> ## Documentation Index
> Fetch the complete documentation index at: https://private-7c7dfe99-parallel-read-in-order-multi-part.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

> ClickHouse 中支持的输入和输出数据格式概览

# 输入和输出数据格式

export const CloudNotSupportedBadge = () => {
  return <a href="https://clickhouse.com/docs/products/cloud/guides/cloud-compatibility#list-of-unsupported-features" className="cloudNotSupportedBadge">
            <div className="cloudNotSupportedIcon">
            <svg width="16" height="16" viewBox="0 0 16 16" fill="none" xmlns="http://www.w3.org/2000/svg">
                <path strokeWidth="1.5" d="M6.33366 12.6666L12.3739 12.6667C13.6593 12.6667 14.7073 11.6187 14.7073 10.3334C14.7073 9.04804 13.6593 8.00003 12.3739 8.00003C12.3739 8.00003 12.3337 7.66659 12.0003 7.33325M10.667 5.33322C8.00033 2.33325 4.45395 4.78537 4.14195 6.68203C2.55728 6.7627 1.29395 8.06203 1.29395 9.6667C1.29395 11.3234 2.66699 12.6666 4.00033 12.6666" stroke="currentColor" strokeLinecap="round" strokeLinejoin="round" />
                <path strokeWidth="1.5" d="M2.66699 14L12.0003 4.66663" stroke="currentColor" strokeLinecap="round" strokeLinejoin="round" />
            </svg>

        </div>
            ClickHouse Cloud 不支持此功能
        </a>;
};

ClickHouse 支持大多数已知的文本和二进制数据格式。这使其能够轻松集成到几乎任何现有的数据管道中，
从而充分发挥 ClickHouse 的优势。

<h2 id="input-formats">
  输入格式
</h2>

输入格式用于：

* 解析提供给 `INSERT` 语句的数据
* 从 `File`、`URL` 或 `HDFS` 等以文件为后端的表执行 `SELECT` 查询
* 读取字典

选择合适的输入格式，对于在 ClickHouse 中高效摄取数据至关重要。支持的格式超过 70 种，
选择性能最佳的格式会显著影响插入速度、CPU 和内存占用，以及整体系统
效率。为了帮助你更好地在这些选项之间做出选择，我们对不同格式的摄取性能进行了基准测试，并总结出以下关键结论：

* **[Native](/zh/reference/formats/Native) 格式是效率最高的输入格式**，具备最佳压缩效果、最低
  资源占用，以及最小的服务端处理开销。
* **压缩至关重要** - LZ4 能以极低的 CPU 代价减小数据体积，而 ZSTD 则能提供更高的压缩率，但
  会增加 CPU 使用。
* **预排序的影响较为有限**，因为 ClickHouse 本身已经能够高效完成排序。
* **批处理可显著提升效率** - 更大的批次可减少插入开销并提高吞吐量。

如需深入了解测试结果和最佳实践，
请阅读完整的[基准分析](https://www.clickhouse.com/blog/clickhouse-input-format-matchup-which-is-fastest-most-efficient)。
如需查看完整测试结果，请访问 [FastFormats](https://fastformats.clickhouse.com/) 在线仪表板。

<h2 id="output-formats">
  输出格式
</h2>

支持输出的格式可用于：

* 格式化 `SELECT` 查询结果
* 对以文件为后端的表执行 `INSERT` 操作

<h2 id="formats-overview">
  格式概述
</h2>

支持的格式如下：

| 格式 | 输入 | 输出 |
| - | - | - |
| [TabSeparated](/zh/reference/formats/TabSeparated/TabSeparated) | ✔ | ✔ |
| [TabSeparatedRaw](/zh/reference/formats/TabSeparated/TabSeparatedRaw) | ✔ | ✔ |
| [TabSeparatedWithNames](/zh/reference/formats/TabSeparated/TabSeparatedWithNames) | ✔ | ✔ |
| [TabSeparatedWithNamesAndTypes](/zh/reference/formats/TabSeparated/TabSeparatedWithNamesAndTypes) | ✔ | ✔ |
| [TabSeparatedRawWithNames](/zh/reference/formats/TabSeparated/TabSeparatedRawWithNames) | ✔ | ✔ |
| [TabSeparatedRawWithNamesAndTypes](/zh/reference/formats/TabSeparated/TabSeparatedRawWithNamesAndTypes) | ✔ | ✔ |
| [Template](/zh/reference/formats/Template/Template) | ✔ | ✔ |
| [TemplateIgnoreSpaces](/zh/reference/formats/Template/TemplateIgnoreSpaces) | ✔ | ✗ |
| [CSV](/zh/reference/formats/CSV/CSV) | ✔ | ✔ |
| [CSVWithNames](/zh/reference/formats/CSV/CSVWithNames) | ✔ | ✔ |
| [CSVWithNamesAndTypes](/zh/reference/formats/CSV/CSVWithNamesAndTypes) | ✔ | ✔ |
| [CustomSeparated](/zh/reference/formats/CustomSeparated/CustomSeparated) | ✔ | ✔ |
| [CustomSeparatedIgnoreSpaces](/zh/reference/formats/CustomSeparated/CustomSeparatedIgnoreSpaces) | ✔ | ✗ |
| [CustomSeparatedIgnoreSpacesWithNames](/zh/reference/formats/CustomSeparated/CustomSeparatedIgnoreSpacesWithNames) | ✔ | ✗ |
| [CustomSeparatedIgnoreSpacesWithNamesAndTypes](/zh/reference/formats/CustomSeparated/CustomSeparatedIgnoreSpacesWithNamesAndTypes) | ✔ | ✗ |
| [CustomSeparatedWithNames](/zh/reference/formats/CustomSeparated/CustomSeparatedWithNames) | ✔ | ✔ |
| [CustomSeparatedWithNamesAndTypes](/zh/reference/formats/CustomSeparated/CustomSeparatedWithNamesAndTypes) | ✔ | ✔ |
| [HiveText](/zh/reference/formats/HiveText) | ✔ | ✔ |
| [SQLInsert](/zh/reference/formats/SQLInsert) | ✗ | ✔ |
| [Values](/zh/reference/formats/Values) | ✔ | ✔ |
| [Vertical](/zh/reference/formats/Vertical) | ✗ | ✔ |
| [JSON](/zh/reference/formats/JSON/JSON) | ✔ | ✔ |
| [JSONAsString](/zh/reference/formats/JSON/JSONAsString) | ✔ | ✗ |
| [JSONAsObject](/zh/reference/formats/JSON/JSONAsObject) | ✔ | ✗ |
| [JSONStrings](/zh/reference/formats/JSON/JSONStrings) | ✗ | ✔ |
| [JSONColumns](/zh/reference/formats/JSON/JSONColumns) | ✔ | ✔ |
| [JSONColumnsWithMetadata](/zh/reference/formats/JSON/JSONColumnsWithMetadata) | ✔ | ✔ |
| [JSONCompact](/zh/reference/formats/JSON/JSONCompact) | ✔ | ✔ |
| [JSONCompactStrings](/zh/reference/formats/JSON/JSONCompactStrings) | ✗ | ✔ |
| [JSONCompactColumns](/zh/reference/formats/JSON/JSONCompactColumns) | ✔ | ✔ |
| [JSONEachRow](/zh/reference/formats/JSON/JSONEachRow) | ✔ | ✔ |
| [JSONLines](/zh/reference/formats/JSON/JSONLines) | ✔ | ✔ |
| [PrettyJSONEachRow](/zh/reference/formats/JSON/PrettyJSONEachRow) | ✗ | ✔ |
| [JSONEachRowWithProgress](/zh/reference/formats/JSON/JSONEachRowWithProgress) | ✗ | ✔ |
| [JSONStringsEachRow](/zh/reference/formats/JSON/JSONStringsEachRow) | ✔ | ✔ |
| [JSONStringsEachRowWithProgress](/zh/reference/formats/JSON/JSONStringsEachRowWithProgress) | ✗ | ✔ |
| [JSONCompactEachRow](/zh/reference/formats/JSON/JSONCompactEachRow) | ✔ | ✔ |
| [JSONCompactEachRowWithNames](/zh/reference/formats/JSON/JSONCompactEachRowWithNames) | ✔ | ✔ |
| [JSONCompactEachRowWithNamesAndTypes](/zh/reference/formats/JSON/JSONCompactEachRowWithNamesAndTypes) | ✔ | ✔ |
| [JSONCompactEachRowWithProgress](/zh/reference/formats/JSON/JSONCompactEachRowWithProgress) | ✗ | ✔ |
| [JSONCompactStringsEachRow](/zh/reference/formats/JSON/JSONCompactStringsEachRow) | ✔ | ✔ |
| [JSONCompactStringsEachRowWithNames](/zh/reference/formats/JSON/JSONCompactStringsEachRowWithNames) | ✔ | ✔ |
| [JSONCompactStringsEachRowWithNamesAndTypes](/zh/reference/formats/JSON/JSONCompactStringsEachRowWithNamesAndTypes) | ✔ | ✔ |
| [JSONCompactStringsEachRowWithProgress](/zh/reference/formats/JSON/JSONCompactStringsEachRowWithProgress) | ✗ | ✔ |
| [JSONObjectEachRow](/zh/reference/formats/JSON/JSONObjectEachRow) | ✔ | ✔ |
| [BSONEachRow](/zh/reference/formats/BSONEachRow) | ✔ | ✔ |
| [TSKV](/zh/reference/formats/TabSeparated/TSKV) | ✔ | ✔ |
| [Pretty](/zh/reference/formats/Pretty/Pretty) | ✗ | ✔ |
| [PrettyNoEscapes](/zh/reference/formats/Pretty/PrettyNoEscapes) | ✗ | ✔ |
| [PrettyMonoBlock](/zh/reference/formats/Pretty/PrettyMonoBlock) | ✗ | ✔ |
| [PrettyNoEscapesMonoBlock](/zh/reference/formats/Pretty/PrettyNoEscapesMonoBlock) | ✗ | ✔ |
| [PrettyCompact](/zh/reference/formats/Pretty/PrettyCompact) | ✗ | ✔ |
| [PrettyCompactNoEscapes](/zh/reference/formats/Pretty/PrettyCompactNoEscapes) | ✗ | ✔ |
| [PrettyCompactMonoBlock](/zh/reference/formats/Pretty/PrettyCompactMonoBlock) | ✗ | ✔ |
| [PrettyCompactNoEscapesMonoBlock](/zh/reference/formats/Pretty/PrettyCompactNoEscapesMonoBlock) | ✗ | ✔ |
| [PrettySpace](/zh/reference/formats/Pretty/PrettySpace) | ✗ | ✔ |
| [PrettySpaceNoEscapes](/zh/reference/formats/Pretty/PrettySpaceNoEscapes) | ✗ | ✔ |
| [PrettySpaceMonoBlock](/zh/reference/formats/Pretty/PrettySpaceMonoBlock) | ✗ | ✔ |
| [PrettySpaceNoEscapesMonoBlock](/zh/reference/formats/Pretty/PrettySpaceNoEscapesMonoBlock) | ✗ | ✔ |
| [Prometheus](/zh/reference/formats/Prometheus) | ✗ | ✔ |
| [Protobuf](/zh/reference/formats/Protobuf/Protobuf) | ✔ | ✔ |
| [ProtobufSingle](/zh/reference/formats/Protobuf/ProtobufSingle) | ✔ | ✔ |
| [ProtobufList](/zh/reference/formats/Protobuf/ProtobufList) | ✔ | ✔ |
| [Avro](/zh/reference/formats/Avro/Avro) | ✔ | ✔ |
| [AvroConfluent](/zh/reference/formats/Avro/AvroConfluent) | ✔ | ✔ |
| [Parquet](/zh/reference/formats/Parquet/Parquet) | ✔ | ✔ |
| [ParquetMetadata](/zh/reference/formats/Parquet/ParquetMetadata) | ✔ | ✗ |
| [Arrow](/zh/reference/formats/Arrow/Arrow) | ✔ | ✔ |
| [ArrowStream](/zh/reference/formats/Arrow/ArrowStream) | ✔ | ✔ |
| [ORC](/zh/reference/formats/ORC) | ✔ | ✔ |
| [One](/zh/reference/formats/One) | ✔ | ✗ |
| [Npy](/zh/reference/formats/Npy) | ✔ | ✔ |
| [RowBinary](/zh/reference/formats/RowBinary/RowBinary) | ✔ | ✔ |
| [RowBinaryWithNames](/zh/reference/formats/RowBinary/RowBinaryWithNames) | ✔ | ✔ |
| [RowBinaryWithNamesAndTypes](/zh/reference/formats/RowBinary/RowBinaryWithNamesAndTypes) | ✔ | ✔ |
| [RowBinaryWithDefaults](/zh/reference/formats/RowBinary/RowBinaryWithDefaults) | ✔ | ✗ |
| [RowBinaryWithNamesAndTypesAndDefaults](/zh/reference/formats/RowBinary/RowBinaryWithNamesAndTypesAndDefaults) | ✔ | ✗ |
| [Native](/zh/reference/formats/Native) | ✔ | ✔ |
| [Buffers](/zh/reference/formats/Buffers) | ✔ | ✔ |
| [Null](/zh/reference/formats/Null) | ✗ | ✔ |
| [Hash](/zh/reference/formats/Hash) | ✗ | ✔ |
| [XML](/zh/reference/formats/XML) | ✗ | ✔ |
| [CapnProto](/zh/reference/formats/CapnProto) | ✔ | ✔ |
| [LineAsString](/zh/reference/formats/LineAsString/LineAsString) | ✔ | ✔ |
| [LineAsStringWithNames](/zh/reference/formats/LineAsString/LineAsStringWithNames) | ✗ | ✔ |
| [LineAsStringWithNamesAndTypes](/zh/reference/formats/LineAsString/LineAsStringWithNamesAndTypes) | ✗ | ✔ |
| [Regexp](/zh/reference/formats/Regexp) | ✔ | ✗ |
| [RawBLOB](/zh/reference/formats/RawBLOB) | ✔ | ✔ |
| [MsgPack](/zh/reference/formats/MsgPack) | ✔ | ✔ |
| [MySQLDump](/zh/reference/formats/MySQLDump) | ✔ | ✗ |
| [MySQLWire](/zh/reference/formats/MySQLWire) | ✗ | ✔ |
| [PostgreSQLWire](/zh/reference/formats/PostgreSQLWire) | ✗ | ✔ |
| [ODBCDriver2](/zh/reference/formats/ODBCDriver2) | ✗ | ✔ |
| [GeoJSON](/zh/reference/formats/GeoJSON) | ✔ | ✔ |
| [DWARF](/zh/reference/formats/DWARF) | ✔ | ✗ |
| [Markdown](/zh/reference/formats/Markdown) | ✗ | ✔ |
| [Form](/zh/reference/formats/Form) | ✔ | ✗ |
| [PNG](/zh/reference/formats/PNG) | ✗ | ✔ |
| [Puffin](/zh/reference/formats/Puffin/Puffin) | ✔ | ✗ |
| [PuffinMetadata](/zh/reference/formats/Puffin/PuffinMetadata) | ✔ | ✗ |

您可以通过 ClickHouse 设置控制某些格式处理参数。更多信息，请参阅 [设置](/zh/reference/settings/formats) 部分。

JSON 特有的其他设置记录在 [JSON 格式设置](/zh/reference/formats/JSON/format-settings) 页面中。

<h2 id="formatschema">
  格式 schema
</h2>

包含格式 schema 的文件名由设置 `format_schema` 指定。
使用 `Cap'n Proto` 和 `Protobuf` 这两种格式之一时，必须设置此项。
格式 schema 由文件名和该文件中的消息类型名称组成，二者以冒号分隔，
例如 `schemafile.proto:MessageType`。
如果文件使用该格式的标准扩展名 (例如 `Protobuf` 的 `.proto`) ，
则可以省略扩展名，此时格式 schema 形如 `schemafile:MessageType`。

如果你通过处于交互模式的[客户端](/zh/concepts/features/interfaces/client)输入或输出数据，则格式 schema 中指定的文件名
可以包含绝对路径，或相对于客户端当前目录的路径。
如果你在[批次模式](/zh/concepts/features/interfaces/client#batch-mode)下使用客户端，出于安全原因，schema 的路径必须是相对路径。

如果你通过 [HTTP 接口](/zh/concepts/features/interfaces/http) 输入或输出数据，则格式 schema 中指定的文件名
应位于服务器配置中 [format\_schema\_path](/zh/reference/settings/server-settings/settings/format#format_schema_path)
指定的目录下。

<h2 id="skippingerrors">
  跳过错误
</h2>

某些格式 (如 `CSV`、`TabSeparated`、`TSKV`、`JSONEachRow`、`Template`、`CustomSeparated` 和 `Protobuf`) 在发生解析错误时，可以跳过损坏的行，并从下一行的开头继续解析。请参阅设置 [input\_format\_allow\_errors\_num](/zh/reference/settings/formats/input-format#input_format_allow_errors_num) 和
[input\_format\_allow\_errors\_ratio](/zh/reference/settings/formats/input-format#input_format_allow_errors_ratio)。
限制：

* 发生解析错误时，`JSONEachRow` 会跳过直到换行符 (或 EOF) 之前的所有数据，因此各行必须以 `\n` 分隔，才能正确统计错误数。
* `Template` 和 `CustomSeparated` 通过最后一列后的分隔符以及行间分隔符来定位下一行的开头，因此只有当其中至少一个非空时，才支持跳过错误。