fabulator-grammar 0.0.1 → 0.0.3

data/lib/fabulator/grammar/token_parser.rb

@@ -0,0 +1,733 @@
+#
+# DO NOT MODIFY!!!!
+# This file is automatically generated by Racc 1.4.6
+# from Racc grammer file "".
+#
+
+require 'racc/parser.rb'
+module Fabulator
+  module Grammar
+    class TokenParser < Racc::Parser
+
+module_eval(<<'...end tokens.racc/module_eval...', 'tokens.racc', 70)
+  require 'fabulator/grammar'
+
+  def parse(t)
+    @source = t
+    @curpos = 0 
+    @col = 0
+    @line = 0   
+
+    @yydebug = true
+      
+    @last_token = nil
+
+    @brackets = 0
+  
+    do_parse
+  end
+    
+  def on_error(*args)
+    raise Fabulator::Grammar::ParserError.new("unable to parse '#{args[1]}' near line #{@line + 1}, column #{@col}")
+  end
+
+  @@ops = {
+    #'[{' => :LB_LC,
+    #'}]' => :RC_RB,
+    #'[[' => :LB_LB,
+    #']]' => :RB_RB,
+    '['  => :LB,
+    ']'  => :RB,
+    '('  => :LP,
+    ')'  => :RP,
+    #'{'  => :LC,
+    #'}'  => :RC,
+    #'#'  => :HASH,
+    '$'  => :DOLLAR,
+    '^'  => :CARET,
+    #'&'  => :AND,
+    '*'  => :STAR,
+    '+'  => :PLUS,
+    '-'  => :MINUS,
+    '?'  => :QUESTION,
+    '.'  => :DOT,
+    '|'  => :PIPE,
+    ','  => :COMMA,
+    ':'  => :COLON
+  }
+
+
+  @@regex = {
+     :simple_tokens => %r{^(#{Regexp.union(@@ops.keys.sort_by{|a| a.length}.reverse.collect{ |k| k })})},
+    :ncname => %r{(?:[a-zA-Z_][-a-zA-Z0-9_.]*)}
+  }
+
+ #puts @@regex[:simple_tokens]
+
+  @@regex[:qname] = %r{((?:#{@@regex[:ncname]}:)?#{@@regex[:ncname]})}
+  @@regex[:general] = Regexp.compile(%{^#{@@regex[:qname]}|#{@@regex[:simple_tokens]}})
+
+  def next_token
+    @token = nil
+    white_space = 0
+    new_line = 0
+    while @curpos < @source.length && @source[@curpos..@curpos] =~ /\s/ do
+      if @source[@curpos..@curpos] =~ /\n/
+        new_line = new_line + 1
+        @line = @line + 1
+        @col = 0
+      else
+        @col = @col + 1
+      end
+      @curpos = @curpos + 1
+      white_space = white_space + 1   
+    end
+
+    # skip comments delimited by (:  :)
+    # comments can be nested
+    # these are XPath 2.0 comments
+    #
+    if @curpos < @source.length && @source[@curpos..@curpos+1] == '(:'
+      comment_depth = 1
+      @curpos = @curpos + 2
+      @col = @col + 2
+      while comment_depth > 0 && @curpos < @source.length
+        if @source[@curpos..@curpos+1] == '(:'
+          comment_depth = comment_depth + 1
+          @curpos = @curpos + 1
+          @col = @col + 1
+        end
+        if @source[@curpos..@curpos+1] == ':)'
+          comment_depth = comment_depth - 1
+          @curpos = @curpos + 1
+          @col = @col + 1
+        end
+        @curpos = @curpos + 1
+        @col = @col + 1
+      end
+      white_space = white_space + 1
+    end
+
+    while @curpos < @source.length && @source[@curpos..@curpos] =~ /\s/ do
+      if @source[@curpos..@curpos] =~ /\n/
+        new_line = new_line + 1
+        @line = @line + 1
+        @col = 0
+      else
+        @col = @col + 1
+      end
+      @curpos = @curpos + 1
+      white_space = white_space + 1
+    end
+
+    if @curpos >= @source.length
+      @last_token = nil
+      return [ false, false ]
+    end
+
+#    case @source[@curpos..@curpos]
+#      when '<': @token = [ :LT, '<' ]
+#      when '>': @token = [ :GT, '>' ]
+#      when '[': @token = [ :LB, '[' ]
+#      when ']': @token = [ :RB, ']' ]
+#      when '(': @token = [ :LP, '(' ]
+#      when ')': @token = [ :RP, ')' ]
+#      when '{': @token = [ :LC, '{' ]
+#      when '}': @token = [ :RC, '}' ]
+#      when ':': @token = [ :COLON, ':' ]
+#      when ',': @token = [ :COMMA, ',' ]
+#      when '|': @token = [ :PIPE, '|' ]
+#      when '*': @token = [ :STAR, '*' ]
+#      when '+': @token = [ :PLUS, '+' ]
+#      when '.': @token = [ :DOT,  '.' ]
+#      when '?': @token = [ :QUESTION, '?' ]
+#      when '$': @token = [ :DOLLAR, '$' ]
+#      when '^': @token = [ :CARET, '^' ]
+#    end
+
+    res = @@regex[:simple_tokens].match(@source[@curpos..@source.length-1])
+    if !res.nil?
+      if !res[1].nil?
+        @token = [ @@ops[res[1]], res[1] ]
+      end
+    end
+
+    if @token.nil?
+      # get longest sequence of non-special characters
+      # if it's all digits, report INTEGER
+      # if it's a qname, report QNAME
+      # otherwise, report TEXT
+      @source[@curpos..@source.length-1] =~ /^(((\\.)|[^ \$\^\[\]\{\}\(\):,|*+.?])+)*/
+      text = $1
+      bits = text.split(/\\/)
+      text = bits.join('')
+      @curpos += bits.size - 1
+      @col += bits.size - 1
+      if text.length > 0
+        if @source[@curpos+text.length .. @curpos+text.length] =~ /[*?+\{]/
+# TODO: make sure we backtrack properly if the last character is escaped
+          text = text[0..text.length-2]
+          @token = [ :TEXT, text ]
+        else
+          case text
+            when /^\d+$/: @token = [ :INTEGER, text ]
+            when /^#{@@regex[:ncname]}$/: @token = [ :NCNAME, text ]
+            else @token = [ :TEXT, text ]
+          end
+        end
+      end
+    end
+
+    if @token.nil?
+      puts "Uh oh... we don't know what to do: #{@source[@curpos .. @source.length-1]}"
+    else
+      @curpos += @token[1].length
+      @col += @token[1].length
+    end
+
+    if @token[0] == :LB
+      if @brackets == 0 && @source[@curpos..@source.length-1] =~ /^\s*\d/
+        @token[0] = :LLB
+      end
+      @brackets += 1
+    elsif @token[0] == :RB
+      @brackets -= 1
+    elsif @brackets > 1
+      @token[0] = :CHAR_TEXT
+    end
+
+    return @token
+  end
+...end tokens.racc/module_eval...
+##### State transition tables begin ###
+
+racc_action_table = [
+    11,    14,    17,     6,    19,     7,    11,    14,    17,     8,
+    19,    10,    13,    16,    28,     8,    55,    10,    13,    16,
+    65,    29,    31,    32,    54,     7,    35,    64,    36,    35,
+    42,    36,    34,    46,    35,    34,    36,    35,    45,    36,
+    34,    38,    35,    34,    36,    44,    38,    47,    34,    60,
+    61,    58,    57,    43,    59,    53,    52,    53,    52,    50,
+    41,     4,    26,    56,    25,     4,    23,    21,     4,    66,
+    67,    68,    69,    70 ]
+
+racc_action_check = [
+    22,    22,    22,     1,    22,     1,     3,     3,     3,    22,
+     3,    22,    22,    22,    18,     3,    46,     3,     3,     3,
+    54,    18,    18,    18,    46,    27,    52,    54,    52,    38,
+    27,    38,    52,    32,    53,    38,    53,    19,    31,    19,
+    53,    19,    35,    19,    35,    29,    35,    34,    35,    49,
+    49,    49,    48,    28,    49,    48,    48,    39,    39,    37,
+    25,    17,    11,    47,     8,     7,     6,     4,     0,    58,
+    61,    64,    65,    68 ]
+
+racc_action_pointer = [
+    65,     3,   nil,     2,    64,   nil,    66,    62,    52,   nil,
+   nil,    58,   nil,   nil,   nil,   nil,   nil,    58,     3,    31,
+   nil,   nil,    -4,   nil,   nil,    47,   nil,    23,    34,    26,
+   nil,    19,    17,   nil,    34,    36,   nil,    50,    23,    47,
+   nil,   nil,   nil,   nil,   nil,   nil,     7,    51,    45,    40,
+   nil,   nil,    20,    28,    11,   nil,   nil,   nil,    55,   nil,
+   nil,    56,   nil,   nil,    62,    53,   nil,   nil,    54,   nil,
+   nil ]
+
+racc_action_default = [
+   -11,   -48,    -1,    -3,    -7,   -11,   -48,   -11,   -35,   -31,
+   -32,    -9,    -5,   -33,   -16,   -12,   -34,   -11,   -14,   -48,
+   -15,    -8,    -4,    71,    -2,   -48,   -10,   -48,   -39,   -37,
+   -13,   -41,   -48,   -20,   -48,   -48,   -27,   -48,   -48,   -19,
+    -6,   -36,   -17,   -40,   -38,   -42,   -48,   -48,   -48,   -48,
+   -18,   -21,   -48,   -48,   -48,   -43,   -25,   -26,   -48,   -28,
+   -24,   -48,   -22,   -23,   -48,   -45,   -30,   -29,   -44,   -46,
+   -47 ]
+
+racc_goto_table = [
+    12,     1,    51,    30,    39,    37,    22,    24,    49,   nil,
+   nil,   nil,   nil,   nil,   nil,   nil,    62,    63,    27,    40,
+    48 ]
+
+racc_goto_check = [
+     5,     1,    12,     8,    11,    10,     3,     2,    13,   nil,
+   nil,   nil,   nil,   nil,   nil,   nil,    12,    12,     1,     5,
+    11 ]
+
+racc_goto_pointer = [
+   nil,     1,     0,     1,   nil,    -3,   nil,   nil,   -15,   nil,
+   -14,   -15,   -36,   -28,   nil ]
+
+racc_goto_default = [
+   nil,   nil,     2,     3,     5,   nil,    15,    18,   nil,    20,
+   nil,   nil,    33,   nil,     9 ]
+
+racc_reduce_table = [
+  0, 0, :racc_error,
+  1, 22, :_reduce_1,
+  3, 22, :_reduce_2,
+  1, 23, :_reduce_3,
+  2, 23, :_reduce_4,
+  2, 23, :_reduce_5,
+  3, 23, :_reduce_6,
+  1, 25, :_reduce_7,
+  2, 25, :_reduce_8,
+  1, 26, :_reduce_9,
+  2, 26, :_reduce_10,
+  0, 24, :_reduce_11,
+  2, 24, :_reduce_12,
+  2, 27, :_reduce_13,
+  1, 27, :_reduce_14,
+  1, 28, :_reduce_15,
+  1, 28, :_reduce_16,
+  3, 28, :_reduce_17,
+  3, 28, :_reduce_18,
+  1, 31, :_reduce_none,
+  1, 32, :_reduce_20,
+  2, 32, :_reduce_21,
+  3, 32, :_reduce_22,
+  3, 32, :_reduce_23,
+  3, 33, :_reduce_24,
+  3, 33, :_reduce_25,
+  3, 33, :_reduce_26,
+  0, 34, :_reduce_27,
+  2, 34, :_reduce_28,
+  3, 34, :_reduce_29,
+  3, 34, :_reduce_30,
+  1, 30, :_reduce_31,
+  1, 30, :_reduce_32,
+  1, 30, :_reduce_33,
+  1, 30, :_reduce_34,
+  1, 35, :_reduce_35,
+  3, 35, :_reduce_36,
+  1, 29, :_reduce_37,
+  2, 29, :_reduce_38,
+  1, 29, :_reduce_39,
+  2, 29, :_reduce_40,
+  1, 29, :_reduce_41,
+  2, 29, :_reduce_42,
+  3, 29, :_reduce_43,
+  5, 29, :_reduce_44,
+  4, 29, :_reduce_45,
+  5, 29, :_reduce_46,
+  6, 29, :_reduce_47 ]
+
+racc_reduce_n = 48
+
+racc_shift_n = 71
+
+racc_token_table = {
+  false => 0,
+  :error => 1,
+  :PIPE => 2,
+  :CARET => 3,
+  :DOLLAR => 4,
+  :DOT => 5,
+  :LP => 6,
+  :RP => 7,
+  :LB => 8,
+  :RB => 9,
+  :MINUS => 10,
+  :PLUS => 11,
+  :COLON => 12,
+  :NCNAME => 13,
+  :CHAR_TEXT => 14,
+  :TEXT => 15,
+  :INTEGER => 16,
+  :COMMA => 17,
+  :STAR => 18,
+  :QUESTION => 19,
+  :LLB => 20 }
+
+racc_nt_base = 21
+
+racc_use_result_var = true
+
+Racc_arg = [
+  racc_action_table,
+  racc_action_check,
+  racc_action_default,
+  racc_action_pointer,
+  racc_goto_table,
+  racc_goto_check,
+  racc_goto_default,
+  racc_goto_pointer,
+  racc_nt_base,
+  racc_reduce_table,
+  racc_token_table,
+  racc_shift_n,
+  racc_reduce_n,
+  racc_use_result_var ]
+
+Racc_token_to_s_table = [
+  "$end",
+  "error",
+  "PIPE",
+  "CARET",
+  "DOLLAR",
+  "DOT",
+  "LP",
+  "RP",
+  "LB",
+  "RB",
+  "MINUS",
+  "PLUS",
+  "COLON",
+  "NCNAME",
+  "CHAR_TEXT",
+  "TEXT",
+  "INTEGER",
+  "COMMA",
+  "STAR",
+  "QUESTION",
+  "LLB",
+  "$start",
+  "rules",
+  "anchored_rule",
+  "rule",
+  "left_anchor",
+  "right_anchor",
+  "sequence",
+  "atom",
+  "sequence_qualifiers",
+  "text",
+  "atom_expr",
+  "char_set_expr",
+  "char_set",
+  "char_set_text",
+  "qname" ]
+
+Racc_debug_parser = false
+
+##### State transition tables end #####
+
+# reduce 0 omitted
+
+module_eval(<<'.,.,', 'tokens.racc', 5)
+  def _reduce_1(val, _values, result)
+     result = Fabulator::Grammar::Expr::Token.new; result.add_alternative(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 6)
+  def _reduce_2(val, _values, result)
+     result = val[0]; result.add_alternative(val[2]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 8)
+  def _reduce_3(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 9)
+  def _reduce_4(val, _values, result)
+     result = val[1]; result.anchor_start(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 10)
+  def _reduce_5(val, _values, result)
+     result = val[0]; result.anchor_end(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 11)
+  def _reduce_6(val, _values, result)
+     result = val[1]; result.anchor_start(val[0]); result.anchor_end(val[2]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 13)
+  def _reduce_7(val, _values, result)
+     result = '^' 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 14)
+  def _reduce_8(val, _values, result)
+     result = '^^' 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 16)
+  def _reduce_9(val, _values, result)
+     result = '$' 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 17)
+  def _reduce_10(val, _values, result)
+     result = '$$' 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 19)
+  def _reduce_11(val, _values, result)
+     result = Fabulator::Grammar::Expr::TokenAlternative.new; 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 20)
+  def _reduce_12(val, _values, result)
+     result = val[0]; result.add_sequence(val[1]); 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 22)
+  def _reduce_13(val, _values, result)
+     result = Fabulator::Grammar::Expr::Sequence.new(nil, val[0], val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 23)
+  def _reduce_14(val, _values, result)
+     result = Fabulator::Grammar::Expr::Sequence.new(nil, val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 25)
+  def _reduce_15(val, _values, result)
+     result = Fabulator::Grammar::Expr::Text.new(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 26)
+  def _reduce_16(val, _values, result)
+     result = Fabulator::Grammar::Expr::Any.new 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 27)
+  def _reduce_17(val, _values, result)
+     result = val[1] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 28)
+  def _reduce_18(val, _values, result)
+     result = val[1] 
+    result
+  end
+.,.,
+
+# reduce 19 omitted
+
+module_eval(<<'.,.,', 'tokens.racc', 33)
+  def _reduce_20(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 34)
+  def _reduce_21(val, _values, result)
+     result = Fabulator::Grammar::Expr::CharSet.new; result.universal; result.but_not(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 35)
+  def _reduce_22(val, _values, result)
+     result = val[0].or(val[2]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 36)
+  def _reduce_23(val, _values, result)
+     result = val[0].but_not(val[2]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 38)
+  def _reduce_24(val, _values, result)
+     result = Fabulator::Grammar::Expr::CharSet.new(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 39)
+  def _reduce_25(val, _values, result)
+     result = Fabulator::Grammar::Expr::CharClass.new(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 40)
+  def _reduce_26(val, _values, result)
+     result = val[1] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 42)
+  def _reduce_27(val, _values, result)
+     result = '' 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 43)
+  def _reduce_28(val, _values, result)
+     result = val[0] + val[1] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 44)
+  def _reduce_29(val, _values, result)
+     result = val[0] + '-' + val[2] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 45)
+  def _reduce_30(val, _values, result)
+     result = val[0] + '+' + val[2] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 47)
+  def _reduce_31(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 48)
+  def _reduce_32(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 49)
+  def _reduce_33(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 50)
+  def _reduce_34(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 52)
+  def _reduce_35(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 53)
+  def _reduce_36(val, _values, result)
+     result = val[0] + ':' + val[2] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 55)
+  def _reduce_37(val, _values, result)
+     result = [ :zero_or_more ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 56)
+  def _reduce_38(val, _values, result)
+     result = [ :zero_or_more, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 57)
+  def _reduce_39(val, _values, result)
+     result = [ :one_or_more ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 58)
+  def _reduce_40(val, _values, result)
+     result = [ :one_or_more, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 59)
+  def _reduce_41(val, _values, result)
+     result = [ :zero_or_one ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 60)
+  def _reduce_42(val, _values, result)
+     result = [ :zero_or_one, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 61)
+  def _reduce_43(val, _values, result)
+     result = [ :exact, val[1].to_i ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 62)
+  def _reduce_44(val, _values, result)
+     result = [ :range, val[1].to_i, val[3].to_i ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 63)
+  def _reduce_45(val, _values, result)
+     result = [ :range, val[1], '' ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 64)
+  def _reduce_46(val, _values, result)
+     result = [ :min, :range, val[1], '' ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'tokens.racc', 65)
+  def _reduce_47(val, _values, result)
+     result = [ :min, :range, val[1].to_i, val[3].to_i ] 
+    result
+  end
+.,.,
+
+def _reduce_none(val, _values, result)
+  val[0]
+end
+
+    end   # class TokenParser
+    end   # module Grammar
+  end   # module Fabulator