From 0460c173aa9c2f95f0df73ca8fe6d13c0d8f6ce2 Mon Sep 17 00:00:00 2001 From: hegotit <153287866+hegotit@users.noreply.github.com> Date: Tue, 13 Aug 2024 15:33:52 +0800 Subject: [PATCH] feat: Support specifying Rime config directory via command args --- others/script/main.go | 5 + others/script/rime/check.go | 4 +- others/script/rime/cn_en.go | 649 +++++++++++++++--------------- others/script/rime/dir_linux.go | 18 + others/script/rime/dir_mac.go | 24 +- others/script/rime/dir_windows.go | 24 +- others/script/rime/pinyin.go | 2 +- others/script/rime/rime.go | 65 ++- 8 files changed, 417 insertions(+), 374 deletions(-) create mode 100644 others/script/rime/dir_linux.go diff --git a/others/script/main.go b/others/script/main.go index 3ef4c54..f718b35 100644 --- a/others/script/main.go +++ b/others/script/main.go @@ -68,6 +68,11 @@ SORT: } func areYouOK() { + if rime.AutoConfirm { + fmt.Println("Auto confirm enabled. Skipping prompt.") + return + } + fmt.Println("Are you OK:") var isOK string _, _ = fmt.Scanf("%s", &isOK) diff --git a/others/script/rime/check.go b/others/script/rime/check.go index 229dcb5..8db5693 100644 --- a/others/script/rime/check.go +++ b/others/script/rime/check.go @@ -28,7 +28,7 @@ var ( ) // 初始化特殊词汇列表、需要注音列表、错别字列表、拼音列表 -func init() { +func initCheck() { // 特殊词汇列表，不进行任何检查 specialWords.Add("狄尔斯–阿尔德反应") specialWords.Add("特里斯坦–达库尼亚") @@ -102,7 +102,7 @@ func init() { text, code := parts[0], parts[1] hanPinyin[text] = append(hanPinyin[text], code) } - // 给 hanPinyin 补充不再字表的读音，和过滤列表 hanPinyinFilter + // 给 hanPinyin 补充不在字表的读音，和过滤列表 hanPinyinFilter file4, err := os.Open(汉字拼音映射TXT) if err != nil { log.Fatalln(err) diff --git a/others/script/rime/cn_en.go b/others/script/rime/cn_en.go index 261261f..268efa4 100644 --- a/others/script/rime/cn_en.go +++ b/others/script/rime/cn_en.go @@ -118,330 +118,341 @@ type schema struct { file *os.File } -var doublePinyin = schema{ - name: "cn_en_double_pinyin", - desc: "自然码双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_double_pinyin.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "aa", - "-e-": "ee", - "-o-": "oo", - "-ai-": "ai", - "-ei-": "ei", - "-ou-": "ou", - "-an-": "an", - "-en-": "en", - "-ang-": "ah", - "-eng-": "eg", - "-ao-": "ao", - "-er-": "er", - // zh ch sh - "zh": "v", - "ch": "i", - "sh": "u", - // 韵母 - "iu": "q", - "ia": "w", - "ua": "w", - "uan": "r", - "ue": "t", - "ve": "t", - "ing": "y", - "uai": "y", - "uo": "o", - "un": "p", - "iong": "s", - "ong": "s", - "iang": "d", - "uang": "d", - "en": "f", - "eng": "g", - "ang": "h", - "an": "j", - "ao": "k", - "ai": "l", - "ei": "z", - "ie": "x", - "iao": "c", - "ui": "v", - "ou": "b", - "in": "n", - "ian": "m", - }, -} +var ( + doublePinyin schema + doublePinyinFlypy schema + doublePinyinMSPY schema + doublePinyinSogou schema + doublePinyinZiGuang schema + doublePinyinABC schema +) -var doublePinyinFlypy = schema{ - name: "cn_en_flypy", - desc: "小鹤双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_flypy.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "aa", - "-e-": "ee", - "-o-": "oo", - "-ai-": "ai", - "-ei-": "ei", - "-ou-": "ou", - "-an-": "an", - "-en-": "en", - "-ang-": "ah", - "-eng-": "eg", - "-ao-": "ao", - "-er-": "er", - // zh ch sh - "zh": "v", - "ch": "i", - "sh": "u", - // 韵母 - "iu": "q", - "ei": "w", - "uan": "r", - "ue": "t", - "ve": "t", - "un": "y", - "uo": "o", - "ie": "p", - "iong": "s", - "ong": "s", - "ai": "d", - "en": "f", - "eng": "g", - "ang": "h", - "an": "j", - "ing": "k", - "uai": "k", - "iang": "l", - "uang": "l", - "ou": "z", - "ia": "x", - "ua": "x", - "ao": "c", - "ui": "v", - "in": "b", - "iao": "n", - "ian": "m", - }, -} +func initSchemas() { + doublePinyin = schema{ + name: "cn_en_double_pinyin", + desc: "自然码双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_double_pinyin.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "aa", + "-e-": "ee", + "-o-": "oo", + "-ai-": "ai", + "-ei-": "ei", + "-ou-": "ou", + "-an-": "an", + "-en-": "en", + "-ang-": "ah", + "-eng-": "eg", + "-ao-": "ao", + "-er-": "er", + // zh ch sh + "zh": "v", + "ch": "i", + "sh": "u", + // 韵母 + "iu": "q", + "ia": "w", + "ua": "w", + "uan": "r", + "ue": "t", + "ve": "t", + "ing": "y", + "uai": "y", + "uo": "o", + "un": "p", + "iong": "s", + "ong": "s", + "iang": "d", + "uang": "d", + "en": "f", + "eng": "g", + "ang": "h", + "an": "j", + "ao": "k", + "ai": "l", + "ei": "z", + "ie": "x", + "iao": "c", + "ui": "v", + "ou": "b", + "in": "n", + "ian": "m", + }, + } -var doublePinyinMSPY = schema{ - name: "cn_en_mspy", - desc: "微软双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_mspy.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "oa", - "-e-": "oe", - "-o-": "oo", - "-ai-": "ol", - "-ei-": "oz", - "-ou-": "ob", - "-an-": "oj", - "-en-": "of", - "-ang-": "oh", - "-eng-": "og", - "-ao-": "ok", - "-er-": "or", - // zh ch sh - "zh": "v", - "ch": "i", - "sh": "u", - // 韵母 - "iu": "q", - "ia": "w", - "ua": "w", - "er": "r", - "uan": "r", - "ue": "t", - "uai": "y", - "uo": "o", - "un": "p", - "iong": "s", - "ong": "s", - "iang": "d", - "uang": "d", - "en": "f", - "eng": "g", - "ang": "h", - "an": "j", - "ao": "k", - "ai": "l", - "ing": ";", - "ei": "z", - "ie": "x", - "iao": "c", - "ui": "v", - "ve": "v", - "ou": "b", - "in": "n", - "ian": "m", - }, -} + doublePinyinFlypy = schema{ + name: "cn_en_flypy", + desc: "小鹤双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_flypy.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "aa", + "-e-": "ee", + "-o-": "oo", + "-ai-": "ai", + "-ei-": "ei", + "-ou-": "ou", + "-an-": "an", + "-en-": "en", + "-ang-": "ah", + "-eng-": "eg", + "-ao-": "ao", + "-er-": "er", + // zh ch sh + "zh": "v", + "ch": "i", + "sh": "u", + // 韵母 + "iu": "q", + "ei": "w", + "uan": "r", + "ue": "t", + "ve": "t", + "un": "y", + "uo": "o", + "ie": "p", + "iong": "s", + "ong": "s", + "ai": "d", + "en": "f", + "eng": "g", + "ang": "h", + "an": "j", + "ing": "k", + "uai": "k", + "iang": "l", + "uang": "l", + "ou": "z", + "ia": "x", + "ua": "x", + "ao": "c", + "ui": "v", + "in": "b", + "iao": "n", + "ian": "m", + }, + } -var doublePinyinSogou = schema{ - name: "cn_en_sogou", - desc: "搜狗双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_sogou.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "oa", - "-e-": "oe", - "-o-": "oo", - "-ai-": "ol", - "-ei-": "oz", - "-ou-": "ob", - "-an-": "oj", - "-en-": "of", - "-ang-": "oh", - "-eng-": "og", - "-ao-": "ok", - "-er-": "or", - // zh ch sh - "zh": "v", - "ch": "i", - "sh": "u", - // 韵母 - "iu": "q", - "ia": "w", - "ua": "w", - "er": "r", - "uan": "r", - "ue": "t", - "uai": "y", - "uo": "o", - "un": "p", - "iong": "s", - "ong": "s", - "iang": "d", - "uang": "d", - "en": "f", - "eng": "g", - "ang": "h", - "an": "j", - "ao": "k", - "ai": "l", - "ing": ";", - "ei": "z", - "ie": "x", - "iao": "c", - "ui": "v", - "ve": "t", - "ou": "b", - "in": "n", - "ian": "m", - }, -} + doublePinyinMSPY = schema{ + name: "cn_en_mspy", + desc: "微软双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_mspy.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "oa", + "-e-": "oe", + "-o-": "oo", + "-ai-": "ol", + "-ei-": "oz", + "-ou-": "ob", + "-an-": "oj", + "-en-": "of", + "-ang-": "oh", + "-eng-": "og", + "-ao-": "ok", + "-er-": "or", + // zh ch sh + "zh": "v", + "ch": "i", + "sh": "u", + // 韵母 + "iu": "q", + "ia": "w", + "ua": "w", + "er": "r", + "uan": "r", + "ue": "t", + "uai": "y", + "uo": "o", + "un": "p", + "iong": "s", + "ong": "s", + "iang": "d", + "uang": "d", + "en": "f", + "eng": "g", + "ang": "h", + "an": "j", + "ao": "k", + "ai": "l", + "ing": ";", + "ei": "z", + "ie": "x", + "iao": "c", + "ui": "v", + "ve": "v", + "ou": "b", + "in": "n", + "ian": "m", + }, + } -var doublePinyinZiGuang = schema{ - name: "cn_en_ziguang", - desc: "紫光双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_ziguang.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "oa", - "-e-": "oe", - "-o-": "oo", - "-ai-": "op", - "-ei-": "ok", - "-ou-": "oz", - "-an-": "or", - "-en-": "ow", - "-ang-": "os", - "-eng-": "ot", - "-ao-": "oq", - "-er-": "oj", - // zh ch sh - "zh": "u", - "ch": "a", - "sh": "i", - // 韵母 - "ao": "q", - "en": "w", - "an": "r", - "eng": "t", - "in": "y", - "uai": "y", - "uo": "o", - "ai": "p", - "ang": "s", - "ie": "d", - "ian": "f", - "iang": "g", - "uang": "g", - "iong": "h", - "ong": "h", - "er": "j", - "iu": "j", - "ei": "k", - "uan": "l", - "ing": ";", - "ou": "z", - "ia": "x", - "ua": "x", - "iao": "b", - "ue": "n", - "ui": "n", - "un": "m", - }, -} + doublePinyinSogou = schema{ + name: "cn_en_sogou", + desc: "搜狗双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_sogou.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "oa", + "-e-": "oe", + "-o-": "oo", + "-ai-": "ol", + "-ei-": "oz", + "-ou-": "ob", + "-an-": "oj", + "-en-": "of", + "-ang-": "oh", + "-eng-": "og", + "-ao-": "ok", + "-er-": "or", + // zh ch sh + "zh": "v", + "ch": "i", + "sh": "u", + // 韵母 + "iu": "q", + "ia": "w", + "ua": "w", + "er": "r", + "uan": "r", + "ue": "t", + "uai": "y", + "uo": "o", + "un": "p", + "iong": "s", + "ong": "s", + "iang": "d", + "uang": "d", + "en": "f", + "eng": "g", + "ang": "h", + "an": "j", + "ao": "k", + "ai": "l", + "ing": ";", + "ei": "z", + "ie": "x", + "iao": "c", + "ui": "v", + "ve": "t", + "ou": "b", + "in": "n", + "ian": "m", + }, + } -var doublePinyinABC = schema{ - name: "cn_en_abc", - desc: "智能 ABC 双拼", - combinationType: "unique", - path: filepath.Join(RimeDir, "en_dicts/cn_en_abc.txt"), - mapping: map[string]string{ - // 零声母 - "-a-": "oa", - "-e-": "oe", - "-o-": "oo", - "-ai-": "ol", - "-ei-": "oq", - "-ou-": "ob", - "-an-": "oj", - "-en-": "of", - "-ang-": "oh", - "-eng-": "og", - "-ao-": "ok", - "-er-": "or", - // zh ch sh - "zh": "a", - "ch": "e", - "sh": "v", - // 韵母 - "ao": "k", - "en": "f", - "an": "j", - "eng": "g", - "in": "c", - "uai": "c", - "uo": "o", - "ai": "l", - "ang": "h", - "ie": "x", - "ian": "w", - "iang": "t", - "uang": "t", - "iong": "s", - "ong": "s", - "er": "r", - "iu": "r", - "ei": "q", - "uan": "p", - "ing": "y", - "ou": "b", - "ia": "d", - "ua": "d", - "iao": "z", - "ue": "m", - "ui": "m", - "un": "n", - }, + doublePinyinZiGuang = schema{ + name: "cn_en_ziguang", + desc: "紫光双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_ziguang.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "oa", + "-e-": "oe", + "-o-": "oo", + "-ai-": "op", + "-ei-": "ok", + "-ou-": "oz", + "-an-": "or", + "-en-": "ow", + "-ang-": "os", + "-eng-": "ot", + "-ao-": "oq", + "-er-": "oj", + // zh ch sh + "zh": "u", + "ch": "a", + "sh": "i", + // 韵母 + "ao": "q", + "en": "w", + "an": "r", + "eng": "t", + "in": "y", + "uai": "y", + "uo": "o", + "ai": "p", + "ang": "s", + "ie": "d", + "ian": "f", + "iang": "g", + "uang": "g", + "iong": "h", + "ong": "h", + "er": "j", + "iu": "j", + "ei": "k", + "uan": "l", + "ing": ";", + "ou": "z", + "ia": "x", + "ua": "x", + "iao": "b", + "ue": "n", + "ui": "n", + "un": "m", + }, + } + + doublePinyinABC = schema{ + name: "cn_en_abc", + desc: "智能 ABC 双拼", + combinationType: "unique", + path: filepath.Join(RimeDir, "en_dicts/cn_en_abc.txt"), + mapping: map[string]string{ + // 零声母 + "-a-": "oa", + "-e-": "oe", + "-o-": "oo", + "-ai-": "ol", + "-ei-": "oq", + "-ou-": "ob", + "-an-": "oj", + "-en-": "of", + "-ang-": "oh", + "-eng-": "og", + "-ao-": "ok", + "-er-": "or", + // zh ch sh + "zh": "a", + "ch": "e", + "sh": "v", + // 韵母 + "ao": "k", + "en": "f", + "an": "j", + "eng": "g", + "in": "c", + "uai": "c", + "uo": "o", + "ai": "l", + "ang": "h", + "ie": "x", + "ian": "w", + "iang": "t", + "uang": "t", + "iong": "s", + "ong": "s", + "er": "r", + "iu": "r", + "ei": "q", + "uan": "p", + "ing": "y", + "ou": "b", + "ia": "d", + "ua": "d", + "iao": "z", + "ue": "m", + "ui": "m", + "un": "n", + }, + } } // CnEn 从 others/cn_en.txt 生成全拼和各个双拼的中英混输词库 diff --git a/others/script/rime/dir_linux.go b/others/script/rime/dir_linux.go new file mode 100644 index 0000000..0b2cbeb --- /dev/null +++ b/others/script/rime/dir_linux.go @@ -0,0 +1,18 @@ +//go:build linux +// +build linux + +package rime + +import ( + "log" + "os/user" + "path/filepath" +) + +func getRimeDirForPlatform() string { + u, err := user.Current() + if err != nil { + log.Fatalln(err) + } + return filepath.Join(u.HomeDir, ".config", "rime") +} diff --git a/others/script/rime/dir_mac.go b/others/script/rime/dir_mac.go index 1f21f43..4cc8610 100644 --- a/others/script/rime/dir_mac.go +++ b/others/script/rime/dir_mac.go @@ -5,26 +5,14 @@ package rime import ( "log" - "os/user" "path/filepath" - "runtime" ) -// 获取 macOS/Windows Rime 配置目录 -func getRimeDir() string { - var dir string - switch runtime.GOOS { - case "darwin": // macOS - u, err := user.Current() - if err != nil { - log.Fatalln(err) - } - dir = filepath.Join(u.HomeDir, "Library/Rime") - // case "windows": // Windows - // dir = getWeaselDir() - default: - log.Fatalf("Unsupported OS: %s so far", runtime.GOOS) +// 获取 macOS Rime 配置目录 +func getRimeDirForPlatform() string { + u, err := user.Current() + if err != nil { + log.Fatalln(err) } - - return dir + return filepath.Join(u.HomeDir, "Library/Rime") } diff --git a/others/script/rime/dir_windows.go b/others/script/rime/dir_windows.go index ec6dbf9..df85fb5 100644 --- a/others/script/rime/dir_windows.go +++ b/others/script/rime/dir_windows.go @@ -7,31 +7,11 @@ import ( "golang.org/x/sys/windows/registry" "log" "os" - "os/user" "path/filepath" - "runtime" ) -// 获取 macOS/Windows Rime 配置目录 -func getRimeDir() string { - var dir string - switch runtime.GOOS { - case "darwin": // macOS - u, err := user.Current() - if err != nil { - log.Fatalln(err) - } - dir = filepath.Join(u.HomeDir, "Library/Rime") - case "windows": // Windows - dir = getWeaselDir() - default: - log.Fatalf("Unsupported OS: %s so far", runtime.GOOS) - } - - return dir -} - -func getWeaselDir() string { +// 获取 Windows Rime 配置目录 +func getRimeDirForPlatform() string { keyPath := `Software\Rime\Weasel` valueName := "RimeUserDir" diff --git a/others/script/rime/pinyin.go b/others/script/rime/pinyin.go index 927ecd6..84a8446 100644 --- a/others/script/rime/pinyin.go +++ b/others/script/rime/pinyin.go @@ -238,7 +238,7 @@ var onlyOne = map[string]string{ "给": "gei", } -func init() { +func initPinyin() { // 从 base、ext 准备结巴的词典和词组拼音映射 for _, dictPath := range []string{BasePath, ExtPath} { file, err := os.Open(dictPath) diff --git a/others/script/rime/rime.go b/others/script/rime/rime.go index efd1477..52fdb21 100644 --- a/others/script/rime/rime.go +++ b/others/script/rime/rime.go @@ -2,6 +2,7 @@ package rime import ( "bufio" + "flag" "fmt" mapset "github.com/deckarep/golang-set/v2" "log" @@ -21,26 +22,66 @@ type lemma struct { } var ( - mark = "# +_+" // 词库中的标记符号，表示从这行开始进行检查或排序 - RimeDir = getRimeDir() // Rime 配置目录 + mark = "# +_+" // 词库中的标记符号，表示从这行开始进行检查或排序 + RimeDir string + EmojiMapPath string + EmojiPath string + HanziPath string + BasePath string + ExtPath string + TencentPath string + HanziSet mapset.Set[string] + BaseSet mapset.Set[string] + ExtSet mapset.Set[string] + TencentSet mapset.Set[string] + 需要注音TXT string + 错别字TXT string + 汉字拼音映射TXT string + AutoConfirm bool +) + +func init() { + // 定义命令行参数 + flag.StringVar(&RimeDir, "rime_path", "", "Specify the Rime configuration directory") + flag.BoolVar(&AutoConfirm, "auto_confirm", false, "Automatically confirm the prompt") + flag.Parse() + + RimeDir = getRimeDir(RimeDir) // Rime 配置目录 EmojiMapPath = filepath.Join(RimeDir, "others/emoji-map.txt") - EmojiPath = filepath.Join(RimeDir, "opencc/emoji.txt") + EmojiPath = filepath.Join(RimeDir, "opencc/emoji.txt") - HanziPath = filepath.Join(RimeDir, "cn_dicts/8105.dict.yaml") - BasePath = filepath.Join(RimeDir, "cn_dicts/base.dict.yaml") - ExtPath = filepath.Join(RimeDir, "cn_dicts/ext.dict.yaml") + HanziPath = filepath.Join(RimeDir, "cn_dicts/8105.dict.yaml") + BasePath = filepath.Join(RimeDir, "cn_dicts/base.dict.yaml") + ExtPath = filepath.Join(RimeDir, "cn_dicts/ext.dict.yaml") TencentPath = filepath.Join(RimeDir, "cn_dicts/tencent.dict.yaml") - HanziSet = readToSet(HanziPath) - BaseSet = readToSet(BasePath) - ExtSet = readToSet(ExtPath) + HanziSet = readToSet(HanziPath) + BaseSet = readToSet(BasePath) + ExtSet = readToSet(ExtPath) TencentSet = readToSet(TencentPath) - 需要注音TXT = filepath.Join(RimeDir, "others/script/rime/需要注音.txt") - 错别字TXT = filepath.Join(RimeDir, "others/script/rime/错别字.txt") + 需要注音TXT = filepath.Join(RimeDir, "others/script/rime/需要注音.txt") + 错别字TXT = filepath.Join(RimeDir, "others/script/rime/错别字.txt") 汉字拼音映射TXT = filepath.Join(RimeDir, "others/script/rime/汉字拼音映射.txt") -) + + initCheck() + initSchemas() + initPinyin() +} + +func getRimeDir(rimePath string) string { + if rimePath != "" { + absPath, err := filepath.Abs(rimePath) + if err != nil { + log.Fatalf("Failed to get absolute path: %v", err) + } + // 使用传入的路径 + return absPath + } + + return getRimeDirForPlatform() +} // 将所有词库读入 set，供检查或排序使用 func readToSet(dictPath string) mapset.Set[string] {