feat: 英文词中数字和标点自动转写 (#326)

2023-06-07 01:36:45 +08:00 · 2023-06-07 01:36:45 +08:00 · 3f67fc6a0f
commit 3f67fc6a0f
parent 141400af9b
2 changed files with 204 additions and 58 deletions
--- a/en_dicts/en_ext.dict.yaml
+++ b/en_dicts/en_ext.dict.yaml
@ -124,10 +124,10 @@ Phoebe	Phoebe

 # 带权重的系列
 iPhone	iPhone	999
-iPhone 14	iPhone	4
-iPhone 14 Plus	iPhone	3
-iPhone 14 Pro	iPhone	2
-iPhone 14 Pro Max	iPhone	1
+iPhone 14	iPhone14	4
+iPhone 14 Plus	iPhone14Plus	3
+iPhone 14 Pro	iPhone14Pro	2
+iPhone 14 Pro Max	iPhone14ProMax	1
 iPhone SE	iPhoneSE

 Mac OS X	MacOSX	999
@ -185,21 +185,21 @@ Ventura	Ventura
 Sonoma	Sonoma

 Windows	Windows	999
-Windows 95	Windows	1
-Windows 98	Windows	2
+Windows 95	Windows95	1
+Windows 98	Windows98	2
 Windows Me	Windows	3
-Windows 2000	Windows	4
+Windows 2000	Windows2000	4
 Windows XP	Windows	5
 Windows XP	WindowsXP
-Windows Vista	Windows	6
-Windows 7	Windows	7
-Windows 8	Windows	8
-Windows 10	Windows	10
-Windows 11	Windows	11
+Windows Vista	WindowsVista	6
+Windows 7	Windows7	7
+Windows 8	Windows8	8
+Windows 10	Windows10	10
+Windows 11	Windows11	11

-ERC-20	ERC
-ERC-721	ERC
-ERC-1155	ERC
+ERC-20	ERC-20
+ERC-721	ERC-721
+ERC-1155	ERC-1155


 # +_+
@ -284,10 +284,10 @@ Secure Shell	SSH
 Secure Shell	SecureShell
 ISO	ISO
 International Organization for Standardization	ISO
-ISO 8601	ISO
+ISO 8601	ISO8601
 RFC	RFC
 Request for Comments	RFC
-RFC 3339	RFC
+RFC 3339	RFC3339
 RFCs	RFCs
 IETF	IETF
 Internet Engineering Task Force	IETF
@ -439,12 +439,8 @@ VIP	VIP
 D.VA	DVA
 Logo	Logo
 Telnet	Telnet
-IPv4	IPv
-IPv6	IPv
-IPv4	IPvs
-IPv6	IPvl
-IPv4	IPvsi
-IPv6	IPvliu
+IPv4	IPv4
+IPv6	IPv6
 Unix	Unix
 Rick and Morty	RickandMorty
 Tokyo	Tokyo
@ -561,11 +557,9 @@ Structured Query Language	StructuredQueryLanguage
 SQLite	SQLite
 Sourcetree	Sourcetree
 Spiritfarer	Spiritfarer
-TCP/IP	tcpip
+TCP/IP	TCP/IP
 Twitter	Twitter
-V2EX	Ver
-V2EX	VerEX
-V2EX	VtoEX
+V2EX	V2EX
 VSCode	VSCode
 Vue	Vue
 Vue.js	Vuejs
@ -767,10 +761,8 @@ MIPS	MIPS
 Moke	Moke
 Moneywiz	Moneywiz
 Moom	Moom
-MP3	MPs
-MP3	MPsan
-MP4	MPs
-MP4	MPsi
+MP3	MP3
+MP4	MP4
 MplayerX	MplayerX
 Netflix	Netflix
 Noizio	Noizio
@ -1106,7 +1098,7 @@ Shottr	Shottr
 turbo	turbo
 Turbo Boost	TurboBoost
 Turbo Boost Switcher	TurboBoostSwitcher
-P2P	ptop
+P2P	P2P
 OPPO	OPPO
 vivo	vivo
 Uno	Uno
@ -1143,11 +1135,11 @@ Minions	Minions
 Rush B	RushB
 Rush A	RushA
 AK	AK
-AK47	AK
-AK74	AK
+AK47	AK47
+AK74	AK74
 Mastercard	Mastercard
 COVID	COVID
-COVID-19	COVID
+COVID-19	COVID-19
 Wallpaper Engine	WallpaperEngine
 Erlang	Erlang
 FAQ	FAQ
@ -1626,7 +1618,8 @@ DVD	DVD
 CD	CD
 eSIM	eSIM
 UTF	UTF
-UTF-8	UTF
+UTF-8	UTF-8
+UTF-16	UTF-16
 ASCII	ASCII
 ANSI	ANSI
 IEEE	IEEE
@ -1648,6 +1641,7 @@ ribonucleic acid	RNA
 mRNA	mRNA
 messenger RNA	mRNA
 C++	Cpp
+C++	C++
 Python	Python
 Perl	Perl
 Swift	Swift
@ -2040,8 +2034,8 @@ WWII	WWII
 World War II	WWII
 WWIII	WWIII
 World War III	WWIII
-PL/pgSQL	PLpgSQL
-Procedural Language / PostGres Structured Query Language	PLpgSQL
+PL/pgSQL	PL/pgSQL
+Procedural Language / PostGres Structured Query Language	PL/pgSQL
 IQ	IQ
 Intelligence quotient	IQ
 quotient	quotient
@ -2198,3 +2192,8 @@ Apple Vision	AppleVision
 Apple Vision Pro	AppleVisionPro
 Vision Pro	VisionPro
 shithead	shithead
+i18n	i18n
+internationalization	i18n
+GPT-4	GPT-4
+a11y	a11y
+accessibility	a11y
--- a/melt_eng.schema.yaml
+++ b/melt_eng.schema.yaml
@ -41,12 +41,53 @@ engine:
    - uniquifier

 speller:
-  alphabet: zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA-_
+  alphabet: zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA
  delimiter: " '"
  algebra:
-    # 为编码派生新的拼写：
-    # 删除特殊字符
-    - derive/['\-_+,.]+//
+    # 根据当前所用中文方案修改 __include 的值
+    # algebra_rime_ice ; algebra_flypy ; algebra_mspy ; algebra_ziguang ; algebra_double_pinyin
+    __include: algebra_rime_ice
+    __append:
+      # 数字派生
+      - derive/1([4-7|9])/$1teen/
+      - derive/11/eleven/
+      - derive/12/twelve/
+      - derive/13/thirteen/
+      - derive/15/fifteen/
+      - derive/18/eighteen/
+      - derive/0/o/ # 如 1000 -> oneOOO
+      - derive/0/O/
+      - derive/0/zero/
+      - derive/1/one/
+      - derive/10/ten/
+      - derive/2/to/ #
+      - derive/2/two/
+      - derive/3/three/
+      - derive/4/for/
+      - derive/4/four/
+      - derive/5/five/
+      - derive/6/six/
+      - derive/7/seven/
+      - derive/8/eight/
+      - derive/9/nine/
+      # 符号派生
+      - derive/\+/plus/
+      - derive/\./dot/
+      - derive/@/at/
+      - derive/-/hyphen/
+      - derive / slash
+      # 派生无单个特殊字符的拼写
+      - derive/[.]//
+      - derive/[+]//
+      - derive/[@]//
+      - derive/[-]//
+      - derive/[_]//
+      # 派生无任何非字母数字字符的拼写
+      - derive/[^a-zA-Z0-9]//
+      # 禁用非英文、数字开头的编码
+      - erase/^[\W|_].+$/
+      # 当开启自动调频时，保证大写+小写开头的词排在全大写的前面
+      - derive/^([A-Z][A-Z]+)[a-z]+/$1/
      # 全小写
      - derive/^.+$/\L$0/
      # 全大写
@ -63,6 +104,7 @@ speller:
      - derive/^([a-z]{8})/\U$1/
      - derive/^([a-z]{9})/\U$1/
      - derive/^([a-z]{10})/\U$1/
+
 translator:
  dictionary: melt_eng
  spelling_hints: 9
@ -75,3 +117,108 @@ punctuator:

 recognizer:
  import_preset: default
+
+algebra_rime_ice:
+  - derive/(?<!\d)1([1-9])(?!\d)/shi$1/
+  - derive/([1-9])0000(?!0)/$1wan/
+  - derive/([1-9])000(?!0)/$1qian/
+  - derive/([1-9])00(?!0)/$1bai/
+  - derive/([2-9])0(?!0)/$1shi/
+  - derive/(?<!\d)([2-9])([1-9])(?!\d)/$1shi$2/
+  - derive/\./dian/
+  - derive/10/shi/
+  - derive/0/ling/
+  - derive/1/yi/
+  - derive/2/er/
+  - derive/2/liang/
+  - derive/3/san/
+  - derive/4/si/
+  - derive/5/wu/
+  - derive/6/liu/
+  - derive/7/qi/
+  - derive/8/ba/
+  - derive/9/jiu/
+
+algebra_flypy:
+  - derive/(?<!\d)1([1-9])(?!\d)/ui$1/
+  - derive/([1-9])0000(?!0)/$1wj/
+  - derive/([1-9])000(?!0)/$1qm/
+  - derive/([1-9])00(?!0)/$1bd/
+  - derive/([2-9])0(?!0)/$1ui/
+  - derive/(?<!\d)([2-9])([1-9])(?!\d)/$1ui$2/
+  - derive/\./dm/
+  - derive/10/ui/
+  - derive/0/lk/
+  - derive/1/yi/
+  - derive/2/er/
+  - derive/2/ll/
+  - derive/3/sj/
+  - derive/4/si/
+  - derive/5/wu/
+  - derive/6/lq/
+  - derive/7/qi/
+  - derive/8/ba/
+  - derive/9/jq/
+
+algebra_mspy:
+  - derive/(?<!\d)1([1-9])(?!\d)/ui$1/
+  - derive/([1-9])0000(?!0)/$1wj/
+  - derive/([1-9])000(?!0)/$1qm/
+  - derive/([1-9])00(?!0)/$1bl/
+  - derive/([2-9])0(?!0)/$1ui/
+  - derive/(?<!\d)([2-9])([1-9])(?!\d)/$1ui$2/
+  - derive/\./dm/
+  - derive/10/ui/
+  - derive/0/l;/
+  - derive/1/yi/
+  - derive/2/er/
+  - derive/2/ld/
+  - derive/3/sj/
+  - derive/4/si/
+  - derive/5/wu/
+  - derive/6/lq/
+  - derive/7/qi/
+  - derive/8/ba/
+  - derive/9/jq/
+
+algebra_ziguang:
+  - derive/(?<!\d)1([1-9])(?!\d)/ii$1/
+  - derive/([1-9])0000(?!0)/$1wr/
+  - derive/([1-9])000(?!0)/$1qf/
+  - derive/([1-9])00(?!0)/$1bp/
+  - derive/([2-9])0(?!0)/$1ii/
+  - derive/(?<!\d)([2-9])([1-9])(?!\d)/$1ii$2/
+  - derive/\./df/
+  - derive/10/ii/
+  - derive/0/l;/
+  - derive/1/yi/
+  - derive/2/er/
+  - derive/2/lg/
+  - derive/3/sr/
+  - derive/4/si/
+  - derive/5/wu/
+  - derive/6/lj/
+  - derive/7/qi/
+  - derive/8/ba/
+  - derive/9/jj/
+
+algebra_double_pinyin:
+  - derive/(?<!\d)1([1-9])(?!\d)/ui$1/
+  - derive/([1-9])0000(?!0)/$1wj/
+  - derive/([1-9])000(?!0)/$1qm/
+  - derive/([1-9])00(?!0)/$1bl/
+  - derive/([2-9])0(?!0)/$1ui/
+  - derive/(?<!\d)([2-9])([1-9])(?!\d)/$1ui$2/
+  - derive/\./dm/
+  - derive/10/ui/
+  - derive/0/ly/
+  - derive/1/yi/
+  - derive/2/er/
+  - derive/2/ld/
+  - derive/3/sj/
+  - derive/4/si/
+  - derive/5/wu/
+  - derive/6/lq/
+  - derive/7/qi/
+  - derive/8/ba/
+  - derive/9/jq/