(Hide table-of-contents)


netCSV is a Dyalog namespace that contains useful methods for reading large comma-separated values (CSV) files representing tabular data directly into a DataTable. The DataTable can be used later for binding or querying. The large CSV file will not transit in the APL workspace and may not cause a WORKSPACE FULL ERROR. The netCSV namespace is using the assembly LumenWorks.Framework.IO.dll (version 3.9.0) that is the work of Sébastien Lorion and Paul Hatcher and must be installed in the same directory has dyalog.exe (as shown by +2 ⎕NQ '.' 'GetEnvironment' 'Dyalog').

A CSV file representing tabular data can be compared to a matrix in APL. Each row of the matrix is a 'record' of the CSV file and each column of the matrix is a 'field' of the CSV file. Each record is separated with ⎕UCS 13 10 (carriage return, linefeed) and each field is separated by a 'field separator' (typically a comma or semicolon).

The CSV format is still popular because it is simple and it can be read by Microsoft Excel. The problem is that not all countries are using the same decimal separator and field separator and those need to be known when parsing a file that was not written on the same computer. To obtain the default decimal separator and field separator on your computer that will be used by Excel you can do:


The first record of the CSV file may or may not contain a 'Header' record that is the name of each columns (fields). Since this namespace is dealing with DataTable (where all columns needs to have a name) it is assume that the first record will always contains the name of the fields.

For a small size file or a variables in the workspace you can use the functions explained in CsvToApl. For more Dyalog methods related to a DataTable you can explore netDataTable.


SimpleCsvFileToDT will parse a CSV file to a DataTable where all the columns will be of type String. There is no conversion to a date or to a number that will be made. The only thing that is required to know is the field separator (in the following example it is ';'). For example if you have a file named Client.csv like this:

100;Michael;Son;Ray;Actuary;Fort Worth
101;Emmanuel;Nick;Williams;Earth Moving Equipment Mechanic;Ontario
104;Kerri;Jewell;Bennett;Production Manager;Jacksonville
106;Heath;Herbert;Glass;Aeronautical Engineer;Seattle
107;Randall;Hans;Tanner;Air Traffic Controller;Corona
108;Damien;;Washington;Chartered Secretary;Allentown
109;Charlene;Alice;Thompson;Committee Clerk;Salem
111;Richard;Gerardo;Riddle;Assayer Sampler;Akron

You can do the following to obtain a DataTable:

      dt ← ';' SimpleCsvFileToDT 'd:\Clients.csv'  ⍝ ';' is the field separator

 System.String  System.String  System.String  System.String  System.String  System.String

      ShowDT dt

You can use Converters on the .Net side to format differently the character representation of a date or number.


CsvFileToDT will parse a CSV file to a DataTable while preserving the data type of each column. The method will analyse the first record to determine the type (either DateTime, Double or String) of each columns. For the file msft.csv that looks like this:


You can do the following to obtain a DataTable where the type of each columns will be preserved:

      dt ← ',.' CsvFileToDT 'd:\msft.csv'  ⍝ ',' is the field separator and '.' is the decimal separator

 System.DateTime  System.Double  System.Double  System.Double  System.Double  System.Double

      ShowDT dt

Sometimes the automatic detection will not work and you will want to explicitly specify the type of each columns. This can be done by inserting the type (either Boolean, Byte, Decimal, Double, Single, Int64, Int16, String, DateTime) within curly braces in the header record of the CSV file. The following file ZipCode.csv is using that feature:

ZIP Code{String},ZIP Code Type{String},City{String},City Type{String},State{String},State Code{String},Area Code{Int16},Latitude{Double},Longitude{Double}
00501,U,Holtsville,D,New York,NY,631,40.8147,-73.0451
00544,U,Holtsville,D,New York,NY,631,40.8128,-73.0480
00601,S,Adjuntas,D,Puerto Rico,PR,787,18.1650,-66.7240
00601,S,Jard de Adjuntas,N,Puerto Rico,PR,787,18.1650,-66.7240
00601,S,Urb San Joaquin,N,Puerto Rico,PR,787,18.1650,-66.7240
00602,S,Aguada,D,Puerto Rico,PR,787,18.3825,-67.1871
00602,S,Alts de Aguada,N,Puerto Rico,PR,787,18.3825,-67.1871
00602,S,Comunidad Las Flores,N,Puerto Rico,PR,787,18.3825,-67.1871
00602,S,Ext Los Robles,N,Puerto Rico,PR,787,18.3825,-67.1871

And you do the following to obtain the DataTable with the specified types:

      dt ← ',.' CsvFileToDT 'd:\ZipCode.csv'  ⍝ ',' is the field separator and '.' is the decimal separator

 System.String  System.String  System.String  System.String  System.String  System.String  System.Int16  System.Double  System.Double

      ShowDT dt

This curly brace feature is not standard to the CSV format and will not be supported by Microsoft Excel.


The method DTtoCsvFile will save a DataTable to a CSV file. It is using the assembly Syncfusion.XlsIO.Base.dll from Syncfusion. The field delimiter can be specified in the method but the assembly will use the decimal separator of the local culture. It will return the number of rows written to the CSV file.

    noRows ← dt DTtoCsvFile fileName


If you want to construct your own CSV file the methods CsvFileNew, CsvFileAddRecords and AplToCsvRecords can assist you in doing so. The syntax of CsvFileNew is as follow:

    r ← fileName CsvFileNew header

    ⍝ header   = Columns Names separated by a delimiter like ',' or ';'
    ⍝ fileName = Fully qualified file name with extension.
    ⍝ r        = 1 if successfull, 0 (ERROR) if failure

    'd:\Car.csv' CsvFileNew 'Year,Make,Model,Description,Price'


One of the advantage of the CSV file format is that you can add easily an additional record to the file. The method CsvFileAddRecords will add some record(s) to a CSV file. The syntax is as follow:

    r ← fileName CsvFileAddRecords records

    ⍝ records =  Line(s) of text with a field delimiter like ',' or ';'
    ⍝ fileName = Fully qualified file name with extension.
    ⍝ r        = 1 if successfull, 0 (ERROR) if failure

   'd:\Car.csv' CsvFileAddRecords '1996,Jeep,Grand Cherokee,MUST SELL!,4799'


The method AplToCsvRecords will convert an Apl array to individual CSV records (one record per row of the array). The syntax is as follow:

    records ← {FieldDecSep} AplToCsvRecords aplArray
    ⍝ aplArray    = Apl array
    ⍝ FieldDecSep = Field and Decimal Separator (like ',.' or ';,' )
    ⍝ records     = one record per row of aplArray formatted according to the CSV format

   'd:\Car.csv' CsvFileAddRecords AplToCsvRecords 1997 'Ford' 'E350' 'ac, abs, moon' 3000

And the file d:\Car.csv will look like this:

1996,Jeep,Grand Cherokee,MUST SELL!,4799
1997,Ford,E350,"ac, abs, moon",3000

How to install netCSV in your workspace

  1. Download netCSV.v1.1.txt

  2. Do a Select all (Ctrl+A) and a copy (Ctrl+C).
  3. In your workspace execute )ed ⍟ netCSV

  4. Paste (Ctrl+V) the text into the Dyalog editor
  5. Press Escape and ')save' your workspace

Optionally to de-script the namespace you can do:

'netCSV' ⎕SE.SALTUtils.CopyNs netCSV

Download LumenWorks.Framework.IO.dll and install the assembly in the same directory as dyalog.exe

Version Information

 February 2016   - Initial version (1.0)

 April 2016 - version (1.1)
              SimpleCsvFileToDT and CsvFileToDT modified so it will not block the file
              on disk while doing the conversion to a DataTable

Original author:

Pierre Gilbert




<apgil AT SPAMFREE videotron DOT ca>

CategoryDyalog - CategoryDyalogDotNet - CategoryDyalogDotNetUtilities - CategoryDotNet

netCSV (last edited 2016-04-29 13:47:03 by PierreGilbert)