private Schema buildOutputSchema(
final PluginTask task, Schema inputSchema, List<ColumnConfig> jsonColumns) {
ImmutableList.Builder<Column> builder = ImmutableList.builder();
int i = 0;
for (Column inputColumn : inputSchema.getColumns()) {
Column outputColumn = new Column(i++, inputColumn.getName(), inputColumn.getType());
builder.add(outputColumn);
}
for (ColumnConfig columnConfig : jsonColumns) {
String columnName = task.getJoinedColumnsPrefix() + columnConfig.getName();
builder.add(new Column(i++, columnName, columnConfig.getType()));
}
return new Schema(builder.build());
}
@Override
public PageOutput open(
TaskSource taskSource,
final Schema inputSchema,
final Schema outputSchema,
final PageOutput output) {
final PluginTask task = taskSource.loadTask(PluginTask.class);
// create jsonColumns/baseColumn
final List<Column> outputColumns = outputSchema.getColumns();
final List<Column> inputColumns = inputSchema.getColumns();
Map<String, Column> inputColumnMap = Maps.newHashMap();
final List<Column> jsonColumns = new ArrayList<>();
for (Column column : outputColumns) {
if (!inputColumns.contains(column)) {
jsonColumns.add(column);
} else {
inputColumnMap.put(column.getName(), column);
}
}
final Column baseColumn = inputColumnMap.get(task.getBaseColumn().getName());
// create timestampParserMap
final HashMap<String, TimestampParser> timestampParserMap = Maps.newHashMap();
for (ColumnConfig jsonColumnConfig : task.getJsonColumns()) {
if (Types.TIMESTAMP.equals(jsonColumnConfig.getType())) {
String format = jsonColumnConfig.getOption().get(String.class, "format");
DateTimeZone timezone = DateTimeZone.forID(task.getTimeZone());
TimestampParser parser = new TimestampParser(task.getJRuby(), format, timezone);
timestampParserMap.put(task.getJoinedColumnsPrefix() + jsonColumnConfig.getName(), parser);
}
}
// get jsonTable
final HashMap<String, HashMap<String, String>> jsonTable = task.getJsonTable();
return new LeftOuterJoinJsonTableFilterFilteredPageOutput(
inputSchema, outputSchema, baseColumn, jsonTable, jsonColumns, timestampParserMap, output);
}
private HashMap<String, HashMap<String, String>> buildJsonTable(
final PluginTask task, List<HashMap<String, String>> jsonData) {
HashMap<String, HashMap<String, String>> jsonTable = Maps.newHashMap();
for (HashMap<String, String> json : jsonData) {
HashMap<String, String> record = Maps.newHashMap();
for (ColumnConfig columnConfig : task.getJsonColumns()) {
String columnKey = task.getJoinedColumnsPrefix() + columnConfig.getName();
String value = json.get(columnConfig.getName());
record.put(columnKey, value);
}
String rowKey = json.get(task.getCounterColumn().getName());
jsonTable.put(rowKey, record);
}
return jsonTable;
}