Function read_csv

datafusion-examples/examples/dataframe/dataframe.rs:110–156

Use the DataFrame API to 1. Read CSV files 2. Optionally specify schema

(ctx: &SessionContext)

/// 1. Read CSV files
/// 2. Optionally specify schema
async fn read_csv(ctx: &SessionContext) -> Result<()> {
    // create example.csv file in a temporary directory
    let dir = tempdir()?;
    let file_path = dir.path().join("example.csv");
    {
        let mut file = File::create(&file_path)?;
        // write CSV data
        file.write_all(
            r#"id,time,vote,unixtime,rating
    a1,"10 6, 2013",3,1381017600,5.0
    a2,"08 9, 2013",2,1376006400,4.5"#
                .as_bytes(),
        )?;
    } // scope closes the file
    let file_path = file_path.to_str().unwrap();

    // You can read a CSV file and DataFusion will infer the schema automatically
    let csv_df = ctx.read_csv(file_path, CsvReadOptions::default()).await?;
    csv_df.show().await?;

    // If you know the types of your data you can specify them explicitly
    let schema = Schema::new(vec![
        Field::new("id", DataType::Utf8, false),
        Field::new("time", DataType::Utf8, false),
        Field::new("vote", DataType::Int32, true),
        Field::new("unixtime", DataType::Int64, false),
        Field::new("rating", DataType::Float32, true),
    ]);
    // Create a csv option provider with the desired schema
    let csv_read_option = CsvReadOptions {
        // Update the option provider with the defined schema
        schema: Some(&schema),
        ..Default::default()
    };
    let csv_df = ctx.read_csv(file_path, csv_read_option).await?;
    csv_df.show().await?;

    // You can also create DataFrames from the result of sql queries
    // and using the `enable_url_table` refer to local files directly
    let dyn_ctx = ctx.clone().enable_url_table();
    let csv_df = dyn_ctx
        .sql(&format!("SELECT rating, unixtime FROM '{file_path}'"))
        .await?;
    csv_df.show().await?;

    Ok(())
}

/// Use the DataFrame API to:
/// 1. Read in-memory data.