sportdb-parser 0.7.0 → 0.7.2

data/lib/sportdb/parser/parser_runtime.rb

@@ -0,0 +1,379 @@
+
+# frozen_string_literal: true
+#--
+# Copyright (c) 1999-2006 Minero Aoki
+#
+# This program is free software.
+# You can distribute/modify this program under the same terms of ruby.
+#
+# As a special exception, when this code is copied by Racc
+# into a Racc output file, you may use that output file
+# without restriction.
+#++
+
+
+module Racc
+  class ParseError < StandardError; end
+end
+unless defined?(::ParseError)
+  ParseError = Racc::ParseError # :nodoc:
+end
+
+
+
+module Racc
+  VERSION   = '1.8.1'
+  Version = VERSION
+
+
+  class Parser
+    def _racc_setup
+      @yydebug = false unless self.class::Racc_debug_parser
+      @yydebug = false unless defined?(@yydebug)
+      if @yydebug
+        @racc_debug_out = $stderr unless defined?(@racc_debug_out)
+        @racc_debug_out ||= $stderr
+      end
+      arg = self.class::Racc_arg
+      arg[13] = true if arg.size < 14
+      arg
+    end
+
+    def _racc_init_sysvars
+      @racc_state  = [0]
+      @racc_tstack = []
+      @racc_vstack = []
+
+      @racc_t = nil
+      @racc_val = nil
+
+      @racc_read_next = true
+
+      @racc_user_yyerror = false
+      @racc_error_status = 0
+    end
+
+    # The entry point of the parser. This method is used with #next_token.
+    # If Racc wants to get token (and its value), calls next_token.
+    #
+    # Example:
+    #     def parse
+    #       @q = [[1,1],
+    #             [2,2],
+    #             [3,3],
+    #             [false, '$']]
+    #       do_parse
+    #     end
+    #
+    #     def next_token
+    #       @q.shift
+    #     end
+    def do_parse
+        _racc_do_parse_rb(_racc_setup(), false)
+    end
+
+    # The method to fetch next token.
+    # If you use #do_parse method, you must implement #next_token.
+    #
+    # The format of return value is [TOKEN_SYMBOL, VALUE].
+    # +token-symbol+ is represented by Ruby's symbol by default, e.g. :IDENT
+    # for 'IDENT'.  ";" (String) for ';'.
+    #
+    # The final symbol (End of file) must be false.
+    def next_token
+      raise NotImplementedError, "#{self.class}\#next_token is not defined"
+    end
+
+    def _racc_do_parse_rb(arg, in_debug)
+      action_table, action_check, action_default, action_pointer,
+      _,            _,            _,              _,
+      _,            _,            token_table,    * = arg
+
+      _racc_init_sysvars
+      tok = act = i = nil
+
+      catch(:racc_end_parse) {
+        while true
+          if i = action_pointer[@racc_state[-1]]
+            if @racc_read_next
+              if @racc_t != 0   # not EOF
+                tok, @racc_val = next_token()
+                unless tok      # EOF
+                  @racc_t = 0
+                else
+                  @racc_t = (token_table[tok] or 1)   # error token
+                end
+                racc_read_token(@racc_t, tok, @racc_val) if @yydebug
+                @racc_read_next = false
+              end
+            end
+            i += @racc_t
+            unless i >= 0 and
+                   act = action_table[i] and
+                   action_check[i] == @racc_state[-1]
+              act = action_default[@racc_state[-1]]
+            end
+          else
+            act = action_default[@racc_state[-1]]
+          end
+          while act = _racc_evalact(act, arg)
+            ;
+          end
+        end
+      }
+    end
+
+
+
+    ###
+    ### common
+    ###
+
+    def _racc_evalact(act, arg)
+      action_table, action_check, _, action_pointer,
+      _,            _,            _, _,
+      _,            _,            _, shift_n,
+      reduce_n,     * = arg
+      nerr = 0   # tmp
+
+      if act > 0 and act < shift_n
+        #
+        # shift
+        #
+        if @racc_error_status > 0
+          @racc_error_status -= 1 unless @racc_t <= 1 # error token or EOF
+        end
+        @racc_vstack.push @racc_val
+        @racc_state.push act
+        @racc_read_next = true
+        if @yydebug
+          @racc_tstack.push @racc_t
+          racc_shift @racc_t, @racc_tstack, @racc_vstack
+        end
+
+      elsif act < 0 and act > -reduce_n
+        #
+        # reduce
+        #
+        code = catch(:racc_jump) {
+          @racc_state.push _racc_do_reduce(arg, act)
+          false
+        }
+        if code
+          case code
+          when 1 # yyerror
+            @racc_user_yyerror = true   # user_yyerror
+            return -reduce_n
+          when 2 # yyaccept
+            return shift_n
+          else
+            raise '[Racc Bug] unknown jump code'
+          end
+        end
+
+      elsif act == shift_n
+        #
+        # accept
+        #
+        racc_accept if @yydebug
+        throw :racc_end_parse, @racc_vstack[0]
+
+      elsif act == -reduce_n
+        #
+        # error
+        #
+        case @racc_error_status
+        when 0
+          unless arg[21]    # user_yyerror
+            nerr += 1
+            on_error @racc_t, @racc_val, @racc_vstack
+          end
+        when 3
+          if @racc_t == 0   # is $
+            # We're at EOF, and another error occurred immediately after
+            # attempting auto-recovery
+            throw :racc_end_parse, nil
+          end
+          @racc_read_next = true
+        end
+        @racc_user_yyerror = false
+        @racc_error_status = 3
+        while true
+          if i = action_pointer[@racc_state[-1]]
+            i += 1   # error token
+            if  i >= 0 and
+                (act = action_table[i]) and
+                action_check[i] == @racc_state[-1]
+              break
+            end
+          end
+          throw :racc_end_parse, nil if @racc_state.size <= 1
+          @racc_state.pop
+          @racc_vstack.pop
+          if @yydebug
+            @racc_tstack.pop
+            racc_e_pop @racc_state, @racc_tstack, @racc_vstack
+          end
+        end
+        return act
+
+      else
+        raise "[Racc Bug] unknown action #{act.inspect}"
+      end
+
+      racc_next_state(@racc_state[-1], @racc_state) if @yydebug
+
+      nil
+    end
+
+    def _racc_do_reduce(arg, act)
+      _,          _,            _,            _,
+      goto_table, goto_check,   goto_default, goto_pointer,
+      nt_base,    reduce_table, _,            _,
+      _,          use_result,   * = arg
+
+      state = @racc_state
+      vstack = @racc_vstack
+      tstack = @racc_tstack
+
+      i = act * -3
+      len       = reduce_table[i]
+      reduce_to = reduce_table[i+1]
+      method_id = reduce_table[i+2]
+      void_array = []
+
+      tmp_t = tstack[-len, len] if @yydebug
+      tmp_v = vstack[-len, len]
+      tstack[-len, len] = void_array if @yydebug
+      vstack[-len, len] = void_array
+      state[-len, len]  = void_array
+
+      # tstack must be updated AFTER method call
+      if use_result
+        vstack.push __send__(method_id, tmp_v, vstack, tmp_v[0])
+      else
+        vstack.push __send__(method_id, tmp_v, vstack)
+      end
+      tstack.push reduce_to
+
+      racc_reduce(tmp_t, reduce_to, tstack, vstack) if @yydebug
+
+      k1 = reduce_to - nt_base
+      if i = goto_pointer[k1]
+        i += state[-1]
+        if i >= 0 and (curstate = goto_table[i]) and goto_check[i] == k1
+          return curstate
+        end
+      end
+      goto_default[k1]
+    end
+
+    # This method is called when a parse error is found.
+    #
+    # ERROR_TOKEN_ID is an internal ID of token which caused error.
+    # You can get string representation of this ID by calling
+    # #token_to_str.
+    #
+    # ERROR_VALUE is a value of error token.
+    #
+    # value_stack is a stack of symbol values.
+    # DO NOT MODIFY this object.
+    #
+    # This method raises ParseError by default.
+    #
+    # If this method returns, parsers enter "error recovering mode".
+    def on_error(t, val, vstack)
+      raise ParseError, sprintf("parse error on value %s (%s)",
+                                val.inspect, token_to_str(t) || '?')
+    end
+
+    # Enter error recovering mode.
+    # This method does not call #on_error.
+    def yyerror
+      throw :racc_jump, 1
+    end
+
+    # Exit parser.
+    # Return value is +Symbol_Value_Stack[0]+.
+    def yyaccept
+      throw :racc_jump, 2
+    end
+
+    # Leave error recovering mode.
+    def yyerrok
+      @racc_error_status = 0
+    end
+
+    # For debugging output
+    def racc_read_token(t, tok, val)
+      @racc_debug_out.print 'read    '
+      @racc_debug_out.print tok.inspect, '(', racc_token2str(t), ') '
+      @racc_debug_out.puts val.inspect
+      @racc_debug_out.puts
+    end
+
+    def racc_shift(tok, tstack, vstack)
+      @racc_debug_out.puts "shift   #{racc_token2str tok}"
+      racc_print_stacks tstack, vstack
+      @racc_debug_out.puts
+    end
+
+    def racc_reduce(toks, sim, tstack, vstack)
+      out = @racc_debug_out
+      out.print 'reduce '
+      if toks.empty?
+        out.print ' <none>'
+      else
+        toks.each {|t| out.print ' ', racc_token2str(t) }
+      end
+      out.puts " --> #{racc_token2str(sim)}"
+      racc_print_stacks tstack, vstack
+      @racc_debug_out.puts
+    end
+
+    def racc_accept
+      @racc_debug_out.puts 'accept'
+      @racc_debug_out.puts
+    end
+
+    def racc_e_pop(state, tstack, vstack)
+      @racc_debug_out.puts 'error recovering mode: pop token'
+      racc_print_states state
+      racc_print_stacks tstack, vstack
+      @racc_debug_out.puts
+    end
+
+    def racc_next_state(curstate, state)
+      @racc_debug_out.puts  "goto    #{curstate}"
+      racc_print_states state
+      @racc_debug_out.puts
+    end
+
+    def racc_print_stacks(t, v)
+      out = @racc_debug_out
+      out.print '        ['
+      t.each_index do |i|
+        out.print ' (', racc_token2str(t[i]), ' ', v[i].inspect, ')'
+      end
+      out.puts ' ]'
+    end
+
+    def racc_print_states(s)
+      out = @racc_debug_out
+      out.print '        ['
+      s.each {|st| out.print ' ', st }
+      out.puts ' ]'
+    end
+
+    def racc_token2str(tok)
+      self.class::Racc_token_to_s_table[tok] or
+          raise "[Racc Bug] can't convert token #{tok} to string"
+    end
+
+    # Convert internal ID of token symbol to the string.
+    def token_to_str(t)
+      self.class::Racc_token_to_s_table[t]
+    end
+
+  end
+end

data/lib/sportdb/parser/racc_parser.rb

@@ -5,61 +5,70 @@
 class RaccMatchParser
 
 
-def initialize( txt,  debug: false )
-    ## puts "==> txt:"
-    ## puts txt
- 
-    @tree   = [] 
+
+  def initialize( txt,  debug: false )
+    @tree   = []
     @errors = []
 
-    ### todo:
-    ##  -  pass along debug flag
     lexer = SportDb::Lexer.new( txt, debug: debug )
     ##  note - use tokenize_with_errors and add/collect tokenize errors
     @tokens, @errors = lexer.tokenize_with_errors
     ## pp @tokens
-    
-    ## quick hack - convert to racc format single char literal tokens e.g. '@' etc.
-    @tokens = @tokens.map do |tok|
-                 if tok.size == 1
-                   [tok[0].to_s, tok[0].to_s]
-                 else 
-                   tok
-                 end
-               end
+
   end
 
 
   def debug( value ) @debug = value; end   ### fix: use setter-style e.g. debug=(value) !!!
   def debug?()  @debug == true; end
 
+  ###
+  ###  fix-fix-fix   rename   to _trace
+  ##                   check lexer - add [debug] Parser -  or such!!!
+  ##
   ## debug - trace / print message
   def trace( msg )
-     puts "  [parse] " + msg    if debug?
+    puts " fix-fix-fix use _trace() in parser!!"
+    puts "  [parse] " + msg
   end
 
-
+  def _trace( *args )
+    ## if debug?
+      print "[DEBUG] Parser -- "
+      args.each { |arg| puts args }
+    ## end
+  end
 
 
   def next_token
     tok = @tokens.shift
-    trace( "next_token => #{tok.pretty_inspect}" )
+   _trace( "next_token => #{tok.pretty_inspect}" )
+
+    ##  convert to racc format single char literal tokens e.g. '@' etc.
+    ##                  note - literal token MUST be string (NOT symbol)
+    ##    note - racc expects array with to items
+    ##               -  item[0] is the token id
+    ##               -  item[1] is the token value
+
+    ## note - returns nil for end-of-file !!!
+    tok = [tok.type, tok]   if tok
+
     tok
   end
-  
+
+
 #  on_error do |error_token_id, error_value, value_stack|
 #      puts "Parse error on token: #{error_token_id}, value: #{error_value}"
-#  end  
+#  end
 
-  def parse_with_errors    
-     trace( "start parse:" )
+  def parse_with_errors
+     _trace( "start parse:" )
      do_parse
      [@tree, @errors]
   end
 
   def parse  ## convenience shortcut (ignores errors)
     tree, _ = parse_with_errors
-    tree 
+    tree
   end
 
 
@@ -69,20 +78,15 @@ def initialize( txt,  debug: false )
 
   def on_error(error_token_id, error_value, value_stack)
     ## auto-add error_token (as string)
-    error_token = Racc_token_to_s_table[error_token_id] 
+    error_token = Racc_token_to_s_table[error_token_id]
     args = [error_token, error_token_id, error_value, value_stack]
+
     puts
     puts "!! on parse error:"
     puts "args=#{args.pretty_inspect}"
 
-    @errors << "parse error on token: #{error_token} (#{error_token_id}) with value: #{error_value}, stack: #{value_stack.pretty_inspect}" 
-    ## exit 1  ##   exit for now  -  get and print more info about context etc.!!
+    @errors << "parse error on token: #{error_token} >#{error_value.pretty_inspect}<, stack: #{value_stack.pretty_inspect}"
   end
 
 
-=begin
-on_error do |error_token_id, error_value, value_stack|
-    puts "Parse error on token: #{error_token_id}, value: #{error_value}"
-end
-=end
 end   # class RaccMatchParser