CSV Module

This XQuery Module contains a single function to parse CSV input. CSV (comma-separated values) is a popular representation for tabular data, exported e. g. from Excel.

Conventions

All functions and errors in this module are assigned to the http://basex.org/modules/csv namespace, which is statically bound to the csv prefix.

Conversion

XML: Direct, Attributes

If the direct or attributes format is chosen, a CSV string is converted to XML:

• The resulting XML document has a csv root element.
• Rows are represented via record elements.
• Fields are represented via entry elements. The value of a field is represented as text node.
• If the header option is set to true, the first text line is parsed as table header:
  • If format is set to direct, the field names are encoded, as described in the Conversion Module, and used as element names.
  • Otherwise, if format is attributes, the field names will be stored in name attributes.

A little advice: in the Database Creation dialog of the GUI, if you select CSV Parsing and switch to the Parsing tab, you can see the effects of some of the conversion options.

XQuery

With the xquery format, CSV records are converted to a sequence of arrays:

• The resulting value will be a map with a records and an optional names key.
• Records are organized as a sequence of arrays. A single array contains the entries of a single record.
• The column names will be available if header option is set to true.

The CSV map can e.g. be accessed as follows:

• $csv?records[5] returns all entries of the 5th record (row)
• $csv?records?(2) returns all entries of the 2nd field (column)
• $csv?names?* returns the names of all fields (if available)
• Return enumerated strings for all records:

for $record at $pos in $csv?records
return $pos || ". " || string-join($record?*, ', ')

The resulting representation consumes less memory than XML-based formats, and values can be directly accessed without conversion. Thus, it is recommendable for very large inputs and for efficient ad-hoc processing.

Options

In the following table, all available options are listed. The Excel column lists recommended options for data that is processed with Excel or Open/Libre Office.

Option	Description	Allowed	Default	Direction	Excel
separator	Defines the character which separates the values of a single record.	comma, semicolon, colon, tab, space or a single character	comma	parse, serialize	semicolon or comma, depending on the region
header	Indicates if the first line of the parsed or serialized CSV data is a table header.	yes, no	no	parse, serialize
format	Specifies the format of the XML data: With direct conversion, field names are represented as element names With attributes conversion, field names are stored in name attributes With xquery conversion, the input is converted to an XQuery map	direct, attributes, xquery	direct	parse, serialize
lax	Specifies if a lax approach is used to convert QNames to JSON names.	yes, no	yes	parse, serialize	no
quotes	Specifies how quotes are parsed: Parsing: If the option is enabled, quotes at the start and end of a value will be treated as control characters. Separators and newlines within the quotes will be adopted without change. Serialization: If the option is enabled, the value will be wrapped with quotes if it contains characters that might be treated as control characters. A quote character in the value will be encoded according to the rules of the backslashes option.	yes, no	yes	parse, serialize	yes
backslashes	Specifies how quotes and other characters are escaped: Parsing: If the option is enabled, \r, n and \t will be replaced with the corresponding control characters. All other escaped characters will be adopted as literals (e.g.: \" → "). If the option is disabled, two consecutive quotes will be replaced with a single quote (unless quotes is enabled and the quote is the first or last character of a value). Serialization: If the option is enabled, \r, n, \t, " and the separator character will be encoded with a backslash. If the option is disabled, quotes will be duplicated.	yes, no	no	parse, serialize	no
allow	Introduced with Version 9.7: In Excel, a value will be evaluated if it starts with the character -, +, =, @, \t or \r. A regular expression can be specified to reject data that will be handled differently than expected by an application, or that may be malicious (see https://owasp.org/www-community/attacks/CSV_Injection for more details).	string		serialize	[^-+=\t\r].*|[-+]\d*([,.]\d+)?

Functions

csv:parse

Signatures	csv:parse($string as xs:string?) as item()? csv:parse($string as xs:string?, $options as map(*)?) as item()?
Summary	Converts the CSV $string to an XQuery value. The $options argument can be used to control the way the input is converted.
Errors	parse: the specified input cannot be parsed as CSV document.

csv:doc

Signatures	csv:doc($uri as xs:string?) as item()? csv:doc($uri as xs:string?, $options as map(*)?) as item()?
Summary	Fetches the CSV document referred to by the given $uri and converts it to an XQuery value. The $options argument can be used to control the way the input is converted.
Errors	parse: the specified input cannot be parsed as CSV document. options: the specified options are conflicting.

csv:serialize

Signatures	csv:serialize($input as item()?) as xs:string csv:serialize($input as item()?, $options as map(*)?) as xs:string
Summary	Serializes the specified $input as CSV, using the specified $options, and returns the result as string. Values can also be serialized as CSV with the standard Serialization feature of XQuery: The parameter method needs to be set to csv, and the options presented in this article need to be assigned to the csv parameter.
Errors	serialize: the input cannot be serialized.

Examples

Example 1: Converts CSV data to XML, interpreting the first row as table header:

Input addressbook.csv:

Name,First Name,Address,City
Huber,Sepp,Hauptstraße 13,93547 Hintertupfing

Query:

let $text := file:read-text('addressbook.csv')
return csv:parse($text, map { 'header': true() })

Result:

<csv>
  <record>
    <Name>Huber</Name>
    <First_Name>Sepp</First_Name>
    <Address>Hauptstraße 13</Address>
    <City>93547 Hintertupfing</City>
  </record>
</csv>

Example 2: Converts some CSV data to XML and back, and checks if the input and output are equal. The expected result is true:

Query:

let $options := map { 'lax': false() }
let $input := file:read-text('some-data.csv')
let $output := $input => csv:parse($options) => csv:serialize($options)
return $input eq $output

Example 3: Converts CSV data to XQuery and returns distinct column values:

Query:

let $text := ``[Name,City
Jack,Chicago
Jack,Washington
John,New York
]``
let $options := map { 'format': 'xquery', 'header': true() }
let $csv := csv:parse($text, $options)
return (
  'Distinct values:',
  let $records := $csv('records')
  for $name at $pos in $csv('names')?*
  let $values := $records?($pos)
  return (
    '* ' || $name || ': ' || string-join(distinct-values($values), ', ')
  )
)

Result:

Distinct values:
* Name: Jack, John
* City: Chicago, Washington, New York

Errors

Code	Description
parse	The input cannot be parsed.
serialize	The node cannot be serialized.

Changelog

Version 9.7

• Added: Options: allow option.

Version 9.4

• Added: csv:doc

Version 9.1

• Updated: csv:parse can be called with empty sequence.

Version 9.0

• Added: xquery option
• Removed: map option
• Updated: error codes updated; errors now use the module namespace

Version 8.6

• Updated: Options: improved Excel compatibility

Version 8.0

• Added: backslashes option

Version 7.8

• Updated: csv:parse now returns a document node instead of an element, or an XQuery map if format is set to map.
• Added: format and lax options

The module was introduced with Version 7.7.2.