private void decodeArrayString(
int index, Type field, Group pqGroup, String arrayString, char delim) throws IOException {
// for parquet, we only have one-dimention array
// anotation support: decimal, time, timestamp
String[] splits = FormatHandlerUtil.getArraySplits(arrayString.toCharArray(), delim);
for (String elementString : splits) {
switch (field.asPrimitiveType().getPrimitiveTypeName()) {
case BOOLEAN:
pqGroup.add(index, Boolean.parseBoolean(elementString));
break;
case INT32:
if (columnSchemas.get(index).getType() == GPDBWritable.DATE) {
pqGroup.add(
index,
(int)
FormatHandlerUtil.getTimeDiff(
elementString, "1970-01-01", "yyyy-mm-dd", 24 * 60 * 60 * 1000));
} else if (columnSchemas.get(index).getType() == GPDBWritable.TIME) {
pqGroup.add(
index,
(int) FormatHandlerUtil.getTimeDiff(elementString, "00:00:00", "mm:hh:ss", 1));
} else {
pqGroup.add(index, Integer.parseInt(elementString));
}
break;
case INT64:
if (columnSchemas.get(index).getType() == GPDBWritable.TIMESTAMP) {
pqGroup.add(
index,
FormatHandlerUtil.getTimeDiff(
elementString, "1970-01-01 00:00:00", "yyyy-mm-dd mm:hh:ss", 1));
} else {
pqGroup.add(index, Long.parseLong(elementString));
}
break;
case FLOAT:
pqGroup.add(index, Float.parseFloat(elementString));
break;
case DOUBLE:
pqGroup.add(index, Double.parseDouble(elementString));
break;
case INT96:
case BINARY:
case FIXED_LEN_BYTE_ARRAY:
OriginalType type = field.getOriginalType();
if (type == OriginalType.UTF8 || type == OriginalType.JSON) {
pqGroup.add(index, elementString);
} else if (type == OriginalType.DECIMAL) {
pqGroup.add(index, Binary.fromByteArray(elementString.getBytes()));
} else if (type == OriginalType.INTERVAL) {
pqGroup.add(
index, Binary.fromByteArray(FormatHandlerUtil.getParquetInterval(elementString)));
} else {
pqGroup.add(
index,
Binary.fromByteArray(FormatHandlerUtil.octString2byteArray(elementString).array()));
}
break;
default:
throw new IOException(
"internal error, you should not be here, pqtype:"
+ field.asPrimitiveType().getPrimitiveTypeName());
}
}
}
private void fillElement(int index, int colType, Group pqGroup, GPDBWritable gw, Type field)
throws IOException {
switch (colType) {
case GPDBWritable.BPCHAR:
case GPDBWritable.CHAR:
case GPDBWritable.DATE:
case GPDBWritable.NUMERIC:
case GPDBWritable.TIME:
case GPDBWritable.TIMESTAMP:
case GPDBWritable.VARCHAR:
case GPDBWritable.TEXT:
// utf8 or array
if (field.getRepetition() == Repetition.REPEATED) {
decodeArrayString(
index, field, pqGroup, gw.getString(index), columnSchemas.get(index).getDelim());
} else {
int gpdbType = columnSchemas.get(index).getType();
PrimitiveTypeName priType = field.asPrimitiveType().getPrimitiveTypeName();
OriginalType originalType = field.getOriginalType();
if (gpdbType == GPDBWritable.NUMERIC && priType == PrimitiveTypeName.INT32) {
pqGroup.add(index, Integer.parseInt(gw.getString(index)));
} else if (gpdbType == GPDBWritable.NUMERIC && priType == PrimitiveTypeName.INT64) {
pqGroup.add(index, Long.parseLong(gw.getString(index)));
} else if (gpdbType == GPDBWritable.DATE && priType == PrimitiveTypeName.INT32) {
pqGroup.add(
index,
(int)
FormatHandlerUtil.getTimeDiff(
gw.getString(index), "1970-01-01", "yyyy-mm-dd", 24 * 60 * 60 * 1000));
} else if (gpdbType == GPDBWritable.TIME && priType == PrimitiveTypeName.INT32) {
pqGroup.add(
index,
(int)
FormatHandlerUtil.getTimeDiff(gw.getString(index), "00:00:00", "mm:hh:ss", 1));
} else if (gpdbType == GPDBWritable.TIMESTAMP && priType == PrimitiveTypeName.INT64) {
pqGroup.add(
index,
FormatHandlerUtil.getTimeDiff(
gw.getString(index), "1970-01-01 00:00:00", "yyyy-mm-dd mm:hh:ss", 1));
} else if (gpdbType == GPDBWritable.INTERVAL && originalType == OriginalType.INTERVAL) {
// interval is complex, we just use string, for now, we just support 'postgres'
// style interval
// 1 year 2 mons -3 days +04:05:06.00901
byte[] interval = FormatHandlerUtil.getParquetInterval(gw.getString(index));
pqGroup.add(index, Binary.fromByteArray(interval));
} else {
pqGroup.add(index, gw.getString(index));
}
}
break;
case GPDBWritable.BYTEA:
pqGroup.add(index, Binary.fromByteArray(gw.getBytes(index)));
break;
case GPDBWritable.REAL:
pqGroup.add(index, gw.getFloat(index));
break;
case GPDBWritable.BIGINT:
pqGroup.add(index, gw.getLong(index));
break;
case GPDBWritable.BOOLEAN:
pqGroup.add(index, gw.getBoolean(index));
break;
case GPDBWritable.FLOAT8:
pqGroup.add(index, gw.getDouble(index));
break;
case GPDBWritable.INTEGER:
pqGroup.add(index, gw.getInt(index));
break;
case GPDBWritable.SMALLINT:
pqGroup.add(index, gw.getShort(index));
break;
default:
throw new IOException("internal error, not support type, typeId:" + colType);
}
}
