Adds an script of making sqls of daxu

diff --git a/dx2sql.rb b/dx2sql.rb
new file mode 100644 (file)
index 0000000..bbb6d09
--- /dev/null
+++ b/dx2sql.rb
@@ -0,0 +1,135 @@
+# -*- coding: utf-8 -*-
+# YAMLからsqlを生成する
+
+require 'yaml'
+require 'erb'
+#require File.join(File.dirname(__FILE__), 'lib', 'sw_seals')
+#require File.join(File.dirname(__FILE__), 'lib', 'unicode')
+
+SOURCE_DIR = File.join(File.dirname(__FILE__), '..', 'yml', 'daxu')
+TARGET_DIR = File.join(File.dirname(__FILE__), '..', 'sql', 'daxu')
+
+VOLUMES = {
+  1 => '一',
+  2 => '二',
+  3 => '三',
+  4 => '四',
+  5 => '五',
+  6 => '六',
+  7 => '七',
+  8 => '八',
+  9 => '九',
+  10 => '十',
+  11 => '十一',
+  12 => '十二',
+  13 => '十三',
+  14 => '十四',
+}
+
+class SqlConverter
+ extend ERB::DefMethod
+  @@rad_total = 0
+  @@volume = 0
+  def initialize(filename)
+    source = File.join(SOURCE_DIR, filename)
+    target = File.basename(filename, 'yml') + 'sql'
+    @out = File.open(File.join(TARGET_DIR, target), "w")
+    @radical = ''
+    @rad_count = 0
+    check(File.open(source))
+  end
+  def check(file)
+    YAML::parse_documents(file) do |syck|
+#      printf("%s\n", get_volume_title(@@volume))
+      create_volume(@@volume)
+      doc = syck.transform
+      doc.each do |wordinfo|
+#        check_radical(wordinfo)
+        check_word_info(wordinfo)
+      end
+    end
+    printf("Radicals: %i\n", @rad_count)
+  end
+  def check_word_info(wordinfo)
+    wordid = sprintf("%04i", wordinfo['id'])
+    word = wordinfo['wd']
+    radical = wordinfo['rd']
+    sw = wordinfo['sw']
+    pinyin = wordinfo['py']
+    fanqie = wordinfo['fn']
+    exp = wordinfo['ex']
+    dexp = wordinfo['dex']
+    xexp = wordinfo['xex']
+    check_radical_info(radical)
+    create_wordhead(word, wordid, sw, pinyin, fanqie, exp, @@rad_total)
+  end
+  def check_radical_info(radical)
+    if @radical != radical then
+      @radical = radical
+      @rad_count += 1
+      @@rad_total += 1
+      create_radical(radical, @@rad_total, @@volume)
+    end
+  end
+
+  # Temp
+  def check_radical(wordinfo)
+    id = sprintf("%04i", wordinfo['id'])
+    word = wordinfo['wd']
+    radical = wordinfo['rd']
+    if word == radical then
+      @radical = radical
+      @rad_count += 1
+      @@rad_total += 1
+    elsif radical != @radical then
+      printf("%s: Missing a radical %s of %s\n", id, radical, word)
+    end
+  end
+  def get_volume_title(num)
+    return sprintf("說文解字第%s篇", VOLUMES[num])
+  end
+
+  # Create SQL statements
+  def create_volume(num)
+    title = get_volume_title(num)
+    insert = "INSERT INTO dx_volumes (title, num)\n" +
+      "  VALUES ('%s', '%s');\n"
+    @out.printf(insert, title, num)
+  end
+  def create_radical(name, num, volume_num)
+    insert = "INSERT INTO dx_radicals (name, num, volume_id)\n" +
+      "  SELECT '%s', '%s', v.id FROM dx_volumes v WHERE v.num = '%s';\n"
+    @out.printf(insert, name, num, volume_num)
+  end
+  def create_wordhead(name, wordid, sw, pinyin, fanqie, exp, radical_num)
+    # TODO: 重文データの追加
+    insert = "INSERT INTO dx_wordheads (name, wordid, sw, pinyin, fanqie, exp, radical_id)\n" +
+      "  SELECT '%s', '%s', '%s', '%s', '%s', '%s', r.id FROM dx_radicals r WHERE r.num = '%s';\n"
+    @out.printf(insert, name, wordid, sw, pinyin, fanqie, exp, radical_num)
+  end
+
+  # 現在は未使用(重文データを追加後に使用)
+  def create_character(xmlid, wordid)
+    insert = "INSERT INTO dx_characters (xmlid, wordid)\n" +
+      "  VALUES('%s', '%s');\n"
+    @out.printf(insert, xmlid, wordid)
+  end
+
+  def SqlConverter.rad_total
+    return @@rad_total
+  end
+  def SqlConverter.inc_volume_num
+    return @@volume += 1
+  end
+end
+
+#Dir.foreach(SOURCE_DIR) do |filename|
+Dir.entries(SOURCE_DIR).sort.each do |filename|
+#  if filename =~ /^sw01\.yml$/ then 
+  if filename =~ /^sw.*\.yml$/ then 
+    printf("\n> %s\n", filename)
+    SqlConverter.inc_volume_num
+    converter = SqlConverter.new(filename)
+  end
+end
+printf("Total: %i\n", SqlConverter.rad_total)