cbeta 3.2.0 → 3.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 982e4d10689bc1fa6f71d8c27e880cbfdcf248c5cbdb8994ad65bf6d2c5e5259
-  data.tar.gz: 7c6cfa28c5f48f1f84bb4ad2921e45ac1db1f2039626999a1fd620bbe91c9564
+  metadata.gz: d6371fac001d191b3245ba02d1b3287cedfe46213bea091d9259d9d24c2454b7
+  data.tar.gz: f71e61e1598462fd02fed5eb8e1c0111ee259d1b3bb85424ce869a9f0ea0dad8
 SHA512:
-  metadata.gz: b1ac00d3090b9a19df3e3d25d4cf1c2dc4f50a37e6de3cbbdf4b09a9cd6cb68b1fa7700e825cad8352628f7a9bab8ed34a916abbd4d8380a68d5783e6aef8513
-  data.tar.gz: b1a6f2ebb218737938ae7bc35eb5bc648cc544c8ac964647054335bf15f218e17db7a337e8b529060eccd69094872df6725ebed18b0e7cb0681253f3eccdf23d
+  metadata.gz: fcc4e69ae85ef617ab214e489842d6c82a4c61fa7326c5f3d9ada35f49e71b602bb4f7c4153aeb9e3aa5c978a52ac19cc926842ea2535e6518b4c518e80f274a
+  data.tar.gz: 837dc5642b1285cf602a66572908d0ac4b3b3a32e7a509ee1c19337ea5f9363b3cbf9b1d59cab65565f45dd39808eff511383521c633ecc08fe480c9c360a6bb

data/lib/cbeta/gaiji.rb

@@ -1,10 +1,12 @@
 require 'json'
+require 'unihan2'
 
 # 存取 CBETA 缺字資料庫
 class CBETA::Gaiji
   
   # 載入 CBETA 缺字資料庫
   def initialize
+    @us = CBETA::UnicodeService.new
     folder = File.join(File.dirname(__FILE__), '../data')
     fn = File.join(folder, 'cbeta_gaiji.json')
     @gaijis = JSON.parse(File.read(fn))
@@ -73,9 +75,12 @@ class CBETA::Gaiji
     
     if gid.start_with? 'CB'
       cb_priority.each do |k|
-        if k == 'PUA'
+        case k
+        when 'PUA'
           return CBETA.pua(gid)
-        elsif g.key? k
+        when 'uni_char', 'norm_uni_char'
+          return g[k] if @us.level2?(g[k])
+        else
           return g[k] unless g[k].empty?
         end
       end

data/lib/cbeta/unicode_service.rb

@@ -0,0 +1,19 @@
+class CBETA::UnicodeService
+  def initialize
+    @u2 = Unihan2.new
+  end
+
+  def level1?(code)
+    return false if code.nil?
+    # Unicode 3.0 以內 在 mobile 可以正確顯示
+    v = @u2.ver(code)
+    raise "Unihan2.ver 回傳 nil, code: #{code}" if v.nil?
+    v <= 3
+  end
+
+  def level2?(code)
+    return false if code.nil?
+    # Unicode 10 以內 在 desktop 有字型可以顯示
+    @u2.ver(code) <= 10
+  end
+end

data/lib/cbeta/xml_document.rb

@@ -3,6 +3,8 @@ require 'nokogiri'
 class CBETA::XMLDocument
   PASS = %w(back graphic mulu rdg sic teiHeader)
 
+  attr_reader :doc
+
   def initialize(string_or_io)
     @doc = Nokogiri::XML(string_or_io)
     @doc.remove_namespaces!
@@ -44,9 +46,21 @@ class CBETA::XMLDocument
     '　'
   end
 
+  def e_caption(e)
+    traverse(e) + "\n"
+  end
+
   def e_cell(e)
     traverse(e) + "\n"
   end
+
+  def e_cit(e)
+    traverse(e)
+  end
+
+  def e_closer(e)
+    traverse(e) + "\n"
+  end
   
   def e_corr(e)
     traverse(e)
@@ -63,6 +77,10 @@ class CBETA::XMLDocument
   def e_div(e)
     traverse(e)
   end
+
+  def e_docAuthor(e)
+    traverse(e)
+  end
   
   def e_docNumber(e)
     traverse(e) + "\n"
@@ -73,6 +91,12 @@ class CBETA::XMLDocument
   end
 
   def e_figure(e)
+    r = traverse(e)
+    r << "\n" unless r.empty?
+    r
+  end
+
+  def e_figDesc(e)
     traverse(e) + "\n"
   end
 
@@ -88,14 +112,19 @@ class CBETA::XMLDocument
       cb_priority = %w(uni_char composition)
     end
 
-    gid = e['ref'][1..-1]
-    r = @gaiji.to_s(gid, cb_priority:)
-    abort "Line:#{__LINE__} 缺字處理失敗:#{gid}" if r.nil?
-    r
+    gid = e['ref'].delete_prefix('#')
+
+    unless @gaiji.key?(gid)
+      raise "在 CBETA 缺字庫中找不到此缺字碼: #{gid}"
+    end
+
+    @gaiji.to_s(gid, cb_priority:)
   end
 
   def e_head(e)
-    traverse(e) + "\n"
+    r = traverse(e)
+    r << "\n" unless r.empty?
+    r
   end
 
   def e_hi(e)
@@ -178,6 +207,14 @@ class CBETA::XMLDocument
     ''
   end
 
+  def e_quote(e)
+    traverse(e)
+  end
+
+  def e_ref(e)
+    traverse(e)
+  end
+
   def e_reg(e)
     r = ''
     choice = e.at_xpath('ancestor::choice')
@@ -189,6 +226,10 @@ class CBETA::XMLDocument
     traverse(e) + "\n"
   end
 
+  def e_seg(e)
+    traverse(e)
+  end
+
   def e_sg(e)
     '(' + traverse(e) + ')'
   end
@@ -266,7 +307,6 @@ class CBETA::XMLDocument
     r
   end
 
-
   def handle_node(e)
     return '' if e.comment?
     return handle_text(e) if e.text?

data/lib/cbeta.rb

@@ -234,4 +234,5 @@ require 'cbeta/p5a_to_simple_html'
 require 'cbeta/p5a_to_text'
 require 'cbeta/p5a_validator'
 require 'cbeta/html_to_text'
+require 'cbeta/unicode_service'
 require 'cbeta/xml_document'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: cbeta
 version: !ruby/object:Gem::Version
-  version: 3.2.0
+  version: 3.3.0
 platform: ruby
 authors:
 - Ray Chou
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-03-09 00:00:00.000000000 Z
+date: 2024-03-14 00:00:00.000000000 Z
 dependencies: []
 description: Ruby gem for use Chinese Buddhist Text resources made by CBETA (http://www.cbeta.org).
 email: zhoubx@gmail.com
@@ -31,6 +31,7 @@ files:
 - lib/cbeta/p5a_to_simple_html.rb
 - lib/cbeta/p5a_to_text.rb
 - lib/cbeta/p5a_validator.rb
+- lib/cbeta/unicode_service.rb
 - lib/cbeta/xml_document.rb
 - lib/data/canons.csv
 - lib/data/categories.json