chemruby 0.9.3

data/lib/chem/db/gspan.rb

@@ -0,0 +1,130 @@
+#
+# = gspan.rb - IO modules for gSpan format
+#
+# Copyright::	Copyright (C) 2005-2006
+#		Tadashi Kadowaki <kadowaki@kuicr.kyoto-u.ac.jp>
+#		Nobuya Tanaka <tanaka@kuicr.kyoto-u.ac.jp>
+
+$: << "/home/tanaka/proj/chemruby/lib/"
+$: << "/home/tanaka/proj/chemruby/ext/"
+
+require 'chem'
+
+module Chem
+
+  class GSpan
+
+    FIRST_NODE = /\((\d+)\)/
+    REG_NODE = / (\d+) \((\d+)+.(\d+)\)/
+
+    # Parse one-lined gSpan formatted string and return
+    # molecule object.
+    def self.parse str, name = ""
+      mol = GSpanMolecule.new
+      mol.name = name
+
+      first_atom = GSpanAtom.new(FIRST_NODE.match(str)[1].to_i)
+      mol.nodes.push(first_atom)
+
+      str.scan(REG_NODE) do |s|
+        bond = GSpanBond.new( s[0].to_i )
+        if s[2] == 'f'
+          from_atom = mol.nodes[ s[1].to_i ]
+          to_atom = GSpanAtom.new( s[3].to_i )
+          mol.nodes.push(to_atom)
+        else # s[2] == 'b'
+          from_atom = mol.nodes[ mol.nodes.size-1 ]
+          to_atom = mol.nodes[ s[3].to_i ]
+        end
+        mol.edges.push([bond, from_atom, to_atom ])
+      end
+
+      mol
+    end
+
+    # Save molecule as gSpan formatted file.
+    # Example :
+    #   Chem::GSpan.save(mols , "filename") # mols : an array of molecules
+    def self.save mols, filename
+      open(filename, "w") do |out|
+        mols.each_with_index do |mol, mol_idx|
+          out.puts "t # #{mol_idx} -1 #{mol.name}"
+          mol.nodes.each_with_index do |node, idx|
+            out.puts "v %d %d" % [idx, node.atomic_number]
+          end
+          mol.edges.each_with_index do |(bond, node1, node2), idx|
+            out.puts "e %d %d %d" % [mol.nodes.index(node1), mol.nodes.index(node2), bond.v]
+          end
+          out.puts
+        end
+        
+      end
+    end
+
+  end
+
+  # Concrete class of gSpan molecule
+  class GSpanMolecule
+
+    include Molecule
+    include Enumerable
+
+    def initialize
+      @nodes = []
+      @edges = []
+    end
+
+  end
+
+  class GSpanAtom
+
+    include Atom
+
+    def initialize element
+      @element = Number2Element[element]
+    end
+
+    def self.parse_line line
+      self.new(line.split[2].to_i)
+    end
+
+  end
+
+  class GSpanBond
+
+    include Bond
+
+    def initialize v
+      @v = v
+    end
+    
+  end
+
+  def self.parse_gspan file
+    t = "t"[0]
+    v = "v"[0]
+    e = "e"[0]
+
+    mols = []
+    mol = GSpanMolecule.new
+
+    open(file).each do |line|
+      case line[0]
+      when t
+        mol = GSpanMolecule.new
+        mols.push(mol)
+      when v
+        mol.nodes.push(GSpanAtom.parse_line(line))
+      when e
+        ary = line.split
+        node1 = mol.nodes[ary[1].to_i]
+        node2 = mol.nodes[ary[2].to_i]
+        mol.edges.push([GSpanBond.new(ary[3].to_i), node1, node2])
+      else
+      end
+    end
+    mols
+  end
+
+
+end

data/lib/chem/db/iupac.rb

@@ -0,0 +1,5 @@
+
+require 'chem/db/iupac/a_1'
+require 'chem/db/iupac/postfix'
+require 'chem/db/iupac/iuparser'
+

data/lib/chem/db/iupac/a_1.rb

@@ -0,0 +1,46 @@
+a_1_1 = [1, "meth",
+      2, "eth",
+      3, "prop",
+      4, "but",
+      5, "pent",
+      6, "hex",
+      7, "hept",
+      8, "oct",
+      9, "non",
+      10, "dec",
+      11, "undec",
+      12, "dodec",
+      13, "tridec",
+      14, "tetradec",
+      15, "pentadec",
+      16, "hexadec",
+      17, "heptadec",
+      18, "octadec",
+      19, "nonadec",
+      20, "icos",
+      21, "henicos",
+      23, "tricos",
+      22, "docos",
+      24, "tetracos",
+      25, "pentacos",
+      26, "hexacos",
+      27, "heptacos",
+      28, "octacos",
+      29, "nonacos",
+      30, "triacont",
+      31, "hentriacont",
+      32, "dotriacont",
+      33, "tritriacont",
+      40, "tetracont",
+      50, "pentacont",
+      60, "hexacont",
+      70, "heptacont",
+      80, "octacont",
+      90, "nonacont",
+      100, "hect",
+      132, "dotriacontahect"]
+
+a = Hash[*a_1_1]
+
+$reg_a_1_1 = /undec|tritriacont|tridec|tricos|triacont|tetradec|tetracos|tetracont|prop|pentadec|pentacos|pentacont|pent|octadec|octacos|octacont|oct|nonadec|nonacos|nonacont|non|meth|icos|hexadec|hexacos|hexacont|hex|heptadec|heptacos|heptacont|hept|hentriacont|henicos|hect|eth|dotriacontahect|dotriacont|dodec|docos|dec|but/
+

data/lib/chem/db/iupac/iuparser.rb

@@ -0,0 +1,226 @@
+#
+# DO NOT MODIFY!!!!
+# This file is automatically generated by racc 1.4.4
+# from racc grammer file "iuparser.ry".
+#
+
+require 'racc/parser'
+
+
+class IupacParser < Racc::Parser
+
+module_eval <<'..end iuparser.ry modeval..id57c30dc50e', 'iuparser.ry', 16
+
+def next_token
+  @q.shift
+end
+
+def parse( line )
+  line.downcase!
+  @q = []
+  
+  until line.empty? do
+    case line
+    when /\A(hex|prop|meth|non|but|dec)/
+      @q.push([:SKELETON, $&])
+    when /\A(ane)/
+      @q.push([:POSTFIX, $&])
+    when /\Ayl/
+      @q.push([:YL, 'yl'])
+    when /\A(di|tri)/
+      @q.push([:GREEK_NUM, $&])
+    when /\A\d+/
+      @q.push([:NUMBER, $&.to_i])
+    when /\A-/
+      @q.push([:HYPHEN, '-'])
+    when /\A,/
+      @q.push([:COMMA, ','])
+    when /\A[()]/
+      @q.push([:ROUND_B, $&])
+    end
+    line = $'
+  end
+  @q.push([:EOL, nil])
+  do_parse
+end
+
+# class IupacParser
+# rule
+#        iupac : name EOF
+#                  {
+#                    puts 'The END'
+#                  }
+#         name : SKELETON POSTFIX
+#                  {
+#                    p val
+#                    result = val
+#                  }
+# end
+
+# ---- inner
+
+# def next_token
+#   @q.shift
+# end
+
+# def parse( line )
+#   @q = []
+  
+# #  line.strip
+#   until line.empty? do
+#     case line
+#     when /hex/# $reg_a_1_1
+#       @q.push [:SKELETON, $&]
+#     when /ane/# $reg_postfix
+#       @q.push [:POSTFIX, $&]
+#     when /\A\d+/
+#       @q.push [:NUMBER, $&.to_i ]
+#     when /\A./
+#       @q.push [$&, $&]
+#     else
+#       raise RuntimeError, 'must not happen'
+#     end
+#     line = $'
+#   end
+#   @q.push [ :EOL, nil ]
+#   p @q
+#   do_parse
+# end
+
+..end iuparser.ry modeval..id57c30dc50e
+
+##### racc 1.4.4 generates ###
+
+racc_reduce_table = [
+ 0, 0, :racc_error,
+ 4, 11, :_reduce_none,
+ 0, 12, :_reduce_none,
+ 3, 12, :_reduce_none,
+ 2, 12, :_reduce_none,
+ 3, 13, :_reduce_none,
+ 1, 13, :_reduce_none,
+ 2, 15, :_reduce_none,
+ 3, 15, :_reduce_none,
+ 1, 14, :_reduce_none,
+ 3, 14, :_reduce_none ]
+
+racc_reduce_n = 11
+
+racc_shift_n = 23
+
+racc_action_table = [
+     5,    18,    16,     8,    16,     9,    10,     9,    10,     9,
+    14,    11,    12,    19,    15,    13,     3,    21,    13,    22 ]
+
+racc_action_check = [
+     2,     9,     8,     2,    14,     2,     2,     8,     8,    14,
+     6,     3,     5,    12,     6,     5,     1,    15,    16,    18 ]
+
+racc_action_pointer = [
+   nil,    16,    -2,    11,   nil,     9,     5,   nil,     0,    -1,
+   nil,   nil,     9,   nil,     2,     9,    12,   nil,    13,   nil,
+   nil,   nil,   nil ]
+
+racc_action_default = [
+    -2,   -11,   -11,   -11,    -4,   -11,   -11,    -6,   -11,   -11,
+    -9,    23,   -11,    -7,   -11,   -11,   -11,    -3,   -11,    -1,
+    -5,   -10,    -8 ]
+
+racc_goto_table = [
+     4,     1,     2,    20,   nil,   nil,    17 ]
+
+racc_goto_check = [
+     3,     1,     2,     5,   nil,   nil,     3 ]
+
+racc_goto_pointer = [
+   nil,     1,     2,    -2,   nil,   -11 ]
+
+racc_goto_default = [
+   nil,   nil,   nil,   nil,     6,     7 ]
+
+racc_token_table = {
+ false => 0,
+ Object.new => 1,
+ :SKELETON => 2,
+ :POSTFIX => 3,
+ :EOL => 4,
+ :HYPHEN => 5,
+ :YL => 6,
+ :GREEK_NUM => 7,
+ :NUMBER => 8,
+ :COMMA => 9 }
+
+racc_use_result_var = true
+
+racc_nt_base = 10
+
+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',
+'SKELETON',
+'POSTFIX',
+'EOL',
+'HYPHEN',
+'YL',
+'GREEK_NUM',
+'NUMBER',
+'COMMA',
+'$start',
+'name',
+'prefix',
+'num_subs',
+'nums',
+'substituent']
+
+Racc_debug_parser = false
+
+##### racc system variables end #####
+
+ # reduce 0 omitted
+
+ # reduce 1 omitted
+
+ # reduce 2 omitted
+
+ # reduce 3 omitted
+
+ # reduce 4 omitted
+
+ # reduce 5 omitted
+
+ # reduce 6 omitted
+
+ # reduce 7 omitted
+
+ # reduce 8 omitted
+
+ # reduce 9 omitted
+
+ # reduce 10 omitted
+
+ def _reduce_none( val, _values, result )
+  result
+ end
+
+end   # class IupacParser
+
+
+# parser = IupacParser.new
+
+# parser.parse("hexane")

data/lib/chem/db/iupac/iuparser.ry

@@ -0,0 +1,97 @@
+class IupacParser
+rule
+        name : prefix SKELETON POSTFIX EOL
+      prefix : 
+             |  prefix HYPHEN num_subs
+             |  prefix num_subs
+    num_subs : nums HYPHEN substituent
+             | substituent
+ substituent : SKELETON YL
+             | GREEK_NUM SKELETON YL
+        nums : NUMBER
+             | nums COMMA NUMBER
+end
+
+---- inner
+
+def next_token
+  @q.shift
+end
+
+def parse( line )
+  line.downcase!
+  @q = []
+  
+  until line.empty? do
+    case line
+    when /\A(hex|prop|meth|non|but|dec)/
+      @q.push([:SKELETON, $&])
+    when /\A(ane)/
+      @q.push([:POSTFIX, $&])
+    when /\Ayl/
+      @q.push([:YL, 'yl'])
+    when /\A(di|tri)/
+      @q.push([:GREEK_NUM, $&])
+    when /\A\d+/
+      @q.push([:NUMBER, $&.to_i])
+    when /\A-/
+      @q.push([:HYPHEN, '-'])
+    when /\A,/
+      @q.push([:COMMA, ','])
+    when /\A[()]/
+      @q.push([:ROUND_B, $&])
+    end
+    line = $'
+  end
+  @q.push([:EOL, nil])
+  do_parse
+end
+
+# class IupacParser
+# rule
+#        iupac : name EOF
+#                  {
+#                    puts 'The END'
+#                  }
+#         name : SKELETON POSTFIX
+#                  {
+#                    p val
+#                    result = val
+#                  }
+# end
+
+# ---- inner
+
+# def next_token
+#   @q.shift
+# end
+
+# def parse( line )
+#   @q = []
+  
+# #  line.strip
+#   until line.empty? do
+#     case line
+#     when /hex/# $reg_a_1_1
+#       @q.push [:SKELETON, $&]
+#     when /ane/# $reg_postfix
+#       @q.push [:POSTFIX, $&]
+#     when /\A\d+/
+#       @q.push [:NUMBER, $&.to_i ]
+#     when /\A./
+#       @q.push [$&, $&]
+#     else
+#       raise RuntimeError, 'must not happen'
+#     end
+#     line = $'
+#   end
+#   @q.push [ :EOL, nil ]
+#   p @q
+#   do_parse
+# end
+
+---- footer
+
+# parser = IupacParser.new
+
+# parser.parse("hexane")