背景

本文基于Spark 4.0

总结

Spark中的 VariantType 类型，用尽量少的字节来存储Json的格式化数据

分析

这里主要介绍 Variant 的存储，我们从VariantBuilder.buildJson方法(把对应的json数据存储为VariantType类型)开始：

public static Variant parseJson(JsonParser parser, boolean allowDuplicateKeys)throws IOException {VariantBuilder builder = new VariantBuilder(allowDuplicateKeys);builder.buildJson(parser);return builder.result();}

这个方法会调用buildJson这个方法：

JsonToken token = parser.currentToken();if (token == null) {throw new JsonParseException(parser, "Unexpected null token");}switch (token) {case START_OBJECT: {ArrayList<FieldEntry> fields = new ArrayList<>();int start = writePos;while (parser.nextToken() != JsonToken.END_OBJECT) {String key = parser.currentName();parser.nextToken();int id = addKey(key);fields.add(new FieldEntry(key, id, writePos - start));buildJson(parser);}finishWritingObject(start, fields);break;}case START_ARRAY: {ArrayList<Integer> offsets = new ArrayList<>();int start = writePos;while (parser.nextToken() != JsonToken.END_ARRAY) {offsets.add(writePos - start);buildJson(parser);}finishWritingArray(start, offsets);break;}case VALUE_STRING:appendString(parser.getText());break;case VALUE_NUMBER_INT:try {appendLong(parser.getLongValue(