X-Git-Url: http://git.chise.org/gitweb/?p=chise%2Fruby.git;a=blobdiff_plain;f=chise%2Flibchise_r.rb;fp=chise%2Flibchise_r.rb;h=42954c56e61b2bb1f21f371730de0fa8e83ef00f;hp=0000000000000000000000000000000000000000;hb=a091d91d69c6783bd58c0a062e7772133d1d7ef0;hpb=532444ed9ee69c56b57e0958769b76dc149ebcd2

diff --git a/chise/libchise_r.rb b/chise/libchise_r.rb
new file mode 100755
index 0000000..42954c5
--- /dev/null
+++ b/chise/libchise_r.rb
@@ -0,0 +1,344 @@
+# Copyright (C) 2002-2004 Kouichirou Eto, All rights reserved.
+# libchise extension compatible library.
+
+require "bdb"
+require "pathname"
+require "fileutils"
+require "chise/config"
+require "chise/path"
+
+module CHISE
+  module ChiseValue; end
+  module TableAccessModule; end
+
+  class DataSource
+    NONE = 0
+    Berkeley_DB = 1
+
+    def initialize(type=Berkeley_DB, loc=nil, subtype=0, modemask=0755)
+      @type = type
+      loc = Config.instance.db_dir if loc.nil?
+      @location = loc.path
+      @subtype = subtype
+      @modemask = modemask
+      @fdb = {}
+      @cdb = {}
+    end
+    attr_reader :type, :location, :subtype, :modemask
+
+    def get_feature(f)
+      @fdb[f] = FeatureTable.new(self, f) if @fdb[f].nil?
+      @fdb[f]
+    end
+
+    def get_ccs(ccs)
+      @cdb[ccs] = CCSTable.new(self, ccs) if @cdb[ccs].nil?
+      @cdb[ccs]
+    end
+
+    def each_feature
+      each_entry("character/feature") {|f| yield(f) }
+    end
+
+    def each_ccs
+      each_entry("character/by_feature") {|f| yield(f) }
+    end
+
+    def load_feature(name, cid)
+      ft = get_feature(name)
+      return nil if ft.nil?
+      ft.get_value(cid)
+    end
+
+    def decode_char(ccs, code_point)
+      ct = get_ccs(ccs)
+      return nil if ct.nil?
+      ct.decode(code_point)
+    end
+
+    private
+    def each_entry(subdir)
+      dir = @location + subdir
+      dir.each_entry {|f|
+	next if f.to_s == "." || f.to_s == ".."
+	next if f.to_s =~ /\.txt\Z/
+	yield(f.unescape_win_filename.unescape.to_s)
+      }
+    end
+  end
+
+  class AttributeTable
+    def initialize(dir, cat, keytype, name, amask, mmask)
+      @name = name
+
+      dbdir  = dir + cat + keytype
+      #qp dbdir.to_s
+      #FileUtils.mkdir_p(dbdir.to_s) unless dbdir.directory?
+      #qp dbdir.to_s, "2"
+      path = dbdir + name.path.escape.escape_win_filename
+      #qp path.basename.to_s, amask, mmask
+#      if /test/ =~ path.to_s
+#	qp path.to_s, amask
+#      end
+
+      if amask == BDB::RDONLY
+	raise unless path.exist?
+      end
+#     @db = BDB::Hash.open(path.to_s, nil, amask, mmask)
+      @db = BDB::Hash.open(path.to_s, nil, amask)
+      at_exit {
+	close
+      }
+    end
+
+    def close
+      return if @db.nil?
+      begin
+	@db.sync
+	@db.close
+	#p ["AttributeTable: close", @name]
+      rescue => e
+	#p e
+      end
+    end
+
+    def get(k)    @db.get(k);    end
+    def put(k, v) @db.put(k, v); end
+
+    def each() @db.each {|k, v| yield(k, v) } end
+  end
+
+  module TableAccessModule
+    def reset
+      @db = nil
+      @access = 0
+    end
+
+    def sync
+      @db.close if @db
+      reset
+    end
+    alias close sync
+
+    private
+    def setup_db(writable=nil)
+      if writable
+	sync if @access & BDB::CREATE == 0
+	@access = BDB::CREATE
+      else
+	@access = BDB::RDONLY
+      end
+
+      return if @db
+
+      begin
+	@db = AttributeTable.new(@ds.location, @category, @keyvalue,
+				 @name, @access, @ds.modemask)
+      rescue => e
+	#qp e
+	@db = nil
+      end
+      #raise if @db.nil?
+    end
+  end
+
+  class FeatureTable
+    include ChiseValue
+    include TableAccessModule
+
+    def initialize(ds, name)
+      @ds, @name = ds, name
+      @category, @keyvalue = "character", "feature"
+      reset
+    end
+
+    def get_value(cid)
+      setup_db
+      return nil if @db.nil?
+      parse_value(@db.get(format_char_id(cid)))
+    end
+
+    def set_value(cid, value)
+      setup_db(true)
+      raise "@db is nil." if @db.nil?
+      @db.put(format_char_id(cid), value)
+    end
+
+    def each
+      setup_db
+      raise "@db is nil." if @db.nil?
+      @db.each {|k, v|
+	yield(parse_c_string(k), v)
+      }
+    end
+  end
+
+  class CCSTable
+    include ChiseValue
+    include TableAccessModule
+
+    def initialize(ds, name)
+      @ds, @name = ds, name
+      @category, @keyvalue = "character", "by_feature"
+      reset
+    end
+
+    def decode(code_point)
+      setup_db
+      return nil if @db.nil?
+      parse_c_string(@db.get(code_point.to_s))
+    end
+
+    def set_decoded_char(code_point, cid)
+      setup_db(true)
+      raise "@db is nil." if @db.nil?
+      @db.put(code_point.to_s, format_char_id(cid))
+    end
+
+    def each
+      setup_db
+      raise "@db is nil." if @db.nil?
+      @db.each {|k, v|
+	yield(parse_value(k), parse_c_string(v))
+      }
+    end
+  end
+
+  module ChiseValue
+    def parse_value(v)
+      return v if v.nil?
+      #return v if v.kind_of?(Integer)
+      return v.to_i if /\A\d+\Z/ =~ v # number?
+      return $1 if /\A"(.+)"\Z/ =~ v # remove surrounding "
+      #return v.sub(/\A\?/, "") if v =~ /\A\?/ # remove ? in the head
+      #return parse_sexp(v) if v =~ /\A\(.+\)\Z/ # parse sexp # not yet
+      v
+    end
+
+    def parse_c_string(str)
+      return nil if str.nil?
+
+      i = 0
+      c = str[i]
+      i += 1
+      len = str.length
+
+      raise unless 2 <= len && c == ?\?
+
+      c = str[i]
+      i += 1
+
+      if (c == ?\\)
+	raise if (len < 3)
+	c = str[i]
+	i += 1
+	if (c == ?^)
+	  raise if (len < 4)
+	  c = str[i]
+	  i += 1
+	  if c == ?\?
+	    return 0x7F
+	  else
+	    return c & (0x80 | 0x1F)
+	  end
+	end
+	# raise # ?
+      end
+
+      if ( c < 0xC0 )
+	cid = c
+	counter = 0
+      elsif ( c < 0xE0 )
+	cid = c & 0x1f
+	counter = 1
+      elsif ( c < 0xF0 )
+	cid = c & 0x0f
+	counter = 2
+      elsif ( c < 0xF8 )
+	cid = c & 0x07
+	counter = 3
+      elsif ( c < 0xFC )
+	cid = c & 0x03
+	counter = 4
+      else
+	cid = c & 0x01
+	counter = 5
+      end
+
+      if (counter + 2 <= len)
+	(0...counter).each {|j|
+	  cid = (cid << 6) | (str[j + i] & 0x3F)
+	}
+	return cid
+      end
+
+      raise
+    end
+
+    def format_char_id(cid)
+      case cid
+      when ?\t  then return "?\t"
+      when ?\n  then return "?\n"
+      when ?\r  then return "?\r"
+      when 0x1C then return "?\^\\"
+      end
+
+      if cid <= 0x1F
+	return "?\\^"+(?@+cid).chr
+      elsif (cid == ?\s) || (cid == ?\") ||
+	  (cid == ?\#) || (cid == ?\') ||
+	  (cid == ?\() || (cid == ?\)) ||
+	  (cid == ?\,) || (cid == ?\.) ||
+	  (cid == ?\;) || (cid == ?\?) ||
+	  (cid == ?\[) || (cid == ?\\) ||
+	  (cid == ?\]) || (cid == ?\`)
+	return "?\\"+cid.chr
+      elsif (cid <= 0x7E)
+	return("?"+cid.chr)
+      elsif (cid == 0x7F)
+	return "?\\^?"+0.chr
+      elsif (cid <= 0x9F)
+	dest = "?\\^"
+	dest += (((cid + ?@) >> 6) | 0xC0).chr
+	dest += (((cid + ?@) & 0x3F) | 0x80).chr
+	return dest
+      elsif (cid <= 0x7FF)
+	dest = "?  "
+	dest[1] = (cid >> 6) | 0xC0
+	dest[2] = (cid & 0x3F) | 0x80
+	return dest
+      elsif (cid <= 0xFFFF)
+	dest = "?   "
+	dest[1] =  (cid >> 12) | 0xE0
+	dest[2] = ((cid >>  6) & 0x3F) | 0x80
+	dest[3] =  (cid        & 0x3F) | 0x80
+	return dest
+      elsif (cid <= 0x1FFFFF)
+	dest = "?    "
+	dest[1] =  (cid >> 18) | 0xF0
+	dest[2] = ((cid >> 12) & 0x3F) | 0x80
+	dest[3] = ((cid >>  6) & 0x3F) | 0x80
+	dest[4] =  (cid        & 0x3F) | 0x80
+	return dest
+      elsif (cid <= 0x3FFFFFF)
+	dest = "?     "
+	dest[1] =  (cid >> 24) | 0xF8
+	dest[2] = ((cid >> 18) & 0x3F) | 0x80
+	dest[3] = ((cid >> 12) & 0x3F) | 0x80
+	dest[4] = ((cid >>  6) & 0x3F) | 0x80
+	dest[5] =  (cid        & 0x3F) | 0x80
+	return dest
+      else
+	dest = "?      "
+	dest[1] =  (cid >> 30) | 0xFC
+	dest[2] = ((cid >> 24) & 0x3F) | 0x80
+	dest[3] = ((cid >> 18) & 0x3F) | 0x80
+	dest[4] = ((cid >> 12) & 0x3F) | 0x80
+	dest[5] = ((cid >>  6) & 0x3F) | 0x80
+	dest[6] =  (cid        & 0x3F) | 0x80
+	return dest
+      end
+      raise
+    end
+  end
+end