;
1027
+ f(b);
1028
+ }
1029
+
1030
+ pop();
1031
+ }
1032
+
1033
+ void
1034
+ parse_wkb(const char **b)
1035
+ {
1036
+ int4 type;
1037
+ uchar xdr = read_wkb_byte(b);
1038
+ int4 localsrid;
1039
+
1040
+ LWDEBUG(2, "parse_wkb");
1041
+
1042
+ swap_order=0;
1043
+
1044
+ if ( xdr != getMachineEndian() )
1045
+ {
1046
+ swap_order=1;
1047
+ }
1048
+
1049
+ type = read_wkb_int(b);
1050
+
1051
+ /* quick exit on error */
1052
+ if ( parser_ferror_occured ) return;
1053
+
1054
+ the_geom.ndims=2;
1055
+ if (type & WKBZOFFSET)
1056
+ {
1057
+ the_geom.hasZ = 1;
1058
+ the_geom.ndims++;
1059
+ }
1060
+ else the_geom.hasZ = 0;
1061
+ if (type & WKBMOFFSET)
1062
+ {
1063
+ the_geom.hasM = 1;
1064
+ the_geom.ndims++;
1065
+ }
1066
+ else the_geom.hasM = 0;
1067
+
1068
+ if (type & WKBSRIDFLAG )
1069
+ {
1070
+ /* local (in-EWKB) srid spec overrides SRID=#; */
1071
+ localsrid = read_wkb_int(b);
1072
+ if ( localsrid != -1 )
1073
+ {
1074
+ if ( the_geom.srid == -1 ) the_geom.alloc_size += 4;
1075
+ the_geom.srid = localsrid;
1076
+ }
1077
+ }
1078
+
1079
+ type &=0x0f;
1080
+
1081
+ if ( the_geom.lwgi ){
1082
+
1083
+ if ( type<= POLYGONTYPE )
1084
+ alloc_stack_tuple(type +9,write_type,1);
1085
+ else
1086
+ alloc_stack_tuple(type,write_type,1);
1087
+ }
1088
+ else{
1089
+ /* If we are writing lwg and are reading wbki */
1090
+ int4 towrite=type;
1091
+ if (towrite >= POINTTYPEI && towrite <= POLYGONTYPEI){
1092
+ towrite-=9;
1093
+ }
1094
+ alloc_stack_tuple(towrite,write_type,1);
1095
+ }
1096
+
1097
+ switch(type ){
1098
+ case POINTTYPE:
1099
+ read_wkb_point(b);
1100
+ break;
1101
+
1102
+ case LINETYPE:
1103
+ read_wkb_linestring(b);
1104
+ break;
1105
+
1106
+ case CIRCSTRINGTYPE:
1107
+ read_wkb_circstring(b);
1108
+ break;
1109
+
1110
+ case POLYGONTYPE:
1111
+ read_wkb_polygon(b);
1112
+ break;
1113
+
1114
+ case COMPOUNDTYPE:
1115
+ read_collection(b,parse_wkb);
1116
+ break;
1117
+
1118
+ case CURVEPOLYTYPE:
1119
+ read_collection(b,parse_wkb);
1120
+ break;
1121
+
1122
+ case MULTIPOINTTYPE:
1123
+ case MULTILINETYPE:
1124
+ case MULTICURVETYPE:
1125
+ case MULTIPOLYGONTYPE:
1126
+ case MULTISURFACETYPE:
1127
+ case COLLECTIONTYPE:
1128
+ read_collection(b,parse_wkb);
1129
+ break;
1130
+
1131
+ case POINTTYPEI:
1132
+ the_geom.from_lwgi=1;
1133
+ read_wkb_point(b);
1134
+ break;
1135
+
1136
+ case LINETYPEI:
1137
+ the_geom.from_lwgi=1;
1138
+ read_wkb_linestring(b);
1139
+ break;
1140
+
1141
+ case POLYGONTYPEI:
1142
+ the_geom.from_lwgi=1;
1143
+ read_wkb_polygon(b);
1144
+ break;
1145
+
1146
+ default:
1147
+ LWGEOM_WKB_PARSER_ERROR(PARSER_ERROR_INVALIDWKBTYPE);
1148
+ }
1149
+
1150
+ the_geom.from_lwgi=0;
1151
+
1152
+ pop();
1153
+ }
1154
+
1155
+
1156
+ void
1157
+ alloc_wkb(const char *parser)
1158
+ {
1159
+ LWDEBUG(2, "alloc_wkb");
1160
+
1161
+ parse_wkb(&parser);
1162
+ }
1163
+
1164
+ /*
1165
+ Parse a string and return a LW_GEOM
1166
+ */
1167
+ int
1168
+ parse_it(LWGEOM_PARSER_RESULT *lwg_parser_result, const char *geometry, int flags, allocator allocfunc, report_error errfunc)
1169
+ {
1170
+ LWDEBUGF(2, "parse_it: %s with parser flags %d", geometry, flags);
1171
+
1172
+ local_malloc = allocfunc;
1173
+ error_func=errfunc;
1174
+
1175
+ parser_ferror_occured = 0;
1176
+
1177
+ /* Setup the inital parser flags and empty the return struct */
1178
+ current_lwg_parser_result = lwg_parser_result;
1179
+ current_parser_check_flags = flags;
1180
+ lwg_parser_result->serialized_lwgeom = NULL;
1181
+ lwg_parser_result->size = 0;
1182
+ lwg_parser_result->wkinput = geometry;
1183
+
1184
+ init_parser(geometry);
1185
+
1186
+ lwg_parse_yyparse();
1187
+
1188
+ close_parser();
1189
+
1190
+ /* Return the parsed geometry */
1191
+ make_serialized_lwgeom(lwg_parser_result);
1192
+
1193
+ return parser_ferror_occured;
1194
+ }
1195
+
1196
+ int
1197
+ parse_lwg(LWGEOM_PARSER_RESULT *lwg_parser_result, const char* geometry, int flags, allocator allocfunc, report_error errfunc)
1198
+ {
1199
+ the_geom.lwgi=0;
1200
+ return parse_it(lwg_parser_result, geometry, flags, allocfunc, errfunc);
1201
+ }
1202
+
1203
+ int
1204
+ parse_lwgi(LWGEOM_PARSER_RESULT *lwg_parser_result, const char* geometry, int flags, allocator allocfunc, report_error errfunc)
1205
+ {
1206
+ the_geom.lwgi=1;
1207
+ return parse_it(lwg_parser_result, geometry, flags, allocfunc, errfunc);
1208
+ }
1209
+
1210
+ void
1211
+ set_zm(char z, char m)
1212
+ {
1213
+ LWDEBUGF(2, "set_zm %d, %d", z, m);
1214
+
1215
+ the_geom.hasZ = z;
1216
+ the_geom.hasM = m;
1217
+ the_geom.ndims = 2+z+m;
1218
+ }
1219
+