=begin
  Copyright (C) 2008 Sam Roberts

  This library is free software; you can redistribute it and/or modify it
  under the same terms as the ruby language itself, see the file COPYING for
  details.
=end

require 'vpim/vpim'

module Vpim
  # Contains regular expression strings for the EBNF of RFC 2425.
  module Bnf #:nodoc:

    # 1*(ALPHA / DIGIT / "-")
    # Note: I think I can add A-Z here, and get rid of the "i" matches elsewhere.
    # Note: added '_' to allowed because its produced by Notes  (X-LOTUS-CHILD_UID:)
    # Note: added '/' to allowed because its produced by KAddressBook (X-messaging/xmpp-All:)
    # Note: added ' ' to allowed because its produced by highrisehq.com (X-GOOGLE TALK:)
    NAME    = '[-a-z0-9_/][-a-z0-9_/ ]*'

    # <"> <Any character except CTLs, DQUOTE> <">
    QSTR    = '"([^"]*)"'
     
    # *<Any character except CTLs, DQUOTE, ";", ":", ",">
    PTEXT   = '([^";:,]+)'
      
    # param-value = ptext / quoted-string
    PVALUE  = "(?:#{QSTR}|#{PTEXT})"
    
    # param = name "=" param-value *("," param-value)
    # Note: v2.1 allows a type or encoding param-value to appear without the type=
    # or the encoding=. This is hideous, but we try and support it, if there
    # is no "=", then $2 will be "", and we will treat it as a v2.1 param.
    PARAM = ";(#{NAME})(=?)((?:#{PVALUE})?(?:,#{PVALUE})*)"

    # V3.0: contentline  =   [group "."]  name *(";" param) ":" value
    # V2.1: contentline  = *( group "." ) name *(";" param) ":" value
    #
    # We accept the V2.1 syntax for backwards compatibility.
    #LINE = "((?:#{NAME}\\.)*)?(#{NAME})([^:]*)\:(.*)"
    LINE = "^((?:#{NAME}\\.)*)?(#{NAME})((?:#{PARAM})*):(.*)$"

    # date = date-fullyear ["-"] date-month ["-"] date-mday
    # date-fullyear = 4 DIGIT
    # date-month = 2 DIGIT
    # date-mday = 2 DIGIT
    DATE = '(\d\d\d\d)-?(\d\d)-?(\d\d)'

    # time = time-hour [":"] time-minute [":"] time-second [time-secfrac] [time-zone]
    # time-hour = 2 DIGIT
    # time-minute = 2 DIGIT
    # time-second = 2 DIGIT
    # time-secfrac = "," 1*DIGIT
    # time-zone = "Z" / time-numzone
    # time-numzone = sign time-hour [":"] time-minute
    TIME = '(\d\d):?(\d\d):?(\d\d)(\.\d+)?(Z|[-+]\d\d:?\d\d)?'

    # integer = (["+"] / "-") 1*DIGIT
    INTEGER = '[-+]?\d+'

    # QSAFE-CHAR = WSP / %x21 / %x23-7E / NON-US-ASCII
    #  ; Any character except CTLs and DQUOTE
    QSAFECHAR = '[ \t\x21\x23-\x7e\x80-\xff]'

    # SAFE-CHAR  = WSP / %x21 / %x23-2B / %x2D-39 / %x3C-7E / NON-US-ASCII
    #   ; Any character except CTLs, DQUOTE, ";", ":", ","
    SAFECHAR = '[ \t\x21\x23-\x2b\x2d-\x39\x3c-\x7e\x80-\xff]'
  end
end

module Vpim
  # Split on \r\n or \n to get the lines, unfold continued lines (they
  # start with ' ' or \t), and return the array of unfolded lines.
  #
  # This also supports the (invalid) encoding convention of allowing empty
  # lines to be inserted for readability - it does this by dropping zero-length
  # lines.
  # 
  # Also supports an the QUOTED-PRINTABLE soft line-break as described here: 
  # http://en.wikipedia.org/wiki/Quoted-printable
  # 
  def Vpim.unfold(card) # :nodoc:
    unfolded = []
    card.each do |line|
      line.chomp!
      # If it's a continuation line, add it to the last.
      # If it's an empty line, drop it from the input.
      if( line =~ /^[ \t]/ )
        unfolded[-1] << line[1, line.size-1]
      elsif (unfolded.last && unfolded.last =~ /;ENCODING=QUOTED-PRINTABLE:.*?=$/)
        unfolded.last << line
      elsif( line =~ /^$/ )
      else
        unfolded << line
      end
    end
    unfolded
  end

  # Convert a +sep+-seperated list of values into an array of values.
  def Vpim.decode_list(value, sep = ',') # :nodoc:
    list = []
    
    value.each_line(sep) do |item|
      item.chomp!(sep)
      list << yield(item)
    end
    list
  end

  # Convert a RFC 2425 date into an array of [year, month, day].
  def Vpim.decode_date(v) # :nodoc:
    unless v =~ %r{^\s*#{Bnf::DATE}\s*$}
      raise Vpim::InvalidEncodingError, "date not valid (#{v})"
    end
    [$1.to_i, $2.to_i, $3.to_i]
  end

  # Convert a RFC 2425 date into a Date object.
  def self.decode_date_to_date(v)
    Date.new(*decode_date(v))
  end

  # Note in the following the RFC2425 allows yyyy-mm-ddThh:mm:ss, but RFC2445
  # does not. I choose to encode to the subset that is valid for both.

  # Encode a Date object as "yyyymmdd".
  def Vpim.encode_date(d) # :nodoc:
     "%0.4d%0.2d%0.2d" % [ d.year, d.mon, d.day ]
  end

  # Encode a Date object as "yyyymmdd".
  def Vpim.encode_time(d) # :nodoc:
     "%0.4d%0.2d%0.2d" % [ d.year, d.mon, d.day ]
  end

  # Encode a Time or DateTime object as "yyyymmddThhmmss"
  def Vpim.encode_date_time(d) # :nodoc:
     "%0.4d%0.2d%0.2dT%0.2d%0.2d%0.2d" % [ d.year, d.mon, d.day, d.hour, d.min, d.sec ]
  end

  # Convert a RFC 2425 time into an array of [hour,min,sec,secfrac,timezone]
  def Vpim.decode_time(v) # :nodoc:
    unless match = %r{^\s*#{Bnf::TIME}\s*$}.match(v)
      raise Vpim::InvalidEncodingError, "time '#{v}' not valid"
    end
    hour, min, sec, secfrac, tz = match.to_a[1..5]

    [hour.to_i, min.to_i, sec.to_i, secfrac ? secfrac.to_f : 0, tz]
  end

  def self.array_datetime_to_time(dtarray) #:nodoc:
    # We get [ year, month, day, hour, min, sec, usec, tz ]
    begin
      tz = (dtarray.pop == "Z") ? :gm : :local
      Time.send(tz, *dtarray)
    rescue ArgumentError => e
      raise Vpim::InvalidEncodingError, "#{tz} #{e} (#{dtarray.join(', ')})"
    end
  end

  # Convert a RFC 2425 time into an array of Time objects.
  def Vpim.decode_time_to_time(v) # :nodoc:
    array_datetime_to_time(decode_date_time(v))
  end

  # Convert a RFC 2425 date-time into an array of [year,mon,day,hour,min,sec,secfrac,timezone]
  def Vpim.decode_date_time(v) # :nodoc:
    unless match = %r{^\s*#{Bnf::DATE}T#{Bnf::TIME}\s*$}.match(v)
      raise Vpim::InvalidEncodingError, "date-time '#{v}' not valid"
    end
    year, month, day, hour, min, sec, secfrac, tz = match.to_a[1..8]

    [
      # date
      year.to_i, month.to_i, day.to_i,
      # time
      hour.to_i, min.to_i, sec.to_i, secfrac ? secfrac.to_f : 0, tz
    ]
  end

  def Vpim.decode_date_time_to_datetime(v) #:nodoc:
    year, month, day, hour, min, sec, secfrac, tz = Vpim.decode_date_time(v)
    # TODO - DateTime understands timezones, so we could decode tz and use it.
    DateTime.civil(year, month, day, hour, min, sec, 0)
  end

  # Vpim.decode_boolean
  #
  # float
  #
  # float_list
=begin
=end

  # Convert an RFC2425 INTEGER value into an Integer
  def Vpim.decode_integer(v) # :nodoc:
    unless match = %r{\s*#{Bnf::INTEGER}\s*}.match(v)
      raise Vpim::InvalidEncodingError, "integer not valid (#{v})"
    end
    v.to_i
  end

  #
  # integer_list

  # Convert a RFC2425 date-list into an array of dates.
  def Vpim.decode_date_list(v) # :nodoc:
    Vpim.decode_list(v) do |date|
      date.strip!
      if date.length > 0
        Vpim.decode_date(date)
      end
    end.compact
  end

  # Convert a RFC 2425 time-list into an array of times.
  def Vpim.decode_time_list(v) # :nodoc:
    Vpim.decode_list(v) do |time|
      time.strip!
      if time.length > 0
        Vpim.decode_time(time)
      end
    end.compact
  end

  # Convert a RFC 2425 date-time-list into an array of date-times.
  def Vpim.decode_date_time_list(v) # :nodoc:
    Vpim.decode_list(v) do |datetime|
      datetime.strip!
      if datetime.length > 0
        Vpim.decode_date_time(datetime)
      end
    end.compact
  end

  # Convert RFC 2425 text into a String.
  # \\ -> \
  # \n -> NL
  # \N -> NL
  # \, -> ,
  # \; -> ;
  #
  # I've seen double-quote escaped by iCal.app. Hmm. Ok, if you aren't supposed
  # to escape anything but the above, everything else is ambiguous, so I'll
  # just support it.
  def Vpim.decode_text(v) # :nodoc:
    # FIXME - I think this should trim leading and trailing space
    v.gsub(/\\(.)/) do
      case $1
      when 'n', 'N' 
        "\n"
      else
        $1
      end
    end
  end
  
  def Vpim.encode_text(v) #:nodoc:
    v.to_str.gsub(/([\\,;\n])/) { $1 == "\n" ? "\\n" : "\\"+$1 }
  end

  # v is an Array of String, or just a single String
  def Vpim.encode_text_list(v, sep = ",") #:nodoc:
    begin
      v.to_ary.map{ |t| Vpim.encode_text(t) }.join(sep)
    rescue
      Vpim.encode_text(v)
    end
  end

  # Convert a +sep+-seperated list of TEXT values into an array of values.
  def Vpim.decode_text_list(value, sep = ',') # :nodoc:
    # Need to do in two stages, as best I can find.
    list = value.scan(/([^#{sep}\\]*(?:\\.[^#{sep}\\]*)*)#{sep}/).map do |v|
      Vpim.decode_text(v.first)
    end
    if value.match(/([^#{sep}\\]*(?:\\.[^#{sep}\\]*)*)$/)
      list << $1
    end
    list
  end

  # param-value = paramtext / quoted-string
  # paramtext  = *SAFE-CHAR
  # quoted-string      = DQUOTE *QSAFE-CHAR DQUOTE
  def Vpim.encode_paramtext(value)
    case value
    when %r{\A#{Bnf::SAFECHAR}*\z}
      value
    else
      raise Vpim::Unencodable, "paramtext #{value.inspect}"
    end
  end

  def Vpim.encode_paramvalue(value)
    case value
    when %r{\A#{Bnf::SAFECHAR}*\z}
      value
    when %r{\A#{Bnf::QSAFECHAR}*\z}
      '"' + value + '"'
    else
      raise Vpim::Unencodable, "param-value #{value.inspect}"
    end
  end


  # Unfold the lines in +card+, then return an array of one Field object per
  # line.
  def Vpim.decode(card) #:nodoc:
      content = Vpim.unfold(card).collect { |line| DirectoryInfo::Field.decode(line) }
  end


  # Expand an array of fields into its syntactic entities. Each entity is a sequence
  # of fields where the sequences is delimited by a BEGIN/END field. Since
  # BEGIN/END delimited entities can be nested, we build a tree. Each entry in
  # the array is either a Field or an array of entries (where each entry is
  # either a Field, or an array of entries...).
  def Vpim.expand(src) #:nodoc:
    # output array to expand the src to
    dst = []
    # stack used to track our nesting level, as we see begin/end we start a
    # new/finish the current entity, and push/pop that entity from the stack
    current = [ dst ]

    for f in src
      if f.name? 'BEGIN'
        e = [ f ]

        current.last.push(e)
        current.push(e)

      elsif f.name? 'END'
        current.last.push(f)

        unless current.last.first.value? current.last.last.value
          raise "BEGIN/END mismatch (#{current.last.first.value} != #{current.last.last.value})"
        end

        current.pop

      else
        current.last.push(f)
      end
    end

    dst
  end

  # Split an array into an array of all the fields at the outer level, and
  # an array of all the inner arrays of fields. Return the array [outer,
  # inner].
  def Vpim.outer_inner(fields) #:nodoc:
    # TODO - use Enumerable#partition
    # seperate into the outer-level fields, and the arrays of component
    # fields
    outer = []
    inner = []
    fields.each do |line|
      case line
      when Array; inner << line
      else;       outer << line
      end
    end
    return outer, inner
  end

end