Class FasterCSV
In: lib/faster_csv.rb
Parent: Object

This class provides a complete interface to CSV files and data. It offers tools to enable you to read and write to and from Strings or IO objects, as needed.

Reading

From a File

A Line at a Time

  FasterCSV.foreach("path/to/file.csv") do |row|
    # use row here...
  end

All at Once

  arr_of_arrs = FasterCSV.read("path/to/file.csv")

From a String

A Line at a Time

  FasterCSV.parse("CSV,data,String") do |row|
    # use row here...
  end

All at Once

  arr_of_arrs = FasterCSV.parse("CSV,data,String")

Writing

To a File

  FasterCSV.open("path/to/file.csv", "w") do |csv|
    csv << ["row", "of", "CSV", "data"]
    csv << ["another", "row"]
    # ...
  end

To a String

  csv_string = FasterCSV.generate do |csv|
    csv << ["row", "of", "CSV", "data"]
    csv << ["another", "row"]
    # ...
  end

Convert a Single Line

  csv_string = ["CSV", "data"].to_csv   # to CSV
  csv_array  = "CSV,String".parse_csv   # from CSV

Shortcut Interface

  FCSV             { |csv_out| csv_out << %w{my data here} }  # to $stdout
  FCSV(csv = "")   { |csv_str| csv_str << %w{my data here} }  # to a String
  FCSV($stderr)    { |csv_err| csv_err << %w{my data here} }  # to $stderr

Methods

<<   add_row   build_csv_interface   convert   dump   each   filter   foreach   generate   generate_line   gets   header_convert   header_row?   inspect   instance   load   new   open   parse   parse_line   puts   read   read   readline   readlines   readlines   rewind   shift   table  

Included Modules

Enumerable

Classes and Modules

Class FasterCSV::MalformedCSVError
Class FasterCSV::Row
Class FasterCSV::Table

Constants

VERSION = "1.5.1".freeze   The version of the installed library.
FieldInfo = Struct.new(:index, :line, :header)   A FieldInfo Struct contains details about a field‘s position in the data source it was read from. FasterCSV will pass this Struct to some blocks that make decisions based on field structure. See FasterCSV.convert_fields() for an example.
index:The zero-based index of the field in its row.
line:The line of the data source this row is from.
header:The header for the column, when available.
DateMatcher = / \A(?: (\w+,?\s+)?\w+\s+\d{1,2},?\s+\d{2,4} | \d{4}-\d{2}-\d{2} )\z /x   A Regexp used to find and convert some common Date formats.
DateTimeMatcher = / \A(?: (\w+,?\s+)?\w+\s+\d{1,2}\s+\d{1,2}:\d{1,2}:\d{1,2},?\s+\d{2,4} | \d{4}-\d{2}-\d{2}\s\d{2}:\d{2}:\d{2} )\z /x   A Regexp used to find and convert some common DateTime formats.
Converters = { :integer => lambda { |f| Integer(f) rescue f }, :float => lambda { |f| Float(f) rescue f }, :numeric => [:integer, :float], :date => lambda { |f| f =~ DateMatcher ? (Date.parse(f) rescue f) : f   This Hash holds the built-in converters of FasterCSV that can be accessed by name. You can select Converters with FasterCSV.convert() or through the options Hash passed to FasterCSV::new().
:integer:Converts any field Integer() accepts.
:float:Converts any field Float() accepts.
:numeric:A combination of :integer and :float.
:date:Converts any field Date::parse() accepts.
:date_time:Converts any field DateTime::parse() accepts.
:all:All built-in converters. A combination of :date_time and :numeric.

This Hash is intetionally left unfrozen and users should feel free to add values to it that can be accessed by all FasterCSV objects.

To add a combo field, the value should be an Array of names. Combo fields can be nested with other combo fields.

HeaderConverters = { :downcase => lambda { |h| h.downcase }, :symbol => lambda { |h| h.downcase.tr(" ", "_").delete("^a-z0-9_").to_sym   This Hash holds the built-in header converters of FasterCSV that can be accessed by name. You can select HeaderConverters with FasterCSV.header_convert() or through the options Hash passed to FasterCSV::new().
:downcase:Calls downcase() on the header String.
:symbol:The header String is downcased, spaces are replaced with underscores, non-word characters are dropped, and finally to_sym() is called.

This Hash is intetionally left unfrozen and users should feel free to add values to it that can be accessed by all FasterCSV objects.

To add a combo field, the value should be an Array of names. Combo fields can be nested with other combo fields.

DEFAULT_OPTIONS = { :col_sep => ",", :row_sep => :auto, :quote_char => '"', :converters => nil, :unconverted_fields => nil, :headers => false, :return_headers => false, :header_converters => nil, :skip_blanks => false, :force_quotes => false }.freeze   The options used when no overrides are given by calling code. They are:
:col_sep:","
:row_sep::auto
:quote_char:’"’
:converters:nil
:unconverted_fields:nil
:headers:false
:return_headers:false
:header_converters:nil
:skip_blanks:false
:force_quotes:false

Attributes

lineno  [R]  The line number of the last row read from this file. Fields with nested line-end characters will not affect this count.

Public Class methods

This method will build a drop-in replacement for many of the standard CSV methods. It allows you to write code like:

  begin
    require "faster_csv"
    FasterCSV.build_csv_interface
  rescue LoadError
    require "csv"
  end
  # ... use CSV here ...

This is not a complete interface with completely identical behavior. However, it is intended to be close enough that you won‘t notice the difference in most cases. CSV methods supported are:

Be warned that this interface is slower than vanilla FasterCSV due to the extra layer of method calls. Depending on usage, this can slow it down to near CSV speeds.

[Source]

# File lib/faster_csv.rb, line 853
  def self.build_csv_interface
    Object.const_set(:CSV, Class.new).class_eval do
      def self.foreach(path, rs = :auto, &block)  # :nodoc:
        FasterCSV.foreach(path, :row_sep => rs, &block)
      end
      
      def self.generate_line(row, fs = ",", rs = "")  # :nodoc:
        FasterCSV.generate_line(row, :col_sep => fs, :row_sep => rs)
      end
      
      def self.open(path, mode, fs = ",", rs = :auto, &block)  # :nodoc:
        if block and mode.include? "r"
          FasterCSV.open(path, mode, :col_sep => fs, :row_sep => rs) do |csv|
            csv.each(&block)
          end
        else
          FasterCSV.open(path, mode, :col_sep => fs, :row_sep => rs, &block)
        end
      end
      
      def self.parse(str_or_readable, fs = ",", rs = :auto, &block)  # :nodoc:
        FasterCSV.parse(str_or_readable, :col_sep => fs, :row_sep => rs, &block)
      end
      
      def self.parse_line(src, fs = ",", rs = :auto)  # :nodoc:
        FasterCSV.parse_line(src, :col_sep => fs, :row_sep => rs)
      end
      
      def self.readlines(path, rs = :auto)  # :nodoc:
        FasterCSV.readlines(path, :row_sep => rs)
      end
    end
  end

This method allows you to serialize an Array of Ruby objects to a String or File of CSV data. This is not as powerful as Marshal or YAML, but perhaps useful for spreadsheet and database interaction.

Out of the box, this method is intended to work with simple data objects or Structs. It will serialize a list of instance variables and/or Struct.members().

If you need need more complicated serialization, you can control the process by adding methods to the class to be serialized.

A class method csv_meta() is responsible for returning the first row of the document (as an Array). This row is considered to be a Hash of the form key_1,value_1,key_2,value_2,… FasterCSV::load() expects to find a class key with a value of the stringified class name and FasterCSV::dump() will create this, if you do not define this method. This method is only called on the first object of the Array.

The next method you can provide is an instance method called csv_headers(). This method is expected to return the second line of the document (again as an Array), which is to be used to give each column a header. By default, FasterCSV::load() will set an instance variable if the field header starts with an @ character or call send() passing the header as the method name and the field value as an argument. This method is only called on the first object of the Array.

Finally, you can provide an instance method called csv_dump(), which will be passed the headers. This should return an Array of fields that can be serialized for this object. This method is called once for every object in the Array.

The io parameter can be used to serialize to a File, and options can be anything FasterCSV::new() accepts.

[Source]

# File lib/faster_csv.rb, line 922
  def self.dump(ary_of_objs, io = "", options = Hash.new)
    obj_template = ary_of_objs.first
    
    csv = FasterCSV.new(io, options)
    
    # write meta information
    begin
      csv << obj_template.class.csv_meta
    rescue NoMethodError
      csv << [:class, obj_template.class]
    end

    # write headers
    begin
      headers = obj_template.csv_headers
    rescue NoMethodError
      headers = obj_template.instance_variables.sort
      if obj_template.class.ancestors.find { |cls| cls.to_s =~ /\AStruct\b/ }
        headers += obj_template.members.map { |mem| "#{mem}=" }.sort
      end
    end
    csv << headers
    
    # serialize each object
    ary_of_objs.each do |obj|
      begin
        csv << obj.csv_dump(headers)
      rescue NoMethodError
        csv << headers.map do |var|
          if var[0] == ?@
            obj.instance_variable_get(var)
          else
            obj[var[0..-2]]
          end
        end
      end
    end
    
    if io.is_a? String
      csv.string
    else
      csv.close
    end
  end

This method is a convenience for building Unix-like filters for CSV data. Each row is yielded to the provided block which can alter it as needed. After the block returns, the row is appended to output altered or not.

The input and output arguments can be anything FasterCSV::new() accepts (generally String or IO objects). If not given, they default to ARGF and $stdout.

The options parameter is also filtered down to FasterCSV::new() after some clever key parsing. Any key beginning with :in_ or :input_ will have that leading identifier stripped and will only be used in the options Hash for the input object. Keys starting with :out_ or :output_ affect only output. All other keys are assigned to both objects.

The :output_row_sep option defaults to $INPUT_RECORD_SEPARATOR ($/).

[Source]

# File lib/faster_csv.rb, line 991
  def self.filter(*args)
    # parse options for input, output, or both
    in_options, out_options = Hash.new, {:row_sep => $INPUT_RECORD_SEPARATOR}
    if args.last.is_a? Hash
      args.pop.each do |key, value|
        case key.to_s
        when /\Ain(?:put)?_(.+)\Z/
          in_options[$1.to_sym] = value
        when /\Aout(?:put)?_(.+)\Z/
          out_options[$1.to_sym] = value
        else
          in_options[key]  = value
          out_options[key] = value
        end
      end
    end
    # build input and output wrappers
    input   = FasterCSV.new(args.shift || ARGF,    in_options)
    output  = FasterCSV.new(args.shift || $stdout, out_options)
    
    # read, yield, write
    input.each do |row|
      yield row
      output << row
    end
  end

This method is intended as the primary interface for reading CSV files. You pass a path and any options you wish to set for the read. Each row of file will be passed to the provided block in turn.

The options parameter can be anything FasterCSV::new() understands.

[Source]

# File lib/faster_csv.rb, line 1025
  def self.foreach(path, options = Hash.new, &block)
    open(path, "rb", options) do |csv|
      csv.each(&block)
    end
  end

This method wraps a String you provide, or an empty default String, in a FasterCSV object which is passed to the provided block. You can use the block to append CSV rows to the String and when the block exits, the final String will be returned.

Note that a passed String is modfied by this method. Call dup() before passing if you need a new String.

The options parameter can be anthing FasterCSV::new() understands.

[Source]

# File lib/faster_csv.rb, line 1046
  def self.generate(*args)
    # add a default empty String, if none was given
    if args.first.is_a? String
      io = StringIO.new(args.shift)
      io.seek(0, IO::SEEK_END)
      args.unshift(io)
    else
      args.unshift("")
    end
    faster_csv = new(*args)  # wrap
    yield faster_csv         # yield for appending
    faster_csv.string        # return final String
  end

This method is a shortcut for converting a single row (Array) into a CSV String.

The options parameter can be anthing FasterCSV::new() understands.

The :row_sep option defaults to $INPUT_RECORD_SEPARATOR ($/) when calling this method.

[Source]

# File lib/faster_csv.rb, line 1069
  def self.generate_line(row, options = Hash.new)
    options = {:row_sep => $INPUT_RECORD_SEPARATOR}.merge(options)
    (new("", options) << row).string
  end

This method will return a FasterCSV instance, just like FasterCSV::new(), but the instance will be cached and returned for all future calls to this method for the same data object (tested by Object#object_id()) with the same options.

If a block is given, the instance is passed to the block and the return value becomes the return value of the block.

[Source]

# File lib/faster_csv.rb, line 1083
  def self.instance(data = $stdout, options = Hash.new)
    # create a _signature_ for this method call, data object and options
    sig = [data.object_id] +
          options.values_at(*DEFAULT_OPTIONS.keys.sort_by { |sym| sym.to_s })
    
    # fetch or create the instance for this signature
    @@instances ||= Hash.new
    instance    =   (@@instances[sig] ||= new(data, options))

    if block_given?
      yield instance  # run block, if given, returning result
    else
      instance        # or return the instance
    end
  end

This method is the reading counterpart to FasterCSV::dump(). See that method for a detailed description of the process.

You can customize loading by adding a class method called csv_load() which will be passed a Hash of meta information, an Array of headers, and an Array of fields for the object the method is expected to return.

Remember that all fields will be Strings after this load. If you need something else, use options to setup converters or provide a custom csv_load() implementation.

[Source]

# File lib/faster_csv.rb, line 1111
  def self.load(io_or_str, options = Hash.new)
    csv = FasterCSV.new(io_or_str, options)
    
    # load meta information
    meta = Hash[*csv.shift]
    cls  = meta["class"].split("::").inject(Object) do |c, const|
      c.const_get(const)
    end
    
    # load headers
    headers = csv.shift
    
    # unserialize each object stored in the file
    results = csv.inject(Array.new) do |all, row|
      begin
        obj = cls.csv_load(meta, headers, row)
      rescue NoMethodError
        obj = cls.allocate
        headers.zip(row) do |name, value|
          if name[0] == ?@
            obj.instance_variable_set(name, value)
          else
            obj.send(name, value)
          end
        end
      end
      all << obj
    end
    
    csv.close unless io_or_str.is_a? String
    
    results
  end

This constructor will wrap either a String or IO object passed in data for reading and/or writing. In addition to the FasterCSV instance methods, several IO methods are delegated. (See FasterCSV::open() for a complete list.) If you pass a String for data, you can later retrieve it (after writing to it, for example) with FasterCSV.string().

Note that a wrapped String will be positioned at at the beginning (for reading). If you want it at the end (for writing), use FasterCSV::generate(). If you want any other positioning, pass a preset StringIO object instead.

You may set any reading and/or writing preferences in the options Hash. Available options are:

:col_sep:The String placed between each field.
:row_sep:The String appended to the end of each row. This can be set to the special :auto setting, which requests that FasterCSV automatically discover this from the data. Auto-discovery reads ahead in the data looking for the next "\r\n", "\n", or "\r" sequence. A sequence will be selected even if it occurs in a quoted field, assuming that you would have the same line endings there. If none of those sequences is found, data is ARGF, STDIN, STDOUT, or STDERR, or the stream is only available for output, the default $INPUT_RECORD_SEPARATOR ($/) is used. Obviously, discovery takes a little time. Set manually if speed is important. Also note that IO objects should be opened in binary mode on Windows if this feature will be used as the line-ending translation can cause problems with resetting the document position to where it was before the read ahead.
:quote_char:The character used to quote fields. This has to be a single character String. This is useful for application that incorrectly use as the quote character instead of the correct ". FasterCSV will always consider a double sequence this character to be an escaped quote.
:encoding:The encoding to use when parsing the file. Defaults to your $KDOCE setting. Valid values: `n’ or `N’ for none, `e’ or `E’ for EUC, `s’ or `S’ for SJIS, and `u’ or `U’ for UTF-8 (see Regexp.new()).
:field_size_limit:This is a maximum size FasterCSV will read ahead looking for the closing quote for a field. (In truth, it reads to the first line ending beyond this size.) If a quote cannot be found within the limit FasterCSV will raise a MalformedCSVError, assuming the data is faulty. You can use this limit to prevent what are effectively DoS attacks on the parser. However, this limit can cause a legitimate parse to fail and thus is set to nil, or off, by default.
:converters:An Array of names from the Converters Hash and/or lambdas that handle custom conversion. A single converter doesn‘t have to be in an Array.
:unconverted_fields:If set to true, an unconverted_fields() method will be added to all returned rows (Array or FasterCSV::Row) that will return the fields as they were before convertion. Note that :headers supplied by Array or String were not fields of the document and thus will have an empty Array attached.
:headers:If set to :first_row or true, the initial row of the CSV file will be treated as a row of headers. If set to an Array, the contents will be used as the headers. If set to a String, the String is run through a call of FasterCSV::parse_line() with the same :col_sep, :row_sep, and :quote_char as this instance to produce an Array of headers. This setting causes FasterCSV.shift() to return rows as FasterCSV::Row objects instead of Arrays and FasterCSV.read() to return FasterCSV::Table objects instead of an Array of Arrays.
:return_headers:When false, header rows are silently swallowed. If set to true, header rows are returned in a FasterCSV::Row object with identical headers and fields (save that the fields do not go through the converters).
:write_headers:When true and :headers is set, a header row will be added to the output.
:header_converters:Identical in functionality to :converters save that the conversions are only made to header rows.
:skip_blanks:When set to a true value, FasterCSV will skip over any rows with no content.
:force_quotes:When set to a true value, FasterCSV will quote all CSV fields it creates.

See FasterCSV::DEFAULT_OPTIONS for the default settings.

Options cannot be overriden in the instance methods for performance reasons, so be sure to set what you want here.

[Source]

# File lib/faster_csv.rb, line 1400
  def initialize(data, options = Hash.new)
    # build the options for this read/write
    options = DEFAULT_OPTIONS.merge(options)
    
    # create the IO object we will read from
    @io = if data.is_a? String then StringIO.new(data) else data end
    
    init_separators(options)
    init_parsers(options)
    init_converters(options)
    init_headers(options)
    
    unless options.empty?
      raise ArgumentError, "Unknown options:  #{options.keys.join(', ')}."
    end
    
    # track our own lineno since IO gets confused about line-ends is CSV fields
    @lineno = 0
  end

This method opens an IO object, and wraps that with FasterCSV. This is intended as the primary interface for writing a CSV file.

You may pass any args Ruby‘s open() understands followed by an optional Hash containing any options FasterCSV::new() understands.

This method works like Ruby‘s open() call, in that it will pass a FasterCSV object to a provided block and close it when the block termminates, or it will return the FasterCSV object when no block is provided. (Note: This is different from the standard CSV library which passes rows to the block. Use FasterCSV::foreach() for that behavior.)

An opened FasterCSV object will delegate to many IO methods, for convenience. You may call:

  • binmode()
  • close()
  • close_read()
  • close_write()
  • closed?()
  • eof()
  • eof?()
  • fcntl()
  • fileno()
  • flush()
  • fsync()
  • ioctl()
  • isatty()
  • pid()
  • pos()
  • reopen()
  • seek()
  • stat()
  • sync()
  • sync=()
  • tell()
  • to_i()
  • to_io()
  • tty?()

[Source]

# File lib/faster_csv.rb, line 1190
  def self.open(*args)
    # find the +options+ Hash
    options = if args.last.is_a? Hash then args.pop else Hash.new end
    # default to a binary open mode
    args << "rb" if args.size == 1
    # wrap a File opened with the remaining +args+
    csv     = new(File.open(*args), options)
    
    # handle blocks like Ruby's open(), not like the CSV library
    if block_given?
      begin
        yield csv
      ensure
        csv.close
      end
    else
      csv
    end
  end

This method can be used to easily parse CSV out of a String. You may either provide a block which will be called with each row of the String in turn, or just use the returned Array of Arrays (when no block is given).

You pass your str to read from, and an optional options Hash containing anything FasterCSV::new() understands.

[Source]

# File lib/faster_csv.rb, line 1222
  def self.parse(*args, &block)
    csv = new(*args)
    if block.nil?  # slurp contents, if no block is given
      begin
        csv.read
      ensure
        csv.close
      end
    else           # or pass each row to a provided block
      csv.each(&block)
    end
  end

This method is a shortcut for converting a single line of a CSV String into a into an Array. Note that if line contains multiple rows, anything beyond the first row is ignored.

The options parameter can be anthing FasterCSV::new() understands.

[Source]

# File lib/faster_csv.rb, line 1242
  def self.parse_line(line, options = Hash.new)
    new(line, options).shift
  end

Use to slurp a CSV file into an Array of Arrays. Pass the path to the file and any options FasterCSV::new() understands.

[Source]

# File lib/faster_csv.rb, line 1250
  def self.read(path, options = Hash.new)
    open(path, "rb", options) { |csv| csv.read }
  end

Alias for FasterCSV::read().

[Source]

# File lib/faster_csv.rb, line 1255
  def self.readlines(*args)
    read(*args)
  end

A shortcut for:

  FasterCSV.read( path, { :headers           => true,
                          :converters        => :numeric,
                          :header_converters => :symbol }.merge(options) )

[Source]

# File lib/faster_csv.rb, line 1266
  def self.table(path, options = Hash.new)
    read( path, { :headers           => true,
                  :converters        => :numeric,
                  :header_converters => :symbol }.merge(options) )
  end

Public Instance methods

The primary write method for wrapped Strings and IOs, row (an Array or FasterCSV::Row) is converted to CSV and appended to the data source. When a FasterCSV::Row is passed, only the row‘s fields() are appended to the output.

The data source must be open for writing.

[Source]

# File lib/faster_csv.rb, line 1452
  def <<(row)
    # make sure headers have been assigned
    if header_row? and [Array, String].include? @use_headers.class
      parse_headers  # won't read data for Array or String
      self << @headers if @write_headers
    end
    
    # Handle FasterCSV::Row objects and Hashes
    row = case row
          when self.class::Row then row.fields
          when Hash            then @headers.map { |header| row[header] }
          else                      row
          end

    @headers =  row if header_row?
    @lineno  += 1

    @io << row.map(&@quote).join(@col_sep) + @row_sep  # quote and separate
    
    self  # for chaining
  end
add_row(row)

Alias for #<<

You can use this method to install a FasterCSV::Converters built-in, or provide a block that handles a custom conversion.

If you provide a block that takes one argument, it will be passed the field and is expected to return the converted value or the field itself. If your block takes two arguments, it will also be passed a FieldInfo Struct, containing details about the field. Again, the block should return a converted field or the field itself.

[Source]

# File lib/faster_csv.rb, line 1491
  def convert(name = nil, &converter)
    add_converter(:converters, self.class::Converters, name, &converter)
  end

Yields each row of the data source in turn.

Support for Enumerable.

The data source must be open for reading.

[Source]

# File lib/faster_csv.rb, line 1522
  def each
    while row = shift
      yield row
    end
  end
gets()

Alias for shift

Identical to FasterCSV.convert(), but for header rows.

Note that this method must be called before header rows are read to have any effect.

[Source]

# File lib/faster_csv.rb, line 1506
  def header_convert(name = nil, &converter)
    add_converter( :header_converters,
                   self.class::HeaderConverters,
                   name,
                   &converter )
  end

Returns true if the next row read will be a header row.

[Source]

# File lib/faster_csv.rb, line 1544
  def header_row?
    @use_headers and @headers.nil?
  end

Returns a simplified description of the key FasterCSV attributes.

[Source]

# File lib/faster_csv.rb, line 1669
  def inspect
    str = "<##{self.class} io_type:"
    # show type of wrapped IO
    if    @io == $stdout then str << "$stdout"
    elsif @io == $stdin  then str << "$stdin"
    elsif @io == $stderr then str << "$stderr"
    else                      str << @io.class.to_s
    end
    # show IO.path(), if available
    if @io.respond_to?(:path) and (p = @io.path)
      str << " io_path:#{p.inspect}"
    end
    # show other attributes
    %w[ lineno     col_sep     row_sep
        quote_char skip_blanks encoding ].each do |attr_name|
      if a = instance_variable_get("@#{attr_name}")
        str << " #{attr_name}:#{a.inspect}"
      end
    end
    if @use_headers
      str << " headers:#{(@headers || true).inspect}"
    end
    str << ">"
  end
puts(row)

Alias for #<<

Slurps the remaining rows and returns an Array of Arrays.

The data source must be open for reading.

[Source]

# File lib/faster_csv.rb, line 1533
  def read
    rows = to_a
    if @use_headers
      Table.new(rows)
    else
      rows
    end
  end
readline()

Alias for shift

readlines()

Alias for read

Rewinds the underlying IO object and resets FasterCSV‘s lineno() counter.

[Source]

# File lib/faster_csv.rb, line 1435
  def rewind
    @headers = nil
    @lineno  = 0
    
    @io.rewind
  end

The primary read method for wrapped Strings and IOs, a single row is pulled from the data source, parsed and returned as an Array of fields (if header rows are not used) or a FasterCSV::Row (when header rows are used).

The data source must be open for reading.

[Source]

# File lib/faster_csv.rb, line 1555
  def shift
    #########################################################################
    ### This method is purposefully kept a bit long as simple conditional ###
    ### checks are faster than numerous (expensive) method calls.         ###
    #########################################################################
    
    # handle headers not based on document content
    if header_row? and @return_headers and
       [Array, String].include? @use_headers.class
      if @unconverted_fields
        return add_unconverted_fields(parse_headers, Array.new)
      else
        return parse_headers
      end
    end
    
    # begin with a blank line, so we can always add to it
    line = String.new

    # 
    # it can take multiple calls to <tt>@io.gets()</tt> to get a full line,
    # because of \r and/or \n characters embedded in quoted fields
    # 
    loop do
      # add another read to the line
      begin
        line  += @io.gets(@row_sep)
      rescue
        return nil
      end
      # copy the line so we can chop it up in parsing
      parse =  line.dup
      parse.sub!(@parsers[:line_end], "")
      
      # 
      # I believe a blank line should be an <tt>Array.new</tt>, not 
      # CSV's <tt>[nil]</tt>
      # 
      if parse.empty?
        @lineno += 1
        if @skip_blanks
          line = ""
          next
        elsif @unconverted_fields
          return add_unconverted_fields(Array.new, Array.new)
        elsif @use_headers
          return FasterCSV::Row.new(Array.new, Array.new)
        else
          return Array.new
        end
      end

      # parse the fields with a mix of String#split and regular expressions
      csv           = Array.new
      current_field = String.new
      field_quotes  = 0
      parse.split(@col_sep, -1).each do |match|
        if current_field.empty? && match.count(@quote_and_newlines).zero?
          csv           << (match.empty? ? nil : match)
        elsif(current_field.empty? ? match[0] : current_field[0]) == @quote_char[0]
          current_field << match
          field_quotes += match.count(@quote_char)
          if field_quotes % 2 == 0
            in_quotes = current_field[@parsers[:quoted_field], 1]
            raise MalformedCSVError unless in_quotes
            current_field = in_quotes
            current_field.gsub!(@quote_char * 2, @quote_char) # unescape contents
            csv           << current_field
            current_field =  String.new
            field_quotes  =  0
          else # we found a quoted field that spans multiple lines
            current_field << @col_sep
          end
        elsif match.count("\r\n").zero?
          raise MalformedCSVError, "Illegal quoting on line #{lineno + 1}."
        else
          raise MalformedCSVError, "Unquoted fields do not allow " +
                                   "\\r or \\n (line #{lineno + 1})."
        end
      end

      # if parse is empty?(), we found all the fields on the line...
      if field_quotes % 2 == 0
        @lineno += 1

        # save fields unconverted fields, if needed...
        unconverted = csv.dup if @unconverted_fields

        # convert fields, if needed...
        csv = convert_fields(csv) unless @use_headers or @converters.empty?
        # parse out header rows and handle FasterCSV::Row conversions...
        csv = parse_headers(csv)  if     @use_headers

        # inject unconverted fields and accessor, if requested...
        if @unconverted_fields and not csv.respond_to? :unconverted_fields
          add_unconverted_fields(csv, unconverted)
        end

        # return the results
        break csv
      end
      # if we're not empty?() but at eof?(), a quoted field wasn't closed...
      if @io.eof?
        raise MalformedCSVError, "Unclosed quoted field on line #{lineno + 1}."
      elsif @field_size_limit and current_field.size >= @field_size_limit
        raise MalformedCSVError, "Field size exceeded on line #{lineno + 1}."
      end
      # otherwise, we need to loop and pull some more data to complete the row
    end
  end

[Validate]