at_coder_friends 0.6.2 → 0.6.7

data/lib/at_coder_friends/generator/main.rb

@@ -14,8 +14,14 @@ module AtCoderFriends
       def process(pbm)
         generators = ctx.config.dig('generators') || []
         generators.each do |gen_name|
-          gen_obj = load_obj(gen_name)
-          gen_obj.process(pbm)
+          begin
+            gen_obj = load_obj(gen_name)
+            gen_obj.process(pbm)
+          rescue StandardError => e
+            puts "an error occurred in generator:#{gen_name}."
+            puts e.to_s
+            puts e.backtrace
+          end
         end
       end
 

data/lib/at_coder_friends/generator/ruby_builtin.rb

@@ -3,43 +3,20 @@
 module AtCoderFriends
   module Generator
     # generates Ruby source from problem description
-    class RubyBuiltin
+    class RubyBuiltin < Base
       ACF_HOME = File.realpath(File.join(__dir__, '..', '..', '..'))
       TMPL_DIR = File.join(ACF_HOME, 'templates')
-      DEFAULT_TMPL = File.join(TMPL_DIR, 'ruby_builtin_default.rb')
-      INTERACTIVE_TMPL = File.join(TMPL_DIR, 'ruby_builtin_interactive.rb')
+      DEFAULT_TMPL = File.join(TMPL_DIR, 'ruby_builtin.rb.erb')
+      ATTRS = Attributes.new(:rb, DEFAULT_TMPL)
 
-      attr_reader :cfg, :pbm
-
-      def initialize(cfg = nil)
-        @cfg = cfg || {}
-      end
-
-      def process(pbm)
-        src = generate(pbm)
-        pbm.add_src(:rb, src)
-      end
-
-      def generate(pbm)
-        @pbm = pbm
-        File
-          .read(select_template)
-          .gsub('### URL ###', pbm.url)
-          .gsub('### CONSTS ###', gen_consts.join("\n"))
-          .gsub('### DCLS ###', gen_decls.join("\n"))
-          .gsub('### OUTPUT ###', gen_output)
+      def attrs
+        ATTRS
       end
 
-      def select_template(interactive = pbm.options.interactive)
-        interactive ? interactive_template : default_template
-      end
-
-      def default_template
-        cfg['default_template'] || DEFAULT_TMPL
-      end
-
-      def interactive_template
-        cfg['interactive_template'] || INTERACTIVE_TMPL
+      def render(src)
+        src = embed_lines(src, '### CONSTS ###', gen_consts)
+        src = embed_lines(src, '### DCLS ###', gen_decls)
+        src
       end
 
       def gen_consts(constants = pbm.constants)
@@ -64,35 +41,45 @@ module AtCoderFriends
         when :harray
           gen_harray_decl(inpdef)
         when :varray
-          if inpdef.names.size == 1
-            gen_varray_1_decl(inpdef)
-          else
-            gen_varray_n_decl(inpdef)
-          end
+          gen_varray_decl(inpdef)
         when :matrix
           gen_matrix_decl(inpdef)
+        when :varray_matrix, :matrix_varray
+          gen_cmb_decl(inpdef)
+        when :vmatrix
+          gen_vmatrix_decl(inpdef)
+        when :hmatrix
+          gen_hmatrix_decl(inpdef)
         end
       end
 
       def gen_single_decl(inpdef)
         names = inpdef.names
         dcl = names.join(', ')
-        expr = gen_expr(inpdef.item, names.size > 1)
+        expr = gen_expr(inpdef, names.size > 1)
         "#{dcl} = #{expr}"
       end
 
       def gen_harray_decl(inpdef)
         v = inpdef.names[0]
         dcl = "#{v}s"
-        expr = gen_expr(inpdef.item, true)
+        expr = gen_expr(inpdef, true)
         "#{dcl} = #{expr}"
       end
 
+      def gen_varray_decl(inpdef)
+        if inpdef.names.size == 1
+          gen_varray_1_decl(inpdef)
+        else
+          gen_varray_n_decl(inpdef)
+        end
+      end
+
       def gen_varray_1_decl(inpdef)
         v = inpdef.names[0]
         sz = inpdef.size[0]
         dcl = "#{v}s"
-        expr = gen_expr(inpdef.item, false)
+        expr = gen_expr(inpdef, false)
         "#{dcl} = Array.new(#{sz}) { #{expr} }"
       end
 
@@ -100,7 +87,7 @@ module AtCoderFriends
         names = inpdef.names
         sz = inpdef.size[0]
         dcl = names.map { |v| "#{v}s[i]" }.join(', ')
-        expr = gen_expr(inpdef.item, true)
+        expr = gen_expr(inpdef, true)
         ret = []
         ret += names.map { |v| "#{v}s = Array.new(#{sz})" }
         ret << "#{sz}.times do |i|"
@@ -113,28 +100,87 @@ module AtCoderFriends
         v = inpdef.names[0]
         sz = inpdef.size[0]
         decl = "#{v}ss"
-        expr = gen_expr(inpdef.item, true)
+        expr = gen_expr(inpdef, true)
         "#{decl} = Array.new(#{sz}) { #{expr} }"
       end
 
-      def gen_expr(item, split)
-        case item
+      def gen_cmb_decl(inpdef)
+        mx = inpdef.container == :varray_matrix ? -1 : 0
+        vs = inpdef.names.map { |v| "#{v}s" }
+        vs[mx] += 's'
+        sz = inpdef.size[0]
+        dcls = vs.map { |v| "#{v}[i]" }
+        dcls[mx] = '*' + dcls[mx] unless inpdef.item == :char
+        dcl = dcls.join(', ')
+        expr = gen_cmb_expr(inpdef)
+        ret = []
+        ret += vs.map { |v| "#{v} = Array.new(#{sz})" }
+        ret << "#{sz}.times do |i|"
+        ret << "  #{dcl} = #{expr}"
+        ret << 'end'
+        ret
+      end
+
+      def gen_vmatrix_decl(inpdef)
+        names = inpdef.names
+        sz1, sz2 = inpdef.size
+        dcl = names.map { |v| "#{v}ss[i][j]" }.join(', ')
+        expr = gen_expr(inpdef, true)
+        ret = []
+        ret += names.map do |v|
+          "#{v}ss = Array.new(#{sz1}) { Array.new(#{sz2}) }"
+        end
+        ret << "#{sz1}.times do |i|"
+        ret << "  #{sz2}.times do |j|"
+        ret << "    #{dcl} = #{expr}"
+        ret << '  end'
+        ret << 'end'
+        ret
+      end
+
+      def gen_hmatrix_decl(inpdef)
+        names = inpdef.names
+        sz = inpdef.size[0]
+        dcl = names.map { |v| "#{v}ss[i]" }.join(', ')
+        expr = gen_expr(inpdef, true)
+        ret = []
+        ret += names.map { |v| "#{v}ss = Array.new(#{sz})" }
+        ret << "#{sz}.times do |i|"
+        ret << "  #{dcl} = #{expr}.each_slice(#{names.size}).to_a.transpose"
+        ret << 'end'
+        ret
+      end
+
+      def gen_expr(inpdef, split)
+        read = gen_read(inpdef.delim)
+        case inpdef.item
         when :number
-          split ? 'gets.split.map(&:to_i)' : 'gets.to_i'
+          split ? "#{read}.split.map(&:to_i)" : "#{read}.to_i"
+        when :decimal
+          split ? "#{read}.split.map(&:to_f)" : "#{read}.to_f"
         when :string
-          split ? 'gets.chomp.split' : 'gets.chomp'
+          split ? "#{read}.chomp.split" : "#{read}.chomp"
         when :char
           'gets.chomp'
         end
       end
 
-      def gen_output(vs = pbm.options.binary_values)
-        if vs
-          "puts cond ? '#{vs[0]}' : '#{vs[1]}'"
-        else
-          'puts ans'
+      def gen_cmb_expr(inpdef)
+        read = gen_read(inpdef.delim)
+        case inpdef.item
+        when :number
+          "#{read}.split.map(&:to_i)"
+        when :decimal
+          "#{read}.split.map(&:to_f)"
+        when :string, :char
+          "#{read}.chomp.split"
         end
       end
+
+      def gen_read(delim)
+        sub = delim.chars.map { |d| ".gsub('#{d}', ' ')" }.join
+        "gets#{sub}"
+      end
     end
   end
 end

data/lib/at_coder_friends/parser/constraints.rb

@@ -64,6 +64,7 @@ module AtCoderFriends
           .gsub('\\lvert', '|')
           .gsub('\\rvert', '|')
           .gsub('\\mathit', '')
+          .gsub('\\mathrm', '')
           .gsub('\\times', '*')
           .gsub(/\\begin(\{[^{}]*\})*/, '')
           .gsub(/\\end(\{[^{}]*\})*/, '')

data/lib/at_coder_friends/parser/input_format.rb

@@ -2,195 +2,463 @@
 
 module AtCoderFriends
   module Parser
-    InputFormatMatcher = Struct.new(:container, :item, :pat, :gen_names, :gen_pat2) do
-      attr_reader :names, :size
+    module InputFormatConstants
+      SECTIONS = [Problem::SECTION_IN_FMT, Problem::SECTION_IO_FMT].freeze
+      DELIMS = %w[- / :].freeze
+      RE_SINGLE = /[A-Za-z{][A-Za-z_0-9{}]*/.freeze
+      RE_ITEM = /\{*[A-Za-z]+(?:_[A-Za-z]+)*\}*/.freeze
+      RE_0 = /[0-9]+/.freeze
+      RE_00 = /[01][,_]?[01]/.freeze
+      RE_99 = /[0-9]+[,_]?[0-9]+/.freeze
+      ADD_TAG = ->(tag, re) { /(?<#{tag}>#{re})/ }
+      TO_SUFFIX = ->(re) { /(_\{*#{re}\}*|\{+#{re}\}+)/ }
+      TO_SUFFIX_STR = ->(str) { "(_\\{*#{str}\\}*|\\{+#{str}\\}+)" }
+      RE_IX = TO_SUFFIX[/\S+?/]
+      RE_IX_0 = TO_SUFFIX[ADD_TAG['ix0', RE_0]]
+      RE_IX_00 = TO_SUFFIX[ADD_TAG['ix0', RE_00]]
+      RE_IX_99 = TO_SUFFIX[RE_99]
+      RE_SZ = TO_SUFFIX[ADD_TAG['sz', /\S+?/]]
+      RE_SZ_0 = TO_SUFFIX[ADD_TAG['sz', RE_0]]
+      RE_SZ_00 = TO_SUFFIX[ADD_TAG['sz', RE_00]]
+      RE_SZ_99 = TO_SUFFIX[ADD_TAG['sz', RE_99]]
+      RE_SZ_REF = TO_SUFFIX_STR['\k<sz>']
+      RE_SZ2_0 = TO_SUFFIX[ADD_TAG['sz2', RE_0]]
+      RE_SZ2_REF = TO_SUFFIX_STR['\k<sz2>']
+      RE_BLOCK = /(?<bl>\{(?:[^{}]|\g<bl>)*\})/.freeze
+      DIMENSION_TBL = {
+        single: 0,
+        varray: 1,
+        harray: 1,
+        matrix: 2,
+        varray_matrix: 2,
+        matrix_varray: 2,
+        vmatrix: 2,
+        hmatrix: 2
+      }.freeze
+    end
+
+    # utilities for input format parser
+    module InputFormatUtils
+      include InputFormatConstants
+
+      # 1) &npsp;, fill-width space -> half width space
+      # 2) {i, j}->{i,j} for nested {}
+      def normalize_fmt(str)
+        str
+          .tr('０-９Ａ-Ｚａ-ｚ', '0-9A-Za-z')
+          .gsub(/[[:space:]]/) { |c| c.gsub(/[^\n]/, ' ') } # 1)
+          .gsub(%r{<var>([^<>]+)</var>}i, '\1') # <sub><var>N</var></sub>
+          .gsub(%r{<sup>([^<>]+)</sup>}i, '^\1')
+          .gsub(%r{<sub>([^<>]+)</sub>}i, '_{\1}')
+          .gsub(%r{<sub>([^<>]+)</sub>}i, '_{\1}') # for nested<sub>
+          .gsub(/<("[^"]*"|'[^']*'|[^'"<>])*>/, '')
+          .gsub('&amp;', '&')
+          .gsub('&gt;', '>')
+          .gsub('&lt;', '<')
+          .gsub('\\ ', ' ')
+          .gsub(/\\hspace\{\d+pt\}/, ' ')
+          .gsub('\\(', '')
+          .gsub('\\)', '')
+          .gsub('\\lvert', '|')
+          .gsub('\\rvert', '|')
+          .gsub('\\mathit', '')
+          .gsub('\\mathrm', '')
+          .gsub('\\times', '*')
+          .gsub(/\\begin(\{[^{}]*\})*/, '')
+          .gsub(/\\end(\{[^{}]*\})*/, '')
+          .gsub(/\\[cdlv]?dots/, '..')
+          .gsub(/\{\}/, ' ')
+          .gsub('−', '-') # full width hyphen
+          .gsub(/[・．：‥⋮︙…]+/, '..')
+          .gsub(/[\\$']/, '') # s' -> s
+          .gsub(/[&~|]/, ' ') # |S| -> S
+          .gsub(/^\s*[.:][\s.:]*$/, '..')
+          .tr('()', '{}')
+          .gsub(/#{RE_BLOCK}/) { |w| w.delete(' ') } # 2)
+          .split("\n")
+          .map(&:strip)
+      end
+
+      def extract_delim(str)
+        # a-b, a/b, a:b -> a b
+        str = str.dup
+        dlms =
+          DELIMS.select { |c| str.gsub!(/#{c}(#{RE_SINGLE})/, ' \1') }.join
+        [str, dlms]
+      end
+
+      def normalize_name(s)
+        s.delete('{},').gsub(/(\A_+|_+\z)/, '')
+      end
+
+      def normalize_names(names)
+        names.map { |nm| normalize_name(nm) }
+      end
+
+      def normalize_size(container, size, ix0)
+        sz = size_array(container, size)
+        sz0 = size_array(container, ix0)
+
+        sz.map.with_index do |s, i|
+          if sz0[i] == '0'
+            # 0 -> 1,  N-1 -> N, N-2 -> N-1 if 0 origin
+            s.gsub(/\A0\z/, '1').gsub(/-1\z/, '').gsub(/-2\z/, '-1')
+          else
+            s
+          end
+        end
+      end
+
+      # split size by container dimension
+      def size_array(container, size)
+        (
+          case DIMENSION_TBL[container]
+          when 2
+            split_size(size)
+          when 1
+            [size]
+          when 0
+            []
+          end
+        ).map { |s| normalize_name(s) }
+      end
+
+      def split_size(str)
+        str = str.gsub(/(\A\{|\}\z)/, '') while str =~ /\A#{RE_BLOCK}\z/
+
+        sz = str.split(',')
+        return sz if sz.size == 2
+
+        sz = str.scan(/(?<nbl>[^{}]+)|#{RE_BLOCK}/).flatten.compact
+        return sz if sz.size == 2
+
+        str = str.delete('{},')
+
+        sz = str.scan(/[^_](?:_[^_])?/)
+        return sz if sz.size == 2
+
+        sz = str.split('_')
+        return sz if sz.size == 2
+
+        [str[0] || '_', str[1..-1] || '_']
+      end
+    end
+
+    # holds regular expressions and matches it with input format string
+    class InputFormatMatcher
+      include InputFormatUtils
+
+      attr_reader :container, :item, :pat, :gen_names, :gen_pat2
+      attr_reader :names, :pat2, :size, :delim, :ix0
+
+      def initialize(
+        container: nil, item: nil,
+        pat: nil, gen_names: nil, gen_pat2: nil
+      )
+        @container = container
+        @item = item
+        @pat = pat
+        @gen_names = gen_names
+        @gen_pat2 = gen_pat2
+      end
 
       def match(str)
+        str, dlm = extract_delim(str)
         return false unless (m1 = pat.match(str))
 
         @names = gen_names.call(m1)
         @pat2 = gen_pat2&.call(names)
         @size = m1.names.include?('sz') && m1['sz'] || ''
+        @ix0 = m1.names.include?('ix0') && m1['ix0'] || size
+        @delim = dlm
         true
       end
 
       def match2(str)
-        return false unless @pat2
+        return false unless pat2
+
+        str, _dlm = extract_delim(str)
         return true if /\A\.+\z/ =~ str
-        return false unless (m2 = @pat2.match(str))
+        return false unless (m2 = pat2.match(str))
 
         m2.names.include?('sz') && @size = m2['sz']
         true
       end
+
+      def to_inpdef
+        Problem::InputFormat.new(
+          container: container, item: item,
+          names: normalize_names(names),
+          size: normalize_size(container, size, ix0),
+          delim: delim
+        )
+      end
     end
 
-    module InputFormatConstants
-      SECTIONS = [
-        Problem::SECTION_IN_FMT,
-        Problem::SECTION_IO_FMT
-      ].freeze
-      ITEM_PAT = /\{*[A-Za-z]+(?:_[A-Za-z]+)*\}*/.freeze
-      SZ = '(_(?<sz>\S+)|{(?<sz>\S+)})'
-      SINGLE_PAT = /([A-Za-z{][A-Za-z_0-9{}]*)/.freeze
-      MATCHERS = [
-        InputFormatMatcher.new(
-          :matrix, :number,
+    # matcher constants
+    module InputFormatMatcherConstants
+      include InputFormatConstants
+
+      MATRIX_MATCHER = InputFormatMatcher.new(
+        container: :matrix,
+        pat:
           /
             \A
-            (?<v>#{ITEM_PAT})[_{]\{*[01][,_]?[01]\}*
-            (\s+\k<v>[_{]\S+)*
-            (\s+\.+)?
-            (\s+\k<v>#{SZ})+
+            (?<v>#{RE_ITEM})#{RE_IX_00}
+            (\s+(\.+|\k<v>#{RE_IX}))*
+            \s+\k<v>#{RE_SZ}
             \z
           /x,
-          ->(m) { [m[:v]] },
-          lambda { |((v))|
+        gen_names: ->(m) { [m[:v]] },
+        gen_pat2:
+          lambda { |(v)|
             /
               \A
-              #{v}[_{]\S+
-              (\s+#{v}#{SZ})*
-              (\s+\.+)?
-              (\s+#{v}#{SZ})*
+              #{v}#{RE_IX}
+              (\s+(\.+|#{v}#{RE_IX}))*
+              \s+(\.+|#{v}#{RE_SZ})
               \z
             /x
           }
-        ),
-        InputFormatMatcher.new(
-          :matrix, :char,
+      )
+      MATRIX_CHAR_MATCHER = InputFormatMatcher.new(
+        container: :matrix,
+        item: :char,
+        pat:
           /
             \A
-            (?<v>#{ITEM_PAT})[_{]\{*[01][,_]?[01]\}*
-            (\k<v>[_{]\S+)*
-            (\s*\.+\s*)?
-            (\k<v>#{SZ})+
+            (?<v>#{RE_ITEM})#{RE_IX_00}
+            (\s*\.+\s*|\k<v>#{RE_IX})*
+            \k<v>#{RE_SZ}
             \z
           /x,
-          ->(m) { [m[:v]] },
-          lambda { |((v))|
+        gen_names: ->(m) { [m[:v]] },
+        gen_pat2:
+          lambda { |(v)|
             /
               \A
-              (#{v}[_{]\S+)+
-              (\s*\.+\s*)?
-              (#{v}#{SZ})+
+              (#{v}#{RE_IX})+
+              (\s*\.+\s*|#{v}#{RE_IX})*
+              (\s*\.+\s*|#{v}#{RE_SZ})
               \z
             /x
           }
-        ),
-        InputFormatMatcher.new(
-          :harray, :number,
+      )
+      HARRAY_MATCHER = InputFormatMatcher.new(
+        container: :harray,
+        pat:
           /
             \A
-            (?<v>#{ITEM_PAT})[_{]\{*[0-9]\}*
-            (\s+\k<v>[_{]\S+)*
-            (\s+\.+)?
-            (\s+\k<v>#{SZ})+
+            (?<v>#{RE_ITEM})#{RE_IX_0}
+            (\s+(\.+|\k<v>#{RE_IX}))*
+            \s+\k<v>#{RE_SZ}
             \z
           /x,
-          ->(m) { [m[:v]] },
-          nil
-        ),
-        InputFormatMatcher.new(
-          :harray, :char,
+        gen_names: ->(m) { [m[:v]] }
+      )
+      HARRAY_CHAR_MATCHER = InputFormatMatcher.new(
+        container: :harray,
+        item: :char,
+        pat:
           /
             \A
-            (?<v>#{ITEM_PAT})[_{]\{*[0-9]\}*
-            (\k<v>[_{]\S+)*
-            (\s*\.+\s*)?
-            (\k<v>#{SZ})+
+            (?<v>#{RE_ITEM})#{RE_IX_0}
+            (\s*\.+\s*|\k<v>#{RE_IX})*
+            \k<v>#{RE_SZ}
             \z
           /x,
-          ->(m) { [m[:v]] },
-          nil
-        ),
-        InputFormatMatcher.new(
-          :varray, :number,
+        gen_names: ->(m) { [m[:v]] }
+      )
+      VARRAY_MATRIX_MATCHER = InputFormatMatcher.new(
+        container: :varray_matrix,
+        pat:
           /
             \A
-            #{ITEM_PAT}[_{]\{*(?<sz>[0-9]+)\}*
-            (\s+#{ITEM_PAT}[_{]\{*\k<sz>\}*)*
+            (?<vs>#{RE_ITEM}#{RE_SZ2_0} (\s+#{RE_ITEM}#{RE_SZ2_REF})*)
+            \s+(?<m>#{RE_ITEM})#{RE_IX_00}
+            (\s+(\.+|\k<m>#{RE_IX}))*
+            \s+\k<m>#{RE_SZ}
             \z
           /x,
-          ->(m) { m[0].split.map { |w| w.scan(ITEM_PAT)[0] } },
+        gen_names:
+          ->(m) { [*m[:vs].split.map { |w| w.scan(RE_ITEM)[0] }, m[:m]] },
+        gen_pat2:
           lambda { |vs|
-            pat2 = vs.map { |v| v + SZ }.join('\s+')
-            /\A#{pat2}\z/
+            ws = vs[0..-2].map { |v| v + RE_IX.source }.join('\s+')
+            m = vs[-1]
+            /
+              \A
+              #{ws}
+              \s+#{m}#{RE_IX}
+              (\s+(\.+|#{m}#{RE_IX}))*
+              \s+(\.+|#{m}#{RE_SZ})
+              \z
+            /x
           }
-        ),
-        InputFormatMatcher.new(
-          :single, :number,
-          /\A(.*\s)?#{SINGLE_PAT}(\s.*)?\z/,
-          ->(m) { m[0].split.select { |w| w =~ /\A#{SINGLE_PAT}\z/ } },
-          nil
-        )
+      )
+      VARRAY_MATRIX_CHAR_MATCHER = InputFormatMatcher.new(
+        container: :varray_matrix,
+        item: :char,
+        pat:
+          /
+            \A
+            (?<vs>#{RE_ITEM}#{RE_SZ2_0} (\s+#{RE_ITEM}#{RE_SZ2_REF})*)
+            \s+(?<m>#{RE_ITEM})#{RE_IX_00}
+            (\s*\.+\s*|\k<m>#{RE_IX})*
+            \k<m>#{RE_SZ} \z
+          /x,
+        gen_names:
+          ->(m) { [*m[:vs].split.map { |w| w.scan(RE_ITEM)[0] }, m[:m]] },
+        gen_pat2:
+          lambda { |vs|
+            ws = vs[0..-2].map { |v| v + RE_IX.source }.join('\s+')
+            m = vs[-1]
+            /
+              \A
+              #{ws}
+              \s+#{m}#{RE_IX}
+              (\s*\.+\s*|#{m}#{RE_IX})*
+              (\s*\.+\s*|#{m}#{RE_SZ})
+              \z
+            /x
+          }
+      )
+      MATRIX_VARRAY_MATCHER = InputFormatMatcher.new(
+        container: :matrix_varray,
+        pat:
+          /
+            \A
+            (?<m>#{RE_ITEM})#{RE_IX_00}
+            (\s+(\.+|\k<m>#{RE_IX}))*
+            \s+\k<m>#{RE_SZ}
+            \s+(?<vs>#{RE_ITEM}#{RE_SZ2_0} (\s+#{RE_ITEM}#{RE_SZ2_REF})*)
+            \z
+          /x,
+        gen_names:
+          ->(m) { [m[:m], *m[:vs].split.map { |w| w.scan(RE_ITEM)[0] }] },
+        gen_pat2:
+          lambda { |vs|
+            m = vs[0]
+            ws = vs[1..-1].map { |v| v + RE_IX.source }.join('\s+')
+            /
+              \A
+              #{m}#{RE_IX}
+              (\s+(\.+|#{m}#{RE_IX}))*
+              \s+(\.+|#{m}#{RE_SZ})
+              \s+#{ws}
+              \z
+            /x
+          }
+      )
+      VMATRIX_MATCHER = InputFormatMatcher.new(
+        container: :vmatrix,
+        pat:
+          /
+            \A
+            #{RE_ITEM}#{RE_SZ_00} (\s+#{RE_ITEM}#{RE_SZ_REF})*
+            \z
+          /x,
+        gen_names: ->(m) { m[0].split.map { |w| w.scan(RE_ITEM)[0] } },
+        gen_pat2:
+          lambda { |vs|
+            ws = [
+              vs[0] + RE_SZ.source,
+              *vs[1..-1]&.map { |v| v + RE_IX.source }
+            ].join('\s+')
+            /\A#{ws}\z/
+          }
+      )
+      HMATRIX_MATCHER = InputFormatMatcher.new(
+        container: :hmatrix,
+        pat:
+          /
+            \A
+            #{RE_ITEM}#{RE_IX_00}
+            (\s+(\.+|#{RE_ITEM}#{RE_IX_99}))*
+            \s+#{RE_ITEM}#{RE_SZ_99}
+            \z
+          /x,
+        gen_names:
+          ->(m) { m[0].split.map { |w| w.scan(RE_ITEM)[0] }.uniq },
+        gen_pat2:
+          lambda { |vs|
+            ws1 = vs.map { |v| v + RE_IX.source }.join('\s+')
+            ws2 = [
+              vs[0] + RE_SZ.source,
+              *vs[1..-1]&.map { |v| v + RE_IX.source }
+            ].join('\s+')
+            /
+              \A
+              #{ws1} (\s+(\.+|#{ws1}))* \s+(\.+|#{ws2})
+              \z
+            /x
+          }
+      )
+      VARRAY_MATCHER = InputFormatMatcher.new(
+        container: :varray,
+        pat:
+          /
+            \A
+            #{RE_ITEM}#{RE_SZ_0} (\s+#{RE_ITEM}#{RE_SZ_REF})*
+            \z
+          /x,
+        gen_names:
+          ->(m) { m[0].split.map { |w| w.scan(RE_ITEM)[0] } },
+        gen_pat2:
+          lambda { |vs|
+            ws = [
+              vs[0] + RE_SZ.source,
+              *vs[1..-1]&.map { |v| v + RE_IX.source }
+            ].join('\s+')
+            /\A#{ws}\z/
+          }
+      )
+      SINGLE_MATCHER = InputFormatMatcher.new(
+        container: :single,
+        pat: /\A(.*\s)?#{RE_SINGLE}(\s.*)?\z/,
+        gen_names: ->(m) { m[0].split.select { |w| w =~ /\A#{RE_SINGLE}\z/ } }
+      )
+      MATCHERS = [
+        MATRIX_MATCHER,
+        MATRIX_CHAR_MATCHER,
+        HARRAY_MATCHER,
+        HARRAY_CHAR_MATCHER,
+        VARRAY_MATRIX_MATCHER,
+        VARRAY_MATRIX_CHAR_MATCHER,
+        MATRIX_VARRAY_MATCHER,
+        VMATRIX_MATCHER,
+        HMATRIX_MATCHER,
+        VARRAY_MATCHER,
+        SINGLE_MATCHER
       ].freeze
     end
 
     # parses input data format and generates input definitons
     module InputFormat
+      extend InputFormatUtils
       include InputFormatConstants
+      include InputFormatMatcherConstants
 
       module_function
 
       def process(pbm)
         return unless (str = find_fmt(pbm))
 
-        inpdefs = parse(str, pbm.samples)
-        pbm.formats = inpdefs
+        inpdefs = parse(str)
+        pbm.formats_src = inpdefs
       end
 
       def find_fmt(pbm)
         str = nil
         SECTIONS.any? do |key|
-          str = pbm.sections[key]&.code_block_html
-          str && !str.empty?
+          (str = pbm.sections[key]&.code_block_html) && !str.empty?
         end
         str
       end
 
-      def parse(str, smps)
+      def parse(str)
         lines = normalize_fmt(str)
-        inpdefs = parse_fmt(lines)
-        normalize_defs!(inpdefs)
-        smpx = max_smp(smps)
-        smpx && match_smp!(inpdefs, smpx)
-        inpdefs
-      end
-
-      def normalize_fmt(fmt)
-        # 1) &npsp; , fill-width space -> half width space
-        # 2) {i, j}->{i,j} for nested {}
-        fmt
-          .tr('０-９Ａ-Ｚａ-ｚ', '0-9A-Za-z')
-          .gsub(/[[:space:]]/) { |c| c.gsub(/[^\n]/, ' ') } # 1)
-          .gsub(%r{<var>([^<>]+)</var>}i, '\1') # <sub><var>N</var></sub>
-          .gsub(%r{<sup>([^<>]+)</sup>}i, '^\1')
-          .gsub(%r{<sub>([^<>]+)</sub>}i, '_{\1}')
-          .gsub(%r{<sub>([^<>]+)</sub>}i, '_{\1}') # for nested<sub>
-          .gsub(/<("[^"]*"|'[^']*'|[^'"<>])*>/, '')
-          .gsub('&amp;', '&')
-          .gsub('&gt;', '>')
-          .gsub('&lt;', '<')
-          .gsub('\\ ', ' ')
-          .gsub('\\(', '')
-          .gsub('\\)', '')
-          .gsub('\\lvert', '|')
-          .gsub('\\rvert', '|')
-          .gsub('\\mathit', '')
-          .gsub('\\times', '*')
-          .gsub(/\\begin(\{[^{}]*\})*/, '')
-          .gsub(/\\end(\{[^{}]*\})*/, '')
-          .gsub(/\\[cdlv]?dots/, '..')
-          .gsub(/\{\}/, ' ')
-          .gsub('−', '-') # fill width hyphen
-          .gsub(/[・：‥⋮︙…]+/, '..')
-          .gsub(/[\\$']/, '') # s' -> s
-          .gsub(/[&~|]/, ' ') # |S| -> S
-          .gsub(%r{[-/:](#{SINGLE_PAT})}, ' \1') # a-b, a/b, a:b -> a b
-          .gsub(/^\s*[.:][\s.:]*$/, '..')
-          .tr('()', '{}')
-          .gsub(/(?<bl>\{(?:[^{}]|\g<bl>)*\})/) { |w| w.delete(' ') } # 2)
-          .split("\n")
-          .map(&:strip)
+        parse_fmt(lines)
       end
 
       def parse_fmt(lines)
@@ -199,85 +467,18 @@ module AtCoderFriends
           if matcher
             next if matcher.match2(line)
 
-            ret.last.size = matcher.size
+            ret << matcher.to_inpdef
           end
           if (matcher = MATCHERS.find { |m| m.match(line) })
-            ret << Problem::InputFormat.new(
-              matcher.container, matcher.item, matcher.names, ''
-            )
           elsif !line.empty?
             puts "unknown format: #{line}"
-            # ret << Problem::InputFormat.new(:unknown, nil, line)
+            ret << unknown_fmt(line)
           end
         end
       end
 
-      def normalize_defs!(inpdefs)
-        inpdefs.each do |inpdef|
-          inpdef.names = normalize_names(inpdef.names)
-          inpdef.size = normalize_size(inpdef.container, inpdef.size)
-        end
-      end
-
-      def normalize_names(names)
-        return names unless names.is_a?(Array)
-
-        names.map { |nm| nm.delete('{}').gsub(/(\A_+|_+\z)/, '') }
-      end
-
-      def normalize_size(container, size)
-        sz =
-          case container
-          when :matrix
-            matrix_size(size)
-          when :harray, :varray
-            [size]
-          else
-            []
-          end
-        sz.map do |w|
-          w
-            .delete('{},')
-            .gsub(/(\A_+|(_|-1)+\z)/, '') # extra underscores, N-1 -> N
-        end
-      end
-
-      def matrix_size(str)
-        sz = str.scan(/([^{}]+|\{[^{}]+\}})/).flatten
-        return sz if sz.size == 2
-
-        sz = str.split(',')
-        return sz if sz.size == 2
-
-        sz = str.split('_')
-        return sz if sz.size == 2
-
-        str = str.delete('{},')
-        len = str.size
-        if len.positive? && len.even?
-          return str.chars.each_slice(len / 2).map(&:join)
-        end
-
-        [str[0] || '_', str[1..-1] || '_']
-      end
-
-      def max_smp(smps)
-        smps
-          .select { |smp| smp.ext == :in }
-          .max_by { |smp| smp.txt.size }
-          &.txt
-      end
-
-      def match_smp!(inpdefs, smp)
-        lines = smp.split("\n")
-        inpdefs.each_with_index do |inpdef, i|
-          break if i >= lines.size
-          next if inpdef.item != :number
-
-          inpdef.item = :string if lines[i].split[0] =~ /[^\-0-9]/
-          break if %i[varray matrix].include?(inpdef.container)
-        end
-        inpdefs
+      def unknown_fmt(line)
+        Problem::InputFormat.new(container: :unknown, item: line)
       end
     end
   end