postgres/src/backend/utils/adt/json.c

/*-------------------------------------------------------------------------
 *
 * json.c
 *		JSON data type support.
 *
 * Portions Copyright (c) 1996-2023, PostgreSQL Global Development Group
 * Portions Copyright (c) 1994, Regents of the University of California
 *
 * IDENTIFICATION
 *	  src/backend/utils/adt/json.c
 *
 *-------------------------------------------------------------------------
 */
#include "postgres.h"

#include "catalog/pg_proc.h"
#include "catalog/pg_type.h"
#include "common/hashfn.h"
#include "funcapi.h"
#include "libpq/pqformat.h"
#include "miscadmin.h"
#include "parser/parse_coerce.h"
#include "utils/array.h"
#include "utils/builtins.h"
#include "utils/date.h"
#include "utils/datetime.h"
#include "utils/json.h"
#include "utils/jsonfuncs.h"
#include "utils/lsyscache.h"
#include "utils/typcache.h"

typedef enum					/* type categories for datum_to_json */
{
	JSONTYPE_NULL,				/* null, so we didn't bother to identify */
	JSONTYPE_BOOL,				/* boolean (built-in types only) */
	JSONTYPE_NUMERIC,			/* numeric (ditto) */
	JSONTYPE_DATE,				/* we use special formatting for datetimes */
	JSONTYPE_TIMESTAMP,
	JSONTYPE_TIMESTAMPTZ,
	JSONTYPE_JSON,				/* JSON itself (and JSONB) */
	JSONTYPE_ARRAY,				/* array */
	JSONTYPE_COMPOSITE,			/* composite */
	JSONTYPE_CAST,				/* something with an explicit cast to JSON */
	JSONTYPE_OTHER				/* all else */
} JsonTypeCategory;


/*
 * Support for fast key uniqueness checking.
 *
 * We maintain a hash table of used keys in JSON objects for fast detection
 * of duplicates.
 */
/* Common context for key uniqueness check */
typedef struct HTAB *JsonUniqueCheckState;	/* hash table for key names */

/* Hash entry for JsonUniqueCheckState */
typedef struct JsonUniqueHashEntry
{
	const char *key;
	int			key_len;
	int			object_id;
} JsonUniqueHashEntry;

/* Context struct for key uniqueness check during JSON building */
typedef struct JsonUniqueBuilderState
{
	JsonUniqueCheckState check; /* unique check */
	StringInfoData skipped_keys;	/* skipped keys with NULL values */
	MemoryContext mcxt;			/* context for saving skipped keys */
} JsonUniqueBuilderState;


/* State struct for JSON aggregation */
typedef struct JsonAggState
{
	StringInfo	str;
	JsonTypeCategory key_category;
	Oid			key_output_func;
	JsonTypeCategory val_category;
	Oid			val_output_func;
	JsonUniqueBuilderState unique_check;
} JsonAggState;

static void composite_to_json(Datum composite, StringInfo result,
							  bool use_line_feeds);
static void array_dim_to_json(StringInfo result, int dim, int ndims, int *dims,
							  Datum *vals, bool *nulls, int *valcount,
							  JsonTypeCategory tcategory, Oid outfuncoid,
							  bool use_line_feeds);
static void array_to_json_internal(Datum array, StringInfo result,
								   bool use_line_feeds);
static void json_categorize_type(Oid typoid,
								 JsonTypeCategory *tcategory,
								 Oid *outfuncoid);
static void datum_to_json(Datum val, bool is_null, StringInfo result,
						  JsonTypeCategory tcategory, Oid outfuncoid,
						  bool key_scalar);
static void add_json(Datum val, bool is_null, StringInfo result,
					 Oid val_type, bool key_scalar);
static text *catenate_stringinfo_string(StringInfo buffer, const char *addon);

/*
 * Input.
 */
Datum
json_in(PG_FUNCTION_ARGS)
{
	char	   *json = PG_GETARG_CSTRING(0);
	text	   *result = cstring_to_text(json);
	JsonLexContext *lex;

	/* validate it */
	lex = makeJsonLexContext(result, false);
	if (!pg_parse_json_or_errsave(lex, &nullSemAction, fcinfo->context))
		PG_RETURN_NULL();

	/* Internal representation is the same as text */
	PG_RETURN_TEXT_P(result);
}

/*
 * Output.
 */
Datum
json_out(PG_FUNCTION_ARGS)
{
	/* we needn't detoast because text_to_cstring will handle that */
	Datum		txt = PG_GETARG_DATUM(0);

	PG_RETURN_CSTRING(TextDatumGetCString(txt));
}

/*
 * Binary send.
 */
Datum
json_send(PG_FUNCTION_ARGS)
{
	text	   *t = PG_GETARG_TEXT_PP(0);
	StringInfoData buf;

	pq_begintypsend(&buf);
	pq_sendtext(&buf, VARDATA_ANY(t), VARSIZE_ANY_EXHDR(t));
	PG_RETURN_BYTEA_P(pq_endtypsend(&buf));
}

/*
 * Binary receive.
 */
Datum
json_recv(PG_FUNCTION_ARGS)
{
	StringInfo	buf = (StringInfo) PG_GETARG_POINTER(0);
	char	   *str;
	int			nbytes;
	JsonLexContext *lex;

	str = pq_getmsgtext(buf, buf->len - buf->cursor, &nbytes);

	/* Validate it. */
	lex = makeJsonLexContextCstringLen(str, nbytes, GetDatabaseEncoding(), false);
	pg_parse_json_or_ereport(lex, &nullSemAction);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(str, nbytes));
}

/*
 * Determine how we want to print values of a given type in datum_to_json.
 *
 * Given the datatype OID, return its JsonTypeCategory, as well as the type's
 * output function OID.  If the returned category is JSONTYPE_CAST, we
 * return the OID of the type->JSON cast function instead.
 */
static void
json_categorize_type(Oid typoid,
					 JsonTypeCategory *tcategory,
					 Oid *outfuncoid)
{
	bool		typisvarlena;

	/* Look through any domain */
	typoid = getBaseType(typoid);

	*outfuncoid = InvalidOid;

	/*
	 * We need to get the output function for everything except date and
	 * timestamp types, array and composite types, booleans, and non-builtin
	 * types where there's a cast to json.
	 */

	switch (typoid)
	{
		case BOOLOID:
			*tcategory = JSONTYPE_BOOL;
			break;

		case INT2OID:
		case INT4OID:
		case INT8OID:
		case FLOAT4OID:
		case FLOAT8OID:
		case NUMERICOID:
			getTypeOutputInfo(typoid, outfuncoid, &typisvarlena);
			*tcategory = JSONTYPE_NUMERIC;
			break;

		case DATEOID:
			*tcategory = JSONTYPE_DATE;
			break;

		case TIMESTAMPOID:
			*tcategory = JSONTYPE_TIMESTAMP;
			break;

		case TIMESTAMPTZOID:
			*tcategory = JSONTYPE_TIMESTAMPTZ;
			break;

		case JSONOID:
		case JSONBOID:
			getTypeOutputInfo(typoid, outfuncoid, &typisvarlena);
			*tcategory = JSONTYPE_JSON;
			break;

		default:
			/* Check for arrays and composites */
			if (OidIsValid(get_element_type(typoid)) || typoid == ANYARRAYOID
				|| typoid == ANYCOMPATIBLEARRAYOID || typoid == RECORDARRAYOID)
				*tcategory = JSONTYPE_ARRAY;
			else if (type_is_rowtype(typoid))	/* includes RECORDOID */
				*tcategory = JSONTYPE_COMPOSITE;
			else
			{
				/* It's probably the general case ... */
				*tcategory = JSONTYPE_OTHER;
				/* but let's look for a cast to json, if it's not built-in */
				if (typoid >= FirstNormalObjectId)
				{
					Oid			castfunc;
					CoercionPathType ctype;

					ctype = find_coercion_pathway(JSONOID, typoid,
												  COERCION_EXPLICIT,
												  &castfunc);
					if (ctype == COERCION_PATH_FUNC && OidIsValid(castfunc))
					{
						*tcategory = JSONTYPE_CAST;
						*outfuncoid = castfunc;
					}
					else
					{
						/* non builtin type with no cast */
						getTypeOutputInfo(typoid, outfuncoid, &typisvarlena);
					}
				}
				else
				{
					/* any other builtin type */
					getTypeOutputInfo(typoid, outfuncoid, &typisvarlena);
				}
			}
			break;
	}
}

/*
 * Turn a Datum into JSON text, appending the string to "result".
 *
 * tcategory and outfuncoid are from a previous call to json_categorize_type,
 * except that if is_null is true then they can be invalid.
 *
 * If key_scalar is true, the value is being printed as a key, so insist
 * it's of an acceptable type, and force it to be quoted.
 */
static void
datum_to_json(Datum val, bool is_null, StringInfo result,
			  JsonTypeCategory tcategory, Oid outfuncoid,
			  bool key_scalar)
{
	char	   *outputstr;
	text	   *jsontext;

	check_stack_depth();

	/* callers are expected to ensure that null keys are not passed in */
	Assert(!(key_scalar && is_null));

	if (is_null)
	{
		appendStringInfoString(result, "null");
		return;
	}

	if (key_scalar &&
		(tcategory == JSONTYPE_ARRAY ||
		 tcategory == JSONTYPE_COMPOSITE ||
		 tcategory == JSONTYPE_JSON ||
		 tcategory == JSONTYPE_CAST))
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("key value must be scalar, not array, composite, or json")));

	switch (tcategory)
	{
		case JSONTYPE_ARRAY:
			array_to_json_internal(val, result, false);
			break;
		case JSONTYPE_COMPOSITE:
			composite_to_json(val, result, false);
			break;
		case JSONTYPE_BOOL:
			outputstr = DatumGetBool(val) ? "true" : "false";
			if (key_scalar)
				escape_json(result, outputstr);
			else
				appendStringInfoString(result, outputstr);
			break;
		case JSONTYPE_NUMERIC:
			outputstr = OidOutputFunctionCall(outfuncoid, val);

			/*
			 * Don't call escape_json for a non-key if it's a valid JSON
			 * number.
			 */
			if (!key_scalar && IsValidJsonNumber(outputstr, strlen(outputstr)))
				appendStringInfoString(result, outputstr);
			else
				escape_json(result, outputstr);
			pfree(outputstr);
			break;
		case JSONTYPE_DATE:
			{
				char		buf[MAXDATELEN + 1];

				JsonEncodeDateTime(buf, val, DATEOID, NULL);
				appendStringInfo(result, "\"%s\"", buf);
			}
			break;
		case JSONTYPE_TIMESTAMP:
			{
				char		buf[MAXDATELEN + 1];

				JsonEncodeDateTime(buf, val, TIMESTAMPOID, NULL);
				appendStringInfo(result, "\"%s\"", buf);
			}
			break;
		case JSONTYPE_TIMESTAMPTZ:
			{
				char		buf[MAXDATELEN + 1];

				JsonEncodeDateTime(buf, val, TIMESTAMPTZOID, NULL);
				appendStringInfo(result, "\"%s\"", buf);
			}
			break;
		case JSONTYPE_JSON:
			/* JSON and JSONB output will already be escaped */
			outputstr = OidOutputFunctionCall(outfuncoid, val);
			appendStringInfoString(result, outputstr);
			pfree(outputstr);
			break;
		case JSONTYPE_CAST:
			/* outfuncoid refers to a cast function, not an output function */
			jsontext = DatumGetTextPP(OidFunctionCall1(outfuncoid, val));
			outputstr = text_to_cstring(jsontext);
			appendStringInfoString(result, outputstr);
			pfree(outputstr);
			pfree(jsontext);
			break;
		default:
			outputstr = OidOutputFunctionCall(outfuncoid, val);
			escape_json(result, outputstr);
			pfree(outputstr);
			break;
	}
}

/*
 * Encode 'value' of datetime type 'typid' into JSON string in ISO format using
 * optionally preallocated buffer 'buf'.  Optional 'tzp' determines time-zone
 * offset (in seconds) in which we want to show timestamptz.
 */
char *
JsonEncodeDateTime(char *buf, Datum value, Oid typid, const int *tzp)
{
	if (!buf)
		buf = palloc(MAXDATELEN + 1);

	switch (typid)
	{
		case DATEOID:
			{
				DateADT		date;
				struct pg_tm tm;

				date = DatumGetDateADT(value);

				/* Same as date_out(), but forcing DateStyle */
				if (DATE_NOT_FINITE(date))
					EncodeSpecialDate(date, buf);
				else
				{
					j2date(date + POSTGRES_EPOCH_JDATE,
						   &(tm.tm_year), &(tm.tm_mon), &(tm.tm_mday));
					EncodeDateOnly(&tm, USE_XSD_DATES, buf);
				}
			}
			break;
		case TIMEOID:
			{
				TimeADT		time = DatumGetTimeADT(value);
				struct pg_tm tt,
						   *tm = &tt;
				fsec_t		fsec;

				/* Same as time_out(), but forcing DateStyle */
				time2tm(time, tm, &fsec);
				EncodeTimeOnly(tm, fsec, false, 0, USE_XSD_DATES, buf);
			}
			break;
		case TIMETZOID:
			{
				TimeTzADT  *time = DatumGetTimeTzADTP(value);
				struct pg_tm tt,
						   *tm = &tt;
				fsec_t		fsec;
				int			tz;

				/* Same as timetz_out(), but forcing DateStyle */
				timetz2tm(time, tm, &fsec, &tz);
				EncodeTimeOnly(tm, fsec, true, tz, USE_XSD_DATES, buf);
			}
			break;
		case TIMESTAMPOID:
			{
				Timestamp	timestamp;
				struct pg_tm tm;
				fsec_t		fsec;

				timestamp = DatumGetTimestamp(value);
				/* Same as timestamp_out(), but forcing DateStyle */
				if (TIMESTAMP_NOT_FINITE(timestamp))
					EncodeSpecialTimestamp(timestamp, buf);
				else if (timestamp2tm(timestamp, NULL, &tm, &fsec, NULL, NULL) == 0)
					EncodeDateTime(&tm, fsec, false, 0, NULL, USE_XSD_DATES, buf);
				else
					ereport(ERROR,
							(errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
							 errmsg("timestamp out of range")));
			}
			break;
		case TIMESTAMPTZOID:
			{
				TimestampTz timestamp;
				struct pg_tm tm;
				int			tz;
				fsec_t		fsec;
				const char *tzn = NULL;

				timestamp = DatumGetTimestampTz(value);

				/*
				 * If a time zone is specified, we apply the time-zone shift,
				 * convert timestamptz to pg_tm as if it were without a time
				 * zone, and then use the specified time zone for converting
				 * the timestamp into a string.
				 */
				if (tzp)
				{
					tz = *tzp;
					timestamp -= (TimestampTz) tz * USECS_PER_SEC;
				}

				/* Same as timestamptz_out(), but forcing DateStyle */
				if (TIMESTAMP_NOT_FINITE(timestamp))
					EncodeSpecialTimestamp(timestamp, buf);
				else if (timestamp2tm(timestamp, tzp ? NULL : &tz, &tm, &fsec,
									  tzp ? NULL : &tzn, NULL) == 0)
				{
					if (tzp)
						tm.tm_isdst = 1;	/* set time-zone presence flag */

					EncodeDateTime(&tm, fsec, true, tz, tzn, USE_XSD_DATES, buf);
				}
				else
					ereport(ERROR,
							(errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE),
							 errmsg("timestamp out of range")));
			}
			break;
		default:
			elog(ERROR, "unknown jsonb value datetime type oid %u", typid);
			return NULL;
	}

	return buf;
}

/*
 * Process a single dimension of an array.
 * If it's the innermost dimension, output the values, otherwise call
 * ourselves recursively to process the next dimension.
 */
static void
array_dim_to_json(StringInfo result, int dim, int ndims, int *dims, Datum *vals,
				  bool *nulls, int *valcount, JsonTypeCategory tcategory,
				  Oid outfuncoid, bool use_line_feeds)
{
	int			i;
	const char *sep;

	Assert(dim < ndims);

	sep = use_line_feeds ? ",\n " : ",";

	appendStringInfoChar(result, '[');

	for (i = 1; i <= dims[dim]; i++)
	{
		if (i > 1)
			appendStringInfoString(result, sep);

		if (dim + 1 == ndims)
		{
			datum_to_json(vals[*valcount], nulls[*valcount], result, tcategory,
						  outfuncoid, false);
			(*valcount)++;
		}
		else
		{
			/*
			 * Do we want line feeds on inner dimensions of arrays? For now
			 * we'll say no.
			 */
			array_dim_to_json(result, dim + 1, ndims, dims, vals, nulls,
							  valcount, tcategory, outfuncoid, false);
		}
	}

	appendStringInfoChar(result, ']');
}

/*
 * Turn an array into JSON.
 */
static void
array_to_json_internal(Datum array, StringInfo result, bool use_line_feeds)
{
	ArrayType  *v = DatumGetArrayTypeP(array);
	Oid			element_type = ARR_ELEMTYPE(v);
	int		   *dim;
	int			ndim;
	int			nitems;
	int			count = 0;
	Datum	   *elements;
	bool	   *nulls;
	int16		typlen;
	bool		typbyval;
	char		typalign;
	JsonTypeCategory tcategory;
	Oid			outfuncoid;

	ndim = ARR_NDIM(v);
	dim = ARR_DIMS(v);
	nitems = ArrayGetNItems(ndim, dim);

	if (nitems <= 0)
	{
		appendStringInfoString(result, "[]");
		return;
	}

	get_typlenbyvalalign(element_type,
						 &typlen, &typbyval, &typalign);

	json_categorize_type(element_type,
						 &tcategory, &outfuncoid);

	deconstruct_array(v, element_type, typlen, typbyval,
					  typalign, &elements, &nulls,
					  &nitems);

	array_dim_to_json(result, 0, ndim, dim, elements, nulls, &count, tcategory,
					  outfuncoid, use_line_feeds);

	pfree(elements);
	pfree(nulls);
}

/*
 * Turn a composite / record into JSON.
 */
static void
composite_to_json(Datum composite, StringInfo result, bool use_line_feeds)
{
	HeapTupleHeader td;
	Oid			tupType;
	int32		tupTypmod;
	TupleDesc	tupdesc;
	HeapTupleData tmptup,
			   *tuple;
	int			i;
	bool		needsep = false;
	const char *sep;

	sep = use_line_feeds ? ",\n " : ",";

	td = DatumGetHeapTupleHeader(composite);

	/* Extract rowtype info and find a tupdesc */
	tupType = HeapTupleHeaderGetTypeId(td);
	tupTypmod = HeapTupleHeaderGetTypMod(td);
	tupdesc = lookup_rowtype_tupdesc(tupType, tupTypmod);

	/* Build a temporary HeapTuple control structure */
	tmptup.t_len = HeapTupleHeaderGetDatumLength(td);
	tmptup.t_data = td;
	tuple = &tmptup;

	appendStringInfoChar(result, '{');

	for (i = 0; i < tupdesc->natts; i++)
	{
		Datum		val;
		bool		isnull;
		char	   *attname;
		JsonTypeCategory tcategory;
		Oid			outfuncoid;
		Form_pg_attribute att = TupleDescAttr(tupdesc, i);

		if (att->attisdropped)
			continue;

		if (needsep)
			appendStringInfoString(result, sep);
		needsep = true;

		attname = NameStr(att->attname);
		escape_json(result, attname);
		appendStringInfoChar(result, ':');

		val = heap_getattr(tuple, i + 1, tupdesc, &isnull);

		if (isnull)
		{
			tcategory = JSONTYPE_NULL;
			outfuncoid = InvalidOid;
		}
		else
			json_categorize_type(att->atttypid, &tcategory, &outfuncoid);

		datum_to_json(val, isnull, result, tcategory, outfuncoid, false);
	}

	appendStringInfoChar(result, '}');
	ReleaseTupleDesc(tupdesc);
}

/*
 * Append JSON text for "val" to "result".
 *
 * This is just a thin wrapper around datum_to_json.  If the same type will be
 * printed many times, avoid using this; better to do the json_categorize_type
 * lookups only once.
 */
static void
add_json(Datum val, bool is_null, StringInfo result,
		 Oid val_type, bool key_scalar)
{
	JsonTypeCategory tcategory;
	Oid			outfuncoid;

	if (val_type == InvalidOid)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("could not determine input data type")));

	if (is_null)
	{
		tcategory = JSONTYPE_NULL;
		outfuncoid = InvalidOid;
	}
	else
		json_categorize_type(val_type,
							 &tcategory, &outfuncoid);

	datum_to_json(val, is_null, result, tcategory, outfuncoid, key_scalar);
}

/*
 * SQL function array_to_json(row)
 */
Datum
array_to_json(PG_FUNCTION_ARGS)
{
	Datum		array = PG_GETARG_DATUM(0);
	StringInfo	result;

	result = makeStringInfo();

	array_to_json_internal(array, result, false);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(result->data, result->len));
}

/*
 * SQL function array_to_json(row, prettybool)
 */
Datum
array_to_json_pretty(PG_FUNCTION_ARGS)
{
	Datum		array = PG_GETARG_DATUM(0);
	bool		use_line_feeds = PG_GETARG_BOOL(1);
	StringInfo	result;

	result = makeStringInfo();

	array_to_json_internal(array, result, use_line_feeds);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(result->data, result->len));
}

/*
 * SQL function row_to_json(row)
 */
Datum
row_to_json(PG_FUNCTION_ARGS)
{
	Datum		array = PG_GETARG_DATUM(0);
	StringInfo	result;

	result = makeStringInfo();

	composite_to_json(array, result, false);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(result->data, result->len));
}

/*
 * SQL function row_to_json(row, prettybool)
 */
Datum
row_to_json_pretty(PG_FUNCTION_ARGS)
{
	Datum		array = PG_GETARG_DATUM(0);
	bool		use_line_feeds = PG_GETARG_BOOL(1);
	StringInfo	result;

	result = makeStringInfo();

	composite_to_json(array, result, use_line_feeds);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(result->data, result->len));
}

/*
 * Is the given type immutable when coming out of a JSON context?
 *
 * At present, datetimes are all considered mutable, because they
 * depend on timezone.  XXX we should also drill down into objects
 * and arrays, but do not.
 */
bool
to_json_is_immutable(Oid typoid)
{
	JsonTypeCategory tcategory;
	Oid			outfuncoid;

	json_categorize_type(typoid, &tcategory, &outfuncoid);

	switch (tcategory)
	{
		case JSONTYPE_BOOL:
		case JSONTYPE_JSON:
		case JSONTYPE_NULL:
			return true;

		case JSONTYPE_DATE:
		case JSONTYPE_TIMESTAMP:
		case JSONTYPE_TIMESTAMPTZ:
			return false;

		case JSONTYPE_ARRAY:
			return false;		/* TODO recurse into elements */

		case JSONTYPE_COMPOSITE:
			return false;		/* TODO recurse into fields */

		case JSONTYPE_NUMERIC:
		case JSONTYPE_CAST:
		case JSONTYPE_OTHER:
			return func_volatile(outfuncoid) == PROVOLATILE_IMMUTABLE;
	}

	return false;				/* not reached */
}

/*
 * SQL function to_json(anyvalue)
 */
Datum
to_json(PG_FUNCTION_ARGS)
{
	Datum		val = PG_GETARG_DATUM(0);
	Oid			val_type = get_fn_expr_argtype(fcinfo->flinfo, 0);
	StringInfo	result;
	JsonTypeCategory tcategory;
	Oid			outfuncoid;

	if (val_type == InvalidOid)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("could not determine input data type")));

	json_categorize_type(val_type,
						 &tcategory, &outfuncoid);

	result = makeStringInfo();

	datum_to_json(val, false, result, tcategory, outfuncoid, false);

	PG_RETURN_TEXT_P(cstring_to_text_with_len(result->data, result->len));
}

/*
 * json_agg transition function
 *
 * aggregate input column as a json array value.
 */
static Datum
json_agg_transfn_worker(FunctionCallInfo fcinfo, bool absent_on_null)
{
	MemoryContext aggcontext,
				oldcontext;
	JsonAggState *state;
	Datum		val;

	if (!AggCheckCallContext(fcinfo, &aggcontext))
	{
		/* cannot be called directly because of internal-type argument */
		elog(ERROR, "json_agg_transfn called in non-aggregate context");
	}

	if (PG_ARGISNULL(0))
	{
		Oid			arg_type = get_fn_expr_argtype(fcinfo->flinfo, 1);

		if (arg_type == InvalidOid)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("could not determine input data type")));

		/*
		 * Make this state object in a context where it will persist for the
		 * duration of the aggregate call.  MemoryContextSwitchTo is only
		 * needed the first time, as the StringInfo routines make sure they
		 * use the right context to enlarge the object if necessary.
		 */
		oldcontext = MemoryContextSwitchTo(aggcontext);
		state = (JsonAggState *) palloc(sizeof(JsonAggState));
		state->str = makeStringInfo();
		MemoryContextSwitchTo(oldcontext);

		appendStringInfoChar(state->str, '[');
		json_categorize_type(arg_type, &state->val_category,
							 &state->val_output_func);
	}
	else
	{
		state = (JsonAggState *) PG_GETARG_POINTER(0);
	}

	if (absent_on_null && PG_ARGISNULL(1))
		PG_RETURN_POINTER(state);

	if (state->str->len > 1)
		appendStringInfoString(state->str, ", ");

	/* fast path for NULLs */
	if (PG_ARGISNULL(1))
	{
		datum_to_json((Datum) 0, true, state->str, JSONTYPE_NULL,
					  InvalidOid, false);
		PG_RETURN_POINTER(state);
	}

	val = PG_GETARG_DATUM(1);

	/* add some whitespace if structured type and not first item */
	if (!PG_ARGISNULL(0) && state->str->len > 1 &&
		(state->val_category == JSONTYPE_ARRAY ||
		 state->val_category == JSONTYPE_COMPOSITE))
	{
		appendStringInfoString(state->str, "\n ");
	}

	datum_to_json(val, false, state->str, state->val_category,
				  state->val_output_func, false);

	/*
	 * The transition type for json_agg() is declared to be "internal", which
	 * is a pass-by-value type the same size as a pointer.  So we can safely
	 * pass the JsonAggState pointer through nodeAgg.c's machinations.
	 */
	PG_RETURN_POINTER(state);
}


/*
 * json_agg aggregate function
 */
Datum
json_agg_transfn(PG_FUNCTION_ARGS)
{
	return json_agg_transfn_worker(fcinfo, false);
}

/*
 * json_agg_strict aggregate function
 */
Datum
json_agg_strict_transfn(PG_FUNCTION_ARGS)
{
	return json_agg_transfn_worker(fcinfo, true);
}

/*
 * json_agg final function
 */
Datum
json_agg_finalfn(PG_FUNCTION_ARGS)
{
	JsonAggState *state;

	/* cannot be called directly because of internal-type argument */
	Assert(AggCheckCallContext(fcinfo, NULL));

	state = PG_ARGISNULL(0) ?
		NULL :
		(JsonAggState *) PG_GETARG_POINTER(0);

	/* NULL result for no rows in, as is standard with aggregates */
	if (state == NULL)
		PG_RETURN_NULL();

	/* Else return state with appropriate array terminator added */
	PG_RETURN_TEXT_P(catenate_stringinfo_string(state->str, "]"));
}

/* Functions implementing hash table for key uniqueness check */
static uint32
json_unique_hash(const void *key, Size keysize)
{
	const JsonUniqueHashEntry *entry = (JsonUniqueHashEntry *) key;
	uint32		hash = hash_bytes_uint32(entry->object_id);

	hash ^= hash_bytes((const unsigned char *) entry->key, entry->key_len);

	return DatumGetUInt32(hash);
}

static int
json_unique_hash_match(const void *key1, const void *key2, Size keysize)
{
	const JsonUniqueHashEntry *entry1 = (const JsonUniqueHashEntry *) key1;
	const JsonUniqueHashEntry *entry2 = (const JsonUniqueHashEntry *) key2;

	if (entry1->object_id != entry2->object_id)
		return entry1->object_id > entry2->object_id ? 1 : -1;

	if (entry1->key_len != entry2->key_len)
		return entry1->key_len > entry2->key_len ? 1 : -1;

	return strncmp(entry1->key, entry2->key, entry1->key_len);
}

/*
 * Uniqueness detection support.
 *
 * In order to detect uniqueness during building or parsing of a JSON
 * object, we maintain a hash table of key names already seen.
 */
static void
json_unique_check_init(JsonUniqueCheckState *cxt)
{
	HASHCTL		ctl;

	memset(&ctl, 0, sizeof(ctl));
	ctl.keysize = sizeof(JsonUniqueHashEntry);
	ctl.entrysize = sizeof(JsonUniqueHashEntry);
	ctl.hcxt = CurrentMemoryContext;
	ctl.hash = json_unique_hash;
	ctl.match = json_unique_hash_match;

	*cxt = hash_create("json object hashtable",
					   32,
					   &ctl,
					   HASH_ELEM | HASH_CONTEXT | HASH_FUNCTION | HASH_COMPARE);
}

static void
json_unique_builder_init(JsonUniqueBuilderState *cxt)
{
	json_unique_check_init(&cxt->check);
	cxt->mcxt = CurrentMemoryContext;
	cxt->skipped_keys.data = NULL;
}

static bool
json_unique_check_key(JsonUniqueCheckState *cxt, const char *key, int object_id)
{
	JsonUniqueHashEntry entry;
	bool		found;

	entry.key = key;
	entry.key_len = strlen(key);
	entry.object_id = object_id;

	(void) hash_search(*cxt, &entry, HASH_ENTER, &found);

	return !found;
}

/*
 * On-demand initialization of a throwaway StringInfo.  This is used to
 * read a key name that we don't need to store in the output object, for
 * duplicate key detection when the value is NULL.
 */
static StringInfo
json_unique_builder_get_throwawaybuf(JsonUniqueBuilderState *cxt)
{
	StringInfo	out = &cxt->skipped_keys;

	if (!out->data)
	{
		MemoryContext oldcxt = MemoryContextSwitchTo(cxt->mcxt);

		initStringInfo(out);
		MemoryContextSwitchTo(oldcxt);
	}
	else
		/* Just reset the string to empty */
		out->len = 0;

	return out;
}

/*
 * json_object_agg transition function.
 *
 * aggregate two input columns as a single json object value.
 */
static Datum
json_object_agg_transfn_worker(FunctionCallInfo fcinfo,
							   bool absent_on_null, bool unique_keys)
{
	MemoryContext aggcontext,
				oldcontext;
	JsonAggState *state;
	StringInfo	out;
	Datum		arg;
	bool		skip;
	int			key_offset;

	if (!AggCheckCallContext(fcinfo, &aggcontext))
	{
		/* cannot be called directly because of internal-type argument */
		elog(ERROR, "json_object_agg_transfn called in non-aggregate context");
	}

	if (PG_ARGISNULL(0))
	{
		Oid			arg_type;

		/*
		 * Make the StringInfo in a context where it will persist for the
		 * duration of the aggregate call. Switching context is only needed
		 * for this initial step, as the StringInfo and dynahash routines make
		 * sure they use the right context to enlarge the object if necessary.
		 */
		oldcontext = MemoryContextSwitchTo(aggcontext);
		state = (JsonAggState *) palloc(sizeof(JsonAggState));
		state->str = makeStringInfo();
		if (unique_keys)
			json_unique_builder_init(&state->unique_check);
		else
			memset(&state->unique_check, 0, sizeof(state->unique_check));
		MemoryContextSwitchTo(oldcontext);

		arg_type = get_fn_expr_argtype(fcinfo->flinfo, 1);

		if (arg_type == InvalidOid)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("could not determine data type for argument %d", 1)));

		json_categorize_type(arg_type, &state->key_category,
							 &state->key_output_func);

		arg_type = get_fn_expr_argtype(fcinfo->flinfo, 2);

		if (arg_type == InvalidOid)
			ereport(ERROR,
					(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
					 errmsg("could not determine data type for argument %d", 2)));

		json_categorize_type(arg_type, &state->val_category,
							 &state->val_output_func);

		appendStringInfoString(state->str, "{ ");
	}
	else
	{
		state = (JsonAggState *) PG_GETARG_POINTER(0);
	}

	/*
	 * Note: since json_object_agg() is declared as taking type "any", the
	 * parser will not do any type conversion on unknown-type literals (that
	 * is, undecorated strings or NULLs).  Such values will arrive here as
	 * type UNKNOWN, which fortunately does not matter to us, since
	 * unknownout() works fine.
	 */

	if (PG_ARGISNULL(1))
		ereport(ERROR,
				(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
				 errmsg("null value not allowed for object key")));

	/* Skip null values if absent_on_null */
	skip = absent_on_null && PG_ARGISNULL(2);

	if (skip)
	{
		/*
		 * We got a NULL value and we're not storing those; if we're not
		 * testing key uniqueness, we're done.  If we are, use the throwaway
		 * buffer to store the key name so that we can check it.
		 */
		if (!unique_keys)
			PG_RETURN_POINTER(state);

		out = json_unique_builder_get_throwawaybuf(&state->unique_check);
	}
	else
	{
		out = state->str;

		/*
		 * Append comma delimiter only if we have already output some fields
		 * after the initial string "{ ".
		 */
		if (out->len > 2)
			appendStringInfoString(out, ", ");
	}

	arg = PG_GETARG_DATUM(1);

	key_offset = out->len;

	datum_to_json(arg, false, out, state->key_category,
				  state->key_output_func, true);

	if (unique_keys)
	{
		const char *key = &out->data[key_offset];

		if (!json_unique_check_key(&state->unique_check.check, key, 0))
			ereport(ERROR,
					errcode(ERRCODE_DUPLICATE_JSON_OBJECT_KEY_VALUE),
					errmsg("duplicate JSON key %s", key));

		if (skip)
			PG_RETURN_POINTER(state);
	}

	appendStringInfoString(state->str, " : ");

	if (PG_ARGISNULL(2))
		arg = (Datum) 0;
	else
		arg = PG_GETARG_DATUM(2);

	datum_to_json(arg, PG_ARGISNULL(2), state->str, state->val_category,
				  state->val_output_func, false);

	PG_RETURN_POINTER(state);
}

/*
 * json_object_agg aggregate function
 */
Datum
json_object_agg_transfn(PG_FUNCTION_ARGS)
{
	return json_object_agg_transfn_worker(fcinfo, false, false);
}

/*
 * json_object_agg_strict aggregate function
 */
Datum
json_object_agg_strict_transfn(PG_FUNCTION_ARGS)
{
	return json_object_agg_transfn_worker(fcinfo, true, false);
}

/*
 * json_object_agg_unique aggregate function
 */
Datum
json_object_agg_unique_transfn(PG_FUNCTION_ARGS)
{
	return json_object_agg_transfn_worker(fcinfo, false, true);
}

/*
 * json_object_agg_unique_strict aggregate function
 */
Datum
json_object_agg_unique_strict_transfn(PG_FUNCTION_ARGS)
{
	return json_object_agg_transfn_worker(fcinfo, true, true);
}

/*
 * json_object_agg final function.
 */
Datum
json_object_agg_finalfn(PG_FUNCTION_ARGS)
{
	JsonAggState *state;

	/* cannot be called directly because of internal-type argument */
	Assert(AggCheckCallContext(fcinfo, NULL));

	state = PG_ARGISNULL(0) ? NULL : (JsonAggState *) PG_GETARG_POINTER(0);

	/* NULL result for no rows in, as is standard with aggregates */
	if (state == NULL)
		PG_RETURN_NULL();

	/* Else return state with appropriate object terminator added */
	PG_RETURN_TEXT_P(catenate_stringinfo_string(state->str, " }"));
}

/*
 * Helper function for aggregates: return given StringInfo's contents plus
 * specified trailing string, as a text datum.  We need this because aggregate
 * final functions are not allowed to modify the aggregate state.
 */
static text *
catenate_stringinfo_string(StringInfo buffer, const char *addon)
{
	/* custom version of cstring_to_text_with_len */
	int			buflen = buffer->len;
	int			addlen = strlen(addon);
	text	   *result = (text *) palloc(buflen + addlen + VARHDRSZ);

	SET_VARSIZE(result, buflen + addlen + VARHDRSZ);
	memcpy(VARDATA(result), buffer->data, buflen);
	memcpy(VARDATA(result) + buflen, addon, addlen);

	return result;
}

Datum
json_build_object_worker(int nargs, Datum *args, bool *nulls, Oid *types,
						 bool absent_on_null, bool unique_keys)
{
	int			i;
	const char *sep = "";
	StringInfo	result;
	JsonUniqueBuilderState unique_check;

	if (nargs % 2 != 0)
		ereport(ERROR,
				(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
				 errmsg("argument list must have even number of elements"),
		/* translator: %s is a SQL function name */
				 errhint("The arguments of %s must consist of alternating keys and values.",
						 "json_build_object()")));

	result = makeStringInfo();

	appendStringInfoChar(result, '{');

	if (unique_keys)
		json_unique_builder_init(&unique_check);

	for (i = 0; i < nargs; i += 2)
	{
		StringInfo	out;
		bool		skip;
		int			key_offset;

		/* Skip null values if absent_on_null */
		skip = absent_on_null && nulls[i + 1];

		if (skip)
		{
			/* If key uniqueness check is needed we must save skipped keys */
			if (!unique_keys)
				continue;

			out = json_unique_builder_get_throwawaybuf(&unique_check);
		}
		else
		{
			appendStringInfoString(result, sep);
			sep = ", ";
			out = result;
		}

		/* process key */
		if (nulls[i])
			ereport(ERROR,
					(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
					 errmsg("null value not allowed for object key")));

		/* save key offset before appending it */
		key_offset = out->len;

		add_json(args[i], false, out, types[i], true);

		if (unique_keys)
		{
			/* check key uniqueness after key appending */
			const char *key = &out->data[key_offset];

			if (!json_unique_check_key(&unique_check.check, key, 0))
				ereport(ERROR,
						errcode(ERRCODE_DUPLICATE_JSON_OBJECT_KEY_VALUE),
						errmsg("duplicate JSON key %s", key));

			if (skip)
				continue;
		}

		appendStringInfoString(result, " : ");

		/* process value */
		add_json(args[i + 1], nulls[i + 1], result, types[i + 1], false);
	}

	appendStringInfoChar(result, '}');

	return PointerGetDatum(cstring_to_text_with_len(result->data, result->len));
}

/*
 * SQL function json_build_object(variadic "any")
 */
Datum
json_build_object(PG_FUNCTION_ARGS)
{
	Datum	   *args;
	bool	   *nulls;
	Oid		   *types;

	/* build argument values to build the object */
	int			nargs = extract_variadic_args(fcinfo, 0, true,
											  &args, &types, &nulls);

	if (nargs < 0)
		PG_RETURN_NULL();

	PG_RETURN_DATUM(json_build_object_worker(nargs, args, nulls, types, false, false));
}

/*
 * degenerate case of json_build_object where it gets 0 arguments.
 */
Datum
json_build_object_noargs(PG_FUNCTION_ARGS)
{
	PG_RETURN_TEXT_P(cstring_to_text_with_len("{}", 2));
}

Datum
json_build_array_worker(int nargs, Datum *args, bool *nulls, Oid *types,
						bool absent_on_null)
{
	int			i;
	const char *sep = "";
	StringInfo	result;

	result = makeStringInfo();

	appendStringInfoChar(result, '[');

	for (i = 0; i < nargs; i++)
	{
		if (absent_on_null && nulls[i])
			continue;

		appendStringInfoString(result, sep);
		sep = ", ";
		add_json(args[i], nulls[i], result, types[i], false);
	}

	appendStringInfoChar(result, ']');

	return PointerGetDatum(cstring_to_text_with_len(result->data, result->len));
}

/*
 * SQL function json_build_array(variadic "any")
 */
Datum
json_build_array(PG_FUNCTION_ARGS)
{
	Datum	   *args;
	bool	   *nulls;
	Oid		   *types;

	/* build argument values to build the object */
	int			nargs = extract_variadic_args(fcinfo, 0, true,
											  &args, &types, &nulls);

	if (nargs < 0)
		PG_RETURN_NULL();

	PG_RETURN_DATUM(json_build_array_worker(nargs, args, nulls, types, false));
}

/*
 * degenerate case of json_build_array where it gets 0 arguments.
 */
Datum
json_build_array_noargs(PG_FUNCTION_ARGS)
{
	PG_RETURN_TEXT_P(cstring_to_text_with_len("[]", 2));
}

/*
 * SQL function json_object(text[])
 *
 * take a one or two dimensional array of text as key/value pairs
 * for a json object.
 */
Datum
json_object(PG_FUNCTION_ARGS)
{
	ArrayType  *in_array = PG_GETARG_ARRAYTYPE_P(0);
	int			ndims = ARR_NDIM(in_array);
	StringInfoData result;
	Datum	   *in_datums;
	bool	   *in_nulls;
	int			in_count,
				count,
				i;
	text	   *rval;
	char	   *v;

	switch (ndims)
	{
		case 0:
			PG_RETURN_DATUM(CStringGetTextDatum("{}"));
			break;

		case 1:
			if ((ARR_DIMS(in_array)[0]) % 2)
				ereport(ERROR,
						(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
						 errmsg("array must have even number of elements")));
			break;

		case 2:
			if ((ARR_DIMS(in_array)[1]) != 2)
				ereport(ERROR,
						(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
						 errmsg("array must have two columns")));
			break;

		default:
			ereport(ERROR,
					(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
					 errmsg("wrong number of array subscripts")));
	}

	deconstruct_array_builtin(in_array, TEXTOID, &in_datums, &in_nulls, &in_count);

	count = in_count / 2;

	initStringInfo(&result);

	appendStringInfoChar(&result, '{');

	for (i = 0; i < count; ++i)
	{
		if (in_nulls[i * 2])
			ereport(ERROR,
					(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
					 errmsg("null value not allowed for object key")));

		v = TextDatumGetCString(in_datums[i * 2]);
		if (i > 0)
			appendStringInfoString(&result, ", ");
		escape_json(&result, v);
		appendStringInfoString(&result, " : ");
		pfree(v);
		if (in_nulls[i * 2 + 1])
			appendStringInfoString(&result, "null");
		else
		{
			v = TextDatumGetCString(in_datums[i * 2 + 1]);
			escape_json(&result, v);
			pfree(v);
		}
	}

	appendStringInfoChar(&result, '}');

	pfree(in_datums);
	pfree(in_nulls);

	rval = cstring_to_text_with_len(result.data, result.len);
	pfree(result.data);

	PG_RETURN_TEXT_P(rval);
}

/*
 * SQL function json_object(text[], text[])
 *
 * take separate key and value arrays of text to construct a json object
 * pairwise.
 */
Datum
json_object_two_arg(PG_FUNCTION_ARGS)
{
	ArrayType  *key_array = PG_GETARG_ARRAYTYPE_P(0);
	ArrayType  *val_array = PG_GETARG_ARRAYTYPE_P(1);
	int			nkdims = ARR_NDIM(key_array);
	int			nvdims = ARR_NDIM(val_array);
	StringInfoData result;
	Datum	   *key_datums,
			   *val_datums;
	bool	   *key_nulls,
			   *val_nulls;
	int			key_count,
				val_count,
				i;
	text	   *rval;
	char	   *v;

	if (nkdims > 1 || nkdims != nvdims)
		ereport(ERROR,
				(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
				 errmsg("wrong number of array subscripts")));

	if (nkdims == 0)
		PG_RETURN_DATUM(CStringGetTextDatum("{}"));

	deconstruct_array_builtin(key_array, TEXTOID, &key_datums, &key_nulls, &key_count);
	deconstruct_array_builtin(val_array, TEXTOID, &val_datums, &val_nulls, &val_count);

	if (key_count != val_count)
		ereport(ERROR,
				(errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR),
				 errmsg("mismatched array dimensions")));

	initStringInfo(&result);

	appendStringInfoChar(&result, '{');

	for (i = 0; i < key_count; ++i)
	{
		if (key_nulls[i])
			ereport(ERROR,
					(errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED),
					 errmsg("null value not allowed for object key")));

		v = TextDatumGetCString(key_datums[i]);
		if (i > 0)
			appendStringInfoString(&result, ", ");
		escape_json(&result, v);
		appendStringInfoString(&result, " : ");
		pfree(v);
		if (val_nulls[i])
			appendStringInfoString(&result, "null");
		else
		{
			v = TextDatumGetCString(val_datums[i]);
			escape_json(&result, v);
			pfree(v);
		}
	}

	appendStringInfoChar(&result, '}');

	pfree(key_datums);
	pfree(key_nulls);
	pfree(val_datums);
	pfree(val_nulls);

	rval = cstring_to_text_with_len(result.data, result.len);
	pfree(result.data);

	PG_RETURN_TEXT_P(rval);
}


/*
 * Produce a JSON string literal, properly escaping characters in the text.
 */
void
escape_json(StringInfo buf, const char *str)
{
	const char *p;

	appendStringInfoCharMacro(buf, '"');
	for (p = str; *p; p++)
	{
		switch (*p)
		{
			case '\b':
				appendStringInfoString(buf, "\\b");
				break;
			case '\f':
				appendStringInfoString(buf, "\\f");
				break;
			case '\n':
				appendStringInfoString(buf, "\\n");
				break;
			case '\r':
				appendStringInfoString(buf, "\\r");
				break;
			case '\t':
				appendStringInfoString(buf, "\\t");
				break;
			case '"':
				appendStringInfoString(buf, "\\\"");
				break;
			case '\\':
				appendStringInfoString(buf, "\\\\");
				break;
			default:
				if ((unsigned char) *p < ' ')
					appendStringInfo(buf, "\\u%04x", (int) *p);
				else
					appendStringInfoCharMacro(buf, *p);
				break;
		}
	}
	appendStringInfoCharMacro(buf, '"');
}

/*
 * SQL function json_typeof(json) -> text
 *
 * Returns the type of the outermost JSON value as TEXT.  Possible types are
 * "object", "array", "string", "number", "boolean", and "null".
 *
 * Performs a single call to json_lex() to get the first token of the supplied
 * value.  This initial token uniquely determines the value's type.  As our
 * input must already have been validated by json_in() or json_recv(), the
 * initial token should never be JSON_TOKEN_OBJECT_END, JSON_TOKEN_ARRAY_END,
 * JSON_TOKEN_COLON, JSON_TOKEN_COMMA, or JSON_TOKEN_END.
 */
Datum
json_typeof(PG_FUNCTION_ARGS)
{
	text	   *json;

	JsonLexContext *lex;
	JsonTokenType tok;
	char	   *type;
	JsonParseErrorType result;

	json = PG_GETARG_TEXT_PP(0);
	lex = makeJsonLexContext(json, false);

	/* Lex exactly one token from the input and check its type. */
	result = json_lex(lex);
	if (result != JSON_SUCCESS)
		json_errsave_error(result, lex, NULL);
	tok = lex->token_type;
	switch (tok)
	{
		case JSON_TOKEN_OBJECT_START:
			type = "object";
			break;
		case JSON_TOKEN_ARRAY_START:
			type = "array";
			break;
		case JSON_TOKEN_STRING:
			type = "string";
			break;
		case JSON_TOKEN_NUMBER:
			type = "number";
			break;
		case JSON_TOKEN_TRUE:
		case JSON_TOKEN_FALSE:
			type = "boolean";
			break;
		case JSON_TOKEN_NULL:
			type = "null";
			break;
		default:
			elog(ERROR, "unexpected json token: %d", tok);
	}

	PG_RETURN_TEXT_P(cstring_to_text(type));
}