gren 0.2.3 → 0.2.4

data/History.ja.txt

@@ -1,3 +1,21 @@
+=== 0.2.4 2010/10/09
+
+* mkgrendb
+  * データベースの形式を変更
+    * 古いmkgrendbで作成したデータベースは、 'mkgrendb --full' で再構築をして下さい。
+  * yamlに、ディレクトリではなく単一ファイルの指定もOKに
+
+* grendb
+  * マッチ対象を行では無くファイルとする --match-file モードを追加 
+    * 今までのモードは --match-line
+    * デフォルトのモードを --match-file に変更
+    * 今までの行マッチモードを使う場合は、 -l or --ml or --match-line と指定する
+  * 拡張子をキーとする検索モードを追加
+    * -s, --suffix
+  * --groonga-only オプションを作成
+    * 実際のファイルを辿らず、データベースの情報のみで探索出来るように
+  * 検索結果を更新時刻順で表示するように
+
 === 0.2.3 2010/09/25
 
 * mkgrendb

data/Manifest.txt

@@ -12,6 +12,7 @@ lib/common/grenfiletest.rb
 lib/common/grensnip.rb
 lib/common/platform.rb
 lib/common/string_snip.rb
+lib/common/util.rb
 lib/findgrep/findgrep.rb
 lib/findgrep/result.rb
 lib/gren.rb

data/bin/grendb

@@ -7,5 +7,5 @@ require 'rubygems'
 require File.expand_path(File.dirname(__FILE__) + "/../lib/gren")
 require "grendb/cli"
 
-Version = "0.2.0"
+Version = "0.2.4"
 Grendb::CLI.execute(STDOUT, ARGV)

data/bin/mkgrendb

@@ -7,5 +7,5 @@ require 'rubygems'
 require File.expand_path(File.dirname(__FILE__) + "/../lib/gren")
 require "mkgrendb/cli"
 
-Version = "0.2.3"
+Version = "0.2.4"
 Mkgrendb::CLI.execute(STDOUT, ARGV)

data/lib/common/display_util.rb

@@ -44,5 +44,19 @@ module Gren
       end
     end
     module_function :size_s
+
+    def dump_methods(c)
+      unless c.is_a?(Class)
+        c = c.class
+      end
+      
+      while (true)
+        p c
+        break if (c == Object)
+        puts "↓  " + c.public_instance_methods(false).inspect
+        c = c.superclass
+      end
+    end
+    module_function :dump_methods
   end
 end

data/lib/common/util.rb

@@ -0,0 +1,98 @@
+# -*- coding: utf-8 -*-
+
+module Gren
+  module Util
+    def time_s(time)
+      t = time.truncate
+      h = t / 3600
+      t = t % 3600
+      m = t / 60
+      t = t % 60
+      t += round(time - time.prec_i, 2)
+      
+      if (h > 0 && m > 0)
+        "#{h}h #{m}m #{t}s"
+      elsif (m > 0)
+        "#{m}m #{t}s"
+      else
+        "#{t}sec"
+      end
+    end
+    module_function :time_s
+
+    def round(n, d)
+      (n * 10 ** d).round / 10.0 ** d
+    end
+    module_function :round
+
+    def size_s(size)
+      tb = 1024 ** 4
+      gb = 1024 ** 3
+      mb = 1024 ** 2
+      kb = 1024
+
+      if (size >= tb)
+        round(size / tb.prec_f, 2).to_s + "TB"
+      elsif (size >= gb)
+        round(size / gb.prec_f, 2).to_s + "GB"
+      elsif (size >= mb)
+        round(size / mb.prec_f, 2).to_s + "MB"
+      elsif (size >= kb)
+        round(size / kb.prec_f, 2).to_s + "KB"
+      else
+        size.to_s + "Byte"
+      end
+    end
+    module_function :size_s
+
+    # アルファベットと演算子で表示する数を変える
+    ALPHABET_DISP_NUM = 5
+    OPERATOR_DISP_NUM = 10
+
+    def p_classtree(c)
+      unless c.is_a?(Class)
+        c = c.class
+      end
+      
+      while (true)
+        puts c.name
+        break if (c == Object)
+        p_classtree_sub(c)
+        c = c.superclass
+      end
+    end
+    module_function :p_classtree
+
+    def p_classtree_sub(c)
+      # メソッドの一覧を得る
+      group = c.public_instance_methods(false).sort.partition { |m| m =~ /\w/ }
+      array = group.flatten
+      operator_start_index = group[0].size
+      limit = ALPHABET_DISP_NUM
+
+      print (array.size > limit) ? "｜  " :  "↓  "
+      
+      counter = 0
+      array.each_with_index do |v, index|
+        if (index == operator_start_index)
+          limit = OPERATOR_DISP_NUM
+          counter = 0
+          puts
+          print (array.size - index > limit) ? "｜  " : "↓  "
+        end
+
+        if (counter >= limit)
+          counter = 0
+          puts
+          print (array.size - index > limit) ? "｜  " : "↓  "
+        end
+
+        print v + ", "
+        counter += 1
+      end
+      puts
+    end
+    module_function :p_classtree_sub
+
+  end
+end

data/lib/findgrep/findgrep.rb

@@ -8,6 +8,8 @@ require File.join(File.dirname(__FILE__), '../common/platform')
 require File.join(File.dirname(__FILE__), '../common/grenfiletest')
 require File.join(File.dirname(__FILE__), '../common/grensnip')
 require 'groonga'
+require File.join(File.dirname(__FILE__), '../common/util')
+include Gren
 
 module FindGrep
   class FindGrep
@@ -20,11 +22,14 @@ module FindGrep
                         :isSilent,
                         :debugMode,
                         :filePatterns,
+                        :suffixs,
                         :ignoreFiles,
                         :ignoreDirs,
                         :kcode,
                         :noSnip,
-                        :dbFile)
+                        :dbFile,
+                        :groongaOnly,
+                        :isMatchFile)
 
     DEFAULT_OPTION = Option.new([],
                                 [],
@@ -37,9 +42,12 @@ module FindGrep
                                 [],
                                 [],
                                 [],
+                                [],
                                 Platform.get_shell_kcode,
                                 false,
-                                nil)
+                                nil,
+                                false,
+                                false)
     
     def initialize(patterns, option)
       @patterns = patterns
@@ -112,9 +120,9 @@ module FindGrep
       
       # ドキュメントを検索
       documents = Groonga::Context.default["documents"]
-      
+
       # 全てのパターンを検索
-      records = documents.select do |record|
+      table = documents.select do |record|
         expression = nil
 
         # キーワード
@@ -137,21 +145,62 @@ module FindGrep
           end
         end
 
-        # 検索方法
+        # 拡張子(OR)
+        se = suffix_expression(record) 
+        expression &= se if (se)
+        
+        # 検索式
         expression
       end
       
+      # タイムスタンプでソート
+      records = table.sort([{:key => "timestamp", :order => "descending"}])
+
       # データベースにヒット
       stdout.puts "Found   : #{records.size} records."
 
       # 検索にヒットしたファイルを実際に検索
       records.each do |record|
-        if FileTest.exist? record.path
-          searchFile(stdout, record.path, record.path)
+        if (@option.groongaOnly)
+          searchGroongaOnly(stdout, record)
+        else
+          searchFile(stdout, record.path, record.path) if FileTest.exist?(record.path)
         end
       end
     end
 
+    def and_expression(key, list)
+      sub = nil
+      
+      list.each do |word|
+        e = key =~ word
+        if sub.nil?
+          sub = e
+        else
+          sub &= e
+        end
+      end
+
+      sub
+    end
+
+    def suffix_expression(record)
+      sub = nil
+      
+      @option.suffixs.each do |word|
+        e = record.suffix =~ word
+        if sub.nil?
+          sub = e
+        else
+          sub |= e
+        end
+      end
+
+      sub
+    end
+    private :suffix_expression
+      
+
     def searchFromDir(stdout, dir, depth)
       if (@option.depth != -1 && depth > @option.depth)
         return
@@ -237,37 +286,57 @@ module FindGrep
       @result.search_count += 1
       @result.search_size += FileTest.size(fpath)
 
-      # 検索本体
       @result.search_files << fpath_disp if (@option.debugMode)
 
-      open(fpath, "r") { |file|
-        match_file = false
+      open(fpath, "r") do |file|
+        searchData(stdout, file2data(file), fpath_disp)
+      end
+    end
+    private :searchFile
+
+    def searchGroongaOnly(stdout, record)
+      file_size = record.content.size
+      
+      @result.count += 1
+      @result.size += file_size
+      
+      @result.search_count += 1
+      @result.search_size += file_size
+      
+      @result.search_files << record.path if (@option.debugMode)
+
+      searchData(stdout, record.content, record.path)
+    end
+    private :searchGroongaOnly
 
-        file2data(file).each_with_index { |line, index|
-          result, match_datas = match?(line)
+    def searchData(stdout, data, path)
+      match_file = false
 
-          if ( result )
-            header = "#{fpath_disp}:#{index + 1}:"
-            line = GrenSnip::snip(line, match_datas) unless (@option.noSnip)
+      data.each_with_index { |line, index|
+        result, match_datas = match?(line)
 
-            unless (@option.colorHighlight)
-              stdout.puts header + line
-            else
-              stdout.puts HighLine::BLUE + header + HighLine::CLEAR + GrenSnip::coloring(line, match_datas)
-            end
+        if ( result )
+          header = "#{path}:#{index + 1}:"
+          line = GrenSnip::snip(line, match_datas) unless (@option.noSnip)
 
-            unless match_file
-              @result.match_file_count += 1
-              @result.match_files << fpath_disp if (@option.debugMode)
-              match_file = true
-            end
+          unless (@option.colorHighlight)
+            stdout.puts header + line
+          else
+            stdout.puts HighLine::BLUE + header + HighLine::CLEAR + GrenSnip::coloring(line, match_datas)
+          end
 
-            @result.match_count += 1
+          unless match_file
+            @result.match_file_count += 1
+            @result.match_files << path if (@option.debugMode)
+            match_file = true
+            break if (@option.isMatchFile)
           end
-        }
+
+          @result.match_count += 1
+        end
       }
     end
-    private :searchFile
+    private :searchData
 
     def file2data(file)
         data = file.read
@@ -294,13 +363,25 @@ module FindGrep
 
       or_matchs = []
       @orRegexps.each {|v| or_matchs << v.match(line)}
-
-      result = match_datas.all? && !sub_matchs.any? && (or_matchs.empty? || or_matchs.any?)
+      
+      unless (@option.isMatchFile)
+        result = match_datas.all? && !sub_matchs.any? && (or_matchs.empty? || or_matchs.any?)
+      else
+        result = first_condition(match_datas, sub_matchs, or_matchs)
+      end
       result_match = match_datas + or_matchs
       result_match.delete(nil)
 
       return result, result_match
     end
     private :match?
+
+    def first_condition(match_datas, sub_matchs, or_matchs)
+      unless match_datas.empty?
+        match_datas[0]
+      else
+        or_matchs[0]
+      end
+    end
   end
 end

data/lib/findgrep/result.rb

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 require 'find'
-require File.join(File.dirname(__FILE__), '../common/display_util')
+require File.join(File.dirname(__FILE__), '../common/util')
 include Gren
 
 module FindGrep
@@ -34,8 +34,8 @@ module FindGrep
     end
 
     def print(stdout)
-      stdout.puts "dir   : #{@start_dir} (#{DisplayUtil::time_s(time)})"
-      stdout.puts "files : #{@search_count} in #{@count} (#{DisplayUtil::size_s(@search_size)} in #{DisplayUtil::size_s(@size)})"
+      stdout.puts "dir   : #{@start_dir} (#{Util::time_s(time)})"
+      stdout.puts "files : #{@search_count} in #{@count} (#{Util::size_s(@search_size)} in #{Util::size_s(@size)})"
       stdout.puts "match : #{@match_file_count} files, #{match_count} hit"
     end
 

data/lib/gren.rb

@@ -2,5 +2,5 @@ $:.unshift(File.dirname(__FILE__)) unless
   $:.include?(File.dirname(__FILE__)) || $:.include?(File.expand_path(File.dirname(__FILE__)))
 
 module Gren
-  VERSION = '0.2.3'
+  VERSION = '0.2.4'
 end

data/lib/grendb/cli.rb

@@ -8,14 +8,21 @@ module Grendb
       # オプション
       option = FindGrep::FindGrep::DEFAULT_OPTION
       option.dbFile = ENV['GRENDB_DEFAULT_DB']
+      
+      # デフォルトのマッチモードは'File'
+      option.isMatchFile = true
 
       # オプション解析
       opt = OptionParser.new("#{File.basename($0)} [option] keyword1 [keyword2 ...]")
       opt.on('--db [GREN_DB_FILE]', 'Search from the grendb database.') {|v| option.dbFile = v }
       opt.on('-f KEYWORD', '--file-keyword KEYWORD', 'Path keyword. (Enable multiple call)') {|v| option.filePatterns << v}
+      opt.on('-s SUFFIX', '--suffix SUFFIX',  'Search suffix.') {|v| option.suffixs << v }
       opt.on('-i', '--ignore', 'Ignore case.') {|v| option.ignoreCase = true}
       opt.on('-c', '--color', 'Color highlight.') {|v| option.colorHighlight = true}
       opt.on('--no-snip', 'There being a long line, it does not snip.') {|v| option.noSnip = true }
+      opt.on('-g', '--groonga-only', 'Search only groonga db.') {|v| option.groongaOnly = true }
+      opt.on('--mf', '--match-file', 'Match file. (Default)') {|v| option.isMatchFile = true }
+      opt.on('-l', '--ml', '--match-line', 'Match line, same mode as "gren".') {|v| option.isMatchFile = false }
 
       opt.parse!(arguments)
 

data/lib/mkgrendb/cli.rb

@@ -9,12 +9,14 @@ module Mkgrendb
       isDump = false
       isFull = false
       isDelete = false
+      isReport = false
 
       opt = OptionParser.new "#{File.basename($0)} INPUT_YAML1 [INPUT_YAML2 ...]"
       opt.on('--ddb', "--default-db", "Create or Update default DB. (Plase set ENV['GRENDB_DEFAULT_DB'])") {|v| input_yamls << ENV['GRENDB_DEFAULT_DB']}
       opt.on('--full', "Full update DB. (Delete and create)") {|v| isFull = true }
       opt.on('--delete', "Delete DB. (Not delete yaml)") {|v| isDelete = true }
       opt.on('--dump', "Dump DB.") {|v| isDump = true }
+      opt.on('--report', "Database Report.") {|v| isReport = true }
       opt.parse!(arguments)
 
       input_yamls.concat arguments
@@ -31,6 +33,8 @@ module Mkgrendb
             stdout.puts
           elsif (isDump)
             obj.dump
+          elsif (isReport)
+            obj.report
           else
             obj.update
             stdout.puts

data/lib/mkgrendb/mkgrendb.rb

@@ -5,7 +5,7 @@ require 'pathname'
 require 'rubygems'
 require 'groonga'
 require File.join(File.dirname(__FILE__), '../common/grenfiletest')
-require File.join(File.dirname(__FILE__), '../common/display_util')
+require File.join(File.dirname(__FILE__), '../common/util')
 include Gren
 
 module Mkgrendb
@@ -25,7 +25,12 @@ module Mkgrendb
       db_create(@output_db)
       db_open(@output_db)
       @src["directory"].each do |dir|
-        db_add_dir(File.expand_path(dir))
+        dir = File.expand_path(dir)
+        if (FileTest.directory? dir)
+          db_add_dir(dir)
+        else
+          db_add_file(STDOUT, dir)
+        end
       end
       @end_time = Time.now
       print_result
@@ -46,7 +51,7 @@ module Mkgrendb
 
     def print_result
       puts
-      puts "input_yaml : #{@input_yaml} (#{DisplayUtil::time_s(time)})"
+      puts "input_yaml : #{@input_yaml} (#{Util::time_s(time)})"
       puts "output_db  : #{@output_db}*"
       puts "files      : #{@file_count}"
       puts "updates    : #{@update_count}"
@@ -58,9 +63,11 @@ module Mkgrendb
       documents = Groonga::Context.default["documents"]
       records = documents.select
       records.each do |record|
+        p record
         puts "path : #{record.path}"
+        puts "suffix : #{record.suffix}"
         puts "timestamp : #{record.timestamp.strftime('%Y/%m/%d %H:%M:%S')}"
-        puts "content :", record.content[0..64]
+        puts "content :", record.content ? record.content[0..64] : nil
         puts
       end
     end
@@ -77,6 +84,7 @@ module Mkgrendb
             table.string("path")
             table.text("content")
             table.time("timestamp")
+            table.text("suffix")
           end
 
           schema.create_table("terms",
@@ -85,7 +93,7 @@ module Mkgrendb
                               :default_tokenizer => "TokenBigram") do |table|
             table.index("documents.path", :with_position => true)
             table.index("documents.content", :with_position => true)
-            table.index("documents.timestamp", :with_position => true)
+            table.index("documents.suffix", :with_position => true)
           end
         end
         puts "create     : #{filename} created."
@@ -127,6 +135,7 @@ module Mkgrendb
         :path => filename,
         :content => nil,
         :timestamp => File.mtime(filename),
+        :suffix => File::extname(filename),
       }
       
       # 検索するデータベース

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 2
-  - 3
-  version: 0.2.3
+  - 4
+  version: 0.2.4
 platform: ruby
 authors: 
 - ongaeshi
@@ -14,7 +14,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-09-26 00:00:00 +09:00
+date: 2010-10-10 00:00:00 +09:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -103,6 +103,7 @@ files:
 - lib/common/grensnip.rb
 - lib/common/platform.rb
 - lib/common/string_snip.rb
+- lib/common/util.rb
 - lib/findgrep/findgrep.rb
 - lib/findgrep/result.rb
 - lib/gren.rb