From: Peter Geoghegan <pg@heroku.com>
Date: Sun, 16 Mar 2014 07:55:24 +0000 (-0700)
Subject: Notes on pair deduplication
X-Git-Url: http://git.postgresql.org/gitweb/static/gitweb.js?a=commitdiff_plain;h=0fdda610eec93bb115fc298884a5597bd7da5141;p=users%2Fandresfreund%2Fpostgres.git

Notes on pair deduplication
---

diff --git a/src/backend/utils/adt/jsonb_util.c b/src/backend/utils/adt/jsonb_util.c
index 584f1b0271..ea6bea0b87 100644
--- a/src/backend/utils/adt/jsonb_util.c
+++ b/src/backend/utils/adt/jsonb_util.c
@@ -983,8 +983,9 @@ deepContains(JsonbIterator ** val, JsonbIterator ** mContained)
 }
 
 /*
- * Convert a Postgres text array to a JsonbSortedArray, with de-duplicated key
- * elements.
+ * Convert a Postgres text array to a Jsonb array, sorted and with
+ * de-duplicated key elements.  This is used for searching for items in the
+ * array.
  */
 JsonbValue *
 arrayToJsonbSortedArray(ArrayType *a)
@@ -995,6 +996,8 @@ arrayToJsonbSortedArray(ArrayType *a)
 	JsonbValue *result;
 	int			i,
 				j;
+	Size		maxn = Min(MaxAllocSize / sizeof(JsonbPair),
+						   JENTRY_POSMASK / sizeof(JsonbPair));
 
 	/* Extract data for sorting */
 	deconstruct_array(a, TEXTOID, -1, false, 'i', &key_datums, &key_nulls,
@@ -1010,11 +1013,11 @@ arrayToJsonbSortedArray(ArrayType *a)
 	 * assumption.  Therefore, use an explicit check rather than relying on
 	 * palloc() to complain.
 	 */
-	if (key_count > MaxAllocSize / sizeof(JsonbPair))
+	if (key_count > maxn)
 		ereport(ERROR,
 				(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
 				 errmsg("number of pairs (%d) exceeds the maximum allowed (%zu)",
-						key_count, MaxAllocSize / sizeof(JsonbPair))));
+						key_count, maxn)));
 
 	result = palloc(sizeof(JsonbValue));
 	result->type = jbvArray;
@@ -1664,7 +1667,7 @@ lengthCompareJsonbStringValue(const void *a, const void *b, void *binequal)
  * same "arg setting" hack will be applied here in respect of the pair's key
  * values.
  *
- * N.B: String comparions here are "length-wise"
+ * N.B: String comparisons here are "length-wise"
  *
  * Pairs with equals keys are ordered such that the order field is respected.
  */
diff --git a/src/include/utils/jsonb.h b/src/include/utils/jsonb.h
index 0c360db65f..04a74a2e51 100644
--- a/src/include/utils/jsonb.h
+++ b/src/include/utils/jsonb.h
@@ -195,7 +195,14 @@ struct JsonbValue
 	};
 };
 
-/* Pair within an Object */
+/*
+ * Pair within an Object.
+ *
+ * Pairs with duplicate keys are de-duplicated.  We store the order for the
+ * benefit of doing so in a well-defined way with respect to the original
+ * observed order (which is "last observed wins").  This is only stored when
+ * originally constructing a Jsonb.
+ */
 struct JsonbPair
 {
 	JsonbValue	key;			/* Must be a jbvString */