path: root/doc/io_streams.rdoc

== \IO Streams

Ruby supports processing data as \IO streams;
that is, as data that may be read, re-read, written, re-written,
and traversed via iteration.

Core classes with such support include:

- IO, and its derived class File.
- {StringIO}[rdoc-ref:StringIO]: for processing a string.
- {ARGF}[rdoc-ref:ARGF]: for processing files cited on the command line.

Pre-existing stream objects that are referenced by constants include:

- $stdin: read-only instance of \IO.
- $stdout: write-only instance of \IO.
- $stderr: read-only instance of \IO.
- \ARGF: read-only instance of \ARGF.

You can create stream objects:

- \File:

  - File.new: returns a new \File object.
  - File.open: passes a new \File object to given the block.

- \IO:

  - IO.new: returns a new \IO object for the given integer file descriptor.
  - IO.open: passes a new \IO object to the given block.
  - IO.popen: returns a new \IO object that is connected to the $stdin
    and $stdout of a newly-launched subprocess.
  - Kernel#open: returns a new \IO object connected to a given source:
    stream, file, or subprocess.

- \StringIO:

  - StringIO.new: returns a new \StringIO object.
  - StringIO.open: passes a new \StringIO object to the given block.

(You cannot create an \ARGF object, but one already exists.)

=== About the Examples

Many examples here use these variables:

  # English text with newlines.
  text = <<~EOT
    First line
    Second line

    Fourth line
    Fifth line
  EOT

  # Russian text.
  russian = "\u{442 435 441 442}" # => "тест"

  # Binary data.
  data = "\u9990\u9991\u9992\u9993\u9994"

  # Text file.
  File.write('t.txt', text)

  # File with Russian text.
  File.write('t.rus', russian)

  # File with binary data.
  f = File.new('t.dat', 'wb:UTF-16')
  f.write(data)
  f.close

=== Position

An \IO stream has a nonnegative integer _position_,
which is the byte offset at which the next read or write is to occur;
the relevant methods:

- IO#tell (aliased as +#pos+):
  Returns the current position (in bytes) in the stream:

    f = File.new('t.txt')
    f.tell # => 0
    f.gets # => "First line\n"
    f.tell # => 12
    f.close

- IO#pos=: Sets the position of the stream (in bytes):

    f = File.new('t.txt')
    f.tell     # => 0
    f.pos = 20 # => 20
    f.tell     # => 20
    f.close

- IO#seek: Sets the position of the stream to a given integer +offset+
  (in bytes), with respect to a given constant +whence+, which is one of:

  - +:CUR+ or <tt>IO::SEEK_CUR</tt>:
    Repositions the stream to its current position plus the given +offset+:

      f = File.new('t.txt')
      f.tell            # => 0
      f.seek(20, :CUR)  # => 0
      f.tell            # => 20
      f.seek(-10, :CUR) # => 0
      f.tell            # => 10
      f.close

  - +:END+ or <tt>IO::SEEK_END</tt>:
    Repositions the stream to its end plus the given +offset+:

      f = File.new('t.txt')
      f.tell            # => 0
      f.seek(0, :END)   # => 0  # Repositions to stream end.
      f.tell            # => 52
      f.seek(-20, :END) # => 0
      f.tell            # => 32
      f.seek(-40, :END) # => 0
      f.tell            # => 12
      f.close

  - +:SET+ or <tt>IO:SEEK_SET</tt>:
    Repositions the stream to the given +offset+:

      f = File.new('t.txt')
      f.tell            # => 0
      f.seek(20, :SET) # => 0
      f.tell           # => 20
      f.seek(40, :SET) # => 0
      f.tell           # => 40
      f.close

- IO#rewind: Positions the stream to the beginning:

    f = File.new('t.txt')
    f.tell     # => 0
    f.gets     # => "First line\n"
    f.tell     # => 12
    f.rewind   # => 0
    f.tell     # => 0
    f.close

=== Lines

Some reader methods in \IO streams are line-oriented;
such a method reads one or more lines,
which are separated by an implicit or explicit line separator.

These methods are included (except as noted) in classes Kernel, IO, File,
and {ARGF}[rdoc-ref:ARGF]:

- IO#each_line: Passes each line to the block; not in Kernel:

    f = File.new('t.txt')
    f.each_line {|line| p line }

  Output:

    "First line\n"
    "Second line\n"
    "\n"
    "Fourth line\n"
    "Fifth line\n"

  The reading may begin mid-line:

    f = File.new('t.txt')
    f.pos = 27
    f.each_line {|line| p line }

  Output:

    "rth line\n"
    "Fifth line\n"

- IO#gets: Returns the next line (which may begin mid-line):

    f = File.new('t.txt')
    f.gets      # => "First line\n"
    f.gets      # => "Second line\n"
    f.pos = 27
    f.gets      # => "rth line\n"
    f.readlines # => ["Fifth line\n"]
    f.gets      # => nil

- IO#readline: Like #gets, but raises an exception at end-of-file;
  not in StringIO.

- IO#readlines: Returns all remaining lines in an array;
  may begin mid-line:

    f = File.new('t.txt')
    f.pos = 19
    f.readlines # => ["ine\n", "\n", "Fourth line\n", "Fifth line\n"]
    f.readlines # => []

Each of these methods may be called with:

- An optional line separator, +sep+.
- An optional line-size limit, +limit+.
- Both +sep+ and +limit+.

==== Line Separator

The default line separator is the given by the global variable <tt>$/</tt>,
whose value is by default <tt>"\n"</tt>.
The line to be read next is all data from the current position
to the next line separator:

  f = File.new('t.txt')
  f.gets # => "First line\n"
  f.gets # => "Second line\n"
  f.gets # => "\n"
  f.gets # => "Fourth line\n"
  f.gets # => "Fifth line\n"
  f.close

You can specify a different line separator:

  f = File.new('t.txt')
  f.gets('l')   # => "First l"
  f.gets('li')  # => "ine\nSecond li"
  f.gets('lin') # => "ne\n\nFourth lin"
  f.gets        # => "e\n"
  f.close

There are two special line separators:

- +nil+: The entire stream is read into a single string:

    f = File.new('t.txt')
    f.gets(nil) # => "First line\nSecond line\n\nFourth line\nFifth line\n"
    f.close

- <tt>''</tt> (the empty string): The next "paragraph" is read
  (paragraphs being separated by two consecutive line separators):

    f = File.new('t.txt')
    f.gets('') # => "First line\nSecond line\n\n"
    f.gets('') # => "Fourth line\nFifth line\n"
    f.close

==== Line Limit

The line to be read may be further defined by an optional integer argument +limit+,
which specifies that the number of bytes returned may not be (much) longer
than the given +limit+;
a multi-byte character will not be split, and so a line may be slightly longer
than the given limit.

If +limit+ is not given, the line is determined only by +sep+.

  # Text with 1-byte characters.
  File.new('t.txt') {|f| f.gets(1) }  # => "F"
  File.new('t.txt') {|f| f.gets(2) }  # => "Fi"
  File.new('t.txt') {|f| f.gets(3) }  # => "Fir"
  File.new('t.txt') {|f| f.gets(4) }  # => "Firs"
  # No more than one line.
  File.new('t.txt') {|f| f.gets(10) } # => "First line"
  File.new('t.txt') {|f| f.gets(11) } # => "First line\n"
  File.new('t.txt') {|f| f.gets(12) } # => "First line\n"

  # Text with 2-byte characters, which will not be split.
  File.new('r.rus') {|f| f.gets(1).size } # => 1
  File.new('r.rus') {|f| f.gets(2).size } # => 1
  File.new('r.rus') {|f| f.gets(3).size } # => 2
  File.new('r.rus') {|f| f.gets(4).size } # => 2

==== Line Separator and Line Limit

With arguments +sep+ and +limit+ given,
combines the two behaviors:

- Returns the next line as determined by line separator +sep+.
- But returns no more bytes than are allowed by the limit.

Example:

  File.new('t.txt') {|f| f.gets('li', 20) } # => "First li"
  File.new('t.txt') {|f| f.gets('li', 2) }  # => "Fi"

==== Line Number

A readable \IO stream has a _line_ _number_,
which is the non-negative integer line number
in the stream where the next read will occur.

A new stream is initially has line number +0+.

\Method IO#lineno returns the line number.

Reading lines from a stream usually changes its line number:

  f = File.new('t.txt', 'r')
  f.lineno   # => 0
  f.readline # => "This is line one.\n"
  f.lineno   # => 1
  f.readline # => "This is the second line.\n"
  f.lineno   # => 2
  f.readline # => "Here's the third line.\n"
  f.lineno   # => 3
  f.eof?     # => true
  f.close

Iterating over lines in a stream usually changes its line number:

  f = File.new('t.txt')
  f.each_line do |line|
    p "position=#{f.pos} eof?=#{f.eof?} lineno=#{f.lineno}"
  end
  f.close

Output:

  "position=11 eof?=false lineno=1"
  "position=23 eof?=false lineno=2"
  "position=24 eof?=false lineno=3"
  "position=36 eof?=false lineno=4"
  "position=47 eof?=true lineno=5"

==== Line Options

A number of \IO methods accept optional keyword arguments
that determine how lines in a stream are to be treated:

- +:chomp+: If +true+, line separators are omitted; default is +false+.

=== Open and Closed \IO Streams

A new \IO stream may be open for reading, open for writing, or both.

You can close a stream using these methods:

- IO#close: Closes the stream for both reading and writing.
- IO#close_read (not available in \ARGF): Closes the stream for reading.
- IO#close_write (not available in \ARGF): Closes the stream for writing.

You can query whether a stream is closed using these methods:

- IO#closed?: Returns whether the stream is closed.

=== Stream End-of-File

You can query whether a stream is at end-of-file using this method:

- IO#eof? (also aliased as +#eof+):
  Returns whether the stream is at end-of-file.