Import CSV or other delimited files

This guide will show you how to import data from CSV (and other delimited) files into Deephaven tables with the read_csv method.

read_csv

The two most common ways to load a CSV file are using the file path or the file URL. The header of the CSV file determines the column names.

The basic syntax for read_csv follows:

from deephaven import read_csv

read_csv(path: str) -> Table

Let's dive into some simple examples. So that we don't have to create a new CSV file from scratch, we'll use some CSV files from Deephaven's examples repository. We encourage you to use your own files by replacing the file paths in our queries.

read_csv with a file path

The read_csv method can be used to import a CSV file from a file path. In this example, we will import a CSV file containing R.A. Fisher's classic iris flower dataset commonly used in machine learning applications.

from deephaven import read_csv

iris = read_csv("/data/examples/Iris/csv/iris.csv")

The /data mount point

If you are using Docker-installed Deephaven, you can find a /data folder inside your Deephaven installation's main folder, on the same level as your docker-compose.yml file. This folder is mounted to the /data volume in the running Deephaven container. This means that if the Deephaven console is used to write data to /data/abc/file.csv, that file will be visible at ./data/abc/file.csv on the local file system of your computer.

See Docker data volumes to learn more about the relation between locations in the container and the local file system.

read_csv with a URL

Now, we will import the same CSV file, but this time we will use a URL instead of a file path.

from deephaven import read_csv

iris = read_csv(
    "https://media.githubusercontent.com/media/deephaven/examples/main/Iris/csv/iris.csv"
)

Headerless CSV files

CSV files don't always have headers. The example below uses the headerless DeNiro CSV and includes an additional headless argument.

from deephaven import read_csv

deniro = read_csv(
    "https://media.githubusercontent.com/media/deephaven/examples/main/DeNiro/csv/deniro_headerless.csv",
    headless=True,
)

Because no column names are provided, the table will produce default column names (Column1, Column2, etc.). You can explicitly set the column names, as shown below.

from deephaven import read_csv
import deephaven.dtypes as dht

header = {"Year": dht.int64, "Score": dht.int64, "Title": dht.string}
deniro = read_csv(
    "https://media.githubusercontent.com/media/deephaven/examples/main/DeNiro/csv/deniro_headerless.csv",
    header=header,
    headless=True,
)

Other formats

Tab-delimited data

Deephaven allows you to specify other delimiters as a second argument if your file is not comma-delimited. In the example below, we import a tab-delimited file, which requires a second argument.

from deephaven import read_csv

deniro_tsv = read_csv(
    "https://raw.githubusercontent.com/deephaven/examples/main/DeNiro/csv/deniro.tsv",
    delimiter="\t",
)

Pipe-delimited data

Any character can be used as a delimiter. The pipe character (|) is common. In the example below, we supply the delimiter | as the second argument.

from deephaven import read_csv

deniro_psv = read_csv(
    "https://raw.githubusercontent.com/deephaven/examples/main/DeNiro/csv/deniro.psv",
    delimiter="|",
)

Trim

By default, quoted values that have leading and trailing white space include the white space when reading the CSV file. For example, if " Taxi Driver " is in the CSV file, it will be read as Taxi Driver.

By setting trim to true when reading the CSV file, these leading and trailing white space will be removed. So " Taxi Driver " will be read as Taxi Driver.

Using optional arguments

In the following example, we want to read in the deniro_poorly_formatted.csv file from the Deephaven Examples repo. This file has a number of issues that we need to address, but read_csv can handle them if we give the right input parameters:

from deephaven import read_csv, input_table
import deephaven.dtypes as dht

my_dict = {"ReleaseYear": dht.int16, "Rating": dht.int16, "Movie Title": dht.string}

deniro = read_csv(
    path="/data/examples/DeNiro/csv/deniro_poorly_formatted.csv",
    header=my_dict,
    header_row=1,
    skip_rows=5,
    num_rows=20,
    ignore_empty_lines=True,
    allow_missing_columns=True,
    ignore_excess_columns=True,
    trim=True,
)

In the example above, we:

• Use header to override and replace the files's header row, specified column names, and data types.
• Set header_row to 1 to indicate that the row in position 1 of the file is the header row.
• Use skip_rows to omit the first 5 rows of the CSV file.
• Set num_rows to 20 to limit our table to 20 rows.
• Set ignore_empty_lines to True avoid throwing an exception due to an empty line in the CSV file.
• Set allow_missing_columns to True to avoid throwing an exception due to a missing column in the CSV file. Note that the 1983 entry in our table has null in the missing column.
• Set ignore_excess_columns to True to avoid throwing an exception due to an extra column in the CSV file - 1984's Brazil entry has an extra column containing a brief review that we don't need in our table.
• Set trim to True to remove leading and trailing white space from inside quoted strings - such as our movie titles.

See the read_csv reference documentation for a complete list of optional arguments.

Related documentation

• How to export CSV files
• read_csv