Browse Source

Add tests for edge case Korean names.

scossu 1 year ago
parent
commit
b50a8b8e84
1 changed files with 39 additions and 31 deletions
  1. 39 31
      tests/data/script_samples/korean.csv

+ 39 - 31
tests/data/script_samples/korean.csv

@@ -941,39 +941,47 @@ korean_names,黃仲寶,Hwang Chung-bo,,Hancha; From Y. Lee,
 korean_names,黃俊良,Hwang Chun-nyang,,Hancha; From Y. Lee,
 korean_names,黃信德,Hwang Sin-dŏk,,Hancha; From Y. Lee,
 korean_names,黃允吉,Hwang Yun-gil,,Hancha; From Y. Lee,
+korean_names,남궁 억,Namgung Ŏk,,From Elaine Kim: 2 character last name + 1 character given name,
+korean_names,독고 영,Tokko Yŏng,,From Elaine Kim: 2 character last name + 1 character given name,
+korean_names,선우 정,Sŏnu Chŏng,,From Elaine Kim: 2 character last name + 1 character given name,
+korean_names,선우 종원,Sŏnu Chong-wŏn,,From Elaine Kim: 2 character last name + 2 character given name,
+korean_names,남궁 지영,Namgung Chi-yŏng,,From Elaine Kim: 2 character last name + 2 character given name,
+korean_names,선우 은숙,Sŏnu Ŭn-suk,,From Elaine Kim: 2 character last name + 2 character given name,
+korean_nonames,성철 선사,Sŏngch'ŏl Sŏnsa,"{""capitalize"": ""all""}",From Elaine Kim: No surname,
+korean_nonames,법현,Pŏphyŏn,"{""capitalize"": ""all""}",From Elaine Kim: No surname,
 korean_nonames,민주화 이후 국정 운영,Minjuhwa ihu kukchŏng unyŏng,"{""capitalize"": ""first""}",,
 korean_nonames,曉城 趙 明基 博士 追慕 佛教 史學 論文集,Hyosŏng Cho Myŏng-gi Paksa ch'umo Pulgyo sahak nonmunjip,"{""capitalize"": ""first""}",Not Chinese,
-korean_nonames,결단력,Kyŏltannyŏk,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,상견례,Sanggyŏnnye,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,신여성,Sinnyŏsŏng,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,의견란,Ŭigyŏnnan,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,만석꾼,Mansŏkkun,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,임진란,Imjinnan,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,임진록,Imjinnok,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,옛이야기,Yenniyagi,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,전달자,Chŏndalcha,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,독해법,Tokhaepŏp,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,방지법,Pangjipŏp,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,추진법,Ch'ujinpŏp,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,여행법,Yŏhaengpŏp,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,사랑법,Sarangpŏp,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,호박꽃,Hobakkot,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,공권력,Kongkwŏnnyŏk,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,생산량,Saengsannyang,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,이원론,Iwŏnnon,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,동원령,Tongwŏnnyŏng,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,한여름,Hannyŏrŭm,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,휘발유,Hwiballyu,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,꽃잎,Kkonnip,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,솔잎,Sollip,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,활동가,Hwaltongga,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,별일,Pyŏllil,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,노근리,Nogŭn-ni,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,창원군,Ch'angwŏn-gun,"{""capitalize"": ""first""}",From Elaine,
-korean_nonames,신여자,Sinnyoja,"{""capitalize"": ""first""}",from Elaine,
-korean_nonames,밀당,miltang,"{""capitalize"": ""first""}",from Elaine,
-korean_nonames,말직,malchik,"{""capitalize"": ""first""}",from Elaine,
-korean_nonames,토벌대,t'obŏltae,"{""capitalize"": ""first""}",from Elaine,
+korean_nonames,결단력,Kyŏltannyŏk,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,상견례,Sanggyŏnnye,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,신여성,Sinnyŏsŏng,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,의견란,Ŭigyŏnnan,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,만석꾼,Mansŏkkun,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,임진란,Imjinnan,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,임진록,Imjinnok,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,옛이야기,Yenniyagi,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,전달자,Chŏndalcha,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,독해법,Tokhaepŏp,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,방지법,Pangjipŏp,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,추진법,Ch'ujinpŏp,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,여행법,Yŏhaengpŏp,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,사랑법,Sarangpŏp,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,호박꽃,Hobakkot,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,공권력,Kongkwŏnnyŏk,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,생산량,Saengsannyang,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,이원론,Iwŏnnon,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,동원령,Tongwŏnnyŏng,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,한여름,Hannyŏrŭm,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,휘발유,Hwiballyu,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,꽃잎,Kkonnip,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,솔잎,Sollip,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,활동가,Hwaltongga,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,별일,Pyŏllil,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,노근리,Nogŭn-ni,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,창원군,Ch'angwŏn-gun,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,신여자,Sinnyoja,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,밀당,miltang,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,말직,malchik,"{""capitalize"": ""first""}",From Elaine Kim,
+korean_nonames,토벌대,t'obŏltae,"{""capitalize"": ""first""}",From Elaine Kim,
 korean_nonames,民法 과 法學 의 重要 問題,Minpŏp kwa pŏphak ŭi chungyo munje,"{""capitalize"": ""first""}",Not Chinese,
 korean_nonames,그래도 돈 버는 사람 은 있다,Kŭraedo ton pŏnŭn saram ŭn itta,"{""capitalize"": ""first""}",From K-Romanizer,
 korean_nonames,근대 계몽기 문학 과 독자 의 발견,Kŭndae kyemonggi munhak kwa tokcha ŭi palgyŏn,"{""capitalize"": ""first""}",From K-Romanizer,