快速参考：适用于 SQL 命令的 Snowpark Java APIs

本主题供您快速参考与 SQL 命令对应的一些 Snowpark APIs

（请注意，此处未列出与 SQL 命令对应的完整 APIs 列表。）

执行查询

选择列

To select specific columns, use select.

Example of a SQL StatementExample of Snowpark Code
SELECT id, name FROM sample_product_data;
DataFrame df = session.table("sample_product_data");

DataFrame dfSelectedCols = df.select(Functions.col("id"), Functions.col("name"));

dfSelectedCols.show();

重命名列

To rename a column, use as or alias.

Example of a SQL StatementExample of Snowpark Code
SELECT id AS item_id FROM sample_product_data;
DataFrame df = session.table("sample_product_data");

DataFrame dfRenamedCol = df.select(Functions.col("id").as("item_id"));

dfRenamedCol.show();
DataFrame df = session.table("sample_product_data");

DataFrame dfRenamedCol = df.select(Functions.col("id").alias("item_id"));

dfRenamedCol.show();

筛选数据

To filter data, use filter or where.

Example of a SQL StatementExample of Snowpark Code
SELECT * FROM sample_product_data WHERE id = 1;
DataFrame df = session.table("sample_product_data");

DataFrame dfFilteredRows = df.filter(Functions.col("id").equal_to(Functions.lit(1)));

dfFilteredRows.show();
DataFrame df = session.table("sample_product_data");

DataFrame dfFilteredRows = df.where(Functions.col("id").equal_to(Functions.lit(1)));

dfFilteredRows.show();

对数据进行排序

To sort data, use sort.

Example of a SQL StatementExample of Snowpark Code
SELECT * FROM sample_product_data ORDER BY category_id;
DataFrame df = session.table("sample_product_data");

DataFrame dfSorted = df.sort(Functions.col("category_id"));

dfSorted.show();

限制返回的行数

To limit the number of rows returned, use limit. See Limiting the Number of Rows in a DataFrame.

Example of a SQL StatementExample of Snowpark Code
SELECT * FROM sample_product_data
  ORDER BY category_id LIMIT 2;
DataFrame df = session.table("sample_product_data");

DataFrame dfSorted = df.sort(Functions.col("category_id")).limit(2);

Row[] arrayRows = dfSorted.collect();

执行联接

To perform a join, use join or naturalJoin. See Joining DataFrames.

Example of a SQL StatementExample of Snowpark Code
SELECT * FROM sample_a
  INNER JOIN sample_b
  on sample_a.id_a = sample_b.id_a;
DataFrame dfLhs = session.table("sample_a");

DataFrame dfRhs = session.table("sample_b");

DataFrame dfJoined =
  dfLhs.join(dfRhs, dfLhs.col("id_a").equal_to(dfRhs.col("id_a")));

dfJoined.show();
SELECT * FROM sample_a NATURAL JOIN sample_b;
DataFrame dfLhs = session.table("sample_a");

DataFrame dfRhs = session.table("sample_b");

DataFrame dfJoined = dfLhs.naturalJoin(dfRhs);

dfJoined.show();

查询半结构化数据

To traverse semi-structured data, use subField(“<field_name>”) and subField(<index>). See Working with Semi-Structured Data.

Example of a SQL StatementExample of Snowpark Code
SELECT src:salesperson.name FROM car_sales;
DataFrame df = session.table("car_sales");

DataFrame dfJsonField =
  df.select(Functions.col("src").subField("salesperson").subField("name"));

dfJsonField.show();

对数据进行分组和聚合

To group data, use groupBy. This returns a RelationalGroupedDataFrame object, which you can use to perform the aggregations.

Example of a SQL StatementExample of Snowpark Code
SELECT category_id, count(*)
  FROM sample_product_data GROUP BY category_id;
DataFrame df = session.table("sample_product_data");

DataFrame dfCountPerCategory = df.groupBy(Functions.col("category_id")).count();

dfCountPerCategory.show();

调用窗口函数

To call a window function, use the Window object methods to build a WindowSpec object, which in turn you can use for windowing functions (similar to using ‘<function> OVER … PARTITION BY … ORDER BY’).

Example of a SQL StatementExample of Snowpark Code
SELECT category_id, id, SUM(amount) OVER
  (PARTITION BY category_id ORDER BY product_date)
  FROM sample_product_data ORDER BY product_date;
WindowSpec window = Window.partitionBy(
  Functions.col("category_id")).orderBy(Functions.col("product_date"));

DataFrame df = session.table("sample_product_data");

DataFrame dfCumulativePrices = df.select(
  Functions.col("category_id"), Functions.col("product_date"),
  Functions.sum(Functions.col("amount")).over(window)).sort(Functions.col("product_date"));

dfCumulativePrices.show();

更新、删除和合并行

To update, delete, and merge rows in a table, use Updatable. See Updating, Deleting, and Merging Rows in a Table.

Example of a SQL StatementExample of Snowpark Code
UPDATE sample_product_data
  SET serial_number = 'xyz' WHERE id = 12;
import java.util.HashMap;
import java.util.Map;
...

Map<Column, Column> assignments = new HashMap<>();

assignments.put(Functions.col("serial_number"), Functions.lit("xyz"));

Updatable updatableDf = session.table("sample_product_data");

UpdateResult updateResult =
  updatableDf.update(
    assignments,
    Functions.col("id").equal_to(Functions.lit(12)));

System.out.println("Number of rows updated: " + updateResult.getRowsUpdated());
DELETE FROM sample_product_data
  WHERE category_id = 50;
Updatable updatableDf = session.table("sample_product_data");

DeleteResult deleteResult =
  updatableDf.delete(updatableDf.col("category_id").equal_to(Functions.lit(50)));

System.out.println("Number of rows deleted: " + deleteResult.getRowsDeleted());
MERGE  INTO target_table USING source_table
  ON target_table.id = source_table.id
  WHEN MATCHED THEN
    UPDATE SET target_table.description =
      source_table.description;
import java.util.HashMap;
import java.util.Map;

Map<String, Column> assignments = new HashMap<>();
assignments.put("description", source.col("description"));
MergeResult mergeResult =
  target.merge(source, target.col("id").equal_to(source.col("id")))
  .whenMatched.updateColumn(assignments)
  .collect();

使用暂存区

For more information on working with stages, see Working With Files in a Stage.

从暂存区上传和下载文件

To upload and download files from a stage, use FileOperation. See Uploading and Downloading Files in a Stage.

Example of a SQL StatementExample of Snowpark Code
PUT file:///tmp/*.csv @myStage OVERWRITE = TRUE;
import java.util.HashMap;
import java.util.Map;
...
Map<String, String> putOptions = new HashMap<>();

putOptions.put("OVERWRITE", "TRUE");

PutResult[] putResults = session.file().put(
  "file:///tmp/*.csv", "@myStage", putOptions);

for (PutResult result : putResults) {
  System.out.println(result.getSourceFileName() + ": " + result.getStatus());
}
GET @myStage file:///tmp PATTERN = '.*.csv.gz';
import java.util.HashMap;
import java.util.Map;
...
Map<String, String> getOptions = new HashMap<>();

getOptions.put("PATTERN", "'.*.csv.gz'");

GetResult[] getResults = session.file().get(
"@myStage", "file:///tmp", getOptions);

for (GetResult result : getResults) {
  System.out.println(result.getFileName() + ": " + result.getStatus());
}

从暂存区的文件中读取数据

To read data from files in a stage, use DataFrameReader to create a DataFrame for the data. See Setting Up a DataFrame for Files in a Stage.

Example of a SQL StatementExample of Snowpark Code
CREATE FILE FORMAT snowpark_temp_format TYPE = JSON;

SELECT "$1"[0]['salesperson']['name'] FROM (
  SELECT $1::VARIANT AS "$1" FROM @mystage/car_sales.json(
    FILE_FORMAT => 'snowpark_temp_format')) LIMIT 10;

DROP FILE FORMAT snowpark_temp_format;
DataFrame df = session.read().json(
  "@mystage/car_sales.json").select(
    Functions.col("$1").subField(0).subField("salesperson").subField("name"));

df.show();

将暂存区文件中的数据复制到表中

To copy data from files in a stage to a table, use DataFrameReader to create a CopyableDataFrame for the data, and use the copyInto method to copy the data to the table. See Copying Data from Files into a Table.

Example of a SQL StatementExample of Snowpark Code
COPY INTO new_car_sales
  FROM @mystage/car_sales.json
  FILE_FORMAT = (TYPE = JSON);
CopyableDataFrame dfCopyableDf = session.read().json("@mystage/car_sales.json");
dfCopyableDf.copyInto("new_car_sales");

将 DataFrame 保存到暂存区上的文件

To save a DataFrame to files on a stage, use the DataFrameWriter method named after the format of the files that you want to use. See Saving a DataFrame to Files on a Stage.

Example of a SQL StatementExample of Snowpark Code
COPY INTO @mystage/saved_data.json
  FROM (  SELECT  *  FROM (car_sales) )
  FILE_FORMAT = ( TYPE = JSON COMPRESSION = 'none' )
  OVERWRITE = TRUE
  DETAILED_OUTPUT = TRUE
DataFrame df = session.table("car_sales");

WriteFileResult writeFileResult = df.write().mode(
  SaveMode.Overwrite).option(
  "DETAILED_OUTPUT", "TRUE").option(
  "compression", "none").json(
  "@mystage/saved_data.json");

创建和调用用户定义的函数 (UDFs)

To create an anonymous UDF, use Functions.udf.

To create a temporary or permanent UDF that you can call by name, use UDFRegistration.registerTemporary or UDFRegistration.registerPermanent.

To call a permanent UDF by name, use Functions.callUDF.

For details, see Creating User-Defined Functions (UDFs) for DataFrames in Java and Calling scalar user-defined functions (UDFs).

Example of a SQL StatementExample of Snowpark Code
CREATE FUNCTION <temp_function_name>
  RETURNS INT
  LANGUAGE JAVA
  ...
  AS
  ...;

SELECT ...,
  <temp_function_name>(quantity) AS doublenum
  FROM sample_product_data;
UserDefinedFunction doubleUdf =
  Functions.udf(
    (Integer x) -> x + x,
    DataTypes.IntegerType,
    DataTypes.IntegerType);

DataFrame df = session.table("sample_product_data");

DataFrame dfWithDoubleNum =
  df.withColumn("doubleNum",
    doubleUdf.apply(Functions.col("quantity")));

dfWithDoubleNum.show();
CREATE FUNCTION <temp_function_name>
  RETURNS INT
  LANGUAGE JAVA
  ...
  AS
  ...;

SELECT ...,
  <temp_function_name>(quantity) AS doublenum
  FROM sample_product_data;
UserDefinedFunction doubleUdf =
  session
    .udf()
    .registerTemporary(
      "doubleUdf",
      (Integer x) -> x + x,
      DataTypes.IntegerType,
      DataTypes.IntegerType);

DataFrame df = session.table("sample_product_data");

DataFrame dfWithDoubleNum =
  df.withColumn("doubleNum",
    Functions.callUDF("doubleUdf", Functions.col("quantity")));
dfWithDoubleNum.show();
CREATE FUNCTION doubleUdf(arg1 INT)
  RETURNS INT
  LANGUAGE JAVA
  ...
  AS
  ...;

SELECT ...,
  doubleUdf(quantity) AS doublenum
  FROM sample_product_data;
UserDefinedFunction doubleUdf =
  session
    .udf()
    .registerPermanent(
      "doubleUdf",
      (Integer x) -> x + x,
      DataTypes.IntegerType,
      DataTypes.IntegerType,
      "mystage");

DataFrame df = session.table("sample_product_data");

DataFrame dfWithDoubleNum =
  df.withColumn("doubleNum",
    Functions.callUDF("doubleUdf", Functions.col("quantity")));
dfWithDoubleNum.show();

创建和调用存储过程

For a guide on creating stored procedures with Snowpark, see Creating stored procedures for DataFrames in Java.

Example of a SQL StatementExample of Snowpark Code
CREATE PROCEDURE <temp_procedure_name>(x INTEGER, y INTEGER)
  RETURNS INTEGER
  LANGUAGE JAVA
  ...
  AS
  $$
  BEGIN
    RETURN x + y;
  END
  $$
  ;

CALL <temp_procedure_name>(2, 3);
StoredProcedure sp =
  session.sproc().registerTemporary((Session session, Integer x, Integer y) -> x + y,
    new DataType[] {DataTypes.IntegerType, DataTypes.IntegerType},
    DataTypes.IntegerType);

  session.storedProcedure(sp, 2, 3).show();
CREATE PROCEDURE sproc(x INTEGER, y INTEGER)
  RETURNS INTEGER
  LANGUAGE JAVA
  ...
  AS
  $$
  BEGIN
   RETURN x + y;
  END
  $$
  ;

CALL sproc(2, 3);
String name = "sproc";

StoredProcedure sp =
  session.sproc().registerTemporary(name,
    (Session session, Integer x, Integer y) -> x + y,
    new DataType[] {DataTypes.IntegerType, DataTypes.IntegerType},
    DataTypes.IntegerType);

  session.storedProcedure(name, 2, 3).show();
CREATE PROCEDURE add_hundred(x INTEGER)
  RETURNS INTEGER
  LANGUAGE JAVA
  ...
  AS
  $$
  BEGIN
   RETURN x + 100;
  END
  $$
  ;

CALL add_hundred(3);
String name = "add_hundred";
String stageName = "sproc_libs";

StoredProcedure sp =
  session.sproc().registerPermanent(
    name,
    (Session session, Integer x) -> x + 100,
    DataTypes.IntegerType,
    DataTypes.IntegerType,
    stageName,
    true);

  session.storedProcedure(name, 3).show();