summaryrefslogtreecommitdiffstats
path: root/src/backend/replication/pgoutput/pgoutput.c
blob: 5fc0defec3a0503a2f483c504bb07a4228afc8d1 (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
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
/*-------------------------------------------------------------------------
 *
 * pgoutput.c
 *		Logical Replication output plugin
 *
 * Copyright (c) 2012-2022, PostgreSQL Global Development Group
 *
 * IDENTIFICATION
 *		  src/backend/replication/pgoutput/pgoutput.c
 *
 *-------------------------------------------------------------------------
 */
#include "postgres.h"

#include "access/tupconvert.h"
#include "catalog/partition.h"
#include "catalog/pg_publication.h"
#include "catalog/pg_publication_rel.h"
#include "commands/defrem.h"
#include "executor/executor.h"
#include "fmgr.h"
#include "nodes/makefuncs.h"
#include "optimizer/optimizer.h"
#include "replication/logical.h"
#include "replication/logicalproto.h"
#include "replication/origin.h"
#include "replication/pgoutput.h"
#include "utils/builtins.h"
#include "utils/inval.h"
#include "utils/lsyscache.h"
#include "utils/memutils.h"
#include "utils/rel.h"
#include "utils/syscache.h"
#include "utils/varlena.h"

PG_MODULE_MAGIC;

extern void _PG_output_plugin_init(OutputPluginCallbacks *cb);

static void pgoutput_startup(LogicalDecodingContext *ctx,
							 OutputPluginOptions *opt, bool is_init);
static void pgoutput_shutdown(LogicalDecodingContext *ctx);
static void pgoutput_begin_txn(LogicalDecodingContext *ctx,
							   ReorderBufferTXN *txn);
static void pgoutput_commit_txn(LogicalDecodingContext *ctx,
								ReorderBufferTXN *txn, XLogRecPtr commit_lsn);
static void pgoutput_change(LogicalDecodingContext *ctx,
							ReorderBufferTXN *txn, Relation rel,
							ReorderBufferChange *change);
static void pgoutput_truncate(LogicalDecodingContext *ctx,
							  ReorderBufferTXN *txn, int nrelations, Relation relations[],
							  ReorderBufferChange *change);
static void pgoutput_message(LogicalDecodingContext *ctx,
							 ReorderBufferTXN *txn, XLogRecPtr message_lsn,
							 bool transactional, const char *prefix,
							 Size sz, const char *message);
static bool pgoutput_origin_filter(LogicalDecodingContext *ctx,
								   RepOriginId origin_id);
static void pgoutput_begin_prepare_txn(LogicalDecodingContext *ctx,
									   ReorderBufferTXN *txn);
static void pgoutput_prepare_txn(LogicalDecodingContext *ctx,
								 ReorderBufferTXN *txn, XLogRecPtr prepare_lsn);
static void pgoutput_commit_prepared_txn(LogicalDecodingContext *ctx,
										 ReorderBufferTXN *txn, XLogRecPtr commit_lsn);
static void pgoutput_rollback_prepared_txn(LogicalDecodingContext *ctx,
										   ReorderBufferTXN *txn,
										   XLogRecPtr prepare_end_lsn,
										   TimestampTz prepare_time);
static void pgoutput_stream_start(struct LogicalDecodingContext *ctx,
								  ReorderBufferTXN *txn);
static void pgoutput_stream_stop(struct LogicalDecodingContext *ctx,
								 ReorderBufferTXN *txn);
static void pgoutput_stream_abort(struct LogicalDecodingContext *ctx,
								  ReorderBufferTXN *txn,
								  XLogRecPtr abort_lsn);
static void pgoutput_stream_commit(struct LogicalDecodingContext *ctx,
								   ReorderBufferTXN *txn,
								   XLogRecPtr commit_lsn);
static void pgoutput_stream_prepare_txn(LogicalDecodingContext *ctx,
										ReorderBufferTXN *txn, XLogRecPtr prepare_lsn);

static bool publications_valid;
static bool in_streaming;

static List *LoadPublications(List *pubnames);
static void publication_invalidation_cb(Datum arg, int cacheid,
										uint32 hashvalue);
static void send_relation_and_attrs(Relation relation, TransactionId xid,
									LogicalDecodingContext *ctx,
									Bitmapset *columns);
static void send_repl_origin(LogicalDecodingContext *ctx,
							 RepOriginId origin_id, XLogRecPtr origin_lsn,
							 bool send_origin);
static void update_replication_progress(LogicalDecodingContext *ctx,
										bool skipped_xact);

/*
 * Only 3 publication actions are used for row filtering ("insert", "update",
 * "delete"). See RelationSyncEntry.exprstate[].
 */
enum RowFilterPubAction
{
	PUBACTION_INSERT,
	PUBACTION_UPDATE,
	PUBACTION_DELETE
};

#define NUM_ROWFILTER_PUBACTIONS (PUBACTION_DELETE+1)

/*
 * Entry in the map used to remember which relation schemas we sent.
 *
 * The schema_sent flag determines if the current schema record for the
 * relation (and for its ancestor if publish_as_relid is set) was already
 * sent to the subscriber (in which case we don't need to send it again).
 *
 * The schema cache on downstream is however updated only at commit time,
 * and with streamed transactions the commit order may be different from
 * the order the transactions are sent in. Also, the (sub) transactions
 * might get aborted so we need to send the schema for each (sub) transaction
 * so that we don't lose the schema information on abort. For handling this,
 * we maintain the list of xids (streamed_txns) for those we have already sent
 * the schema.
 *
 * For partitions, 'pubactions' considers not only the table's own
 * publications, but also those of all of its ancestors.
 */
typedef struct RelationSyncEntry
{
	Oid			relid;			/* relation oid */

	bool		replicate_valid;	/* overall validity flag for entry */

	bool		schema_sent;
	List	   *streamed_txns;	/* streamed toplevel transactions with this
								 * schema */

	/* are we publishing this rel? */
	PublicationActions pubactions;

	/*
	 * ExprState array for row filter. Different publication actions don't
	 * allow multiple expressions to always be combined into one, because
	 * updates or deletes restrict the column in expression to be part of the
	 * replica identity index whereas inserts do not have this restriction, so
	 * there is one ExprState per publication action.
	 */
	ExprState  *exprstate[NUM_ROWFILTER_PUBACTIONS];
	EState	   *estate;			/* executor state used for row filter */
	TupleTableSlot *new_slot;	/* slot for storing new tuple */
	TupleTableSlot *old_slot;	/* slot for storing old tuple */

	/*
	 * OID of the relation to publish changes as.  For a partition, this may
	 * be set to one of its ancestors whose schema will be used when
	 * replicating changes, if publish_via_partition_root is set for the
	 * publication.
	 */
	Oid			publish_as_relid;

	/*
	 * Map used when replicating using an ancestor's schema to convert tuples
	 * from partition's type to the ancestor's; NULL if publish_as_relid is
	 * same as 'relid' or if unnecessary due to partition and the ancestor
	 * having identical TupleDesc.
	 */
	AttrMap    *attrmap;

	/*
	 * Columns included in the publication, or NULL if all columns are
	 * included implicitly.  Note that the attnums in this bitmap are not
	 * shifted by FirstLowInvalidHeapAttributeNumber.
	 */
	Bitmapset  *columns;

	/*
	 * Private context to store additional data for this entry - state for the
	 * row filter expressions, column list, etc.
	 */
	MemoryContext entry_cxt;
} RelationSyncEntry;

/*
 * Maintain a per-transaction level variable to track whether the transaction
 * has sent BEGIN. BEGIN is only sent when the first change in a transaction
 * is processed. This makes it possible to skip sending a pair of BEGIN/COMMIT
 * messages for empty transactions which saves network bandwidth.
 *
 * This optimization is not used for prepared transactions because if the
 * WALSender restarts after prepare of a transaction and before commit prepared
 * of the same transaction then we won't be able to figure out if we have
 * skipped sending BEGIN/PREPARE of a transaction as it was empty. This is
 * because we would have lost the in-memory txndata information that was
 * present prior to the restart. This will result in sending a spurious
 * COMMIT PREPARED without a corresponding prepared transaction at the
 * downstream which would lead to an error when it tries to process it.
 *
 * XXX We could achieve this optimization by changing protocol to send
 * additional information so that downstream can detect that the corresponding
 * prepare has not been sent. However, adding such a check for every
 * transaction in the downstream could be costly so we might want to do it
 * optionally.
 *
 * We also don't have this optimization for streamed transactions because
 * they can contain prepared transactions.
 */
typedef struct PGOutputTxnData
{
	bool		sent_begin_txn; /* flag indicating whether BEGIN has been sent */
} PGOutputTxnData;

/* Map used to remember which relation schemas we sent. */
static HTAB *RelationSyncCache = NULL;

static void init_rel_sync_cache(MemoryContext decoding_context);
static void cleanup_rel_sync_cache(TransactionId xid, bool is_commit);
static RelationSyncEntry *get_rel_sync_entry(PGOutputData *data,
											 Relation relation);
static void rel_sync_cache_relation_cb(Datum arg, Oid relid);
static void rel_sync_cache_publication_cb(Datum arg, int cacheid,
										  uint32 hashvalue);
static void set_schema_sent_in_streamed_txn(RelationSyncEntry *entry,
											TransactionId xid);
static bool get_schema_sent_in_streamed_txn(RelationSyncEntry *entry,
											TransactionId xid);
static void init_tuple_slot(PGOutputData *data, Relation relation,
							RelationSyncEntry *entry);

/* row filter routines */
static EState *create_estate_for_relation(Relation rel);
static void pgoutput_row_filter_init(PGOutputData *data,
									 List *publications,
									 RelationSyncEntry *entry);
static bool pgoutput_row_filter_exec_expr(ExprState *state,
										  ExprContext *econtext);
static bool pgoutput_row_filter(Relation relation, TupleTableSlot *old_slot,
								TupleTableSlot **new_slot_ptr,
								RelationSyncEntry *entry,
								ReorderBufferChangeType *action);

/* column list routines */
static void pgoutput_column_list_init(PGOutputData *data,
									  List *publications,
									  RelationSyncEntry *entry);

/*
 * Specify output plugin callbacks
 */
void
_PG_output_plugin_init(OutputPluginCallbacks *cb)
{
	AssertVariableIsOfType(&_PG_output_plugin_init, LogicalOutputPluginInit);

	cb->startup_cb = pgoutput_startup;
	cb->begin_cb = pgoutput_begin_txn;
	cb->change_cb = pgoutput_change;
	cb->truncate_cb = pgoutput_truncate;
	cb->message_cb = pgoutput_message;
	cb->commit_cb = pgoutput_commit_txn;

	cb->begin_prepare_cb = pgoutput_begin_prepare_txn;
	cb->prepare_cb = pgoutput_prepare_txn;
	cb->commit_prepared_cb = pgoutput_commit_prepared_txn;
	cb->rollback_prepared_cb = pgoutput_rollback_prepared_txn;
	cb->filter_by_origin_cb = pgoutput_origin_filter;
	cb->shutdown_cb = pgoutput_shutdown;

	/* transaction streaming */
	cb->stream_start_cb = pgoutput_stream_start;
	cb->stream_stop_cb = pgoutput_stream_stop;
	cb->stream_abort_cb = pgoutput_stream_abort;
	cb->stream_commit_cb = pgoutput_stream_commit;
	cb->stream_change_cb = pgoutput_change;
	cb->stream_message_cb = pgoutput_message;
	cb->stream_truncate_cb = pgoutput_truncate;
	/* transaction streaming - two-phase commit */
	cb->stream_prepare_cb = pgoutput_stream_prepare_txn;
}

static void
parse_output_parameters(List *options, PGOutputData *data)
{
	ListCell   *lc;
	bool		protocol_version_given = false;
	bool		publication_names_given = false;
	bool		binary_option_given = false;
	bool		messages_option_given = false;
	bool		streaming_given = false;
	bool		two_phase_option_given = false;

	data->binary = false;
	data->streaming = false;
	data->messages = false;
	data->two_phase = false;

	foreach(lc, options)
	{
		DefElem    *defel = (DefElem *) lfirst(lc);

		Assert(defel->arg == NULL || IsA(defel->arg, String));

		/* Check each param, whether or not we recognize it */
		if (strcmp(defel->defname, "proto_version") == 0)
		{
			unsigned long parsed;
			char	   *endptr;

			if (protocol_version_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			protocol_version_given = true;

			errno = 0;
			parsed = strtoul(strVal(defel->arg), &endptr, 10);
			if (errno != 0 || *endptr != '\0')
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("invalid proto_version")));

			if (parsed > PG_UINT32_MAX)
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
						 errmsg("proto_version \"%s\" out of range",
								strVal(defel->arg))));

			data->protocol_version = (uint32) parsed;
		}
		else if (strcmp(defel->defname, "publication_names") == 0)
		{
			if (publication_names_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			publication_names_given = true;

			if (!SplitIdentifierString(strVal(defel->arg), ',',
									   &data->publication_names))
				ereport(ERROR,
						(errcode(ERRCODE_INVALID_NAME),
						 errmsg("invalid publication_names syntax")));
		}
		else if (strcmp(defel->defname, "binary") == 0)
		{
			if (binary_option_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			binary_option_given = true;

			data->binary = defGetBoolean(defel);
		}
		else if (strcmp(defel->defname, "messages") == 0)
		{
			if (messages_option_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			messages_option_given = true;

			data->messages = defGetBoolean(defel);
		}
		else if (strcmp(defel->defname, "streaming") == 0)
		{
			if (streaming_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			streaming_given = true;

			data->streaming = defGetBoolean(defel);
		}
		else if (strcmp(defel->defname, "two_phase") == 0)
		{
			if (two_phase_option_given)
				ereport(ERROR,
						(errcode(ERRCODE_SYNTAX_ERROR),
						 errmsg("conflicting or redundant options")));
			two_phase_option_given = true;

			data->two_phase = defGetBoolean(defel);
		}
		else
			elog(ERROR, "unrecognized pgoutput option: %s", defel->defname);
	}
}

/*
 * Initialize this plugin
 */
static void
pgoutput_startup(LogicalDecodingContext *ctx, OutputPluginOptions *opt,
				 bool is_init)
{
	PGOutputData *data = palloc0(sizeof(PGOutputData));
	static bool publication_callback_registered = false;

	/* Create our memory context for private allocations. */
	data->context = AllocSetContextCreate(ctx->context,
										  "logical replication output context",
										  ALLOCSET_DEFAULT_SIZES);

	data->cachectx = AllocSetContextCreate(ctx->context,
										   "logical replication cache context",
										   ALLOCSET_DEFAULT_SIZES);

	ctx->output_plugin_private = data;

	/* This plugin uses binary protocol. */
	opt->output_type = OUTPUT_PLUGIN_BINARY_OUTPUT;

	/*
	 * This is replication start and not slot initialization.
	 *
	 * Parse and validate options passed by the client.
	 */
	if (!is_init)
	{
		/* Parse the params and ERROR if we see any we don't recognize */
		parse_output_parameters(ctx->output_plugin_options, data);

		/* Check if we support requested protocol */
		if (data->protocol_version > LOGICALREP_PROTO_MAX_VERSION_NUM)
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("client sent proto_version=%d but we only support protocol %d or lower",
							data->protocol_version, LOGICALREP_PROTO_MAX_VERSION_NUM)));

		if (data->protocol_version < LOGICALREP_PROTO_MIN_VERSION_NUM)
			ereport(ERROR,
					(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					 errmsg("client sent proto_version=%d but we only support protocol %d or higher",
							data->protocol_version, LOGICALREP_PROTO_MIN_VERSION_NUM)));

		if (list_length(data->publication_names) < 1)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("publication_names parameter missing")));

		/*
		 * Decide whether to enable streaming. It is disabled by default, in
		 * which case we just update the flag in decoding context. Otherwise
		 * we only allow it with sufficient version of the protocol, and when
		 * the output plugin supports it.
		 */
		if (!data->streaming)
			ctx->streaming = false;
		else if (data->protocol_version < LOGICALREP_PROTO_STREAM_VERSION_NUM)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("requested proto_version=%d does not support streaming, need %d or higher",
							data->protocol_version, LOGICALREP_PROTO_STREAM_VERSION_NUM)));
		else if (!ctx->streaming)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("streaming requested, but not supported by output plugin")));

		/* Also remember we're currently not streaming any transaction. */
		in_streaming = false;

		/*
		 * Here, we just check whether the two-phase option is passed by
		 * plugin and decide whether to enable it at later point of time. It
		 * remains enabled if the previous start-up has done so. But we only
		 * allow the option to be passed in with sufficient version of the
		 * protocol, and when the output plugin supports it.
		 */
		if (!data->two_phase)
			ctx->twophase_opt_given = false;
		else if (data->protocol_version < LOGICALREP_PROTO_TWOPHASE_VERSION_NUM)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("requested proto_version=%d does not support two-phase commit, need %d or higher",
							data->protocol_version, LOGICALREP_PROTO_TWOPHASE_VERSION_NUM)));
		else if (!ctx->twophase)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("two-phase commit requested, but not supported by output plugin")));
		else
			ctx->twophase_opt_given = true;

		/* Init publication state. */
		data->publications = NIL;
		publications_valid = false;

		/*
		 * Register callback for pg_publication if we didn't already do that
		 * during some previous call in this process.
		 */
		if (!publication_callback_registered)
		{
			CacheRegisterSyscacheCallback(PUBLICATIONOID,
										  publication_invalidation_cb,
										  (Datum) 0);
			publication_callback_registered = true;
		}

		/* Initialize relation schema cache. */
		init_rel_sync_cache(CacheMemoryContext);
	}
	else
	{
		/*
		 * Disable the streaming and prepared transactions during the slot
		 * initialization mode.
		 */
		ctx->streaming = false;
		ctx->twophase = false;
	}
}

/*
 * BEGIN callback.
 *
 * Don't send the BEGIN message here instead postpone it until the first
 * change. In logical replication, a common scenario is to replicate a set of
 * tables (instead of all tables) and transactions whose changes were on
 * the table(s) that are not published will produce empty transactions. These
 * empty transactions will send BEGIN and COMMIT messages to subscribers,
 * using bandwidth on something with little/no use for logical replication.
 */
static void
pgoutput_begin_txn(LogicalDecodingContext *ctx, ReorderBufferTXN *txn)
{
	PGOutputTxnData *txndata = MemoryContextAllocZero(ctx->context,
													  sizeof(PGOutputTxnData));

	txn->output_plugin_private = txndata;
}

/*
 * Send BEGIN.
 *
 * This is called while processing the first change of the transaction.
 */
static void
pgoutput_send_begin(LogicalDecodingContext *ctx, ReorderBufferTXN *txn)
{
	bool		send_replication_origin = txn->origin_id != InvalidRepOriginId;
	PGOutputTxnData *txndata = (PGOutputTxnData *) txn->output_plugin_private;

	Assert(txndata);
	Assert(!txndata->sent_begin_txn);

	OutputPluginPrepareWrite(ctx, !send_replication_origin);
	logicalrep_write_begin(ctx->out, txn);
	txndata->sent_begin_txn = true;

	send_repl_origin(ctx, txn->origin_id, txn->origin_lsn,
					 send_replication_origin);

	OutputPluginWrite(ctx, true);
}

/*
 * COMMIT callback
 */
static void
pgoutput_commit_txn(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
					XLogRecPtr commit_lsn)
{
	PGOutputTxnData *txndata = (PGOutputTxnData *) txn->output_plugin_private;
	bool		sent_begin_txn;

	Assert(txndata);

	/*
	 * We don't need to send the commit message unless some relevant change
	 * from this transaction has been sent to the downstream.
	 */
	sent_begin_txn = txndata->sent_begin_txn;
	update_replication_progress(ctx, !sent_begin_txn);
	pfree(txndata);
	txn->output_plugin_private = NULL;

	if (!sent_begin_txn)
	{
		elog(DEBUG1, "skipped replication of an empty transaction with XID: %u", txn->xid);
		return;
	}

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_commit(ctx->out, txn, commit_lsn);
	OutputPluginWrite(ctx, true);
}

/*
 * BEGIN PREPARE callback
 */
static void
pgoutput_begin_prepare_txn(LogicalDecodingContext *ctx, ReorderBufferTXN *txn)
{
	bool		send_replication_origin = txn->origin_id != InvalidRepOriginId;

	OutputPluginPrepareWrite(ctx, !send_replication_origin);
	logicalrep_write_begin_prepare(ctx->out, txn);

	send_repl_origin(ctx, txn->origin_id, txn->origin_lsn,
					 send_replication_origin);

	OutputPluginWrite(ctx, true);
}

/*
 * PREPARE callback
 */
static void
pgoutput_prepare_txn(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
					 XLogRecPtr prepare_lsn)
{
	update_replication_progress(ctx, false);

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_prepare(ctx->out, txn, prepare_lsn);
	OutputPluginWrite(ctx, true);
}

/*
 * COMMIT PREPARED callback
 */
static void
pgoutput_commit_prepared_txn(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
							 XLogRecPtr commit_lsn)
{
	update_replication_progress(ctx, false);

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_commit_prepared(ctx->out, txn, commit_lsn);
	OutputPluginWrite(ctx, true);
}

/*
 * ROLLBACK PREPARED callback
 */
static void
pgoutput_rollback_prepared_txn(LogicalDecodingContext *ctx,
							   ReorderBufferTXN *txn,
							   XLogRecPtr prepare_end_lsn,
							   TimestampTz prepare_time)
{
	update_replication_progress(ctx, false);

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_rollback_prepared(ctx->out, txn, prepare_end_lsn,
									   prepare_time);
	OutputPluginWrite(ctx, true);
}

/*
 * Write the current schema of the relation and its ancestor (if any) if not
 * done yet.
 */
static void
maybe_send_schema(LogicalDecodingContext *ctx,
				  ReorderBufferChange *change,
				  Relation relation, RelationSyncEntry *relentry)
{
	bool		schema_sent;
	TransactionId xid = InvalidTransactionId;
	TransactionId topxid = InvalidTransactionId;

	/*
	 * Remember XID of the (sub)transaction for the change. We don't care if
	 * it's top-level transaction or not (we have already sent that XID in
	 * start of the current streaming block).
	 *
	 * If we're not in a streaming block, just use InvalidTransactionId and
	 * the write methods will not include it.
	 */
	if (in_streaming)
		xid = change->txn->xid;

	if (change->txn->toptxn)
		topxid = change->txn->toptxn->xid;
	else
		topxid = xid;

	/*
	 * Do we need to send the schema? We do track streamed transactions
	 * separately, because those may be applied later (and the regular
	 * transactions won't see their effects until then) and in an order that
	 * we don't know at this point.
	 *
	 * XXX There is a scope of optimization here. Currently, we always send
	 * the schema first time in a streaming transaction but we can probably
	 * avoid that by checking 'relentry->schema_sent' flag. However, before
	 * doing that we need to study its impact on the case where we have a mix
	 * of streaming and non-streaming transactions.
	 */
	if (in_streaming)
		schema_sent = get_schema_sent_in_streamed_txn(relentry, topxid);
	else
		schema_sent = relentry->schema_sent;

	/* Nothing to do if we already sent the schema. */
	if (schema_sent)
		return;

	/*
	 * Send the schema.  If the changes will be published using an ancestor's
	 * schema, not the relation's own, send that ancestor's schema before
	 * sending relation's own (XXX - maybe sending only the former suffices?).
	 */
	if (relentry->publish_as_relid != RelationGetRelid(relation))
	{
		Relation	ancestor = RelationIdGetRelation(relentry->publish_as_relid);

		send_relation_and_attrs(ancestor, xid, ctx, relentry->columns);
		RelationClose(ancestor);
	}

	send_relation_and_attrs(relation, xid, ctx, relentry->columns);

	if (in_streaming)
		set_schema_sent_in_streamed_txn(relentry, topxid);
	else
		relentry->schema_sent = true;
}

/*
 * Sends a relation
 */
static void
send_relation_and_attrs(Relation relation, TransactionId xid,
						LogicalDecodingContext *ctx,
						Bitmapset *columns)
{
	TupleDesc	desc = RelationGetDescr(relation);
	int			i;

	/*
	 * Write out type info if needed.  We do that only for user-created types.
	 * We use FirstGenbkiObjectId as the cutoff, so that we only consider
	 * objects with hand-assigned OIDs to be "built in", not for instance any
	 * function or type defined in the information_schema. This is important
	 * because only hand-assigned OIDs can be expected to remain stable across
	 * major versions.
	 */
	for (i = 0; i < desc->natts; i++)
	{
		Form_pg_attribute att = TupleDescAttr(desc, i);

		if (att->attisdropped || att->attgenerated)
			continue;

		if (att->atttypid < FirstGenbkiObjectId)
			continue;

		/* Skip this attribute if it's not present in the column list */
		if (columns != NULL && !bms_is_member(att->attnum, columns))
			continue;

		OutputPluginPrepareWrite(ctx, false);
		logicalrep_write_typ(ctx->out, xid, att->atttypid);
		OutputPluginWrite(ctx, false);
	}

	OutputPluginPrepareWrite(ctx, false);
	logicalrep_write_rel(ctx->out, xid, relation, columns);
	OutputPluginWrite(ctx, false);
}

/*
 * Executor state preparation for evaluation of row filter expressions for the
 * specified relation.
 */
static EState *
create_estate_for_relation(Relation rel)
{
	EState	   *estate;
	RangeTblEntry *rte;

	estate = CreateExecutorState();

	rte = makeNode(RangeTblEntry);
	rte->rtekind = RTE_RELATION;
	rte->relid = RelationGetRelid(rel);
	rte->relkind = rel->rd_rel->relkind;
	rte->rellockmode = AccessShareLock;
	ExecInitRangeTable(estate, list_make1(rte));

	estate->es_output_cid = GetCurrentCommandId(false);

	return estate;
}

/*
 * Evaluates row filter.
 *
 * If the row filter evaluates to NULL, it is taken as false i.e. the change
 * isn't replicated.
 */
static bool
pgoutput_row_filter_exec_expr(ExprState *state, ExprContext *econtext)
{
	Datum		ret;
	bool		isnull;

	Assert(state != NULL);

	ret = ExecEvalExprSwitchContext(state, econtext, &isnull);

	elog(DEBUG3, "row filter evaluates to %s (isnull: %s)",
		 isnull ? "false" : DatumGetBool(ret) ? "true" : "false",
		 isnull ? "true" : "false");

	if (isnull)
		return false;

	return DatumGetBool(ret);
}

/*
 * Make sure the per-entry memory context exists.
 */
static void
pgoutput_ensure_entry_cxt(PGOutputData *data, RelationSyncEntry *entry)
{
	Relation	relation;

	/* The context may already exist, in which case bail out. */
	if (entry->entry_cxt)
		return;

	relation = RelationIdGetRelation(entry->publish_as_relid);

	entry->entry_cxt = AllocSetContextCreate(data->cachectx,
											 "entry private context",
											 ALLOCSET_SMALL_SIZES);

	MemoryContextCopyAndSetIdentifier(entry->entry_cxt,
									  RelationGetRelationName(relation));
}

/*
 * Initialize the row filter.
 */
static void
pgoutput_row_filter_init(PGOutputData *data, List *publications,
						 RelationSyncEntry *entry)
{
	ListCell   *lc;
	List	   *rfnodes[] = {NIL, NIL, NIL};	/* One per pubaction */
	bool		no_filter[] = {false, false, false};	/* One per pubaction */
	MemoryContext oldctx;
	int			idx;
	bool		has_filter = true;
	Oid			schemaid = get_rel_namespace(entry->publish_as_relid);

	/*
	 * Find if there are any row filters for this relation. If there are, then
	 * prepare the necessary ExprState and cache it in entry->exprstate. To
	 * build an expression state, we need to ensure the following:
	 *
	 * All the given publication-table mappings must be checked.
	 *
	 * Multiple publications might have multiple row filters for this
	 * relation. Since row filter usage depends on the DML operation, there
	 * are multiple lists (one for each operation) to which row filters will
	 * be appended.
	 *
	 * FOR ALL TABLES and FOR TABLES IN SCHEMA implies "don't use row
	 * filter expression" so it takes precedence.
	 */
	foreach(lc, publications)
	{
		Publication *pub = lfirst(lc);
		HeapTuple	rftuple = NULL;
		Datum		rfdatum = 0;
		bool		pub_no_filter = true;

		/*
		 * If the publication is FOR ALL TABLES, or the publication includes a
		 * FOR TABLES IN SCHEMA where the table belongs to the referred
		 * schema, then it is treated the same as if there are no row filters
		 * (even if other publications have a row filter).
		 */
		if (!pub->alltables &&
			!SearchSysCacheExists2(PUBLICATIONNAMESPACEMAP,
								   ObjectIdGetDatum(schemaid),
								   ObjectIdGetDatum(pub->oid)))
		{
			/*
			 * Check for the presence of a row filter in this publication.
			 */
			rftuple = SearchSysCache2(PUBLICATIONRELMAP,
									  ObjectIdGetDatum(entry->publish_as_relid),
									  ObjectIdGetDatum(pub->oid));

			if (HeapTupleIsValid(rftuple))
			{
				/* Null indicates no filter. */
				rfdatum = SysCacheGetAttr(PUBLICATIONRELMAP, rftuple,
										  Anum_pg_publication_rel_prqual,
										  &pub_no_filter);
			}
		}

		if (pub_no_filter)
		{
			if (rftuple)
				ReleaseSysCache(rftuple);

			no_filter[PUBACTION_INSERT] |= pub->pubactions.pubinsert;
			no_filter[PUBACTION_UPDATE] |= pub->pubactions.pubupdate;
			no_filter[PUBACTION_DELETE] |= pub->pubactions.pubdelete;

			/*
			 * Quick exit if all the DML actions are publicized via this
			 * publication.
			 */
			if (no_filter[PUBACTION_INSERT] &&
				no_filter[PUBACTION_UPDATE] &&
				no_filter[PUBACTION_DELETE])
			{
				has_filter = false;
				break;
			}

			/* No additional work for this publication. Next one. */
			continue;
		}

		/* Form the per pubaction row filter lists. */
		if (pub->pubactions.pubinsert && !no_filter[PUBACTION_INSERT])
			rfnodes[PUBACTION_INSERT] = lappend(rfnodes[PUBACTION_INSERT],
												TextDatumGetCString(rfdatum));
		if (pub->pubactions.pubupdate && !no_filter[PUBACTION_UPDATE])
			rfnodes[PUBACTION_UPDATE] = lappend(rfnodes[PUBACTION_UPDATE],
												TextDatumGetCString(rfdatum));
		if (pub->pubactions.pubdelete && !no_filter[PUBACTION_DELETE])
			rfnodes[PUBACTION_DELETE] = lappend(rfnodes[PUBACTION_DELETE],
												TextDatumGetCString(rfdatum));

		ReleaseSysCache(rftuple);
	}							/* loop all subscribed publications */

	/* Clean the row filter */
	for (idx = 0; idx < NUM_ROWFILTER_PUBACTIONS; idx++)
	{
		if (no_filter[idx])
		{
			list_free_deep(rfnodes[idx]);
			rfnodes[idx] = NIL;
		}
	}

	if (has_filter)
	{
		Relation	relation = RelationIdGetRelation(entry->publish_as_relid);

		pgoutput_ensure_entry_cxt(data, entry);

		/*
		 * Now all the filters for all pubactions are known. Combine them when
		 * their pubactions are the same.
		 */
		oldctx = MemoryContextSwitchTo(entry->entry_cxt);
		entry->estate = create_estate_for_relation(relation);
		for (idx = 0; idx < NUM_ROWFILTER_PUBACTIONS; idx++)
		{
			List	   *filters = NIL;
			Expr	   *rfnode;

			if (rfnodes[idx] == NIL)
				continue;

			foreach(lc, rfnodes[idx])
				filters = lappend(filters, stringToNode((char *) lfirst(lc)));

			/* combine the row filter and cache the ExprState */
			rfnode = make_orclause(filters);
			entry->exprstate[idx] = ExecPrepareExpr(rfnode, entry->estate);
		}						/* for each pubaction */
		MemoryContextSwitchTo(oldctx);

		RelationClose(relation);
	}
}

/*
 * Initialize the column list.
 */
static void
pgoutput_column_list_init(PGOutputData *data, List *publications,
						  RelationSyncEntry *entry)
{
	ListCell   *lc;
	bool		first = true;
	Relation	relation = RelationIdGetRelation(entry->publish_as_relid);

	/*
	 * Find if there are any column lists for this relation. If there are,
	 * build a bitmap using the column lists.
	 *
	 * Multiple publications might have multiple column lists for this
	 * relation.
	 *
	 * Note that we don't support the case where the column list is different
	 * for the same table when combining publications. See comments atop
	 * fetch_table_list. But one can later change the publication so we still
	 * need to check all the given publication-table mappings and report an
	 * error if any publications have a different column list.
	 *
	 * FOR ALL TABLES and FOR TABLES IN SCHEMA imply "don't use column list".
	 */
	foreach(lc, publications)
	{
		Publication *pub = lfirst(lc);
		HeapTuple	cftuple = NULL;
		Datum		cfdatum = 0;
		Bitmapset  *cols = NULL;

		/*
		 * If the publication is FOR ALL TABLES then it is treated the same as
		 * if there are no column lists (even if other publications have a
		 * list).
		 */
		if (!pub->alltables)
		{
			bool		pub_no_list = true;

			/*
			 * Check for the presence of a column list in this publication.
			 *
			 * Note: If we find no pg_publication_rel row, it's a publication
			 * defined for a whole schema, so it can't have a column list,
			 * just like a FOR ALL TABLES publication.
			 */
			cftuple = SearchSysCache2(PUBLICATIONRELMAP,
									  ObjectIdGetDatum(entry->publish_as_relid),
									  ObjectIdGetDatum(pub->oid));

			if (HeapTupleIsValid(cftuple))
			{
				/* Lookup the column list attribute. */
				cfdatum = SysCacheGetAttr(PUBLICATIONRELMAP, cftuple,
										  Anum_pg_publication_rel_prattrs,
										  &pub_no_list);

				/* Build the column list bitmap in the per-entry context. */
				if (!pub_no_list)	/* when not null */
				{
					pgoutput_ensure_entry_cxt(data, entry);

					cols = pub_collist_to_bitmapset(cols, cfdatum,
													entry->entry_cxt);

					/*
					 * If column list includes all the columns of the table,
					 * set it to NULL.
					 */
					if (bms_num_members(cols) == RelationGetNumberOfAttributes(relation))
					{
						bms_free(cols);
						cols = NULL;
					}
				}

				ReleaseSysCache(cftuple);
			}
		}

		if (first)
		{
			entry->columns = cols;
			first = false;
		}
		else if (!bms_equal(entry->columns, cols))
			ereport(ERROR,
					errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
					errmsg("cannot use different column lists for table \"%s.%s\" in different publications",
						   get_namespace_name(RelationGetNamespace(relation)),
						   RelationGetRelationName(relation)));
	}							/* loop all subscribed publications */

	RelationClose(relation);
}

/*
 * Initialize the slot for storing new and old tuples, and build the map that
 * will be used to convert the relation's tuples into the ancestor's format.
 */
static void
init_tuple_slot(PGOutputData *data, Relation relation,
				RelationSyncEntry *entry)
{
	MemoryContext oldctx;
	TupleDesc	oldtupdesc;
	TupleDesc	newtupdesc;

	oldctx = MemoryContextSwitchTo(data->cachectx);

	/*
	 * Create tuple table slots. Create a copy of the TupleDesc as it needs to
	 * live as long as the cache remains.
	 */
	oldtupdesc = CreateTupleDescCopyConstr(RelationGetDescr(relation));
	newtupdesc = CreateTupleDescCopyConstr(RelationGetDescr(relation));

	entry->old_slot = MakeSingleTupleTableSlot(oldtupdesc, &TTSOpsHeapTuple);
	entry->new_slot = MakeSingleTupleTableSlot(newtupdesc, &TTSOpsHeapTuple);

	MemoryContextSwitchTo(oldctx);

	/*
	 * Cache the map that will be used to convert the relation's tuples into
	 * the ancestor's format, if needed.
	 */
	if (entry->publish_as_relid != RelationGetRelid(relation))
	{
		Relation	ancestor = RelationIdGetRelation(entry->publish_as_relid);
		TupleDesc	indesc = RelationGetDescr(relation);
		TupleDesc	outdesc = RelationGetDescr(ancestor);

		/* Map must live as long as the session does. */
		oldctx = MemoryContextSwitchTo(CacheMemoryContext);

		entry->attrmap = build_attrmap_by_name_if_req(indesc, outdesc);

		MemoryContextSwitchTo(oldctx);
		RelationClose(ancestor);
	}
}

/*
 * Change is checked against the row filter if any.
 *
 * Returns true if the change is to be replicated, else false.
 *
 * For inserts, evaluate the row filter for new tuple.
 * For deletes, evaluate the row filter for old tuple.
 * For updates, evaluate the row filter for old and new tuple.
 *
 * For updates, if both evaluations are true, we allow sending the UPDATE and
 * if both the evaluations are false, it doesn't replicate the UPDATE. Now, if
 * only one of the tuples matches the row filter expression, we transform
 * UPDATE to DELETE or INSERT to avoid any data inconsistency based on the
 * following rules:
 *
 * Case 1: old-row (no match)    new-row (no match)  -> (drop change)
 * Case 2: old-row (no match)    new row (match)     -> INSERT
 * Case 3: old-row (match)       new-row (no match)  -> DELETE
 * Case 4: old-row (match)       new row (match)     -> UPDATE
 *
 * The new action is updated in the action parameter.
 *
 * The new slot could be updated when transforming the UPDATE into INSERT,
 * because the original new tuple might not have column values from the replica
 * identity.
 *
 * Examples:
 * Let's say the old tuple satisfies the row filter but the new tuple doesn't.
 * Since the old tuple satisfies, the initial table synchronization copied this
 * row (or another method was used to guarantee that there is data
 * consistency).  However, after the UPDATE the new tuple doesn't satisfy the
 * row filter, so from a data consistency perspective, that row should be
 * removed on the subscriber. The UPDATE should be transformed into a DELETE
 * statement and be sent to the subscriber. Keeping this row on the subscriber
 * is undesirable because it doesn't reflect what was defined in the row filter
 * expression on the publisher. This row on the subscriber would likely not be
 * modified by replication again. If someone inserted a new row with the same
 * old identifier, replication could stop due to a constraint violation.
 *
 * Let's say the old tuple doesn't match the row filter but the new tuple does.
 * Since the old tuple doesn't satisfy, the initial table synchronization
 * probably didn't copy this row. However, after the UPDATE the new tuple does
 * satisfy the row filter, so from a data consistency perspective, that row
 * should be inserted on the subscriber. Otherwise, subsequent UPDATE or DELETE
 * statements have no effect (it matches no row -- see
 * apply_handle_update_internal()). So, the UPDATE should be transformed into a
 * INSERT statement and be sent to the subscriber. However, this might surprise
 * someone who expects the data set to satisfy the row filter expression on the
 * provider.
 */
static bool
pgoutput_row_filter(Relation relation, TupleTableSlot *old_slot,
					TupleTableSlot **new_slot_ptr, RelationSyncEntry *entry,
					ReorderBufferChangeType *action)
{
	TupleDesc	desc;
	int			i;
	bool		old_matched,
				new_matched,
				result;
	TupleTableSlot *tmp_new_slot;
	TupleTableSlot *new_slot = *new_slot_ptr;
	ExprContext *ecxt;
	ExprState  *filter_exprstate;

	/*
	 * We need this map to avoid relying on ReorderBufferChangeType enums
	 * having specific values.
	 */
	static const int map_changetype_pubaction[] = {
		[REORDER_BUFFER_CHANGE_INSERT] = PUBACTION_INSERT,
		[REORDER_BUFFER_CHANGE_UPDATE] = PUBACTION_UPDATE,
		[REORDER_BUFFER_CHANGE_DELETE] = PUBACTION_DELETE
	};

	Assert(*action == REORDER_BUFFER_CHANGE_INSERT ||
		   *action == REORDER_BUFFER_CHANGE_UPDATE ||
		   *action == REORDER_BUFFER_CHANGE_DELETE);

	Assert(new_slot || old_slot);

	/* Get the corresponding row filter */
	filter_exprstate = entry->exprstate[map_changetype_pubaction[*action]];

	/* Bail out if there is no row filter */
	if (!filter_exprstate)
		return true;

	elog(DEBUG3, "table \"%s.%s\" has row filter",
		 get_namespace_name(RelationGetNamespace(relation)),
		 RelationGetRelationName(relation));

	ResetPerTupleExprContext(entry->estate);

	ecxt = GetPerTupleExprContext(entry->estate);

	/*
	 * For the following occasions where there is only one tuple, we can
	 * evaluate the row filter for that tuple and return.
	 *
	 * For inserts, we only have the new tuple.
	 *
	 * For updates, we can have only a new tuple when none of the replica
	 * identity columns changed and none of those columns have external data
	 * but we still need to evaluate the row filter for the new tuple as the
	 * existing values of those columns might not match the filter. Also,
	 * users can use constant expressions in the row filter, so we anyway need
	 * to evaluate it for the new tuple.
	 *
	 * For deletes, we only have the old tuple.
	 */
	if (!new_slot || !old_slot)
	{
		ecxt->ecxt_scantuple = new_slot ? new_slot : old_slot;
		result = pgoutput_row_filter_exec_expr(filter_exprstate, ecxt);

		return result;
	}

	/*
	 * Both the old and new tuples must be valid only for updates and need to
	 * be checked against the row filter.
	 */
	Assert(map_changetype_pubaction[*action] == PUBACTION_UPDATE);

	slot_getallattrs(new_slot);
	slot_getallattrs(old_slot);

	tmp_new_slot = NULL;
	desc = RelationGetDescr(relation);

	/*
	 * The new tuple might not have all the replica identity columns, in which
	 * case it needs to be copied over from the old tuple.
	 */
	for (i = 0; i < desc->natts; i++)
	{
		Form_pg_attribute att = TupleDescAttr(desc, i);

		/*
		 * if the column in the new tuple or old tuple is null, nothing to do
		 */
		if (new_slot->tts_isnull[i] || old_slot->tts_isnull[i])
			continue;

		/*
		 * Unchanged toasted replica identity columns are only logged in the
		 * old tuple. Copy this over to the new tuple. The changed (or WAL
		 * Logged) toast values are always assembled in memory and set as
		 * VARTAG_INDIRECT. See ReorderBufferToastReplace.
		 */
		if (att->attlen == -1 &&
			VARATT_IS_EXTERNAL_ONDISK(new_slot->tts_values[i]) &&
			!VARATT_IS_EXTERNAL_ONDISK(old_slot->tts_values[i]))
		{
			if (!tmp_new_slot)
			{
				tmp_new_slot = MakeSingleTupleTableSlot(desc, &TTSOpsVirtual);
				ExecClearTuple(tmp_new_slot);

				memcpy(tmp_new_slot->tts_values, new_slot->tts_values,
					   desc->natts * sizeof(Datum));
				memcpy(tmp_new_slot->tts_isnull, new_slot->tts_isnull,
					   desc->natts * sizeof(bool));
			}

			tmp_new_slot->tts_values[i] = old_slot->tts_values[i];
			tmp_new_slot->tts_isnull[i] = old_slot->tts_isnull[i];
		}
	}

	ecxt->ecxt_scantuple = old_slot;
	old_matched = pgoutput_row_filter_exec_expr(filter_exprstate, ecxt);

	if (tmp_new_slot)
	{
		ExecStoreVirtualTuple(tmp_new_slot);
		ecxt->ecxt_scantuple = tmp_new_slot;
	}
	else
		ecxt->ecxt_scantuple = new_slot;

	new_matched = pgoutput_row_filter_exec_expr(filter_exprstate, ecxt);

	/*
	 * Case 1: if both tuples don't match the row filter, bailout. Send
	 * nothing.
	 */
	if (!old_matched && !new_matched)
		return false;

	/*
	 * Case 2: if the old tuple doesn't satisfy the row filter but the new
	 * tuple does, transform the UPDATE into INSERT.
	 *
	 * Use the newly transformed tuple that must contain the column values for
	 * all the replica identity columns. This is required to ensure that the
	 * while inserting the tuple in the downstream node, we have all the
	 * required column values.
	 */
	if (!old_matched && new_matched)
	{
		*action = REORDER_BUFFER_CHANGE_INSERT;

		if (tmp_new_slot)
			*new_slot_ptr = tmp_new_slot;
	}

	/*
	 * Case 3: if the old tuple satisfies the row filter but the new tuple
	 * doesn't, transform the UPDATE into DELETE.
	 *
	 * This transformation does not require another tuple. The Old tuple will
	 * be used for DELETE.
	 */
	else if (old_matched && !new_matched)
		*action = REORDER_BUFFER_CHANGE_DELETE;

	/*
	 * Case 4: if both tuples match the row filter, transformation isn't
	 * required. (*action is default UPDATE).
	 */

	return true;
}

/*
 * Sends the decoded DML over wire.
 *
 * This is called both in streaming and non-streaming modes.
 */
static void
pgoutput_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
				Relation relation, ReorderBufferChange *change)
{
	PGOutputData *data = (PGOutputData *) ctx->output_plugin_private;
	PGOutputTxnData *txndata = (PGOutputTxnData *) txn->output_plugin_private;
	MemoryContext old;
	RelationSyncEntry *relentry;
	TransactionId xid = InvalidTransactionId;
	Relation	ancestor = NULL;
	Relation	targetrel = relation;
	ReorderBufferChangeType action = change->action;
	TupleTableSlot *old_slot = NULL;
	TupleTableSlot *new_slot = NULL;

	update_replication_progress(ctx, false);

	if (!is_publishable_relation(relation))
		return;

	/*
	 * Remember the xid for the change in streaming mode. We need to send xid
	 * with each change in the streaming mode so that subscriber can make
	 * their association and on aborts, it can discard the corresponding
	 * changes.
	 */
	if (in_streaming)
		xid = change->txn->xid;

	relentry = get_rel_sync_entry(data, relation);

	/* First check the table filter */
	switch (action)
	{
		case REORDER_BUFFER_CHANGE_INSERT:
			if (!relentry->pubactions.pubinsert)
				return;
			break;
		case REORDER_BUFFER_CHANGE_UPDATE:
			if (!relentry->pubactions.pubupdate)
				return;
			break;
		case REORDER_BUFFER_CHANGE_DELETE:
			if (!relentry->pubactions.pubdelete)
				return;
			break;
		default:
			Assert(false);
	}

	/* Avoid leaking memory by using and resetting our own context */
	old = MemoryContextSwitchTo(data->context);

	/* Send the data */
	switch (action)
	{
		case REORDER_BUFFER_CHANGE_INSERT:
			new_slot = relentry->new_slot;
			ExecStoreHeapTuple(&change->data.tp.newtuple->tuple,
							   new_slot, false);

			/* Switch relation if publishing via root. */
			if (relentry->publish_as_relid != RelationGetRelid(relation))
			{
				Assert(relation->rd_rel->relispartition);
				ancestor = RelationIdGetRelation(relentry->publish_as_relid);
				targetrel = ancestor;
				/* Convert tuple if needed. */
				if (relentry->attrmap)
				{
					TupleDesc	tupdesc = RelationGetDescr(targetrel);

					new_slot = execute_attr_map_slot(relentry->attrmap,
													 new_slot,
													 MakeTupleTableSlot(tupdesc, &TTSOpsVirtual));
				}
			}

			/* Check row filter */
			if (!pgoutput_row_filter(targetrel, NULL, &new_slot, relentry,
									 &action))
				break;

			/*
			 * Send BEGIN if we haven't yet.
			 *
			 * We send the BEGIN message after ensuring that we will actually
			 * send the change. This avoids sending a pair of BEGIN/COMMIT
			 * messages for empty transactions.
			 */
			if (txndata && !txndata->sent_begin_txn)
				pgoutput_send_begin(ctx, txn);

			/*
			 * Schema should be sent using the original relation because it
			 * also sends the ancestor's relation.
			 */
			maybe_send_schema(ctx, change, relation, relentry);

			OutputPluginPrepareWrite(ctx, true);
			logicalrep_write_insert(ctx->out, xid, targetrel, new_slot,
									data->binary, relentry->columns);
			OutputPluginWrite(ctx, true);
			break;
		case REORDER_BUFFER_CHANGE_UPDATE:
			if (change->data.tp.oldtuple)
			{
				old_slot = relentry->old_slot;
				ExecStoreHeapTuple(&change->data.tp.oldtuple->tuple,
								   old_slot, false);
			}

			new_slot = relentry->new_slot;
			ExecStoreHeapTuple(&change->data.tp.newtuple->tuple,
							   new_slot, false);

			/* Switch relation if publishing via root. */
			if (relentry->publish_as_relid != RelationGetRelid(relation))
			{
				Assert(relation->rd_rel->relispartition);
				ancestor = RelationIdGetRelation(relentry->publish_as_relid);
				targetrel = ancestor;
				/* Convert tuples if needed. */
				if (relentry->attrmap)
				{
					TupleDesc	tupdesc = RelationGetDescr(targetrel);

					if (old_slot)
						old_slot = execute_attr_map_slot(relentry->attrmap,
														 old_slot,
														 MakeTupleTableSlot(tupdesc, &TTSOpsVirtual));

					new_slot = execute_attr_map_slot(relentry->attrmap,
													 new_slot,
													 MakeTupleTableSlot(tupdesc, &TTSOpsVirtual));
				}
			}

			/* Check row filter */
			if (!pgoutput_row_filter(targetrel, old_slot, &new_slot,
									 relentry, &action))
				break;

			/* Send BEGIN if we haven't yet */
			if (txndata && !txndata->sent_begin_txn)
				pgoutput_send_begin(ctx, txn);

			maybe_send_schema(ctx, change, relation, relentry);

			OutputPluginPrepareWrite(ctx, true);

			/*
			 * Updates could be transformed to inserts or deletes based on the
			 * results of the row filter for old and new tuple.
			 */
			switch (action)
			{
				case REORDER_BUFFER_CHANGE_INSERT:
					logicalrep_write_insert(ctx->out, xid, targetrel,
											new_slot, data->binary,
											relentry->columns);
					break;
				case REORDER_BUFFER_CHANGE_UPDATE:
					logicalrep_write_update(ctx->out, xid, targetrel,
											old_slot, new_slot, data->binary,
											relentry->columns);
					break;
				case REORDER_BUFFER_CHANGE_DELETE:
					logicalrep_write_delete(ctx->out, xid, targetrel,
											old_slot, data->binary,
											relentry->columns);
					break;
				default:
					Assert(false);
			}

			OutputPluginWrite(ctx, true);
			break;
		case REORDER_BUFFER_CHANGE_DELETE:
			if (change->data.tp.oldtuple)
			{
				old_slot = relentry->old_slot;

				ExecStoreHeapTuple(&change->data.tp.oldtuple->tuple,
								   old_slot, false);

				/* Switch relation if publishing via root. */
				if (relentry->publish_as_relid != RelationGetRelid(relation))
				{
					Assert(relation->rd_rel->relispartition);
					ancestor = RelationIdGetRelation(relentry->publish_as_relid);
					targetrel = ancestor;
					/* Convert tuple if needed. */
					if (relentry->attrmap)
					{
						TupleDesc	tupdesc = RelationGetDescr(targetrel);

						old_slot = execute_attr_map_slot(relentry->attrmap,
														 old_slot,
														 MakeTupleTableSlot(tupdesc, &TTSOpsVirtual));
					}
				}

				/* Check row filter */
				if (!pgoutput_row_filter(targetrel, old_slot, &new_slot,
										 relentry, &action))
					break;

				/* Send BEGIN if we haven't yet */
				if (txndata && !txndata->sent_begin_txn)
					pgoutput_send_begin(ctx, txn);

				maybe_send_schema(ctx, change, relation, relentry);

				OutputPluginPrepareWrite(ctx, true);
				logicalrep_write_delete(ctx->out, xid, targetrel,
										old_slot, data->binary,
										relentry->columns);
				OutputPluginWrite(ctx, true);
			}
			else
				elog(DEBUG1, "didn't send DELETE change because of missing oldtuple");
			break;
		default:
			Assert(false);
	}

	if (RelationIsValid(ancestor))
	{
		RelationClose(ancestor);
		ancestor = NULL;
	}

	/* Cleanup */
	MemoryContextSwitchTo(old);
	MemoryContextReset(data->context);
}

static void
pgoutput_truncate(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
				  int nrelations, Relation relations[], ReorderBufferChange *change)
{
	PGOutputData *data = (PGOutputData *) ctx->output_plugin_private;
	PGOutputTxnData *txndata = (PGOutputTxnData *) txn->output_plugin_private;
	MemoryContext old;
	RelationSyncEntry *relentry;
	int			i;
	int			nrelids;
	Oid		   *relids;
	TransactionId xid = InvalidTransactionId;

	update_replication_progress(ctx, false);

	/* Remember the xid for the change in streaming mode. See pgoutput_change. */
	if (in_streaming)
		xid = change->txn->xid;

	old = MemoryContextSwitchTo(data->context);

	relids = palloc0(nrelations * sizeof(Oid));
	nrelids = 0;

	for (i = 0; i < nrelations; i++)
	{
		Relation	relation = relations[i];
		Oid			relid = RelationGetRelid(relation);

		if (!is_publishable_relation(relation))
			continue;

		relentry = get_rel_sync_entry(data, relation);

		if (!relentry->pubactions.pubtruncate)
			continue;

		/*
		 * Don't send partitions if the publication wants to send only the
		 * root tables through it.
		 */
		if (relation->rd_rel->relispartition &&
			relentry->publish_as_relid != relid)
			continue;

		relids[nrelids++] = relid;

		/* Send BEGIN if we haven't yet */
		if (txndata && !txndata->sent_begin_txn)
			pgoutput_send_begin(ctx, txn);

		maybe_send_schema(ctx, change, relation, relentry);
	}

	if (nrelids > 0)
	{
		OutputPluginPrepareWrite(ctx, true);
		logicalrep_write_truncate(ctx->out,
								  xid,
								  nrelids,
								  relids,
								  change->data.truncate.cascade,
								  change->data.truncate.restart_seqs);
		OutputPluginWrite(ctx, true);
	}

	MemoryContextSwitchTo(old);
	MemoryContextReset(data->context);
}

static void
pgoutput_message(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
				 XLogRecPtr message_lsn, bool transactional, const char *prefix, Size sz,
				 const char *message)
{
	PGOutputData *data = (PGOutputData *) ctx->output_plugin_private;
	TransactionId xid = InvalidTransactionId;

	update_replication_progress(ctx, false);

	if (!data->messages)
		return;

	/*
	 * Remember the xid for the message in streaming mode. See
	 * pgoutput_change.
	 */
	if (in_streaming)
		xid = txn->xid;

	/*
	 * Output BEGIN if we haven't yet. Avoid for non-transactional messages.
	 */
	if (transactional)
	{
		PGOutputTxnData *txndata = (PGOutputTxnData *) txn->output_plugin_private;

		/* Send BEGIN if we haven't yet */
		if (txndata && !txndata->sent_begin_txn)
			pgoutput_send_begin(ctx, txn);
	}

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_message(ctx->out,
							 xid,
							 message_lsn,
							 transactional,
							 prefix,
							 sz,
							 message);
	OutputPluginWrite(ctx, true);
}

/*
 * Currently we always forward.
 */
static bool
pgoutput_origin_filter(LogicalDecodingContext *ctx,
					   RepOriginId origin_id)
{
	return false;
}

/*
 * Shutdown the output plugin.
 *
 * Note, we don't need to clean the data->context and data->cachectx as
 * they are child contexts of the ctx->context so they will be cleaned up by
 * logical decoding machinery.
 */
static void
pgoutput_shutdown(LogicalDecodingContext *ctx)
{
	if (RelationSyncCache)
	{
		hash_destroy(RelationSyncCache);
		RelationSyncCache = NULL;
	}
}

/*
 * Load publications from the list of publication names.
 */
static List *
LoadPublications(List *pubnames)
{
	List	   *result = NIL;
	ListCell   *lc;

	foreach(lc, pubnames)
	{
		char	   *pubname = (char *) lfirst(lc);
		Publication *pub = GetPublicationByName(pubname, false);

		result = lappend(result, pub);
	}

	return result;
}

/*
 * Publication syscache invalidation callback.
 *
 * Called for invalidations on pg_publication.
 */
static void
publication_invalidation_cb(Datum arg, int cacheid, uint32 hashvalue)
{
	publications_valid = false;

	/*
	 * Also invalidate per-relation cache so that next time the filtering info
	 * is checked it will be updated with the new publication settings.
	 */
	rel_sync_cache_publication_cb(arg, cacheid, hashvalue);
}

/*
 * START STREAM callback
 */
static void
pgoutput_stream_start(struct LogicalDecodingContext *ctx,
					  ReorderBufferTXN *txn)
{
	bool		send_replication_origin = txn->origin_id != InvalidRepOriginId;

	/* we can't nest streaming of transactions */
	Assert(!in_streaming);

	/*
	 * If we already sent the first stream for this transaction then don't
	 * send the origin id in the subsequent streams.
	 */
	if (rbtxn_is_streamed(txn))
		send_replication_origin = false;

	OutputPluginPrepareWrite(ctx, !send_replication_origin);
	logicalrep_write_stream_start(ctx->out, txn->xid, !rbtxn_is_streamed(txn));

	send_repl_origin(ctx, txn->origin_id, InvalidXLogRecPtr,
					 send_replication_origin);

	OutputPluginWrite(ctx, true);

	/* we're streaming a chunk of transaction now */
	in_streaming = true;
}

/*
 * STOP STREAM callback
 */
static void
pgoutput_stream_stop(struct LogicalDecodingContext *ctx,
					 ReorderBufferTXN *txn)
{
	/* we should be streaming a trasanction */
	Assert(in_streaming);

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_stream_stop(ctx->out);
	OutputPluginWrite(ctx, true);

	/* we've stopped streaming a transaction */
	in_streaming = false;
}

/*
 * Notify downstream to discard the streamed transaction (along with all
 * it's subtransactions, if it's a toplevel transaction).
 */
static void
pgoutput_stream_abort(struct LogicalDecodingContext *ctx,
					  ReorderBufferTXN *txn,
					  XLogRecPtr abort_lsn)
{
	ReorderBufferTXN *toptxn;

	/*
	 * The abort should happen outside streaming block, even for streamed
	 * transactions. The transaction has to be marked as streamed, though.
	 */
	Assert(!in_streaming);

	/* determine the toplevel transaction */
	toptxn = (txn->toptxn) ? txn->toptxn : txn;

	Assert(rbtxn_is_streamed(toptxn));

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_stream_abort(ctx->out, toptxn->xid, txn->xid);
	OutputPluginWrite(ctx, true);

	cleanup_rel_sync_cache(toptxn->xid, false);
}

/*
 * Notify downstream to apply the streamed transaction (along with all
 * it's subtransactions).
 */
static void
pgoutput_stream_commit(struct LogicalDecodingContext *ctx,
					   ReorderBufferTXN *txn,
					   XLogRecPtr commit_lsn)
{
	/*
	 * The commit should happen outside streaming block, even for streamed
	 * transactions. The transaction has to be marked as streamed, though.
	 */
	Assert(!in_streaming);
	Assert(rbtxn_is_streamed(txn));

	update_replication_progress(ctx, false);

	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_stream_commit(ctx->out, txn, commit_lsn);
	OutputPluginWrite(ctx, true);

	cleanup_rel_sync_cache(txn->xid, true);
}

/*
 * PREPARE callback (for streaming two-phase commit).
 *
 * Notify the downstream to prepare the transaction.
 */
static void
pgoutput_stream_prepare_txn(LogicalDecodingContext *ctx,
							ReorderBufferTXN *txn,
							XLogRecPtr prepare_lsn)
{
	Assert(rbtxn_is_streamed(txn));

	update_replication_progress(ctx, false);
	OutputPluginPrepareWrite(ctx, true);
	logicalrep_write_stream_prepare(ctx->out, txn, prepare_lsn);
	OutputPluginWrite(ctx, true);
}

/*
 * Initialize the relation schema sync cache for a decoding session.
 *
 * The hash table is destroyed at the end of a decoding session. While
 * relcache invalidations still exist and will still be invoked, they
 * will just see the null hash table global and take no action.
 */
static void
init_rel_sync_cache(MemoryContext cachectx)
{
	HASHCTL		ctl;
	static bool relation_callbacks_registered = false;

	/* Nothing to do if hash table already exists */
	if (RelationSyncCache != NULL)
		return;

	/* Make a new hash table for the cache */
	ctl.keysize = sizeof(Oid);
	ctl.entrysize = sizeof(RelationSyncEntry);
	ctl.hcxt = cachectx;

	RelationSyncCache = hash_create("logical replication output relation cache",
									128, &ctl,
									HASH_ELEM | HASH_CONTEXT | HASH_BLOBS);

	Assert(RelationSyncCache != NULL);

	/* No more to do if we already registered callbacks */
	if (relation_callbacks_registered)
		return;

	CacheRegisterRelcacheCallback(rel_sync_cache_relation_cb, (Datum) 0);
	CacheRegisterSyscacheCallback(PUBLICATIONRELMAP,
								  rel_sync_cache_publication_cb,
								  (Datum) 0);
	CacheRegisterSyscacheCallback(PUBLICATIONNAMESPACEMAP,
								  rel_sync_cache_publication_cb,
								  (Datum) 0);

	relation_callbacks_registered = true;
}

/*
 * We expect relatively small number of streamed transactions.
 */
static bool
get_schema_sent_in_streamed_txn(RelationSyncEntry *entry, TransactionId xid)
{
	ListCell   *lc;

	foreach(lc, entry->streamed_txns)
	{
		if (xid == (uint32) lfirst_int(lc))
			return true;
	}

	return false;
}

/*
 * Add the xid in the rel sync entry for which we have already sent the schema
 * of the relation.
 */
static void
set_schema_sent_in_streamed_txn(RelationSyncEntry *entry, TransactionId xid)
{
	MemoryContext oldctx;

	oldctx = MemoryContextSwitchTo(CacheMemoryContext);

	entry->streamed_txns = lappend_int(entry->streamed_txns, xid);

	MemoryContextSwitchTo(oldctx);
}

/*
 * Find or create entry in the relation schema cache.
 *
 * This looks up publications that the given relation is directly or
 * indirectly part of (the latter if it's really the relation's ancestor that
 * is part of a publication) and fills up the found entry with the information
 * about which operations to publish and whether to use an ancestor's schema
 * when publishing.
 */
static RelationSyncEntry *
get_rel_sync_entry(PGOutputData *data, Relation relation)
{
	RelationSyncEntry *entry;
	bool		found;
	MemoryContext oldctx;
	Oid			relid = RelationGetRelid(relation);

	Assert(RelationSyncCache != NULL);

	/* Find cached relation info, creating if not found */
	entry = (RelationSyncEntry *) hash_search(RelationSyncCache,
											  (void *) &relid,
											  HASH_ENTER, &found);
	Assert(entry != NULL);

	/* initialize entry, if it's new */
	if (!found)
	{
		entry->replicate_valid = false;
		entry->schema_sent = false;
		entry->streamed_txns = NIL;
		entry->pubactions.pubinsert = entry->pubactions.pubupdate =
			entry->pubactions.pubdelete = entry->pubactions.pubtruncate = false;
		entry->new_slot = NULL;
		entry->old_slot = NULL;
		memset(entry->exprstate, 0, sizeof(entry->exprstate));
		entry->entry_cxt = NULL;
		entry->publish_as_relid = InvalidOid;
		entry->columns = NULL;
		entry->attrmap = NULL;
	}

	/* Validate the entry */
	if (!entry->replicate_valid)
	{
		Oid			schemaId = get_rel_namespace(relid);
		List	   *pubids = GetRelationPublications(relid);

		/*
		 * We don't acquire a lock on the namespace system table as we build
		 * the cache entry using a historic snapshot and all the later changes
		 * are absorbed while decoding WAL.
		 */
		List	   *schemaPubids = GetSchemaPublications(schemaId);
		ListCell   *lc;
		Oid			publish_as_relid = relid;
		int			publish_ancestor_level = 0;
		bool		am_partition = get_rel_relispartition(relid);
		char		relkind = get_rel_relkind(relid);
		List	   *rel_publications = NIL;

		/* Reload publications if needed before use. */
		if (!publications_valid)
		{
			oldctx = MemoryContextSwitchTo(CacheMemoryContext);
			if (data->publications)
			{
				list_free_deep(data->publications);
				data->publications = NIL;
			}
			data->publications = LoadPublications(data->publication_names);
			MemoryContextSwitchTo(oldctx);
			publications_valid = true;
		}

		/*
		 * Reset schema_sent status as the relation definition may have
		 * changed.  Also reset pubactions to empty in case rel was dropped
		 * from a publication.  Also free any objects that depended on the
		 * earlier definition.
		 */
		entry->schema_sent = false;
		list_free(entry->streamed_txns);
		entry->streamed_txns = NIL;
		bms_free(entry->columns);
		entry->columns = NULL;
		entry->pubactions.pubinsert = false;
		entry->pubactions.pubupdate = false;
		entry->pubactions.pubdelete = false;
		entry->pubactions.pubtruncate = false;

		/*
		 * Tuple slots cleanups. (Will be rebuilt later if needed).
		 */
		if (entry->old_slot)
			ExecDropSingleTupleTableSlot(entry->old_slot);
		if (entry->new_slot)
			ExecDropSingleTupleTableSlot(entry->new_slot);

		entry->old_slot = NULL;
		entry->new_slot = NULL;

		if (entry->attrmap)
			free_attrmap(entry->attrmap);
		entry->attrmap = NULL;

		/*
		 * Row filter cache cleanups.
		 */
		if (entry->entry_cxt)
			MemoryContextDelete(entry->entry_cxt);

		entry->entry_cxt = NULL;
		entry->estate = NULL;
		memset(entry->exprstate, 0, sizeof(entry->exprstate));

		/*
		 * Build publication cache. We can't use one provided by relcache as
		 * relcache considers all publications that the given relation is in,
		 * but here we only need to consider ones that the subscriber
		 * requested.
		 */
		foreach(lc, data->publications)
		{
			Publication *pub = lfirst(lc);
			bool		publish = false;

			/*
			 * Under what relid should we publish changes in this publication?
			 * We'll use the top-most relid across all publications. Also
			 * track the ancestor level for this publication.
			 */
			Oid			pub_relid = relid;
			int			ancestor_level = 0;

			/*
			 * If this is a FOR ALL TABLES publication, pick the partition
			 * root and set the ancestor level accordingly.
			 */
			if (pub->alltables)
			{
				publish = true;
				if (pub->pubviaroot && am_partition)
				{
					List	   *ancestors = get_partition_ancestors(relid);

					pub_relid = llast_oid(ancestors);
					ancestor_level = list_length(ancestors);
				}
			}

			if (!publish)
			{
				bool		ancestor_published = false;

				/*
				 * For a partition, check if any of the ancestors are
				 * published.  If so, note down the topmost ancestor that is
				 * published via this publication, which will be used as the
				 * relation via which to publish the partition's changes.
				 */
				if (am_partition)
				{
					Oid			ancestor;
					int			level;
					List	   *ancestors = get_partition_ancestors(relid);

					ancestor = GetTopMostAncestorInPublication(pub->oid,
															   ancestors,
															   &level);

					if (ancestor != InvalidOid)
					{
						ancestor_published = true;
						if (pub->pubviaroot)
						{
							pub_relid = ancestor;
							ancestor_level = level;
						}
					}
				}

				if (list_member_oid(pubids, pub->oid) ||
					list_member_oid(schemaPubids, pub->oid) ||
					ancestor_published)
					publish = true;
			}

			/*
			 * If the relation is to be published, determine actions to
			 * publish, and list of columns, if appropriate.
			 *
			 * Don't publish changes for partitioned tables, because
			 * publishing those of its partitions suffices, unless partition
			 * changes won't be published due to pubviaroot being set.
			 */
			if (publish &&
				(relkind != RELKIND_PARTITIONED_TABLE || pub->pubviaroot))
			{
				entry->pubactions.pubinsert |= pub->pubactions.pubinsert;
				entry->pubactions.pubupdate |= pub->pubactions.pubupdate;
				entry->pubactions.pubdelete |= pub->pubactions.pubdelete;
				entry->pubactions.pubtruncate |= pub->pubactions.pubtruncate;

				/*
				 * We want to publish the changes as the top-most ancestor
				 * across all publications. So we need to check if the already
				 * calculated level is higher than the new one. If yes, we can
				 * ignore the new value (as it's a child). Otherwise the new
				 * value is an ancestor, so we keep it.
				 */
				if (publish_ancestor_level > ancestor_level)
					continue;

				/*
				 * If we found an ancestor higher up in the tree, discard the
				 * list of publications through which we replicate it, and use
				 * the new ancestor.
				 */
				if (publish_ancestor_level < ancestor_level)
				{
					publish_as_relid = pub_relid;
					publish_ancestor_level = ancestor_level;

					/* reset the publication list for this relation */
					rel_publications = NIL;
				}
				else
				{
					/* Same ancestor level, has to be the same OID. */
					Assert(publish_as_relid == pub_relid);
				}

				/* Track publications for this ancestor. */
				rel_publications = lappend(rel_publications, pub);
			}
		}

		entry->publish_as_relid = publish_as_relid;

		/*
		 * Initialize the tuple slot, map, and row filter. These are only used
		 * when publishing inserts, updates, or deletes.
		 */
		if (entry->pubactions.pubinsert || entry->pubactions.pubupdate ||
			entry->pubactions.pubdelete)
		{
			/* Initialize the tuple slot and map */
			init_tuple_slot(data, relation, entry);

			/* Initialize the row filter */
			pgoutput_row_filter_init(data, rel_publications, entry);

			/* Initialize the column list */
			pgoutput_column_list_init(data, rel_publications, entry);
		}

		list_free(pubids);
		list_free(schemaPubids);
		list_free(rel_publications);

		entry->replicate_valid = true;
	}

	return entry;
}

/*
 * Cleanup list of streamed transactions and update the schema_sent flag.
 *
 * When a streamed transaction commits or aborts, we need to remove the
 * toplevel XID from the schema cache. If the transaction aborted, the
 * subscriber will simply throw away the schema records we streamed, so
 * we don't need to do anything else.
 *
 * If the transaction is committed, the subscriber will update the relation
 * cache - so tweak the schema_sent flag accordingly.
 */
static void
cleanup_rel_sync_cache(TransactionId xid, bool is_commit)
{
	HASH_SEQ_STATUS hash_seq;
	RelationSyncEntry *entry;
	ListCell   *lc;

	Assert(RelationSyncCache != NULL);

	hash_seq_init(&hash_seq, RelationSyncCache);
	while ((entry = hash_seq_search(&hash_seq)) != NULL)
	{
		/*
		 * We can set the schema_sent flag for an entry that has committed xid
		 * in the list as that ensures that the subscriber would have the
		 * corresponding schema and we don't need to send it unless there is
		 * any invalidation for that relation.
		 */
		foreach(lc, entry->streamed_txns)
		{
			if (xid == (uint32) lfirst_int(lc))
			{
				if (is_commit)
					entry->schema_sent = true;

				entry->streamed_txns =
					foreach_delete_current(entry->streamed_txns, lc);
				break;
			}
		}
	}
}

/*
 * Relcache invalidation callback
 */
static void
rel_sync_cache_relation_cb(Datum arg, Oid relid)
{
	RelationSyncEntry *entry;

	/*
	 * We can get here if the plugin was used in SQL interface as the
	 * RelSchemaSyncCache is destroyed when the decoding finishes, but there
	 * is no way to unregister the relcache invalidation callback.
	 */
	if (RelationSyncCache == NULL)
		return;

	/*
	 * Nobody keeps pointers to entries in this hash table around outside
	 * logical decoding callback calls - but invalidation events can come in
	 * *during* a callback if we do any syscache access in the callback.
	 * Because of that we must mark the cache entry as invalid but not damage
	 * any of its substructure here.  The next get_rel_sync_entry() call will
	 * rebuild it all.
	 */
	if (OidIsValid(relid))
	{
		/*
		 * Getting invalidations for relations that aren't in the table is
		 * entirely normal.  So we don't care if it's found or not.
		 */
		entry = (RelationSyncEntry *) hash_search(RelationSyncCache, &relid,
												  HASH_FIND, NULL);
		if (entry != NULL)
			entry->replicate_valid = false;
	}
	else
	{
		/* Whole cache must be flushed. */
		HASH_SEQ_STATUS status;

		hash_seq_init(&status, RelationSyncCache);
		while ((entry = (RelationSyncEntry *) hash_seq_search(&status)) != NULL)
		{
			entry->replicate_valid = false;
		}
	}
}

/*
 * Publication relation/schema map syscache invalidation callback
 *
 * Called for invalidations on pg_publication, pg_publication_rel, and
 * pg_publication_namespace.
 */
static void
rel_sync_cache_publication_cb(Datum arg, int cacheid, uint32 hashvalue)
{
	HASH_SEQ_STATUS status;
	RelationSyncEntry *entry;

	/*
	 * We can get here if the plugin was used in SQL interface as the
	 * RelSchemaSyncCache is destroyed when the decoding finishes, but there
	 * is no way to unregister the relcache invalidation callback.
	 */
	if (RelationSyncCache == NULL)
		return;

	/*
	 * There is no way to find which entry in our cache the hash belongs to so
	 * mark the whole cache as invalid.
	 */
	hash_seq_init(&status, RelationSyncCache);
	while ((entry = (RelationSyncEntry *) hash_seq_search(&status)) != NULL)
	{
		entry->replicate_valid = false;
	}
}

/* Send Replication origin */
static void
send_repl_origin(LogicalDecodingContext *ctx, RepOriginId origin_id,
				 XLogRecPtr origin_lsn, bool send_origin)
{
	if (send_origin)
	{
		char	   *origin;

		/*----------
		 * XXX: which behaviour do we want here?
		 *
		 * Alternatives:
		 *  - don't send origin message if origin name not found
		 *    (that's what we do now)
		 *  - throw error - that will break replication, not good
		 *  - send some special "unknown" origin
		 *----------
		 */
		if (replorigin_by_oid(origin_id, true, &origin))
		{
			/* Message boundary */
			OutputPluginWrite(ctx, false);
			OutputPluginPrepareWrite(ctx, true);

			logicalrep_write_origin(ctx->out, origin, origin_lsn);
		}
	}
}

/*
 * Try to update progress and send a keepalive message if too many changes were
 * processed.
 *
 * For a large transaction, if we don't send any change to the downstream for a
 * long time (exceeds the wal_receiver_timeout of standby) then it can timeout.
 * This can happen when all or most of the changes are either not published or
 * got filtered out.
 */
static void
update_replication_progress(LogicalDecodingContext *ctx, bool skipped_xact)
{
	static int	changes_count = 0;

	/*
	 * We don't want to try sending a keepalive message after processing each
	 * change as that can have overhead. Tests revealed that there is no
	 * noticeable overhead in doing it after continuously processing 100 or so
	 * changes.
	 */
#define CHANGES_THRESHOLD 100

	/*
	 * If we are at the end of transaction LSN, update progress tracking.
	 * Otherwise, after continuously processing CHANGES_THRESHOLD changes, we
	 * try to send a keepalive message if required.
	 */
	if (ctx->end_xact || ++changes_count >= CHANGES_THRESHOLD)
	{
		OutputPluginUpdateProgress(ctx, skipped_xact);
		changes_count = 0;
	}
}