class

CSV

Inherits Reference / Object

Provides methods and classes for parsing and generating CSV (comma-separated values) strings.

This module conforms to RFC 4180.

NOTE: To use CSV or its children, you must explicitly import it with require "csv"

Parsing

Several ways of parsing CSV are provided. The most straight-forward, but slow or inefficient for some scenarios, is CSV#parse, which returns an array of arrays of all data.

Rows can be traversed in a linear fashion with CSV#each_row, or using an Iterator.

To parse a CSV in an efficient way, optionally being able to access row values from header names, create an instance of a CSV.

Parsing with `CSV#new`

A CSV instance holds a cursor to the current row in the CSV. The cursor is advanced by invoking #next, which returns true if a next row was found, and false otherwise. A first call to #next is required to position the CSV parser in the first row.

Once positioned in a row, values can be obtained with the several #[] methods, which can accept a header name, column position, or header name pattern as a Regex.

Additionally, a Row object can be obtained with the #row method which provides similar methods and can be converted to an Array or Hash.

Example

require "csv"

csv = CSV.new("Name, Age\nJohn, 20\nPeter, 30", headers: true)
csv.next # => true

csv["Name"]  # => "John"
csv[0]       # => "John"
csv[-2]      # => "John"
csv[/name/i] # => "John"

csv["Age"] # => " 20"

csv.row.to_a # => ["John", " 20"]
csv.row.to_h # => {"Name" => "John", "Age" => " 20"}

csv.next    # => true
csv["Name"] # => "Peter"

csv.next # => false

Building

To create CSV data, check CSV#build and the CSV::Builder class.

Constants

DEFAULT_QUOTE_CHAR = '"'

DEFAULT_SEPARATOR = ','

Constructors

new(string_or_io : String | IO, headers : Bool = false, strip : Bool = false, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR)

Creates a new instance from the given String or IO.

If strip is true, row values are stripped with String#strip before being returned from methods.
If headers is true, row values can be accessed with header names or patterns. Headers are always stripped.

See CSV.parse about the separator and quote_char arguments.

new(string_or_io : String | IO, headers = false, strip = false, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR, &)

Creates a new instance from the given String or IO, and yields it to the given block once for each row in the CSV.

If strip is true, row values are stripped with String#strip before being returned from methods.
If headers is true, row values can be accessed with header names or patterns. Headers are always stripped.

See CSV.parse about the separator and quote_char arguments.

Class methods

build(separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR, quoting : Builder::Quoting = Builder::Quoting::RFC, &) : String

Builds a CSV. This yields a CSV::Builder to the given block.

Takes optional quoting argument to define quote behavior.

require "csv"

result = CSV.build do |csv|
  csv.row "one", "two"
  csv.row "three"
end
result # => "one,two\nthree\n"
result = CSV.build(quoting: CSV::Builder::Quoting::ALL) do |csv|
  csv.row "one", "two"
  csv.row "three"
end
result # => "\"one\",\"two\"\n\"three\"\n"

See: CSV::Builder::Quoting

build(io : IO, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR, quoting : Builder::Quoting = Builder::Quoting::RFC, &) : Nil

Appends CSV data to the given IO. This yields a CSV::Builder that writes to the given IO.

require "csv"

io = IO::Memory.new
io.puts "HEADER"
CSV.build(io) do |csv|
  csv.row "one", "two"
  csv.row "three"
end
io.to_s # => "HEADER\none,two\nthree\n"

each_row(string_or_io : String | IO, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR, &)

Yields each of a CSV's rows as an Array(String).

See CSV.parse about the separator and quote_char arguments.

require "csv"

CSV.each_row("one,two\nthree") do |row|
  puts row
end

Output:

["one", "two"]
["three"]

each_row(string_or_io : String | IO, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR) : Iterator(Array(String))

Returns an Iterator of Array(String) over a CSV's rows.

See CSV.parse about the separator and quote_char arguments.

require "csv"

rows = CSV.each_row("one,two\nthree")
rows.next # => ["one", "two"]
rows.next # => ["three"]

parse(string_or_io : String | IO, separator : Char = DEFAULT_SEPARATOR, quote_char : Char = DEFAULT_QUOTE_CHAR) : Array(Array(String))

Parses a CSV or IO into an array.

Takes optional separator and quote_char arguments for defining non-standard csv cell separators and quote characters.

require "csv"

CSV.parse("one,two\nthree")
# => [["one", "two"], ["three"]]
CSV.parse("one;two\n'three;'", separator: ';', quote_char: '\'')
# => [["one", "two"], ["three;"]]

Instance methods

[](header : String) : String

Returns the current row's value corresponding to the given header name. Raises KeyError if no such header exists. Raises CSV::Error if headers were not requested.

[](column : Int) : String

Returns the current row's value at the given column index. A negative index counts from the end. Raises IndexError if no such column exists.

[](header_pattern : Regex, *, options : Regex::MatchOptions = Regex::MatchOptions::None) : String

Returns the current row's value corresponding to the given header_pattern. Raises KeyError if no such header exists. Raises CSV::Error if headers were not requested.

[]?(header : String) : String | Nil

Returns the current row's value corresponding to the given header name. Returns nil if no such header exists. Raises CSV::Error if headers were not requested.

[]?(column : Int) : String | Nil

Returns the current row's value at the given column index. A negative index counts from the end. Returns nil if no such column exists.

[]?(header_pattern : Regex, *, options : Regex::MatchOptions = Regex::MatchOptions::None) : String | Nil

Returns the current row's value corresponding to the given header_pattern. Returns nil if no such header exists. Raises CSV::Error if headers were not requested.

each

Invokes the block once for each row in this CSV, yielding self.

headers

Returns this CSV headers. Their values are always stripped. Raises CSV::Error if headers were not requested.

next

Advanced the cursor to the next row. Must be called once to position the cursor in the first row. Returns true if a next row was found, false otherwise.

rewind

Rewinds this CSV to the beginning, rewinding the underlying IO if any.

row

Returns the current row as a Row instance.

values_at(*columns : Int)

Returns a tuple of the current row's values at given indices A negative index counts from the end. Raises IndexError if any column doesn't exist The behavior of returning a tuple is similar to Hash#values_at

values_at(*headers : String)

Returns a tuple of the current row's values corresponding to the given headers Raises KeyError if any header doesn't exist. Raises CSV::Error if headers were not requested The behavior of returning a tuple is similar to Hash#values_at

Nested types