fabulator-grammar 0.0.1

data/lib/fabulator/grammar/expr/any.rb

@@ -0,0 +1,7 @@
+module Fabulator::Grammar::Expr
+  class Any
+    def to_regex
+      %r{.}
+    end
+  end
+end

data/lib/fabulator/grammar/expr/char_set.rb

@@ -0,0 +1,45 @@
+module Fabulator::Grammar::Expr
+  class CharSet
+    def initialize(cs)
+      chars = ""
+      ranges = ""
+      if cs[0..0] == '-'
+        chars = '-'
+        cs = cs[1..cs.length-1]
+      end
+      bits = cs.split(/-/) # to pull out ranges
+      if bits.size == 1
+        chars = bits[0]
+      else
+        if bits[0].size > 1
+          chars += b[0..0]
+        end
+        while(bits.size > 1)
+          b = bits.shift
+          if b.size > 2
+            chars += b[1..b.size-2]
+          end
+          ranges += Regexp.quote(b[b.size-1 .. b.size-1]) + '-' + Regexp.quote(bits[0][0..0])
+        end
+        if bits[0].size > 1
+          chars += bits[0][1..bits[0].size-1]
+        end
+      end
+      chars = chars.collect{ |cc| Regexp.quote(cc) }.join('')
+      @set = chars + ranges
+      @inverted = false
+    end
+
+    def inverted
+      @inverted = true
+    end
+
+    def to_regex
+      if @set != ''
+        Regexp.compile("[" + (@inverted ? '^' : '') + @set + "]")
+      else
+        %r{}
+      end
+    end
+  end
+end

data/lib/fabulator/grammar/expr/rule.rb

@@ -0,0 +1,36 @@
+module Fabulator::Grammar::Expr
+  class Rule
+    def initialize
+      @sequences = [ ]
+      @anchor_start = false
+      @anchor_end = false
+    end
+
+    def anchor_start
+      @anchor_start = true
+    end
+
+    def anchor_end
+      @anchor_end = true
+    end
+
+    def add_sequence(s)
+      @sequences << s
+    end
+
+    def to_regex
+      r = %r{#{@sequences.collect{ |s| s.to_regex }}}
+      if @anchor_start
+        if @anchor_end
+          %r{^#{r}$}
+        else
+          %r{^#{r}}
+        end
+      elsif @anchor_end
+        %r{#{r}$}
+      else
+        r
+      end
+    end
+  end
+end

data/lib/fabulator/grammar/expr/rule_ref.rb

@@ -0,0 +1,13 @@
+module Fabulator::Grammar::Expr
+  class RuleRef
+    def initialize(qname)
+      bits = qname.split(/:/,2)
+      @ns_prefix = bits[0]
+      @name = bits[1]
+    end
+
+    def to_regex
+      %r{}
+    end
+  end
+end

data/lib/fabulator/grammar/expr/rules.rb

@@ -0,0 +1,15 @@
+module Fabulator::Grammar::Expr
+  class Rules
+    def initialize
+      @alternatives = [ ]
+    end
+
+    def add_alternative(a)
+      @alternatives << a
+    end
+
+    def to_regex
+      Regexp.union(@alternatives.collect{|a| a.to_regex })
+    end
+  end
+end

data/lib/fabulator/grammar/expr/sequence.rb

@@ -0,0 +1,26 @@
+module Fabulator::Grammar::Expr
+  class Sequence
+    def initialize(sub_seq, count = [ :one ])
+      @sub_sequence = sub_seq
+      @modifiers = ''
+      case (count - [:min]).first
+        when :zero_or_one: @modifiers = '?'
+        when :one_or_more: @modifiers = '+'
+        when :zero_or_more: @modifiers = '*'
+        when :exact:      
+          amt = count.select{ |c| !c.is_a?(Symbol) }
+          @modifiers = '{' + amt.first.to_s + '}'
+        when :range:
+          ends = count.select{ |c| !c.is_a?(Symbol) }
+          @modifiers = '{'+ends.join(",")+'}'
+      end
+      if count.include?(:min) && @modifiers != ''
+        @modifiers += "?"
+      end
+    end
+
+    def to_regex
+      s = %r{#{@sub_sequence.to_regex}#{@modifiers}}
+    end
+  end
+end

data/lib/fabulator/grammar/expr/text.rb

@@ -0,0 +1,11 @@
+module Fabulator::Grammar::Expr
+  class Text
+    def initialize(t)
+      @text = t
+    end
+
+    def to_regex
+      Regexp.escape(@text)
+    end
+  end
+end

data/lib/fabulator/grammar/parser.rb

@@ -0,0 +1,548 @@
+#
+# 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 Parser < Racc::Parser
+
+module_eval(<<'...end regex.racc/module_eval...', 'regex.racc', 48)
+  require 'fabulator/grammar'
+
+  def parse(t, ctx)
+    @source = t
+    @curpos = 0 
+    @context = ctx
+    @line = 0   
+
+    @yydebug = true
+      
+    @last_token = nil
+  
+    do_parse
+  end
+    
+  def on_error(*args)
+    raise Fabulator::Grammar::ParserError.new("unable to parse '#{args[1]}' near line #{@line + 1}, column #{@col}")
+  end
+
+  @@regex = {
+    :ncname => %r{(?:[a-zA-Z_][-a-zA-Z0-9_.]*)}
+  }
+
+  @@regex[:qname] = %r{((?:#{@@regex[:ncname]}:)?#{@@regex[:ncname]})}
+
+  def next_token
+    @token = nil
+    white_space = 0
+    new_line = 0
+    @col = 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
+
+    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 = $1
+      bits = text.split(/\\/)
+      text = bits.join('')
+      @curpos += bits.size - 1
+      if text.length > 0
+        if @source[@curpos+text.length .. @curpos+text.length] =~ /[*?+\{]/
+          text = text[0..text.length-2]
+          @token = [ :TEXT, text ]
+        else
+          case text
+            when /^\d+$/: @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
+    end
+
+    return @token
+  end
+...end regex.racc/module_eval...
+##### State transition tables begin ###
+
+racc_action_table = [
+    31,    45,    25,    26,    28,    29,    35,     6,    46,     8,
+     9,    10,    12,    13,    39,    16,    17,     5,    19,     6,
+     8,     9,    10,    22,    13,    10,    16,    17,    33,    19,
+    34,     8,     9,    10,     8,     9,    10,    42,    43,     3,
+    36,    37,    38,    23,     3,    41,    20,    44,     3,    47,
+    48,    49 ]
+
+racc_action_check = [
+    19,    43,    14,    14,    14,    14,    25,    30,    43,    19,
+    19,    19,     4,     4,    30,     4,     4,     1,     4,     1,
+     4,     4,     4,     7,     7,    13,     7,     7,    23,     7,
+    24,     7,     7,     7,    31,    31,    31,    38,    38,    17,
+    26,    28,    29,    10,     6,    32,     5,    40,     0,    45,
+    46,    47 ]
+
+racc_action_pointer = [
+    45,    17,   nil,   nil,     8,    46,    41,    19,   nil,   nil,
+    28,   nil,   nil,    11,   -14,   nil,   nil,    36,   nil,    -3,
+   nil,   nil,   nil,    14,    24,   -11,    23,   nil,    24,    29,
+     5,    22,    34,   nil,   nil,   nil,   nil,   nil,    17,   nil,
+    36,   nil,   nil,   -12,   nil,    29,    33,    34,   nil,   nil ]
+
+racc_action_default = [
+    -7,   -33,    -1,    -7,    -3,   -33,    -7,    -4,   -18,   -19,
+   -20,    -8,    -5,   -33,   -10,   -17,   -13,    -7,   -12,   -33,
+    50,    -2,    -6,   -33,   -33,   -22,   -26,    -9,   -24,   -33,
+   -33,   -33,   -33,   -21,   -11,   -23,   -27,   -25,   -33,   -14,
+   -33,   -15,   -28,   -33,   -16,   -33,   -30,   -29,   -31,   -32 ]
+
+racc_goto_table = [
+     1,    32,     7,    21,    24,    27,   nil,   nil,   nil,   nil,
+   nil,   nil,   nil,    40,   nil,   nil,   nil,    30 ]
+
+racc_goto_check = [
+     1,     8,     3,     2,     7,     6,   nil,   nil,   nil,   nil,
+   nil,   nil,   nil,     8,   nil,   nil,   nil,     1 ]
+
+racc_goto_pointer = [
+   nil,     0,    -3,    -1,   nil,   nil,    -9,    -9,   -18 ]
+
+racc_goto_default = [
+   nil,   nil,     2,     4,    11,    14,   nil,    15,    18 ]
+
+racc_reduce_table = [
+  0, 0, :racc_error,
+  1, 23, :_reduce_1,
+  3, 23, :_reduce_2,
+  1, 24, :_reduce_3,
+  2, 24, :_reduce_4,
+  2, 24, :_reduce_5,
+  3, 24, :_reduce_6,
+  0, 25, :_reduce_7,
+  2, 25, :_reduce_8,
+  2, 26, :_reduce_9,
+  1, 26, :_reduce_10,
+  3, 27, :_reduce_11,
+  1, 27, :_reduce_12,
+  1, 27, :_reduce_13,
+  3, 27, :_reduce_14,
+  3, 27, :_reduce_15,
+  4, 27, :_reduce_16,
+  1, 30, :_reduce_17,
+  1, 30, :_reduce_18,
+  1, 30, :_reduce_19,
+  1, 29, :_reduce_20,
+  3, 29, :_reduce_21,
+  1, 28, :_reduce_22,
+  2, 28, :_reduce_23,
+  1, 28, :_reduce_24,
+  2, 28, :_reduce_25,
+  1, 28, :_reduce_26,
+  2, 28, :_reduce_27,
+  3, 28, :_reduce_28,
+  5, 28, :_reduce_29,
+  4, 28, :_reduce_30,
+  5, 28, :_reduce_31,
+  6, 28, :_reduce_32 ]
+
+racc_reduce_n = 33
+
+racc_shift_n = 50
+
+racc_token_table = {
+  false => 0,
+  :error => 1,
+  :PIPE => 2,
+  :CARET => 3,
+  :DOLLAR => 4,
+  :LT => 5,
+  :GT => 6,
+  :DOT => 7,
+  :LP => 8,
+  :RP => 9,
+  :LB => 10,
+  :RB => 11,
+  :TEXT => 12,
+  :INTEGER => 13,
+  :NCNAME => 14,
+  :COLON => 15,
+  :STAR => 16,
+  :QUESTION => 17,
+  :PLUS => 18,
+  :LC => 19,
+  :RC => 20,
+  :COMMA => 21 }
+
+racc_nt_base = 22
+
+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",
+  "LT",
+  "GT",
+  "DOT",
+  "LP",
+  "RP",
+  "LB",
+  "RB",
+  "TEXT",
+  "INTEGER",
+  "NCNAME",
+  "COLON",
+  "STAR",
+  "QUESTION",
+  "PLUS",
+  "LC",
+  "RC",
+  "COMMA",
+  "$start",
+  "rules",
+  "anchored_rule",
+  "rule",
+  "sequence",
+  "sub_sequence",
+  "sequence_qualifiers",
+  "qname",
+  "text" ]
+
+Racc_debug_parser = false
+
+##### State transition tables end #####
+
+# reduce 0 omitted
+
+module_eval(<<'.,.,', 'regex.racc', 5)
+  def _reduce_1(val, _values, result)
+     result = Fabulator::Grammar::Expr::Rules.new; result.add_alternative(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 6)
+  def _reduce_2(val, _values, result)
+     result = val[0]; result.add_alternative(val[2]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 8)
+  def _reduce_3(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 9)
+  def _reduce_4(val, _values, result)
+     result = val[1]; result.anchor_start 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 10)
+  def _reduce_5(val, _values, result)
+     result = val[0]; result.anchor_end 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 11)
+  def _reduce_6(val, _values, result)
+     result = val[1]; result.anchor_start; result.anchor_end 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 13)
+  def _reduce_7(val, _values, result)
+     result = Fabulator::Grammar::Expr::Rule.new; 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 14)
+  def _reduce_8(val, _values, result)
+     result = val[0]; result.add_sequence(val[1]); 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 16)
+  def _reduce_9(val, _values, result)
+     result = Fabulator::Grammar::Expr::Sequence.new(val[0], val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 17)
+  def _reduce_10(val, _values, result)
+     result = Fabulator::Grammar::Expr::Sequence.new(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 19)
+  def _reduce_11(val, _values, result)
+     result = Fabulator::Grammar::Expr::RuleRef.new(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 20)
+  def _reduce_12(val, _values, result)
+     result = Fabulator::Grammar::Expr::Text.new(val[0]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 21)
+  def _reduce_13(val, _values, result)
+     result = Fabulator::Grammar::Expr::Any.new 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 22)
+  def _reduce_14(val, _values, result)
+     result = val[1] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 23)
+  def _reduce_15(val, _values, result)
+     result = Fabulator::Grammar::Expr::CharSet.new(val[1]) 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 24)
+  def _reduce_16(val, _values, result)
+     result = Fabulator::Grammar::Expr::CharSet.new(val[2]); result.inverted 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 26)
+  def _reduce_17(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 27)
+  def _reduce_18(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 28)
+  def _reduce_19(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 30)
+  def _reduce_20(val, _values, result)
+     result = val[0] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 31)
+  def _reduce_21(val, _values, result)
+     result = val[0] + ':' + val[2] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 33)
+  def _reduce_22(val, _values, result)
+     result = [ :zero_or_more ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 34)
+  def _reduce_23(val, _values, result)
+     result = [ :zero_or_more, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 35)
+  def _reduce_24(val, _values, result)
+     result = [ :one_or_more ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 36)
+  def _reduce_25(val, _values, result)
+     result = [ :one_or_more, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 37)
+  def _reduce_26(val, _values, result)
+     result = [ :zero_or_one ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 38)
+  def _reduce_27(val, _values, result)
+     result = [ :zero_or_one, :min ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 39)
+  def _reduce_28(val, _values, result)
+     result = [ :exact, val[1].to_i ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 40)
+  def _reduce_29(val, _values, result)
+     result = [ :range, val[1].to_i, val[3].to_i ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 41)
+  def _reduce_30(val, _values, result)
+     result = [ :range, val[1], '' ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 42)
+  def _reduce_31(val, _values, result)
+     result = [ :min, :range, val[1], '' ] 
+    result
+  end
+.,.,
+
+module_eval(<<'.,.,', 'regex.racc', 43)
+  def _reduce_32(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 Parser
+    end   # module Grammar
+  end   # module Fabulator