fix: Use Mongolian dictionary from wooorm/dictionaries (#4116)

dictionaries/mn_MN/checksum.txt

@@ -1,3 +1,4 @@
-5080c13d61bf60137bfb8f8c6b8e96c2db05f5a3  dict/mn-mn.trie
-739b662b85481c70956640f8f1d831a039c26155  src/mn_MN.aff
-13448fd57a5daa61ce27149a6f323ec76f25269a  src/mn_MN.dic
+e90c354ed223fb0a8d2b604b2358e6d62a37db0e  dict/mn-mn.trie
+01b3cc8b21ed8e943b2b22d554f19b6b1abde040  src/additional-words.txt
+b786697b4e3e589f17f97a9af799a7e75be1fd62  src/hunspell/index.aff
+1ffeb349473e17bd44fef97284b606d0f48c566b  src/hunspell/index.dic

dictionaries/mn_MN/cspell-ext.json

@@ -1,44 +1,21 @@
-// cSpell Settings
 {
     "id": "mn-mn",
     "version": "0.2",
     "name": "Mongolian",
     "description": "Mongolian dictionary for cspell.",
     "readonly": true,
-    // List of dictionary files to add to the global list of dictionaries
     "dictionaryDefinitions": [
         {
             "name": "mn-mn",
-            "path": "./dict/mn-mn.trie",
-            "description": "Mongolian dictionary for cspell."
+            "path": "./dict/mn-mn.trie.gz",
+            "description": "Mongolian dictionary."
         }
     ],
-    // Dictionaries to always be used.
-    // Generally left empty
-    "dictionaries": [],
-    // Language Rules to apply to matching files.
-    // Files are matched on `languageId` and `locale`
     "languageSettings": [
         {
-            // VSCode languageId. i.e. typescript, java, go, cpp, javascript, markdown, latex
-            // * will match against any file type.
             "languageId": "*",
-            // Language locale. i.e. en-US, de-AT, or ru. * will match all locales.
-            // Multiple locales can be specified like: "en, en-US" to match both English and English US.
             "locale": "mn,mn-MN",
-            // By default the whole text of a file is included for spell checking
-            // Adding patterns to the "includeRegExpList" to only include matching patterns
-            "includeRegExpList": [],
-            // To exclude patterns, add them to "ignoreRegExpList"
-            "ignoreRegExpList": [],
-            // regex patterns than can be used with ignoreRegExpList or includeRegExpList
-            // Example: "pattern": [{ "name": "mdash", "pattern": "—" }]
-            // This could be included in "ignoreRegExpList": ["mdash"]
-            "patterns": [],
-            // List of dictionaries to enable by name in `dictionaryDefinitions`
-            "dictionaries": ["mn-mn"],
-            // Dictionary definitions can also be supplied here. They are only used iff "languageId" and "locale" match.
-            "dictionaryDefinitions": []
+            "dictionaries": ["mn-mn"]
         }
     ]
 }

dictionaries/mn_MN/cspell-tools.config.yaml

@@ -3,8 +3,9 @@
 targets:
   - name: 'mn-mn'
     sources:
-      - filename: 'src/mn_MN.dic'
-        maxDepth: 1 # This is set to 1 to prevent initial builds from taking too long.
+      - filename: 'src/hunspell/index.dic'
+        maxDepth: 1 # This is set to 1 to prevent builds from taking too long.
+      - src/additional-words.txt
     format: 'trie3'
     targetDirectory: './dict'
     generateNonStrict: true

dictionaries/mn_MN/cspell.json

@@ -8,5 +8,6 @@
     ],
     "import": [
         "./cspell-ext.json"
-    ]
+    ],
+    "ignorePaths": ["src/hunspell"]
 }

dictionaries/mn_MN/package.json

@@ -1,8 +1,7 @@
 {
   "name": "@cspell/dict-mn-mn",
   "version": "1.0.0",
-  "description": "Mongolian dictionary for cspell. -- Private until verified",
-  "private": true,
+  "description": "Mongolian dictionary for cspell.",
   "publishConfig": {
     "access": "public",
     "provenance": true
@@ -14,11 +13,13 @@
   },
   "scripts": {
     "build": "cross-env NODE_OPTIONS=--max_old_space_size=8192 cspell-tools-cli build",
-    "test": "hunspell-reader words -n 1000 -m 0 \"src/mn_MN.dic\" | cspell -c ./cspell-ext.json \"--locale=mn,mn-MN\" \"--languageId=*\" stdin",
+    "test": "pnpm test:words && pnpm test:samples",
+    "test:samples": "cspell samples",
+    "test:words": "hunspell-reader words -n 1000 -m 0 \"src/hunspell/index.dic\" | cspell -c ./cspell-ext.json \"--locale=mn,mn-MN\" \"--languageId=*\" stdin",
     "prepublishOnly": "pnpm run conditional-build && pnpm test",
     "prepare:dictionary": "cspell-tools-cli gzip \"dict/*.trie\"",
-    "conditional-build": "pnpm run sync && pnpm run build --conditional",
-    "sync": "pnpm cpy \"node_modules/dictionary-ru/**\" src/"
+    "conditional-build": "pnpm run sync && pnpm run build --conditional && pnpm run prepare:dictionary",
+    "sync": "pnpm cpy \"node_modules/dictionary-mn/*\" src/hunspell"
   },
   "repository": {
     "type": "git",
@@ -39,10 +40,11 @@
     "url": "https://github.com/streetsidesoftware/cspell-dicts/issues"
   },
   "homepage": "https://github.com/streetsidesoftware/cspell-dicts/blob/main/dictionaries/mn_MN#readme",
-  "devDependencies": {},
-  "dependencies": {},
+  "devDependencies": {
+    "dictionary-mn": "^3.0.0"
+  },
   "files": [
-    "dict/mn-mn.trie",
+    "dict/mn-mn.trie.gz",
     "cspell-ext.json",
     "*.js",
     "*.d.ts"

dictionaries/mn_MN/samples/sample.md

@@ -0,0 +1,14 @@
+# Улаанбаатар - Ulaanbaatar
+
+[Улаанбаатар — Википедиа нэвтэрхий толь](https://mn.wikipedia.org/wiki/%D0%A3%D0%BB%D0%B0%D0%B0%D0%BD%D0%B1%D0%B0%D0%B0%D1%82%D0%B0%D1%80)
+
+Улаанбаатар (ᠤᠯᠠᠭᠠᠨᠪᠠᠭᠠᠲᠤᠷ улаганбагатур, улаан-баа-тар) нь Монгол Улсын нийслэл бөгөөд Монгол Улсын цорын ганц саятан хот юм.
+
+Улаанбаатар хот Монгол Улсын төв хэсэгт хуучнаар Алтан тэвшийн хөндий, өнөөгийнхөөр Туул-Сэлбийн гол бэлчир хөндийд, далайн түвшнээс дээш 1300-1350 метр өндөрт Богд хан, Сонгино хайрхан, Чингэлтэй уул, Баянзүрх дөрвөн уулаар хүрээлэгдэн орших ба 4,704.4 км2 нутаг дэвсгэртэй. 2021 оны эцсээр 1,649,172 хүн оршин сууж байна. Засаг захиргааны хувьд 9 дүүрэг, 204 хороонд хуваагддаг.
+
+1639 онд Ширээт цагаан нуурт Өндөр гэгээнийг залж орд өргөө боссоноор Улаанбаатар хотын өмнөх суурин байгуулагджээ. 1778 он хүртэл олон газар нүүдэллэж байгаад одоогийн байрлалдаа тогтсон. Өмнө нь шашин, худалдааны төв байсан бол 1912 оноос улсын нийслэл болж[4], XX зуунд барилгажиж, аж үйлдвэр хөгжин, XXI зуунд их хотын төрх бүрдэж байна. Улаанбаатар Монгол Улсын улс төр, эдийн засаг, соёл, технологийн төв юм. Түгжрэл утаа маш ихтэй.
+
+<!--
+cspell:locale en,mn
+cspell:words Ulaanbaatar улаганбагатур ᠤᠯᠠᠭᠠᠨᠪᠠᠭᠠᠲᠤᠷ
+-->

dictionaries/mn_MN/src/additional-words.txt

@@ -0,0 +1,8 @@
+# Please add additional words below
+Википедиа
+өнөөгийнхөөр
+түвшнээс
+дэвсгэртэй
+боссоноор
+Википедиа
+байрлалдаа