better fuzzy search

Dockerfile

@@ -5,7 +5,7 @@ COPY . .
 
 RUN apt-get -y update
 RUN apt-get -y upgrade
-RUN apt-get install -y sqlite3 libsqlite3-dev
+RUN apt-get install -y sqlite3 libsqlite3-dev opencc libopencc-dev
 
 RUN bundle install
 

Gemfile

@@ -8,3 +8,4 @@ gem 'concurrent-ruby'
 gem 'discordrb'
 gem 'sqlite3'
 gem 'any_ascii'
+gem 'ropencc'

Gemfile.lock

@@ -17,7 +17,7 @@ GEM
     http-accept (1.7.0)
     http-cookie (1.0.5)
       domain_name (~> 0.5)
-    mime-types (3.5.1)
+    mime-types (3.5.2)
       mime-types-data (~> 3.2015)
     mime-types-data (3.2023.1205)
     netrc (0.11.0)
@@ -29,6 +29,8 @@ GEM
       http-cookie (>= 1.0.2, < 2.0)
       mime-types (>= 1.16, < 4.0)
       netrc (~> 0.8)
+    ropencc (0.0.6)
+      ffi (~> 1.0)
     sqlite3 (1.7.0-x86_64-linux)
     websocket (1.2.10)
     websocket-client-simple (0.8.0)
@@ -43,6 +45,7 @@ DEPENDENCIES
   concurrent-ruby
   discordrb
   rake
+  ropencc
   sqlite3
 
 BUNDLED WITH

README.md

@@ -23,6 +23,7 @@ or copy the useful files to the paths specified in `config.yml`.
 | `LYRICAT_RES_DIR` | The dir containing resource files. The base dir of items in `res` in config. Defaults to `./res`. |
 | `LYRICAT_CONFIG` | The path to config file. Relative to `LYRICAT_DATA_DIR`. Defaults to `config.yml`. |
 | `LYRICAT_DAN` | The path to the file specifying dan courses. Relative to `LYRICAT_DATA_DIR`. Defaults to `dan.yml`. |
+| `LYRICAT_ALIASES` | The path to the file specifying aliases. Relative to `LYRICAT_DATA_DIR`. Defaults to `aliases.yml`. |
 | `LYRICAT_THREAD_COUNT` | The number of threads to use for parallel HTTP requests. Defaults to 8. |
 | `LYRICAT_RETRY_COUNT` | The number of retries when communicating with Lyrica's server. Defaults to 3. |
 | `LYRICAT_STATIC_SESSION_TOKEN` | (Required) The session token used to retrieve leaderboards. |

data/aliases.yml

@@ -1 +1,18 @@
 ---
+songs:
+  源自苍穹: 26
+  蝶舞: 115
+  呱: 225
+  千魔王: 56
+  绿魔王: 50
+  狗咬: 202
+  destroyer: 197
+  d0: 197
+  脑力: 196
+  吃山人: 206
+  吞山人: 206
+  吃山者: 206
+  吞山者: 206
+  小红帽: 109
+  乡野整活: 27
+  神谕: 213

lib/lyricat.rb

@@ -9,6 +9,7 @@
 require 'sqlite3'
 require 'concurrent'
 require 'any_ascii'
+require 'ropencc'
 
 require 'patches'
 

lib/lyricat/song.rb

@@ -263,30 +263,43 @@ def best n, sorted, session_token
 				end.filter { _1[:mr] > 0 }
 			end
 
-			def fuzzy_search lang, query
+			def fuzzy_search lang, query, excluded = []
 				if query.like_int?
 					id = query.to_i
 					return id if LIB[id]
 				end
+				if from_alias = ALIASES[query]
+					return from_alias
+				end
+				get_forms = ->original do
+					ascii = AnyAscii.transliterate original
+					abbr0 = original.upper_letters
+					abbr0 = nil if abbr0.length <= 1
+					abbr1 = original.split.map { _1[0] }.join
+					abbr1 = nil if abbr1.length <= 1
+					abbr2 = original.split(/[^\w]/).map { _1[0] }.join
+					abbr2 = nil if abbr2.length <= 1
+					[original, ascii, abbr0, abbr1, abbr2].compact
+				end
+				query_forms = get_forms.(query)
+				if query =~ /^(.*?)(\d+)$/
+					n = $2.to_i
+					alter_forms = get_forms.($1) if n > 1
+					searched = Set.new
+				end
+				filtered_lib = LIB.reject { excluded.include? _2.song_id }
 				[lang, nil].each do |l|
-					if o = LIB.find { _2.match_name query, true, l }
-						return o[0]
-					end
-					if o = LIB.find { _2.match_roman1 query, true, l }
-						return o[0]
-					end
-					if o = LIB.find { _2.match_roman2 query, true, l }
-						return o[0]
-					end
-					if o = LIB.find { _2.match_roman1 query, false, l }
-						return o[0]
-					end
-					if o = LIB.find { _2.match_roman2 query, false, l }
-						return o[0]
-					end
-					if o = LIB.find { _2.match_name query, false, l }
-						return o[0]
+					match = ->meth, strong = false do
+						filtered_lib.find do |song_id, song|
+							next true if query_forms.any? { song.__send__ meth, _1, strong, l }
+							if alter_forms&.any? { song.__send__ meth, _1, strong, l }
+								searched.add song_id
+								next true if n == searched.size
+							end
+						end&.first
 					end
+					o = match.(:match_name, true) || match.(:match_roman1, true) || match.(:match_roman2, true) || match.(:match_roman1) || match.(:match_roman2) || match.(:match_name)
+					return o if o
 				end
 				nil
 			end
@@ -342,6 +355,10 @@ def select_charts_by_difficulty diff, &block
 			labels[index].delete :index
 			negative
 		end.freeze
+
+		ALIASES = YAML.load_file(File.join Lyricat::DATA_DIR, ENV['LYRICAT_ALIASES'] || 'aliases.yml')['songs']
+		ALIASES.merge! ALIASES.transform_keys { Ropencc.conv 's2t', _1 }
+		ALIASES.freeze
 	end
 
 end

lib/lyricat/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Lyricat
-	VERSION = '0.5.0'
+	VERSION = '0.6.0'
 end