rroonga 5.0.0-x64-mingw32 → 5.0.1-x64-mingw32

data/vendor/local/lib/groonga/scripts/ruby/expression_size_estimator.rb

@@ -0,0 +1,158 @@
+module Groonga
+  class ExpressionSizeEstimator
+    def initialize(expression, table)
+      @expression = expression
+      @table = table
+      @table_size = @table.size
+    end
+
+    def estimate
+      builder = ScanInfoBuilder.new(@expression, Operator::OR, 0)
+      data_list = builder.build
+      return @table_size if data_list.nil?
+
+      or_data_list = group_data_list(data_list)
+      or_sizes = or_data_list.collect do |and_data_list|
+        and_sizes = and_data_list.collect do |data|
+          size = estimate_data(data)
+          if data.logical_op == Operator::AND_NOT
+            size = @table_size - size
+            size = 0 if size < 0
+          end
+          size
+        end
+        and_sizes.min
+      end
+      or_sizes.max
+    end
+
+    private
+    def group_data_list(data_list)
+      or_data_list = [[]]
+      data_list.each do |data|
+        next if data.op == Operator::NOP
+
+        and_data_list = or_data_list.last
+        if and_data_list.empty?
+          and_data_list << data
+        else
+          case data.logical_op
+          when Operator::AND, Operator::AND_NOT
+            and_data_list << data
+          else
+            and_data_list = [data]
+            or_data_list << and_data_list
+          end
+        end
+      end
+      or_data_list
+    end
+
+    def estimate_data(data)
+      search_index = data.search_indexes.first
+      size = nil
+      if search_index
+        case data.op
+        when Operator::MATCH
+          size = estimate_match(data, search_index)
+        when Operator::EQUAL
+          size = estimate_equal(data, search_index)
+        when Operator::LESS,
+             Operator::LESS_EQUAL,
+             Operator::GREATER,
+             Operator::GREATER_EQUAL
+          size = estimate_range(data, search_index)
+        when Operator::CALL
+          procedure = data.args.first
+          if procedure.is_a?(Procedure) and procedure.name == "between"
+            size = estimate_between(data, search_index)
+          end
+        end
+      end
+      size || @table_size
+    end
+
+    def estimate_match(data, search_index)
+      index_column = search_index.index_column
+      if index_column.is_a?(Accessor)
+        # TODO
+        return nil
+      end
+
+      index_column.estimate_size(:query => data.query.value)
+    end
+
+    def estimate_equal(data, search_index)
+      index_column = search_index.index_column
+      if index_column.is_a?(Accessor)
+        # TODO
+        return nil
+      end
+
+      lexicon = index_column.lexicon
+      term_id = lexicon[data.query]
+      return 0 if term_id.nil?
+
+      index_column.estimate_size(:term_id => term_id)
+    end
+
+    def estimate_range(data, search_index)
+      index_column = search_index.index_column
+      if index_column.is_a?(Accessor)
+        # TODO
+        return nil
+      end
+
+      lexicon = index_column.lexicon
+      value = data.query.value
+      options = {}
+      case data.op
+      when Operator::LESS
+        options[:max] = value
+        options[:flags] = TableCursorFlags::LT
+      when Operator::LESS_EQUAL
+        options[:max] = value
+        options[:flags] = TableCursorFlags::LE
+      when Operator::GREATER
+        options[:min] = value
+        options[:flags] = TableCursorFlags::GT
+      when Operator::GREATER_EQUAL
+        options[:min] = value
+        options[:flags] = TableCursorFlags::GE
+      end
+      TableCursor.open(lexicon, options) do |cursor|
+        index_column.estimate_size(:lexicon_cursor => cursor)
+      end
+    end
+
+    def estimate_between(data, search_index)
+      index_column = search_index.index_column
+      if index_column.is_a?(Accessor)
+        # TODO
+        return nil
+      end
+
+      lexicon = index_column.lexicon
+      _, _, min, min_border, max, max_border = data.args
+      options = {
+        :min => min,
+        :max => max,
+        :flags => 0,
+      }
+      if min_border == "include"
+        options[:flags] |= TableCursorFlags::LT
+      else
+        options[:flags] |= TableCursorFlags::LE
+      end
+      if max_border == "include"
+        options[:flags] |= TableCursorFlags::GT
+      else
+        options[:flags] |= TableCursorFlags::GE
+      end
+
+      TableCursor.open(lexicon, options) do |cursor|
+        index_column.estimate_size(:lexicon_cursor => cursor)
+      end
+    end
+  end
+end

data/vendor/local/lib/groonga/scripts/ruby/index_column.rb

@@ -0,0 +1,39 @@
+module Groonga
+  class IndexColumn
+    private :estimate_size_for_term_id
+    private :estimate_size_for_query
+    private :estimate_size_for_lexicon_cursor
+
+    # Estimate the number of matched records for term ID or query.
+    #
+    # @overload estimate_size(:term_id => term_id)
+    #   @return [Integer] the number of matched records for the term ID.
+    #
+    # @overload estimate_size(:query => query)
+    #   @return [Integer] the number of matched records for the query.
+    #
+    # @overload estimate_size(:lexicon_cursor => lexicon_cursor)
+    #   @return [Integer] the number of matched records for the lexicon cursor.
+    #
+    def estimate_size(parameters)
+      term_id = parameters[:term_id]
+      if term_id
+        return estimate_size_for_term_id(term_id)
+      end
+
+      query = parameters[:query]
+      if query
+        return estimate_size_for_query(query, parameters)
+      end
+
+      lexicon_cursor = parameters[:lexicon_cursor]
+      if lexicon_cursor
+        return estimate_size_for_lexicon_cursor(lexicon_cursor)
+      end
+
+      message =
+        "must specify :term_id, :query, :lexicon_cursor: #{parameters.inspect}"
+      raise InvalidArgument, message
+    end
+  end
+end

data/vendor/local/lib/groonga/scripts/ruby/initialize/post.rb

@@ -4,11 +4,15 @@ require "context"
 
 require "writer"
 
+require "object"
 require "database"
+require "index_column"
 require "command"
 require "table_cursor"
 require "index_cursor"
 
+require "expression"
+
 require "plugin_loader"
 
 require "eval_context"

data/vendor/local/lib/groonga/scripts/ruby/initialize/pre.rb

@@ -1 +1,3 @@
 require "backtrace_entry"
+
+require "operator"

data/vendor/local/lib/groonga/scripts/ruby/logger.rb

@@ -9,12 +9,13 @@ module Groonga
         message = "#{error.class}: #{error.message}"
       end
       backtrace = error.backtrace
-      first_raw_entry = backtrace.first
-      if first_raw_entry
-        first_entry = BacktraceEntry.parse(first_raw_entry)
-        file = first_entry.file
-        line = first_entry.line
-        method = first_entry.method
+      last_raw_entry = backtrace.last
+      if last_raw_entry
+        last_entry = BacktraceEntry.parse(last_raw_entry)
+        file = last_entry.file
+        line = last_entry.line
+        method = last_entry.method
+        # message = "#{file}:#{line}:#{method}: #{message}"
       else
         file = ""
         line = 0
@@ -22,8 +23,11 @@ module Groonga
       end
       log(log_level, file, line, method, message)
 
-      backtrace.each do |raw_entry|
+      backtrace.reverse_each.with_index do |raw_entry, i|
+        next if i == 0
         entry = BacktraceEntry.parse(raw_entry)
+        message = entry.message
+        message = raw_entry if message.empty?
         log(log_level, entry.file, entry.line, entry.method, raw_entry)
       end
     end

data/vendor/local/lib/groonga/scripts/ruby/object.rb

@@ -0,0 +1,11 @@
+module Groonga
+  class Object
+    def domain
+      Context.instance[domain_id]
+    end
+
+    def range
+      Context.instance[range_id]
+    end
+  end
+end

data/vendor/local/lib/groonga/scripts/ruby/operator.rb

@@ -0,0 +1,22 @@
+module Groonga
+  class Operator
+    @values = {}
+    class << self
+      def register(operator)
+        const_set(operator.name, operator)
+        @values[operator.value] = operator
+      end
+
+      def find(value)
+        @values[value]
+      end
+    end
+
+    attr_reader :name
+    attr_reader :value
+    def initialize(name, value)
+      @name = name
+      @value = value
+    end
+  end
+end

data/vendor/local/lib/groonga/scripts/ruby/scan_info.rb

@@ -22,8 +22,13 @@ module Groonga
       data.args.each do |arg|
         push_arg(arg)
       end
-      data.indexes.each do |index, section_id, weight|
-        put_index(index, section_id, weight)
+      data.search_indexes.each do |search_index|
+        put_index(search_index.index_column,
+                  search_index.section_id,
+                  search_index.weight,
+                  search_index.scorer,
+                  search_index.scorer_args_expr,
+                  search_index.scorer_args_expr_offset || 0)
       end
     end
   end

data/vendor/local/lib/groonga/scripts/ruby/scan_info_builder.rb

@@ -1,11 +1,6 @@
-module Groonga
-  # TODO: Move me
-  class ExpressionCode
-    module Flags
-      RELATIONAL_EXPRESSION = 0x01
-    end
-  end
+require "scan_info_data"
 
+module Groonga
   class ScanInfoBuilder
     module Status
       START = 0
@@ -39,6 +34,7 @@ module Groonga
       Operator::GEO_WITHINP6,
       Operator::GEO_WITHINP8,
       Operator::TERM_EXTRACT,
+      Operator::REGEXP,
     ]
 
     ARITHMETIC_OPERATORS = [
@@ -294,10 +290,10 @@ module Groonga
 
       return false if data.args[0] != next_data.args[0]
 
-      data_indexes = data.indexes
-      return false if data_indexes.empty?
+      data_search_indexes = data.search_indexes
+      return false if data_search_indexes.empty?
 
-      data_indexes == next_data.indexes
+      data_search_indexes == next_data.search_indexes
     end
 
     def lower_condition?(operator)
@@ -325,7 +321,7 @@ module Groonga
       between_data.op = Operator::CALL
       between_data.logical_op = data.logical_op
       between_data.args = create_between_data_args(data, next_data)
-      between_data.indexes = data.indexes
+      between_data.search_indexes = data.search_indexes
       between_data
     end
 

data/vendor/local/lib/groonga/scripts/ruby/scan_info_data.rb

@@ -1,3 +1,5 @@
+require "scan_info_search_index"
+
 module Groonga
   class ScanInfoData
     attr_accessor :start
@@ -6,18 +8,18 @@ module Groonga
     attr_accessor :logical_op
     attr_accessor :query
     attr_accessor :args
-    attr_accessor :indexes
+    attr_accessor :search_indexes
     attr_accessor :flags
     attr_accessor :max_interval
     attr_accessor :similarity_threshold
     def initialize(start)
       @start = start
       @end = 0
-      @op = 0
+      @op = Operator::NOP
       @logical_op = Operator::OR
       @query = nil
       @args = []
-      @indexes = []
+      @search_indexes = []
       @flags = ScanInfo::Flags::PUSH
       @max_interval = nil
       @similarity_threshold = nil
@@ -100,6 +102,38 @@ module Groonga
           self.query = arg
         end
       end
+      if @op == Operator::REGEXP and not index_searchable_regexp?(@query)
+        @search_indexes.clear
+      end
+    end
+
+    def index_searchable_regexp?(pattern)
+      return false if pattern.nil?
+
+      previous_char = nil
+      pattern.value.each_char do |char|
+        if previous_char == "\\"
+          case char
+          when "Z"
+            return false
+          when "b", "B"
+            return false
+          when "d", "D", "h", "H", "p", "s", "S", "w", "W"
+            return false
+          when "X"
+            return false
+          when "k", "g", "1", "2", "3", "4", "5", "6", "7", "8", "9"
+            return false
+          end
+        else
+          case char
+          when ".", "[", "]", "|", "?", "+", "*", "{", "}", "^", "$", "(", ")"
+            return false
+          end
+        end
+        previous_char = char
+      end
+      true
     end
 
     def match_resolve_index_expression(expression)
@@ -107,31 +141,96 @@ module Groonga
       n_codes = codes.size
       i = 0
       while i < n_codes
-        code = codes[i]
-        value = code.value
-        case value
-        when Accessor
-          match_resolve_index_expression_accessor(code)
-        when FixedSizeColumn, VariableSizeColumn
-          match_resolve_index_expression_data_column(code)
-        when IndexColumn
-          section_id = 0
-          rest_n_codes = n_codes - i
-          if rest_n_codes >= 2 and
-              codes[i + 1].value.is_a?(Bulk) and
-              (codes[i + 1].value.domain == ID::UINT32 or
-               codes[i + 1].value.domain == ID::INT32) and
-              codes[i + 2].op == Operator::GET_MEMBER
-            section_id = codes[i + 1].value.value + 1
-            code = codes[i + 2]
-            i += 2
+        i = match_resolve_index_expression_codes(expression, codes, i, n_codes)
+      end
+    end
+
+    def match_resolve_index_expression_codes(expression, codes, i, n_codes)
+      code = codes[i]
+      value = code.value
+      return i + 1 if value.nil?
+
+      case value
+      when Accessor, Column
+        index_info, offset =
+          match_resolve_index_expression_find_index(expression,
+                                                    codes, i, n_codes)
+        i += offset - 1
+        if index_info
+          if value.is_a?(Accessor)
+            self.flags |= ScanInfo::Flags::ACCESSOR
           end
-          put_index(value, section_id, code.weight)
-        when Table
-          raise ErrorMessage, "invalid match target: <#{value.name}>"
+          weight, offset = codes[i].weight
+          i += offset
+          put_search_index(index_info.index, index_info.section_id, weight)
+       end
+      when Procedure
+        unless value.scorer?
+          message = "procedure must be scorer: #{scorer.name}>"
+          raise ErrorMessage, message
         end
+        scorer = value
         i += 1
+        index_info, offset =
+          match_resolve_index_expression_find_index(expression,
+                                                    codes, i, n_codes)
+        i += offset
+        if index_info
+          scorer_args_expr_offset = 0
+          if codes[i].op != Operator::CALL
+            scorer_args_expr_offset = i
+          end
+          while i < n_codes and codes[i].op != Operator::CALL
+            i += 1
+          end
+          weight, offset = codes[i].weight
+          i += offset
+          search_index = ScanInfoSearchIndex.new(index_info.index,
+                                                 index_info.section_id,
+                                                 weight,
+                                                 scorer,
+                                                 expression,
+                                                 scorer_args_expr_offset)
+          @search_indexes << search_index
+        end
+      when Table
+        raise ErrorMessage, "invalid match target: <#{value.name}>"
       end
+      i + 1
+    end
+
+    def match_resolve_index_expression_find_index(expression, codes, i, n_codes)
+      code = codes[i]
+      value = code.value
+      index_info = nil
+      offset = 1
+      case value
+      when Accessor
+        accessor = value
+        index_info = accessor.find_index(@op)
+        if index_info
+          if accessor.have_next? and index_info.index != accessor.object
+            index_info = IndexInfo.new(accessor, index_info.section_id)
+          end
+        end
+      when FixedSizeColumn, VariableSizeColumn
+        index_info = value.find_index(@op)
+      when IndexColumn
+        index = value
+        section_id = 0
+        rest_n_codes = n_codes - i
+        if rest_n_codes >= 2 and
+          codes[i + 1].value.is_a?(Bulk) and
+          (codes[i + 1].value.domain == ID::UINT32 or
+           codes[i + 1].value.domain == ID::INT32) and
+          codes[i + 2].op == Operator::GET_MEMBER
+          section_id = codes[i + 1].value.value + 1
+          offset += 2
+        end
+        index_info = IndexInfo.new(index, section_id)
+      end
+
+      [index_info, offset]
     end
 
     def match_resolve_index_expression_accessor(expr_code)
@@ -139,10 +238,13 @@ module Groonga
       self.flags |= ScanInfo::Flags::ACCESSOR
       index_info = accessor.find_index(op)
       return if index_info.nil?
+
+      section_id = index_info.section_id
+      weight = expr_code.weight
       if accessor.next
-        put_index(accessor, index_info.section_id, expr_code.weight)
+        put_search_index(accessor, section_id, weight)
       else
-        put_index(index_info.index, index_info.section_id, expr_code.weight)
+        put_search_index(index_info.index, section_id, weight)
       end
     end
 
@@ -150,13 +252,13 @@ module Groonga
       column = expr_code.value
       index_info = column.find_index(op)
       return if index_info.nil?
-      put_index(index_info.index, index_info.section_id, expr_code.weight)
+      put_search_index(index_info.index, index_info.section_id, expr_code.weight)
     end
 
     def match_resolve_index_db_obj(db_obj)
       index_info = db_obj.find_index(op)
       return if index_info.nil?
-      put_index(index_info.index, index_info.section_id, 1)
+      put_search_index(index_info.index, index_info.section_id, 1)
     end
 
     def match_resolve_index_accessor(accessor)
@@ -164,9 +266,9 @@ module Groonga
       index_info = accessor.find_index(op)
       return if index_info.nil?
       if accessor.next
-        put_index(accessor, index_info.section_id, 1)
+        put_search_index(accessor, index_info.section_id, 1)
       else
-        put_index(index_info.index, index_info.section_id, 1)
+        put_search_index(index_info.index, index_info.section_id, 1)
       end
     end
 
@@ -184,18 +286,19 @@ module Groonga
     def call_relational_resolve_index_db_obj(db_obj)
       index_info = db_obj.find_index(op)
       return if index_info.nil?
-      put_index(index_info.index, index_info.section_id, 1)
+      put_search_index(index_info.index, index_info.section_id, 1)
     end
 
     def call_relational_resolve_index_accessor(accessor)
       self.flags |= ScanInfo::Flags::ACCESSOR
       index_info = accessor.find_index(op)
       return if index_info.nil?
-      put_index(index_info.index, index_info.section_id, 1)
+      put_search_index(index_info.index, index_info.section_id, 1)
     end
 
-    def put_index(index, section_id, weight)
-      @indexes << [index, section_id, weight]
+    def put_search_index(index, section_id, weight)
+      search_index = ScanInfoSearchIndex.new(index, section_id, weight)
+      @search_indexes << search_index
     end
   end
 end

data/vendor/local/lib/groonga/scripts/ruby/scan_info_search_index.rb

@@ -0,0 +1,9 @@
+module Groonga
+  class ScanInfoSearchIndex < Struct.new(:index_column,
+                                         :section_id,
+                                         :weight,
+                                         :scorer,
+                                         :scorer_args_expr,
+                                         :scorer_args_expr_offset)
+  end
+end

data/vendor/local/lib/libgroonga.la

@@ -17,7 +17,7 @@ old_library='libgroonga.a'
 inherited_linker_flags=''
 
 # Libraries that this one depends upon.
-dependency_libs=' -L/home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib /home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib/libmsgpack.la /home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib/libonig.la -llz4 -lz -lws2_32'
+dependency_libs=' -L/home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib /home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib/libmsgpack.la /home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib/libonig.la -llz4 -lz -lws2_32'
 
 # Names of additional weak libraries provided by this library
 weak_library_names=''
@@ -38,4 +38,4 @@ dlopen=''
 dlpreopen=''
 
 # Directory that this library needs to be installed in:
-libdir='/home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib'
+libdir='/home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib'

data/vendor/local/lib/libmecab.la

@@ -17,7 +17,7 @@ old_library='libmecab.a'
 inherited_linker_flags=''
 
 # Libraries that this one depends upon.
-dependency_libs=' -L/home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib -lpthread'
+dependency_libs=' -L/home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib -lpthread'
 
 # Names of additional weak libraries provided by this library
 weak_library_names=''
@@ -38,4 +38,4 @@ dlopen=''
 dlpreopen=''
 
 # Directory that this library needs to be installed in:
-libdir='/home/kenhys/work/groonga/groonga.clean/packages/windows/dist-x64/lib'
+libdir='/home/myokoym/work/groonga/groonga.clean/packages/windows/dist-x64/lib'