Fixes for Malay$ and reorderings due to new ICU4J.

diff --git a/src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java b/src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java
index 892c3203a82bed3b66673e2a26a974b3b3d0a382..57e76cc7d435acc8999535fe097cc68661be2380 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java
+++ b/src/com/hughes/android/dictionary/engine/DictionaryBuilderMain.java
@@ -150,7 +150,7 @@ public class DictionaryBuilderMain extends TestCase {
   isoToDedication.put("PT", "Wiktionary-based Portuguese dictionary dedicated to Carlos Melo, one Tough Mudder.");
   isoToDedication.put("RO", "Wiktionary-based Romanian dictionary dedicated to Radu Teodorescu.");
   isoToDedication.put("RU", "Wiktionary-based Russian dictionary dedicated to Maxim Aronin--best friend always!.");
-  isoToDedication.put("SR", "Wiktionary-based Serbian dictionary dedicated to Filip Crnogorac--thanks for the honey.");
+  isoToDedication.put("SR", "Wiktionary-based Serbian dictionary dedicated to Filip Crnogorac--thanks for the honey!");
   isoToDedication.put("ES", "Wiktionary-based Spanish dictionary made especially for Carolina Tropini! <3 XoXoXXXXX!");
   isoToDedication.put("SV", "Wiktionary-based Swedish dictionary dedicated to Kajsa Palmblad--björn kramar!");
   }

diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
index 6341fa9e2c6c19df199e2c5f5f008771c6284ade..3af9a139a63f9e6e21e43af3b6ed1eda7e6e5bfa 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
+++ b/src/com/hughes/android/dictionary/parser/wiktionary/WiktionaryLangs.java
@@ -101,13 +101,14 @@ public class WiktionaryLangs {
     isoCodeToEnWikiName.put("AZ", "Azeri");
     isoCodeToEnWikiName.put("EU", "Basque");
     isoCodeToEnWikiName.put("BR", "Breton");
-    isoCodeToEnWikiName.put("MR", "Burmese");
+    isoCodeToEnWikiName.put("MR", "Marathi");
     isoCodeToEnWikiName.put("FO", "Faroese");
     isoCodeToEnWikiName.put("GL", "Galician");
     isoCodeToEnWikiName.put("KA", "Georgian");
     isoCodeToEnWikiName.put("HT", "Haitian Creole");
     isoCodeToEnWikiName.put("LB", "Luxembourgish");
     isoCodeToEnWikiName.put("MK", "Macedonian");
+    isoCodeToEnWikiName.put("GV", "Manx");
     
     // No longer exists in EN:
     // isoCodeToEnWikiName.put("BS", "Bosnian");
@@ -199,7 +200,10 @@ public class WiktionaryLangs {
           return null;
       }
       if (name.indexOf('|') != -1) {
-          return name.substring(name.indexOf('|'));
+          return name.substring(0, name.indexOf('|'));
+      }
+      if (name.indexOf('$') != -1) {
+          return name.substring(0, name.indexOf('$'));
       }
       return name;  // can be null.
   }

diff --git a/testdata/goldens/wiktionary.cmn_en.quickdic.text b/testdata/goldens/wiktionary.cmn_en.quickdic.text
index 141150d61de3038353c4f90d0f6a6e81a8f81c28..70365bcbf9eab535a3c126389409e02c33785f12 100644 (file)
--- a/testdata/goldens/wiktionary.cmn_en.quickdic.text
+++ b/testdata/goldens/wiktionary.cmn_en.quickdic.text
@@ -763,12 +763,6 @@ Index: cmn cmn->EN
   二月 (èryuè) :: February (second month of the Gregorian calendar) (proper noun)
 ===example===
   (the adjectives are in a dictionary form) 越……越…… (yuè...yuè...) (example: 越<u>热</u>越<u>好</u> yuè rè yuè hǎo "the hotter the better"), 愈……愈…… (yù...yù...) :: the (the + ～comparative, the + comparative) (adverb)
-***略語***
-  縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
-  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
-***略语***
-  縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
-  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
 ===faan1===
   (Cantonese) 南瓜 (naam4 gwaa1),番瓜 (faan1 gwaa1) :: pumpkin (plant) (noun)
   (Cantonese) 南瓜 (naam4 gwaa1),番瓜 (faan1 gwaa1) :: pumpkin (fruit of this plant) (noun)
@@ -1868,6 +1862,12 @@ Index: cmn cmn->EN
 ===lüèyǔ===
   縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
   縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
+***略語***
+  縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
+  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
+***略语***
+  縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
+  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
 ===luh===
   (Min Nan) 啤酒 (bih-luh), 麥仔酒 (be̍h-á-chiú), 米仔酒 (bí-á-chiú) :: beer (alcoholic drink made of malt) (noun)
 ===luk6===
@@ -2619,10 +2619,10 @@ Index: cmn cmn->EN
   赦免 (shèmiǎn), 赦罪 (shèzuì) :: absolve (theology: pronounce free or give absolution from sin) (verb)
 ===sheng===
   女性服务生 (nuxing fuwu-sheng) :: bellgirl (a female bellhop) (noun)
-***省略***
-  減少, 减少 (jiǎnshǎo), 省略 (shěnglüè) :: abate (to deduct, to omit) (verb)
 ===shěnglüè===
   減少, 减少 (jiǎnshǎo), 省略 (shěnglüè) :: abate (to deduct, to omit) (verb)
+***省略***
+  減少, 减少 (jiǎnshǎo), 省略 (shěnglüè) :: abate (to deduct, to omit) (verb)
 ===shēngzhǎng===
   生長, 生长 (shēngzhǎng) :: growth (increase in size) (noun)
 ***生長***
@@ -2895,12 +2895,12 @@ Index: cmn cmn->EN
   蘇聯, 苏联 (Sūlián) :: USSR (Union of Soviet Socialist Republics) (proper noun)
 ===suǒ===
   廁所, 厕所 (cè suǒ) :: can (toilet) (noun)
+===suōlüècí===
+  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
 ***縮略詞***
   縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
 ***缩略词***
   縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
-===suōlüècí===
-  縮寫, 缩写 (suōxiě), 縮略詞, 缩略词 (suōlüècí), 略語, 略语 (lüèyǔ) :: acronym (word formed by initial letters) (noun)
 ===suōxiě===
   縮寫, 缩写 (suōxiě); 簡寫, 简写 (jiǎnxiě); 略語, 略语 (lüèyǔ) :: abbreviation (shortened or contracted form of a word or phrase) (noun)
   縮寫, 缩写 (suōxiě) :: abbreviation (act or result of shortening or reducing) (noun)

diff --git a/testdata/goldens/wiktionary.th_th.quickdic.text b/testdata/goldens/wiktionary.th_th.quickdic.text
index b802d411dd4e23b380c2fd1d6ac13a80282c1590..01f53bcec6e26b927295029bf3def3f693fc9d16 100644 (file)
--- a/testdata/goldens/wiktionary.th_th.quickdic.text
+++ b/testdata/goldens/wiktionary.th_th.quickdic.text
@@ -2,8 +2,6 @@ dictInfo=SomeWikiData
 EntrySource: enwiktionary.thai 4998
 
 Index: TH TH->EN
-***ฯลฯ***
-  ฯลฯ (abbreviation) :: et cetera
 ***๐***
   ๐ (ศูนย์, súún) :: 0 (zero)
 ***๑***
@@ -1758,6 +1756,8 @@ Index: TH TH->EN
 ***กอด***
   กอด (/gɔd/) {th-verb} :: To hug
   กอด (/gɔd/) {th-verb} :: To hold on to someone
+***ฯลฯ***
+  ฯลฯ (abbreviation) :: et cetera
 ===ḹ===
   ฦๅ (ḹ; leu) (letter) :: A letter of the Thai alphabet, considered a vowel.
 ***ล***