From 031b8014901bd101d92af148a841e9b9c8fb59f5 Mon Sep 17 00:00:00 2001 From: haturatu Date: Wed, 9 Oct 2024 01:36:19 +0900 Subject: fi --- geturl.rb | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'geturl.rb') diff --git a/geturl.rb b/geturl.rb index 804ed91..c5422c3 100755 --- a/geturl.rb +++ b/geturl.rb @@ -7,7 +7,7 @@ require 'nokogiri' require 'charlock_holmes' require 'unicode_utils' -$FILE_PATH = './ok' +$FILE_PATH = '/Your/URLs/list/file' $RESULT_FILE = 'Result' $OTHER_ERROR_FILE = 'Other' $CONCURRENCY = 10 @@ -17,7 +17,7 @@ def is_garbled?(text) end def clean_title(title) - title = title.chars.reject { |ch| UnicodeUtils.general_category(ch).start_with?('C') && !['(', ')', '[', ']', '{', '}', '【', '】', '「', '」', '（' ,'）' ].include?(ch) }.join + title = title.chars.reject { |ch| UnicodeUtils.general_category(ch).start_with?('C') && !['(', ')', '[', ']', '{', '}', '【', '】', '【', '】', '「', '」', '（' ,'）' ].include?(ch) }.join title = UnicodeUtils.nfkc(title) title = title.chars.select(&:valid_encoding?).join title.strip -- cgit v1.2.3