Function GinBsonExtractQueryRegex

pg_documentdb/src/opclass/bson_gin_core.c:2683–2761

* GinBsonExtractRegex generates the index term needed to match * $regex on a dotted path and value. * In this case, We return a lower bound value of the min value for that path. * Since we set the PartialMatch value to be true, the min value will be considered * a lower bound for the indexed search. * We also pass 'extraData' being the regex text being queried. This will later be treated * a

 * i.e. a BsonElement.
 */
static Datum *
GinBsonExtractQueryRegex(BsonExtractQueryArgs *args)
{
	int32 *nentries = args->nentries;
	bool **partialmatch = args->partialmatch;
	Pointer **extra_data = args->extra_data;
	pgbsonelement queryElement = args->filterElement;

	if ((queryElement.bsonValue.value_type != BSON_TYPE_UTF8) &&
		(queryElement.bsonValue.value_type != BSON_TYPE_REGEX))
	{
		ereport(ERROR, (errcode(ERRCODE_DOCUMENTDB_BADVALUE), errmsg(
							"$regex must be provided with a string value")));
	}

	int32_t numEntries = 2;
	bool isRegexValue = false;
	if (queryElement.bsonValue.value_type == BSON_TYPE_REGEX)
	{
		numEntries++;
		isRegexValue = true;
	}

	*nentries = numEntries;
	Datum *entries = (Datum *) palloc(sizeof(Datum) * numEntries);
	*partialmatch = (bool *) palloc0(sizeof(bool) * numEntries);
	(*partialmatch)[0] = true;
	(*partialmatch)[1] = false;

	if (isRegexValue)
	{
		(*partialmatch)[2] = false;
	}

	*extra_data = (Pointer *) palloc(sizeof(Pointer) * numEntries);

	/* now create a bson for that path which has the min value for the field */
	/* We set this as string.Empty but this can be set as the prefix expression */
	/* for the regex (i.e. the lowest possible value that can match a regex) */
	pgbsonelement emptyStringElement = queryElement;
	emptyStringElement.bsonValue.value_type = BSON_TYPE_UTF8;
	emptyStringElement.bsonValue.value.v_utf8.str = "";
	emptyStringElement.bsonValue.value.v_utf8.len = 0;

	entries[0] = PointerGetDatum(SerializeBsonIndexTerm(&emptyStringElement,
														&args->termMetadata).
								 indexTermVal);

	entries[1] = GenerateRootTruncatedTerm(&args->termMetadata);

	if (isRegexValue)
	{
		/* Also match for the regex itself */
		entries[2] = PointerGetDatum(SerializeBsonIndexTerm(&queryElement,
															&args->termMetadata).
									 indexTermVal);
	}