summaryrefslogtreecommitdiff
path: root/contrib/tablefunc/tablefunc.c
blob: ea2f2e15b0a3428a1dba73f93d10021a5618aadf (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
/*
 * contrib/tablefunc/tablefunc.c
 *
 *
 * tablefunc
 *
 * Sample to demonstrate C functions which return setof scalar
 * and setof composite.
 * Joe Conway <mail@joeconway.com>
 * And contributors:
 * Nabil Sayegh <postgresql@e-trolley.de>
 *
 * Copyright (c) 2002-2010, PostgreSQL Global Development Group
 *
 * Permission to use, copy, modify, and distribute this software and its
 * documentation for any purpose, without fee, and without a written agreement
 * is hereby granted, provided that the above copyright notice and this
 * paragraph and the following two paragraphs appear in all copies.
 *
 * IN NO EVENT SHALL THE AUTHORS OR DISTRIBUTORS BE LIABLE TO ANY PARTY FOR
 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING
 * LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS
 * DOCUMENTATION, EVEN IF THE AUTHOR OR DISTRIBUTORS HAVE BEEN ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 *
 * THE AUTHORS AND DISTRIBUTORS SPECIFICALLY DISCLAIM ANY WARRANTIES,
 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
 * AND FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
 * ON AN "AS IS" BASIS, AND THE AUTHOR AND DISTRIBUTORS HAS NO OBLIGATIONS TO
 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
 *
 */
#include "postgres.h"

#include <math.h>

#include "catalog/pg_type.h"
#include "fmgr.h"
#include "funcapi.h"
#include "executor/spi.h"
#include "lib/stringinfo.h"
#include "miscadmin.h"
#include "utils/builtins.h"
#include "utils/guc.h"
#include "utils/lsyscache.h"

#include "tablefunc.h"

PG_MODULE_MAGIC;

static HTAB *load_categories_hash(char *cats_sql, MemoryContext per_query_ctx);
static Tuplestorestate *get_crosstab_tuplestore(char *sql,
						HTAB *crosstab_hash,
						TupleDesc tupdesc,
						MemoryContext per_query_ctx,
						bool randomAccess);
static void validateConnectbyTupleDesc(TupleDesc tupdesc, bool show_branch, bool show_serial);
static bool compatCrosstabTupleDescs(TupleDesc tupdesc1, TupleDesc tupdesc2);
static bool compatConnectbyTupleDescs(TupleDesc tupdesc1, TupleDesc tupdesc2);
static void get_normal_pair(float8 *x1, float8 *x2);
static Tuplestorestate *connectby(char *relname,
		  char *key_fld,
		  char *parent_key_fld,
		  char *orderby_fld,
		  char *branch_delim,
		  char *start_with,
		  int max_depth,
		  bool show_branch,
		  bool show_serial,
		  MemoryContext per_query_ctx,
		  bool randomAccess,
		  AttInMetadata *attinmeta);
static Tuplestorestate *build_tuplestore_recursively(char *key_fld,
							 char *parent_key_fld,
							 char *relname,
							 char *orderby_fld,
							 char *branch_delim,
							 char *start_with,
							 char *branch,
							 int level,
							 int *serial,
							 int max_depth,
							 bool show_branch,
							 bool show_serial,
							 MemoryContext per_query_ctx,
							 AttInMetadata *attinmeta,
							 Tuplestorestate *tupstore);
static char *quote_literal_cstr(char *rawstr);

typedef struct
{
	float8		mean;			/* mean of the distribution */
	float8		stddev;			/* stddev of the distribution */
	float8		carry_val;		/* hold second generated value */
	bool		use_carry;		/* use second generated value */
} normal_rand_fctx;

#define xpfree(var_) \
	do { \
		if (var_ != NULL) \
		{ \
			pfree(var_); \
			var_ = NULL; \
		} \
	} while (0)

#define xpstrdup(tgtvar_, srcvar_) \
	do { \
		if (srcvar_) \
			tgtvar_ = pstrdup(srcvar_); \
		else \
			tgtvar_ = NULL; \
	} while (0)

#define xstreq(tgtvar_, srcvar_) \
	(((tgtvar_ == NULL) && (srcvar_ == NULL)) || \
	 ((tgtvar_ != NULL) && (srcvar_ != NULL) && (strcmp(tgtvar_, srcvar_) == 0)))

/* sign, 10 digits, '\0' */
#define INT32_STRLEN	12

/* stored info for a crosstab category */
typedef struct crosstab_cat_desc
{
	char	   *catname;		/* full category name */
	int			attidx;			/* zero based */
} crosstab_cat_desc;

#define MAX_CATNAME_LEN			NAMEDATALEN
#define INIT_CATS				64

#define crosstab_HashTableLookup(HASHTAB, CATNAME, CATDESC) \
do { \
	crosstab_HashEnt *hentry; char key[MAX_CATNAME_LEN]; \
	\
	MemSet(key, 0, MAX_CATNAME_LEN); \
	snprintf(key, MAX_CATNAME_LEN - 1, "%s", CATNAME); \
	hentry = (crosstab_HashEnt*) hash_search(HASHTAB, \
										 key, HASH_FIND, NULL); \
	if (hentry) \
		CATDESC = hentry->catdesc; \
	else \
		CATDESC = NULL; \
} while(0)

#define crosstab_HashTableInsert(HASHTAB, CATDESC) \
do { \
	crosstab_HashEnt *hentry; bool found; char key[MAX_CATNAME_LEN]; \
	\
	MemSet(key, 0, MAX_CATNAME_LEN); \
	snprintf(key, MAX_CATNAME_LEN - 1, "%s", CATDESC->catname); \
	hentry = (crosstab_HashEnt*) hash_search(HASHTAB, \
										 key, HASH_ENTER, &found); \
	if (found) \
		ereport(ERROR, \
				(errcode(ERRCODE_DUPLICATE_OBJECT), \
				 errmsg("duplicate category name"))); \
	hentry->catdesc = CATDESC; \
} while(0)

/* hash table */
typedef struct crosstab_hashent
{
	char		internal_catname[MAX_CATNAME_LEN];
	crosstab_cat_desc *catdesc;
} crosstab_HashEnt;

/*
 * normal_rand - return requested number of random values
 * with a Gaussian (Normal) distribution.
 *
 * inputs are int numvals, float8 mean, and float8 stddev
 * returns setof float8
 */
PG_FUNCTION_INFO_V1(normal_rand);
Datum
normal_rand(PG_FUNCTION_ARGS)
{
	FuncCallContext *funcctx;
	int			call_cntr;
	int			max_calls;
	normal_rand_fctx *fctx;
	float8		mean;
	float8		stddev;
	float8		carry_val;
	bool		use_carry;
	MemoryContext oldcontext;

	/* stuff done only on the first call of the function */
	if (SRF_IS_FIRSTCALL())
	{
		/* create a function context for cross-call persistence */
		funcctx = SRF_FIRSTCALL_INIT();

		/*
		 * switch to memory context appropriate for multiple function calls
		 */
		oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);

		/* total number of tuples to be returned */
		funcctx->max_calls = PG_GETARG_UINT32(0);

		/* allocate memory for user context */
		fctx = (normal_rand_fctx *) palloc(sizeof(normal_rand_fctx));

		/*
		 * Use fctx to keep track of upper and lower bounds from call to call.
		 * It will also be used to carry over the spare value we get from the
		 * Box-Muller algorithm so that we only actually calculate a new value
		 * every other call.
		 */
		fctx->mean = PG_GETARG_FLOAT8(1);
		fctx->stddev = PG_GETARG_FLOAT8(2);
		fctx->carry_val = 0;
		fctx->use_carry = false;

		funcctx->user_fctx = fctx;

		MemoryContextSwitchTo(oldcontext);
	}

	/* stuff done on every call of the function */
	funcctx = SRF_PERCALL_SETUP();

	call_cntr = funcctx->call_cntr;
	max_calls = funcctx->max_calls;
	fctx = funcctx->user_fctx;
	mean = fctx->mean;
	stddev = fctx->stddev;
	carry_val = fctx->carry_val;
	use_carry = fctx->use_carry;

	if (call_cntr < max_calls)	/* do when there is more left to send */
	{
		float8		result;

		if (use_carry)
		{
			/*
			 * reset use_carry and use second value obtained on last pass
			 */
			fctx->use_carry = false;
			result = carry_val;
		}
		else
		{
			float8		normval_1;
			float8		normval_2;

			/* Get the next two normal values */
			get_normal_pair(&normval_1, &normval_2);

			/* use the first */
			result = mean + (stddev * normval_1);

			/* and save the second */
			fctx->carry_val = mean + (stddev * normval_2);
			fctx->use_carry = true;
		}

		/* send the result */
		SRF_RETURN_NEXT(funcctx, Float8GetDatum(result));
	}
	else
		/* do when there is no more left */
		SRF_RETURN_DONE(funcctx);
}

/*
 * get_normal_pair()
 * Assigns normally distributed (Gaussian) values to a pair of provided
 * parameters, with mean 0, standard deviation 1.
 *
 * This routine implements Algorithm P (Polar method for normal deviates)
 * from Knuth's _The_Art_of_Computer_Programming_, Volume 2, 3rd ed., pages
 * 122-126. Knuth cites his source as "The polar method", G. E. P. Box, M. E.
 * Muller, and G. Marsaglia, _Annals_Math,_Stat._ 29 (1958), 610-611.
 *
 */
static void
get_normal_pair(float8 *x1, float8 *x2)
{
	float8		u1,
				u2,
				v1,
				v2,
				s;

	do
	{
		u1 = (float8) random() / (float8) MAX_RANDOM_VALUE;
		u2 = (float8) random() / (float8) MAX_RANDOM_VALUE;

		v1 = (2.0 * u1) - 1.0;
		v2 = (2.0 * u2) - 1.0;

		s = v1 * v1 + v2 * v2;
	} while (s >= 1.0);

	if (s == 0)
	{
		*x1 = 0;
		*x2 = 0;
	}
	else
	{
		s = sqrt((-2.0 * log(s)) / s);
		*x1 = v1 * s;
		*x2 = v2 * s;
	}
}

/*
 * crosstab - create a crosstab of rowids and values columns from a
 * SQL statement returning one rowid column, one category column,
 * and one value column.
 *
 * e.g. given sql which produces:
 *
 *			rowid	cat		value
 *			------+-------+-------
 *			row1	cat1	val1
 *			row1	cat2	val2
 *			row1	cat3	val3
 *			row1	cat4	val4
 *			row2	cat1	val5
 *			row2	cat2	val6
 *			row2	cat3	val7
 *			row2	cat4	val8
 *
 * crosstab returns:
 *					<===== values columns =====>
 *			rowid	cat1	cat2	cat3	cat4
 *			------+-------+-------+-------+-------
 *			row1	val1	val2	val3	val4
 *			row2	val5	val6	val7	val8
 *
 * NOTES:
 * 1. SQL result must be ordered by 1,2.
 * 2. The number of values columns depends on the tuple description
 *	  of the function's declared return type.  The return type's columns
 *	  must match the datatypes of the SQL query's result.  The datatype
 *	  of the category column can be anything, however.
 * 3. Missing values (i.e. not enough adjacent rows of same rowid to
 *	  fill the number of result values columns) are filled in with nulls.
 * 4. Extra values (i.e. too many adjacent rows of same rowid to fill
 *	  the number of result values columns) are skipped.
 * 5. Rows with all nulls in the values columns are skipped.
 */
PG_FUNCTION_INFO_V1(crosstab);
Datum
crosstab(PG_FUNCTION_ARGS)
{
	char	   *sql = text_to_cstring(PG_GETARG_TEXT_PP(0));
	ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
	Tuplestorestate *tupstore;
	TupleDesc	tupdesc;
	int			call_cntr;
	int			max_calls;
	AttInMetadata *attinmeta;
	SPITupleTable *spi_tuptable;
	TupleDesc	spi_tupdesc;
	bool		firstpass;
	char	   *lastrowid;
	int			i;
	int			num_categories;
	MemoryContext per_query_ctx;
	MemoryContext oldcontext;
	int			ret;
	int			proc;

	/* check to see if caller supports us returning a tuplestore */
	if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("set-valued function called in context that cannot accept a set")));
	if (!(rsinfo->allowedModes & SFRM_Materialize))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("materialize mode required, but it is not " \
						"allowed in this context")));

	per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;

	/* Connect to SPI manager */
	if ((ret = SPI_connect()) < 0)
		/* internal error */
		elog(ERROR, "crosstab: SPI_connect returned %d", ret);

	/* Retrieve the desired rows */
	ret = SPI_execute(sql, true, 0);
	proc = SPI_processed;

	/* If no qualifying tuples, fall out early */
	if (ret != SPI_OK_SELECT || proc <= 0)
	{
		SPI_finish();
		rsinfo->isDone = ExprEndResult;
		PG_RETURN_NULL();
	}

	spi_tuptable = SPI_tuptable;
	spi_tupdesc = spi_tuptable->tupdesc;

	/*----------
	 * The provided SQL query must always return three columns.
	 *
	 * 1. rowname
	 *	the label or identifier for each row in the final result
	 * 2. category
	 *	the label or identifier for each column in the final result
	 * 3. values
	 *	the value for each column in the final result
	 *----------
	 */
	if (spi_tupdesc->natts != 3)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("invalid source data SQL statement"),
				 errdetail("The provided SQL must return 3 "
						   "columns: rowid, category, and values.")));

	/* get a tuple descriptor for our result type */
	switch (get_call_result_type(fcinfo, NULL, &tupdesc))
	{
		case TYPEFUNC_COMPOSITE:
			/* success */
			break;
		case TYPEFUNC_RECORD:
			/* failed to determine actual type of RECORD */
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("function returning record called in context "
							"that cannot accept type record")));
			break;
		default:
			/* result type isn't composite */
			elog(ERROR, "return type must be a row type");
			break;
	}

	/*
	 * Check that return tupdesc is compatible with the data we got from SPI,
	 * at least based on number and type of attributes
	 */
	if (!compatCrosstabTupleDescs(tupdesc, spi_tupdesc))
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("return and sql tuple descriptions are " \
						"incompatible")));

	/*
	 * switch to long-lived memory context
	 */
	oldcontext = MemoryContextSwitchTo(per_query_ctx);

	/* make sure we have a persistent copy of the result tupdesc */
	tupdesc = CreateTupleDescCopy(tupdesc);

	/* initialize our tuplestore in long-lived context */
	tupstore =
		tuplestore_begin_heap(rsinfo->allowedModes & SFRM_Materialize_Random,
							  false, work_mem);

	MemoryContextSwitchTo(oldcontext);

	/*
	 * Generate attribute metadata needed later to produce tuples from raw C
	 * strings
	 */
	attinmeta = TupleDescGetAttInMetadata(tupdesc);

	/* total number of tuples to be examined */
	max_calls = proc;

	/* the return tuple always must have 1 rowid + num_categories columns */
	num_categories = tupdesc->natts - 1;

	firstpass = true;
	lastrowid = NULL;

	for (call_cntr = 0; call_cntr < max_calls; call_cntr++)
	{
		bool		skip_tuple = false;
		char	  **values;

		/* allocate and zero space */
		values = (char **) palloc0((1 + num_categories) * sizeof(char *));

		/*
		 * now loop through the sql results and assign each value in sequence
		 * to the next category
		 */
		for (i = 0; i < num_categories; i++)
		{
			HeapTuple	spi_tuple;
			char	   *rowid;

			/* see if we've gone too far already */
			if (call_cntr >= max_calls)
				break;

			/* get the next sql result tuple */
			spi_tuple = spi_tuptable->vals[call_cntr];

			/* get the rowid from the current sql result tuple */
			rowid = SPI_getvalue(spi_tuple, spi_tupdesc, 1);

			/*
			 * If this is the first pass through the values for this rowid,
			 * set the first column to rowid
			 */
			if (i == 0)
			{
				xpstrdup(values[0], rowid);

				/*
				 * Check to see if the rowid is the same as that of the last
				 * tuple sent -- if so, skip this tuple entirely
				 */
				if (!firstpass && xstreq(lastrowid, rowid))
				{
					xpfree(rowid);
					skip_tuple = true;
					break;
				}
			}

			/*
			 * If rowid hasn't changed on us, continue building the output
			 * tuple.
			 */
			if (xstreq(rowid, values[0]))
			{
				/*
				 * Get the next category item value, which is always attribute
				 * number three.
				 *
				 * Be careful to assign the value to the array index based on
				 * which category we are presently processing.
				 */
				values[1 + i] = SPI_getvalue(spi_tuple, spi_tupdesc, 3);

				/*
				 * increment the counter since we consume a row for each
				 * category, but not for last pass because the outer loop will
				 * do that for us
				 */
				if (i < (num_categories - 1))
					call_cntr++;
				xpfree(rowid);
			}
			else
			{
				/*
				 * We'll fill in NULLs for the missing values, but we need to
				 * decrement the counter since this sql result row doesn't
				 * belong to the current output tuple.
				 */
				call_cntr--;
				xpfree(rowid);
				break;
			}
		}

		if (!skip_tuple)
		{
			HeapTuple	tuple;

			/* build the tuple and store it */
			tuple = BuildTupleFromCStrings(attinmeta, values);
			tuplestore_puttuple(tupstore, tuple);
			heap_freetuple(tuple);
		}

		/* Remember current rowid */
		xpfree(lastrowid);
		xpstrdup(lastrowid, values[0]);
		firstpass = false;

		/* Clean up */
		for (i = 0; i < num_categories + 1; i++)
			if (values[i] != NULL)
				pfree(values[i]);
		pfree(values);
	}

	/* let the caller know we're sending back a tuplestore */
	rsinfo->returnMode = SFRM_Materialize;
	rsinfo->setResult = tupstore;
	rsinfo->setDesc = tupdesc;

	/* release SPI related resources (and return to caller's context) */
	SPI_finish();

	return (Datum) 0;
}

/*
 * crosstab_hash - reimplement crosstab as materialized function and
 * properly deal with missing values (i.e. don't pack remaining
 * values to the left)
 *
 * crosstab - create a crosstab of rowids and values columns from a
 * SQL statement returning one rowid column, one category column,
 * and one value column.
 *
 * e.g. given sql which produces:
 *
 *			rowid	cat		value
 *			------+-------+-------
 *			row1	cat1	val1
 *			row1	cat2	val2
 *			row1	cat4	val4
 *			row2	cat1	val5
 *			row2	cat2	val6
 *			row2	cat3	val7
 *			row2	cat4	val8
 *
 * crosstab returns:
 *					<===== values columns =====>
 *			rowid	cat1	cat2	cat3	cat4
 *			------+-------+-------+-------+-------
 *			row1	val1	val2	null	val4
 *			row2	val5	val6	val7	val8
 *
 * NOTES:
 * 1. SQL result must be ordered by 1.
 * 2. The number of values columns depends on the tuple description
 *	  of the function's declared return type.
 * 3. Missing values (i.e. missing category) are filled in with nulls.
 * 4. Extra values (i.e. not in category results) are skipped.
 */
PG_FUNCTION_INFO_V1(crosstab_hash);
Datum
crosstab_hash(PG_FUNCTION_ARGS)
{
	char	   *sql = text_to_cstring(PG_GETARG_TEXT_PP(0));
	char	   *cats_sql = text_to_cstring(PG_GETARG_TEXT_PP(1));
	ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
	TupleDesc	tupdesc;
	MemoryContext per_query_ctx;
	MemoryContext oldcontext;
	HTAB	   *crosstab_hash;

	/* check to see if caller supports us returning a tuplestore */
	if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("set-valued function called in context that cannot accept a set")));
	if (!(rsinfo->allowedModes & SFRM_Materialize) ||
		rsinfo->expectedDesc == NULL)
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("materialize mode required, but it is not " \
						"allowed in this context")));

	per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
	oldcontext = MemoryContextSwitchTo(per_query_ctx);

	/* get the requested return tuple description */
	tupdesc = CreateTupleDescCopy(rsinfo->expectedDesc);

	/*
	 * Check to make sure we have a reasonable tuple descriptor
	 *
	 * Note we will attempt to coerce the values into whatever the return
	 * attribute type is and depend on the "in" function to complain if
	 * needed.
	 */
	if (tupdesc->natts < 2)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("query-specified return tuple and " \
						"crosstab function are not compatible")));

	/* load up the categories hash table */
	crosstab_hash = load_categories_hash(cats_sql, per_query_ctx);

	/* let the caller know we're sending back a tuplestore */
	rsinfo->returnMode = SFRM_Materialize;

	/* now go build it */
	rsinfo->setResult = get_crosstab_tuplestore(sql,
												crosstab_hash,
												tupdesc,
												per_query_ctx,
							 rsinfo->allowedModes & SFRM_Materialize_Random);

	/*
	 * SFRM_Materialize mode expects us to return a NULL Datum. The actual
	 * tuples are in our tuplestore and passed back through rsinfo->setResult.
	 * rsinfo->setDesc is set to the tuple description that we actually used
	 * to build our tuples with, so the caller can verify we did what it was
	 * expecting.
	 */
	rsinfo->setDesc = tupdesc;
	MemoryContextSwitchTo(oldcontext);

	return (Datum) 0;
}

/*
 * load up the categories hash table
 */
static HTAB *
load_categories_hash(char *cats_sql, MemoryContext per_query_ctx)
{
	HTAB	   *crosstab_hash;
	HASHCTL		ctl;
	int			ret;
	int			proc;
	MemoryContext SPIcontext;

	/* initialize the category hash table */
	MemSet(&ctl, 0, sizeof(ctl));
	ctl.keysize = MAX_CATNAME_LEN;
	ctl.entrysize = sizeof(crosstab_HashEnt);
	ctl.hcxt = per_query_ctx;

	/*
	 * use INIT_CATS, defined above as a guess of how many hash table entries
	 * to create, initially
	 */
	crosstab_hash = hash_create("crosstab hash",
								INIT_CATS,
								&ctl,
								HASH_ELEM | HASH_CONTEXT);

	/* Connect to SPI manager */
	if ((ret = SPI_connect()) < 0)
		/* internal error */
		elog(ERROR, "load_categories_hash: SPI_connect returned %d", ret);

	/* Retrieve the category name rows */
	ret = SPI_execute(cats_sql, true, 0);
	proc = SPI_processed;

	/* Check for qualifying tuples */
	if ((ret == SPI_OK_SELECT) && (proc > 0))
	{
		SPITupleTable *spi_tuptable = SPI_tuptable;
		TupleDesc	spi_tupdesc = spi_tuptable->tupdesc;
		int			i;

		/*
		 * The provided categories SQL query must always return one column:
		 * category - the label or identifier for each column
		 */
		if (spi_tupdesc->natts != 1)
			ereport(ERROR,
					(errcode(ERRCODE_SYNTAX_ERROR),
					 errmsg("provided \"categories\" SQL must " \
							"return 1 column of at least one row")));

		for (i = 0; i < proc; i++)
		{
			crosstab_cat_desc *catdesc;
			char	   *catname;
			HeapTuple	spi_tuple;

			/* get the next sql result tuple */
			spi_tuple = spi_tuptable->vals[i];

			/* get the category from the current sql result tuple */
			catname = SPI_getvalue(spi_tuple, spi_tupdesc, 1);

			SPIcontext = MemoryContextSwitchTo(per_query_ctx);

			catdesc = (crosstab_cat_desc *) palloc(sizeof(crosstab_cat_desc));
			catdesc->catname = catname;
			catdesc->attidx = i;

			/* Add the proc description block to the hashtable */
			crosstab_HashTableInsert(crosstab_hash, catdesc);

			MemoryContextSwitchTo(SPIcontext);
		}
	}

	if (SPI_finish() != SPI_OK_FINISH)
		/* internal error */
		elog(ERROR, "load_categories_hash: SPI_finish() failed");

	return crosstab_hash;
}

/*
 * create and populate the crosstab tuplestore using the provided source query
 */
static Tuplestorestate *
get_crosstab_tuplestore(char *sql,
						HTAB *crosstab_hash,
						TupleDesc tupdesc,
						MemoryContext per_query_ctx,
						bool randomAccess)
{
	Tuplestorestate *tupstore;
	int			num_categories = hash_get_num_entries(crosstab_hash);
	AttInMetadata *attinmeta = TupleDescGetAttInMetadata(tupdesc);
	char	  **values;
	HeapTuple	tuple;
	int			ret;
	int			proc;

	/* initialize our tuplestore (while still in query context!) */
	tupstore = tuplestore_begin_heap(randomAccess, false, work_mem);

	/* Connect to SPI manager */
	if ((ret = SPI_connect()) < 0)
		/* internal error */
		elog(ERROR, "get_crosstab_tuplestore: SPI_connect returned %d", ret);

	/* Now retrieve the crosstab source rows */
	ret = SPI_execute(sql, true, 0);
	proc = SPI_processed;

	/* Check for qualifying tuples */
	if ((ret == SPI_OK_SELECT) && (proc > 0))
	{
		SPITupleTable *spi_tuptable = SPI_tuptable;
		TupleDesc	spi_tupdesc = spi_tuptable->tupdesc;
		int			ncols = spi_tupdesc->natts;
		char	   *rowid;
		char	   *lastrowid = NULL;
		bool		firstpass = true;
		int			i,
					j;
		int			result_ncols;

		if (num_categories == 0)
		{
			/* no qualifying category tuples */
			ereport(ERROR,
					(errcode(ERRCODE_SYNTAX_ERROR),
					 errmsg("provided \"categories\" SQL must " \
							"return 1 column of at least one row")));
		}

		/*
		 * The provided SQL query must always return at least three columns:
		 *
		 * 1. rowname	the label for each row - column 1 in the final result
		 * 2. category	the label for each value-column in the final result 3.
		 * value	 the values used to populate the value-columns
		 *
		 * If there are more than three columns, the last two are taken as
		 * "category" and "values". The first column is taken as "rowname".
		 * Additional columns (2 thru N-2) are assumed the same for the same
		 * "rowname", and are copied into the result tuple from the first time
		 * we encounter a particular rowname.
		 */
		if (ncols < 3)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("invalid source data SQL statement"),
					 errdetail("The provided SQL must return 3 " \
							   " columns; rowid, category, and values.")));

		result_ncols = (ncols - 2) + num_categories;

		/* Recheck to make sure we tuple descriptor still looks reasonable */
		if (tupdesc->natts != result_ncols)
			ereport(ERROR,
					(errcode(ERRCODE_SYNTAX_ERROR),
					 errmsg("invalid return type"),
					 errdetail("Query-specified return " \
							   "tuple has %d columns but crosstab " \
							   "returns %d.", tupdesc->natts, result_ncols)));

		/* allocate space */
		values = (char **) palloc(result_ncols * sizeof(char *));

		/* and make sure it's clear */
		memset(values, '\0', result_ncols * sizeof(char *));

		for (i = 0; i < proc; i++)
		{
			HeapTuple	spi_tuple;
			crosstab_cat_desc *catdesc;
			char	   *catname;

			/* get the next sql result tuple */
			spi_tuple = spi_tuptable->vals[i];

			/* get the rowid from the current sql result tuple */
			rowid = SPI_getvalue(spi_tuple, spi_tupdesc, 1);

			/*
			 * if we're on a new output row, grab the column values up to
			 * column N-2 now
			 */
			if (firstpass || !xstreq(lastrowid, rowid))
			{
				/*
				 * a new row means we need to flush the old one first, unless
				 * we're on the very first row
				 */
				if (!firstpass)
				{
					/* rowid changed, flush the previous output row */
					tuple = BuildTupleFromCStrings(attinmeta, values);

					tuplestore_puttuple(tupstore, tuple);

					for (j = 0; j < result_ncols; j++)
						xpfree(values[j]);
				}

				values[0] = rowid;
				for (j = 1; j < ncols - 2; j++)
					values[j] = SPI_getvalue(spi_tuple, spi_tupdesc, j + 1);

				/* we're no longer on the first pass */
				firstpass = false;
			}

			/* look up the category and fill in the appropriate column */
			catname = SPI_getvalue(spi_tuple, spi_tupdesc, ncols - 1);

			if (catname != NULL)
			{
				crosstab_HashTableLookup(crosstab_hash, catname, catdesc);

				if (catdesc)
					values[catdesc->attidx + ncols - 2] =
						SPI_getvalue(spi_tuple, spi_tupdesc, ncols);
			}

			xpfree(lastrowid);
			xpstrdup(lastrowid, rowid);
		}

		/* flush the last output row */
		tuple = BuildTupleFromCStrings(attinmeta, values);

		tuplestore_puttuple(tupstore, tuple);
	}

	if (SPI_finish() != SPI_OK_FINISH)
		/* internal error */
		elog(ERROR, "get_crosstab_tuplestore: SPI_finish() failed");

	tuplestore_donestoring(tupstore);

	return tupstore;
}

/*
 * connectby_text - produce a result set from a hierarchical (parent/child)
 * table.
 *
 * e.g. given table foo:
 *
 *			keyid	parent_keyid pos
 *			------+------------+--
 *			row1	NULL		 0
 *			row2	row1		 0
 *			row3	row1		 0
 *			row4	row2		 1
 *			row5	row2		 0
 *			row6	row4		 0
 *			row7	row3		 0
 *			row8	row6		 0
 *			row9	row5		 0
 *
 *
 * connectby(text relname, text keyid_fld, text parent_keyid_fld
 *			  [, text orderby_fld], text start_with, int max_depth
 *			  [, text branch_delim])
 * connectby('foo', 'keyid', 'parent_keyid', 'pos', 'row2', 0, '~') returns:
 *
 *		keyid	parent_id	level	 branch				serial
 *		------+-----------+--------+-----------------------
 *		row2	NULL		  0		  row2				  1
 *		row5	row2		  1		  row2~row5			  2
 *		row9	row5		  2		  row2~row5~row9	  3
 *		row4	row2		  1		  row2~row4			  4
 *		row6	row4		  2		  row2~row4~row6	  5
 *		row8	row6		  3		  row2~row4~row6~row8 6
 *
 */
PG_FUNCTION_INFO_V1(connectby_text);

#define CONNECTBY_NCOLS					4
#define CONNECTBY_NCOLS_NOBRANCH		3

Datum
connectby_text(PG_FUNCTION_ARGS)
{
	char	   *relname = text_to_cstring(PG_GETARG_TEXT_PP(0));
	char	   *key_fld = text_to_cstring(PG_GETARG_TEXT_PP(1));
	char	   *parent_key_fld = text_to_cstring(PG_GETARG_TEXT_PP(2));
	char	   *start_with = text_to_cstring(PG_GETARG_TEXT_PP(3));
	int			max_depth = PG_GETARG_INT32(4);
	char	   *branch_delim = NULL;
	bool		show_branch = false;
	bool		show_serial = false;
	ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
	TupleDesc	tupdesc;
	AttInMetadata *attinmeta;
	MemoryContext per_query_ctx;
	MemoryContext oldcontext;

	/* check to see if caller supports us returning a tuplestore */
	if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("set-valued function called in context that cannot accept a set")));
	if (!(rsinfo->allowedModes & SFRM_Materialize) ||
		rsinfo->expectedDesc == NULL)
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("materialize mode required, but it is not " \
						"allowed in this context")));

	if (fcinfo->nargs == 6)
	{
		branch_delim = text_to_cstring(PG_GETARG_TEXT_PP(5));
		show_branch = true;
	}
	else
		/* default is no show, tilde for the delimiter */
		branch_delim = pstrdup("~");

	per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
	oldcontext = MemoryContextSwitchTo(per_query_ctx);

	/* get the requested return tuple description */
	tupdesc = CreateTupleDescCopy(rsinfo->expectedDesc);

	/* does it meet our needs */
	validateConnectbyTupleDesc(tupdesc, show_branch, show_serial);

	/* OK, use it then */
	attinmeta = TupleDescGetAttInMetadata(tupdesc);

	/* OK, go to work */
	rsinfo->returnMode = SFRM_Materialize;
	rsinfo->setResult = connectby(relname,
								  key_fld,
								  parent_key_fld,
								  NULL,
								  branch_delim,
								  start_with,
								  max_depth,
								  show_branch,
								  show_serial,
								  per_query_ctx,
							  rsinfo->allowedModes & SFRM_Materialize_Random,
								  attinmeta);
	rsinfo->setDesc = tupdesc;

	MemoryContextSwitchTo(oldcontext);

	/*
	 * SFRM_Materialize mode expects us to return a NULL Datum. The actual
	 * tuples are in our tuplestore and passed back through rsinfo->setResult.
	 * rsinfo->setDesc is set to the tuple description that we actually used
	 * to build our tuples with, so the caller can verify we did what it was
	 * expecting.
	 */
	return (Datum) 0;
}

PG_FUNCTION_INFO_V1(connectby_text_serial);
Datum
connectby_text_serial(PG_FUNCTION_ARGS)
{
	char	   *relname = text_to_cstring(PG_GETARG_TEXT_PP(0));
	char	   *key_fld = text_to_cstring(PG_GETARG_TEXT_PP(1));
	char	   *parent_key_fld = text_to_cstring(PG_GETARG_TEXT_PP(2));
	char	   *orderby_fld = text_to_cstring(PG_GETARG_TEXT_PP(3));
	char	   *start_with = text_to_cstring(PG_GETARG_TEXT_PP(4));
	int			max_depth = PG_GETARG_INT32(5);
	char	   *branch_delim = NULL;
	bool		show_branch = false;
	bool		show_serial = true;
	ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo;
	TupleDesc	tupdesc;
	AttInMetadata *attinmeta;
	MemoryContext per_query_ctx;
	MemoryContext oldcontext;

	/* check to see if caller supports us returning a tuplestore */
	if (rsinfo == NULL || !IsA(rsinfo, ReturnSetInfo))
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("set-valued function called in context that cannot accept a set")));
	if (!(rsinfo->allowedModes & SFRM_Materialize) ||
		rsinfo->expectedDesc == NULL)
		ereport(ERROR,
				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
				 errmsg("materialize mode required, but it is not " \
						"allowed in this context")));

	if (fcinfo->nargs == 7)
	{
		branch_delim = text_to_cstring(PG_GETARG_TEXT_PP(6));
		show_branch = true;
	}
	else
		/* default is no show, tilde for the delimiter */
		branch_delim = pstrdup("~");

	per_query_ctx = rsinfo->econtext->ecxt_per_query_memory;
	oldcontext = MemoryContextSwitchTo(per_query_ctx);

	/* get the requested return tuple description */
	tupdesc = CreateTupleDescCopy(rsinfo->expectedDesc);

	/* does it meet our needs */
	validateConnectbyTupleDesc(tupdesc, show_branch, show_serial);

	/* OK, use it then */
	attinmeta = TupleDescGetAttInMetadata(tupdesc);

	/* OK, go to work */
	rsinfo->returnMode = SFRM_Materialize;
	rsinfo->setResult = connectby(relname,
								  key_fld,
								  parent_key_fld,
								  orderby_fld,
								  branch_delim,
								  start_with,
								  max_depth,
								  show_branch,
								  show_serial,
								  per_query_ctx,
							  rsinfo->allowedModes & SFRM_Materialize_Random,
								  attinmeta);
	rsinfo->setDesc = tupdesc;

	MemoryContextSwitchTo(oldcontext);

	/*
	 * SFRM_Materialize mode expects us to return a NULL Datum. The actual
	 * tuples are in our tuplestore and passed back through rsinfo->setResult.
	 * rsinfo->setDesc is set to the tuple description that we actually used
	 * to build our tuples with, so the caller can verify we did what it was
	 * expecting.
	 */
	return (Datum) 0;
}


/*
 * connectby - does the real work for connectby_text()
 */
static Tuplestorestate *
connectby(char *relname,
		  char *key_fld,
		  char *parent_key_fld,
		  char *orderby_fld,
		  char *branch_delim,
		  char *start_with,
		  int max_depth,
		  bool show_branch,
		  bool show_serial,
		  MemoryContext per_query_ctx,
		  bool randomAccess,
		  AttInMetadata *attinmeta)
{
	Tuplestorestate *tupstore = NULL;
	int			ret;
	MemoryContext oldcontext;

	int			serial = 1;

	/* Connect to SPI manager */
	if ((ret = SPI_connect()) < 0)
		/* internal error */
		elog(ERROR, "connectby: SPI_connect returned %d", ret);

	/* switch to longer term context to create the tuple store */
	oldcontext = MemoryContextSwitchTo(per_query_ctx);

	/* initialize our tuplestore */
	tupstore = tuplestore_begin_heap(randomAccess, false, work_mem);

	MemoryContextSwitchTo(oldcontext);

	/* now go get the whole tree */
	tupstore = build_tuplestore_recursively(key_fld,
											parent_key_fld,
											relname,
											orderby_fld,
											branch_delim,
											start_with,
											start_with, /* current_branch */
											0,	/* initial level is 0 */
											&serial,	/* initial serial is 1 */
											max_depth,
											show_branch,
											show_serial,
											per_query_ctx,
											attinmeta,
											tupstore);

	SPI_finish();

	return tupstore;
}

static Tuplestorestate *
build_tuplestore_recursively(char *key_fld,
							 char *parent_key_fld,
							 char *relname,
							 char *orderby_fld,
							 char *branch_delim,
							 char *start_with,
							 char *branch,
							 int level,
							 int *serial,
							 int max_depth,
							 bool show_branch,
							 bool show_serial,
							 MemoryContext per_query_ctx,
							 AttInMetadata *attinmeta,
							 Tuplestorestate *tupstore)
{
	TupleDesc	tupdesc = attinmeta->tupdesc;
	int			ret;
	int			proc;
	int			serial_column;
	StringInfoData sql;
	char	  **values;
	char	   *current_key;
	char	   *current_key_parent;
	char		current_level[INT32_STRLEN];
	char		serial_str[INT32_STRLEN];
	char	   *current_branch;
	HeapTuple	tuple;

	if (max_depth > 0 && level > max_depth)
		return tupstore;

	initStringInfo(&sql);

	/* Build initial sql statement */
	if (!show_serial)
	{
		appendStringInfo(&sql, "SELECT %s, %s FROM %s WHERE %s = %s AND %s IS NOT NULL AND %s <> %s",
						 key_fld,
						 parent_key_fld,
						 relname,
						 parent_key_fld,
						 quote_literal_cstr(start_with),
						 key_fld, key_fld, parent_key_fld);
		serial_column = 0;
	}
	else
	{
		appendStringInfo(&sql, "SELECT %s, %s FROM %s WHERE %s = %s AND %s IS NOT NULL AND %s <> %s ORDER BY %s",
						 key_fld,
						 parent_key_fld,
						 relname,
						 parent_key_fld,
						 quote_literal_cstr(start_with),
						 key_fld, key_fld, parent_key_fld,
						 orderby_fld);
		serial_column = 1;
	}

	if (show_branch)
		values = (char **) palloc((CONNECTBY_NCOLS + serial_column) * sizeof(char *));
	else
		values = (char **) palloc((CONNECTBY_NCOLS_NOBRANCH + serial_column) * sizeof(char *));

	/* First time through, do a little setup */
	if (level == 0)
	{
		/* root value is the one we initially start with */
		values[0] = start_with;

		/* root value has no parent */
		values[1] = NULL;

		/* root level is 0 */
		sprintf(current_level, "%d", level);
		values[2] = current_level;

		/* root branch is just starting root value */
		if (show_branch)
			values[3] = start_with;

		/* root starts the serial with 1 */
		if (show_serial)
		{
			sprintf(serial_str, "%d", (*serial)++);
			if (show_branch)
				values[4] = serial_str;
			else
				values[3] = serial_str;
		}

		/* construct the tuple */
		tuple = BuildTupleFromCStrings(attinmeta, values);

		/* now store it */
		tuplestore_puttuple(tupstore, tuple);

		/* increment level */
		level++;
	}

	/* Retrieve the desired rows */
	ret = SPI_execute(sql.data, true, 0);
	proc = SPI_processed;

	/* Check for qualifying tuples */
	if ((ret == SPI_OK_SELECT) && (proc > 0))
	{
		HeapTuple	spi_tuple;
		SPITupleTable *tuptable = SPI_tuptable;
		TupleDesc	spi_tupdesc = tuptable->tupdesc;
		int			i;
		StringInfoData branchstr;
		StringInfoData chk_branchstr;
		StringInfoData chk_current_key;

		/* First time through, do a little more setup */
		if (level == 0)
		{
			/*
			 * Check that return tupdesc is compatible with the one we got
			 * from the query, but only at level 0 -- no need to check more
			 * than once
			 */

			if (!compatConnectbyTupleDescs(tupdesc, spi_tupdesc))
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("invalid return type"),
						 errdetail("Return and SQL tuple descriptions are " \
								   "incompatible.")));
		}

		initStringInfo(&branchstr);
		initStringInfo(&chk_branchstr);
		initStringInfo(&chk_current_key);

		for (i = 0; i < proc; i++)
		{
			/* initialize branch for this pass */
			appendStringInfo(&branchstr, "%s", branch);
			appendStringInfo(&chk_branchstr, "%s%s%s", branch_delim, branch, branch_delim);

			/* get the next sql result tuple */
			spi_tuple = tuptable->vals[i];

			/* get the current key and parent */
			current_key = SPI_getvalue(spi_tuple, spi_tupdesc, 1);
			appendStringInfo(&chk_current_key, "%s%s%s", branch_delim, current_key, branch_delim);
			current_key_parent = pstrdup(SPI_getvalue(spi_tuple, spi_tupdesc, 2));

			/* get the current level */
			sprintf(current_level, "%d", level);

			/* check to see if this key is also an ancestor */
			if (strstr(chk_branchstr.data, chk_current_key.data))
				elog(ERROR, "infinite recursion detected");

			/* OK, extend the branch */
			appendStringInfo(&branchstr, "%s%s", branch_delim, current_key);
			current_branch = branchstr.data;

			/* build a tuple */
			values[0] = pstrdup(current_key);
			values[1] = current_key_parent;
			values[2] = current_level;
			if (show_branch)
				values[3] = current_branch;
			if (show_serial)
			{
				sprintf(serial_str, "%d", (*serial)++);
				if (show_branch)
					values[4] = serial_str;
				else
					values[3] = serial_str;
			}

			tuple = BuildTupleFromCStrings(attinmeta, values);

			xpfree(current_key);
			xpfree(current_key_parent);

			/* store the tuple for later use */
			tuplestore_puttuple(tupstore, tuple);

			heap_freetuple(tuple);

			/* recurse using current_key_parent as the new start_with */
			tupstore = build_tuplestore_recursively(key_fld,
													parent_key_fld,
													relname,
													orderby_fld,
													branch_delim,
													values[0],
													current_branch,
													level + 1,
													serial,
													max_depth,
													show_branch,
													show_serial,
													per_query_ctx,
													attinmeta,
													tupstore);

			/* reset branch for next pass */
			resetStringInfo(&branchstr);
			resetStringInfo(&chk_branchstr);
			resetStringInfo(&chk_current_key);
		}

		xpfree(branchstr.data);
		xpfree(chk_branchstr.data);
		xpfree(chk_current_key.data);
	}

	return tupstore;
}

/*
 * Check expected (query runtime) tupdesc suitable for Connectby
 */
static void
validateConnectbyTupleDesc(TupleDesc tupdesc, bool show_branch, bool show_serial)
{
	int			serial_column = 0;

	if (show_serial)
		serial_column = 1;

	/* are there the correct number of columns */
	if (show_branch)
	{
		if (tupdesc->natts != (CONNECTBY_NCOLS + serial_column))
			ereport(ERROR,
					(errcode(ERRCODE_SYNTAX_ERROR),
					 errmsg("invalid return type"),
					 errdetail("Query-specified return tuple has " \
							   "wrong number of columns.")));
	}
	else
	{
		if (tupdesc->natts != CONNECTBY_NCOLS_NOBRANCH + serial_column)
			ereport(ERROR,
					(errcode(ERRCODE_SYNTAX_ERROR),
					 errmsg("invalid return type"),
					 errdetail("Query-specified return tuple has " \
							   "wrong number of columns.")));
	}

	/* check that the types of the first two columns match */
	if (tupdesc->attrs[0]->atttypid != tupdesc->attrs[1]->atttypid)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("First two columns must be the same type.")));

	/* check that the type of the third column is INT4 */
	if (tupdesc->attrs[2]->atttypid != INT4OID)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("Third column must be type %s.",
						   format_type_be(INT4OID))));

	/* check that the type of the fourth column is TEXT if applicable */
	if (show_branch && tupdesc->attrs[3]->atttypid != TEXTOID)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("Fourth column must be type %s.",
						   format_type_be(TEXTOID))));

	/* check that the type of the fifth column is INT4 */
	if (show_branch && show_serial && tupdesc->attrs[4]->atttypid != INT4OID)
		elog(ERROR, "query-specified return tuple not valid for Connectby: "
			 "fifth column must be type %s", format_type_be(INT4OID));

	/* check that the type of the fifth column is INT4 */
	if (!show_branch && show_serial && tupdesc->attrs[3]->atttypid != INT4OID)
		elog(ERROR, "query-specified return tuple not valid for Connectby: "
			 "fourth column must be type %s", format_type_be(INT4OID));

	/* OK, the tupdesc is valid for our purposes */
}

/*
 * Check if spi sql tupdesc and return tupdesc are compatible
 */
static bool
compatConnectbyTupleDescs(TupleDesc ret_tupdesc, TupleDesc sql_tupdesc)
{
	Oid			ret_atttypid;
	Oid			sql_atttypid;

	/* check the key_fld types match */
	ret_atttypid = ret_tupdesc->attrs[0]->atttypid;
	sql_atttypid = sql_tupdesc->attrs[0]->atttypid;
	if (ret_atttypid != sql_atttypid)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("SQL key field datatype does " \
						   "not match return key field datatype.")));

	/* check the parent_key_fld types match */
	ret_atttypid = ret_tupdesc->attrs[1]->atttypid;
	sql_atttypid = sql_tupdesc->attrs[1]->atttypid;
	if (ret_atttypid != sql_atttypid)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("SQL parent key field datatype does " \
						   "not match return parent key field datatype.")));

	/* OK, the two tupdescs are compatible for our purposes */
	return true;
}

/*
 * Check if two tupdescs match in type of attributes
 */
static bool
compatCrosstabTupleDescs(TupleDesc ret_tupdesc, TupleDesc sql_tupdesc)
{
	int			i;
	Form_pg_attribute ret_attr;
	Oid			ret_atttypid;
	Form_pg_attribute sql_attr;
	Oid			sql_atttypid;

	if (ret_tupdesc->natts < 2 ||
		sql_tupdesc->natts < 3)
		return false;

	/* check the rowid types match */
	ret_atttypid = ret_tupdesc->attrs[0]->atttypid;
	sql_atttypid = sql_tupdesc->attrs[0]->atttypid;
	if (ret_atttypid != sql_atttypid)
		ereport(ERROR,
				(errcode(ERRCODE_SYNTAX_ERROR),
				 errmsg("invalid return type"),
				 errdetail("SQL rowid datatype does not match " \
						   "return rowid datatype.")));

	/*
	 * - attribute [1] of the sql tuple is the category; no need to check it -
	 * attribute [2] of the sql tuple should match attributes [1] to [natts]
	 * of the return tuple
	 */
	sql_attr = sql_tupdesc->attrs[2];
	for (i = 1; i < ret_tupdesc->natts; i++)
	{
		ret_attr = ret_tupdesc->attrs[i];

		if (ret_attr->atttypid != sql_attr->atttypid)
			return false;
	}

	/* OK, the two tupdescs are compatible for our purposes */
	return true;
}

/*
 * Return a properly quoted literal value.
 * Uses quote_literal in quote.c
 */
static char *
quote_literal_cstr(char *rawstr)
{
	text	   *rawstr_text;
	text	   *result_text;
	char	   *result;

	rawstr_text = cstring_to_text(rawstr);
	result_text = DatumGetTextP(DirectFunctionCall1(quote_literal,
											  PointerGetDatum(rawstr_text)));
	result = text_to_cstring(result_text);

	return result;
}