--- layout: Conceptual title: DESCRIBE TABLE（描述表结构）- Azure Databricks - Azure Databricks | Azure Docs canonicalUrl: https://docs.azure.cn/zh-cn/databricks/sql/language-manual/sql-ref-syntax-aux-describe-table schema: Conceptual author: rockboyfor breadcrumb_path: /bread/toc.json depot_name: Azure.mooncake-docs description: 了解如何在 Databricks SQL 和 Databricks Runtime 中使用 SQL 语言的 DESCRIBE TABLE 语法。 document_id: 2ce4f122-8453-dd79-28b6-53bbfc8b5e28 document_version_independent_id: 0ed05535-983b-f655-4651-d95d43544845 git_commit_id: 3c39e98586f35073778c9a364fe111cbf4d1aa09 gitcommit: https://github.com/MicrosoftDocs/mc-docs-pr/blob/3c39e98586f35073778c9a364fe111cbf4d1aa09/articles/databricks/sql/language-manual/sql-ref-syntax-aux-describe-table.md locale: zh-cn ms.author: v-edwardchen ms.custom: databricksmigration ms.date: 2026-04-13T00:00:00.0000000Z ms.reviewer: jasonh ms.service: azure-databricks ms.subservice: databricks-sql-analytics ms.topic: language-reference origin.date: 2026-01-20T00:00:00.0000000Z original_content_git_url: https://github.com/MicrosoftDocs/mc-docs-pr/blob/live/articles/databricks/sql/language-manual/sql-ref-syntax-aux-describe-table.md recommendations: false site_name: DocsAzureCN uhfHeaderId: mooncake updated_at: 2026-04-17T07:23:00.0000000Z ms.translationtype: MT ms.contentlocale: zh-cn loc_version: 2026-02-12T09:27:43.6701168Z loc_source_id: Github-85544329#live loc_file_id: Github-85544329.live.Azure.mooncake-docs.articles/databricks/sql/language-manual/sql-ref-syntax-aux-describe-table.md page_type: conceptual toc_rel: ../../toc.json feedback_system: None feedback_product_url: '' feedback_help_link_type: '' feedback_help_link_url: '' word_count: 1449 asset_id: databricks/sql/language-manual/sql-ref-syntax-aux-describe-table item_type: Content cmProducts: - https://authoring-docs-microsoft.poolparty.biz/devrel/545d40c6-c50c-444b-b422-1c707eeab28e - https://authoring-docs-microsoft.poolparty.biz/devrel/cbe4ca68-43ac-4375-aba5-5945a6394c20 spProducts: - https://authoring-docs-microsoft.poolparty.biz/devrel/b908d601-32e8-445a-b044-a507b5d1689e - https://authoring-docs-microsoft.poolparty.biz/devrel/ced846cc-6a3c-4c8f-9dfb-3de0e90e2742 platformId: d602a500-2fa3-e7fc-b411-1f648d0bb4b2 --- # DESCRIBE TABLE（描述表结构）- Azure Databricks - Azure Databricks | Azure Docs 适用于：![已勾选](../../_static/images/icons/check.png) Databricks SQL ![勾选标记为“是”](../../_static/images/icons/check.png) Databricks Runtime 返回表的基本元数据信息。 元数据信息包括列名、列类型和列注释。 可根据需要指定分区规范或列名称，以分别返回与分区或列有关的元数据。 使用 Delta 表时，不会返回所有字段。 元数据以报表形式或 JSON 文档的形式返回。 重要 使用 `DESCRIBE AS JSON` 以编程方式分析/描述输出。 非 JSON 报表格式可能会更改。 ## 语法 ``` { DESC | DESCRIBE } [ TABLE ] [ EXTENDED ] table_name { [ PARTITION clause ] | [ column_name ] } [ AS JSON ] ``` 为了兼容性，`FORMATTED` 可以作为 `EXTENDED`的同义词。 ## 参数 - **`EXTENDED`** 如果指定了该参数，则显示有关指定列的详细信息，包括命令收集的列统计信息和其他元数据（例如架构限定符、所有者和访问时间）。 - **[table_name](sql-ref-names#table-name)** 标识要描述的表。 该名称不能使用 [时态规范或选项规范](sql-ref-names#table-name)。 如果找不到表，Azure Databricks 将引发 [TABLE_OR_VIEW_NOT_FOUND](../../error-messages/table-or-view-not-found-error-class) 错误。 - **[PARTITION 子句](sql-ref-partition#partition)** 一个可选参数，指示 Databricks SQL 为命名分区返回附加元数据。 - **[column_name](sql-ref-names#column-name)** 一个可选参数，具有需说明的列名。 目前不允许指定嵌套列。 单个列支持 JSON 格式。 - **`AS JSON`** **适用于：**![已勾选“是”的](../../_static/images/icons/check.png) Databricks SQL ![已勾选“是”的](../../_static/images/icons/check.png) Databricks Runtime 16.2 及更高版本 （可选）将表元数据作为 JSON 字符串而不是人工可读报表返回。 使用程序分析结果时使用此格式。 仅当指定 `EXTENDED` 格式时才受支持。 参数`partition_spec`和`column_name`互斥，不能同时指定。 ## JSON 格式的输出 指定 `AS JSON` 时，输出将作为 JSON 字符串返回。 支持以下架构： ``` { "table_name": "", "catalog_name": "", "schema_name": "", "namespace": [""], "type": "", "provider": "", "columns": [ { "name": "", "type": , "comment": "", "nullable": , "default": "", "is_measure": , // Included only for measure columns of metric views } ], "partition_values": { "": "" }, "partition_columns": ["col1", "col2"], "location": "", "view_text": "", "view_original_text": "", "view_schema_mode": "", "view_catalog_and_namespace": "", "view_query_output_columns": [""], "comment": "", "table_properties": { "property1": "", "property2": "" }, "statistics": { "num_rows": , "size_in_bytes": , "table_change_stats": { "inserted": , "deleted": , "updated": , "change_percent": } }, "storage_properties": { "property1": "", "property2": "" }, "serde_library": "", "input_format": "", "output_format": "", "num_buckets": , "bucket_columns": [""], "sort_columns": [""], "created_time": "", "created_by": "", "last_access": "", "partition_provider": "", "collation": "", // Available in :re[DBR] 16.4 and above "language": "", // Language such as YAML, included only for metric views "row_filter": { // Available in :re[DBR] 17.0 and above "filter_function": { "catalog_name": "", "schema_name": "", "function_name: "", "specific_name": "" }, "arguments": ["", ""] }, "column_masks": [ // Available in :re[DBR] 17.0 and above {"column_name": "", "mask_function": { "catalog_name": "", "schema_name": "", "function_name: "", "specific_name": "" }, "arguments": ["", ""] } ] "refresh_information": { // Available in :re[DBR] 17.3 and above, included only for materialized views and streaming tables "last_refreshed_at": "", "last_refresh_type": "", "latest_refresh_status": "", "latest_refresh_link": "", "refresh_schedule": "" } } ``` 下面是 ``的架构定义： | SQL 类型 | JSON 表示形式 | | --- | --- | | [TINYINT](data-types/tinyint-type) | `{ "name" : "tinyint" }` | | [SMALLINT](data-types/smallint-type) | `{ "name" : "smallint" }` | | [INT](data-types/int-type) | `{ "name" : "int" }` | | [BIGINT](data-types/bigint-type) | `{ "name" : "bigint" }` | | [浮](data-types/float-type) | `{ "name" : "float" }` | | [加倍](data-types/double-type) | `{ "name" : "double" }` | | [DECIMAL（p，s）](data-types/decimal-type) | `{ "name" : "decimal", "precision": p, "scale": s }` | | [字符串](data-types/string-type) | `{ "name" : "string", "collation": "" }` | | [VARCHAR（n）](data-types/string-type) | `{ "name" : "varchar", "length": n }` | | [CHAR（n）](data-types/string-type) | `{ "name" : "char", "length": n }` | | [二元](data-types/binary-type) | `{ "name" : "binary" }` | | [布尔](data-types/boolean-type) | `{ "name" : "boolean" }` | | DATE | `{ "name" : "date" }` | | [时间戳](data-types/timestamp-type) | `{ "name" : "timestamp_ltz" }` | | [TIMESTAMP_NTZ](data-types/timestamp-ntz-type) | `{ "name" : "timestamp_ntz" }` | | [从起始单位到结束单位的区间](data-types/interval-type) | `{ "name" : "interval", "start_unit": "", "end_unit": "" }` | | [数组<元素类型>](data-types/array-type) | `{ "name" : "array", "element_type": , "element_nullable": }` | | [MAP<key_type、value_type>](data-types/map-type) | `{ "name" : "map", "key_type": , "value_type": , "element_nullable": }` | | [结构<field_name...， ...>](data-types/struct-type) | `{ "name" : "struct", "fields": [ {"name" : "", "type" : , "nullable": , "comment": "", "default": ""}] }` | | [变体](data-types/variant-type) | `{ "name" : "variant" }` | ## 示例 ```sql -- Creates a table `customer`. Assumes current schema is `salesdb`. > CREATE TABLE customer( cust_id INT, state VARCHAR(20), name STRING COMMENT 'Short name' ) USING parquet PARTITIONED BY (state); > INSERT INTO customer PARTITION (state = 'AR') VALUES (100, 'Mike'); -- Returns basic metadata information for unqualified table `customer` > DESCRIBE TABLE customer; col_name data_type comment ----------------------- --------- ---------- cust_id int null name string Short name state string null # Partition Information # col_name data_type comment state string null -- Returns basic metadata information for qualified table `customer` > DESCRIBE TABLE salesdb.customer; col_name data_type comment ----------------------- --------- ---------- cust_id int null name string Short name state string null # Partition Information # col_name data_type comment state string null -- Returns additional metadata such as parent schema, owner, access time etc. > DESCRIBE TABLE EXTENDED customer; col_name data_type comment ---------------------------- ------------------------------ ---------- cust_id int null name string Short name state string null # Partition Information # col_name data_type comment state string null # Detailed Table Information Database default Table customer Owner Created Time Tue Apr 07 22:56:34 JST 2020 Last Access UNKNOWN Created By Type MANAGED Provider parquet Location file:/tmp/salesdb.db/custom... Serde Library org.apache.hadoop.hive.ql.i... InputFormat org.apache.hadoop.hive.ql.i... OutputFormat org.apache.hadoop.hive.ql.i... Partition Provider Catalog -- Returns partition metadata such as partitioning column name, column type and comment. > DESCRIBE TABLE EXTENDED customer PARTITION (state = 'AR'); col_name data_type comment ------------------------------ ------------------------------ ---------- cust_id int null name string Short name state string null # Partition Information # col_name data_type comment state string null # Detailed Partition Inform... Database default Table customer Partition Values [state=AR] Location file:/tmp/salesdb.db/custom... Serde Library org.apache.hadoop.hive.ql.i... InputFormat org.apache.hadoop.hive.ql.i... OutputFormat org.apache.hadoop.hive.ql.i... Storage Properties [serialization.format=1, pa... Partition Parameters {transient_lastDdlTime=1586... Created Time Tue Apr 07 23:05:43 JST 2020 Last Access UNKNOWN Partition Statistics 659 bytes # Storage Information Location file:/tmp/salesdb.db/custom... Serde Library org.apache.hadoop.hive.ql.i... InputFormat org.apache.hadoop.hive.ql.i... OutputFormat org.apache.hadoop.hive.ql.i... ------------------------------ ------------------------------ ---------- -- Returns the metadata for `name` column. -- Optional `TABLE` clause is omitted and column is fully qualified. > DESCRIBE customer salesdb.customer.name; info_name info_value --------- ---------- col_name name data_type string comment Short name - Returns the table metadata in JSON format. > DESCRIBE EXTENDED customer AS JSON; { "table_name":"customer", "catalog_name":"spark_catalog", "schema_name":"default", "namespace":["default"], "columns":[ {"name":"cust_id","type":{"name":"integer"},"nullable":true}, {"name":"name","type":{"name":"string"},"comment":"Short name","nullable":true}, {"name":"state","type":{"name":"varchar","length":20},"nullable":true}], "location": "file:/tmp/salesdb.db/custom...", "created_time":"2020-04-07T14:05:43Z", "last_access":"UNKNOWN", "created_by":"None", "type":"MANAGED", "provider":"parquet", "partition_provider":"Catalog", "partition_columns":["state"]} -- The JSON describe of a metric view > DESCRIBE EXTENDED region_sales_metrics AS JSON; { "table_name":"region_sales_metrics", "catalog_name":"main", "namespace":["default"], "schema_name":"default", "columns":[ {"name":"month","type":{"name":"timestamp_ltz"},"nullable":true}, {"name":"status","type":{"name":"string","collation":"UTF8_BINARY"},"nullable":true}, {"name":"prder_priority","type":{"name":"string","collation":"UTF8_BINARY"},"nullable":true}, {"name":"count_orders","type":{"name":"bigint"},"nullable":false,"is_measure":true}, {"name":"total_revenue","type":{"name":"decimal","precision":28,"scale":2},"nullable":true,"is_measure":true}, {"name":"total_revenue_per_customer","type":{"name":"decimal","precision":38,"scale":12},"nullable":true,"is_measure":true}], "owner":"alf@melmak.et", "created_time":"2025-05-18T23:45:25Z", "last_access":"UNKNOWN", "created_by":"Spark ", "type":"METRIC_VIEW", "comment":"A metric view for regional sales metrics.", "view_text":"\n version: 0.1\n source: samples.tpch.orders\n filter: o_orderdate > '1990-01-01'\n dimensions:\n - name: month\n expr: date_trunc('MONTH', o_orderdate)\n - name: status\n expr: case\n when o_orderstatus = 'O' then 'Open'\n when o_orderstatus = 'P' then 'Processing'\n when o_orderstatus = 'F' then 'Fulfilled'\n end\n - name: prder_priority\n expr: split(o_orderpriority, '-')[1]\n measures:\n - name: count_orders\n expr: count(1)\n - name: total_revenue\n expr: SUM(o_totalprice)\n - name: total_revenue_per_customer\n expr: SUM(o_totalprice) / count(distinct o_custkey)\n ","language":"YAML","table_properties":{"metric_view.from.name":"samples.tpch.orders","metric_view.from.type":"ASSET","metric_view.where":"o_orderdate > '1990-01-01'"}, "view_creation_spark_configuration":{ ... }, "collation":"UTF8_BINARY"} ``` ## 描述细节 ``` DESCRIBE DETAIL [schema_name.]table_name ``` 返回架构、分区、表大小等方面的信息。 例如，对于 Delta 表，可以看到表的当前 [读取器和编写器版本](../../delta/feature-compatibility) 。 请参阅 [使用 describe detail 来查看表格详细信息](../../delta/table-details) 以获取详细的架构信息。