cantonese 0.1.2 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ae7740aa252f2be334b9f68abfcf916fe692320f
-  data.tar.gz: a4dbd9c30c8bea7fdfd197cd43c801f0c7a53d6c
+  metadata.gz: 4ddaa4f96c77b0c35bb61e83b2feef43a234f844
+  data.tar.gz: 5684236f0d3ba3626de4a10ebdc1ac9eeeac3734
 SHA512:
-  metadata.gz: baaa4749abad128a550b0a56d3157c2380a06cc9c0974c234d1356772a2d6bfe3974493a664d7a5b47a357817b001e5ff3c1ede3a9faa4e076423d457a4b2690
-  data.tar.gz: a1c385ecedc1f7175080a6059263215db08170b74db5c004fb0ddc9a2ba8699c52ae9eeacedbcf48809117e68f16140862207ea5dbf0d39b673f865f0fd1216b
+  metadata.gz: a02f42b7aac35e3e2071114a868be195d592f2a93eb087f304afa058700d62664d5595160ea57d208b49bb2c3eb0f7cbc4b973b3e70b41933ee9682d87be0f81
+  data.tar.gz: efc2ba3e781a25a0a2cf72c75d52eda78298a6d0818cbbc382f1832042a240ccfce9989ba6f2a10331806eb911c40bde627e80e731f16c345fe82d0e9098cf03

data/lib/cantonese/scraper/word_scraper.rb

@@ -30,14 +30,15 @@ module Cantonese
         classified  = doc.search("//*[@class = 't' and .='字音分類:']/following-sibling::td[1]").text rescue nil
         big5        = doc.search("//*[@class = 't' and .='大五碼:']/following-sibling::td[1]").text rescue nil
         chanjie     = doc.search("//*[@class = 't' and .='倉頡碼:']/following-sibling::td[1]").text rescue nil
-        frequency   = doc.search("//*[@class = 't' and .='頻序 / 頻次:']/following-sibling::td[1]").text rescue nil
+        rank_and_frequency   = doc.search("//*[@class = 't' and .='頻序 / 頻次:']/following-sibling::td[1]").text rescue nil
         combination = doc.search("//text()[.='配搭點:']/following-sibling::a").collect{|a| a.text}
+        rank, frequency = rank_and_frequency.split("/").collect{|word| word.strip.to_i }
 
         syllable    = doc.search('//form/table[1]/tr[position()>1]').collect do |row|
           sound = row.search("./td[1]")
-          initial = sound.xpath("./*[@color='red']").text rescue nil
-          final = sound.xpath("./*[@color='green']").text rescue nil
-          tone = sound.xpath("./*[@color='blue']").text rescue nil
+          initial = sound.xpath("./*[@color='red']").text rescue ""
+          final = sound.xpath("./*[@color='green']").text rescue ""
+          tone = sound.xpath("./*[@color='blue']").text rescue ""
           sound_text = sound.text
           pronunciation = "http://humanum.arts.cuhk.edu.hk/Lexis/lexi-can/sound/#{sound_text}.wav"
 
@@ -52,6 +53,7 @@ module Cantonese
           end
 
           {
+            :full => "#{initial}#{final}#{tone}",
             :initial => initial, 
             :final => final, 
             :tone => tone,
@@ -68,6 +70,7 @@ module Cantonese
           :classified => classified,
           :big5 => big5,
           :chanjie => chanjie,
+          :rank => rank,
           :frequency => frequency,
           :syllable => syllable,
           :combination => combination

data/lib/cantonese/version.rb

@@ -1,3 +1,3 @@
 module Cantonese
-  VERSION = "0.1.2"
+  VERSION = "0.2.0"
 end

data/spec/scraper/word_scraper_spec.rb

@@ -15,11 +15,14 @@ describe Cantonese::Scraper::WordScraper do
 
       expect(word[:big5]).to eq("A440")
       expect(word[:chanjie]).to eq("一")
-      expect(word[:frequency]).to eq("2 / 166396")
+
+      expect(word[:rank]).to eq(2)
+      expect(word[:frequency]).to eq(166396)
       expect(word[:combination]).to be_a(Array)
       expect(word[:combination]).to be_include("山")
 
       expect(word[:syllable]).to be_a(Array)
+      expect(word[:syllable][0][:full]).to eq("jat1")
       expect(word[:syllable][0][:initial]).to eq("j")
       expect(word[:syllable][0][:final]).to eq("at")
       expect(word[:syllable][0][:tone]).to eq("1")
@@ -40,17 +43,20 @@ describe Cantonese::Scraper::WordScraper do
 
       expect(word[:big5]).to eq("A569")
       expect(word[:chanjie]).to eq("一弓口")
-      expect(word[:frequency]).to eq("36 / 40754")
+      expect(word[:rank]).to eq(36)
+      expect(word[:frequency]).to eq(40754)
       expect(word[:combination]).to be_a(Array)
       expect(word[:combination]).to be_include("不")
 
       expect(word[:syllable]).to be_a(Array)
+      expect(word[:syllable][0][:full]).to eq("hak1")
       expect(word[:syllable][0][:initial]).to eq("h")
       expect(word[:syllable][0][:final]).to eq("ak")
       expect(word[:syllable][0][:tone]).to eq("1")
       expect(word[:syllable][0][:examples]).to be_a(Array)
       expect(word[:syllable][0][:examples]).to be_include("可汗")
 
+      expect(word[:syllable][1][:full]).to eq("ho2")
       expect(word[:syllable][1][:initial]).to eq("h")
       expect(word[:syllable][1][:final]).to eq("o")
       expect(word[:syllable][1][:tone]).to eq("2")

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: cantonese
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.2.0
 platform: ruby
 authors:
 - Francis Chong