feat: Update Spanish dictionary (#1077)

* feat: Update Spanish dictionary * docs: add information for add new words * feat: create additional_words.txt for spanish dictionary * Move Additional words to the right place * Update additional_words.txt * Add suggestion maps from `index.aff` * Do not spell check hunspell src Co-authored-by: Fernando López Moreno <50037244+koficoud@users.noreply.github.com>
streetsidesoftware · May 4, 2022 · beca796 · beca796
1 parent 9cbbbee
commit beca796
Show file tree

Hide file tree

Showing 20 changed files with 13,296 additions and 22,001 deletions.
diff --git a/dictionaries/es_ES/LICENSE b/dictionaries/es_ES/LICENSE
diff --git a/dictionaries/es_ES/README.md b/dictionaries/es_ES/README.md
@@ -39,6 +39,10 @@ Building is only necessary if you want to modify the contents of the dictionary.
 npm run build
 ```
 
+## Adding Words
+
+Please add any words to [src/additional_words.txt](./src/additional_words.txt) by making a pull request.
+
 ## Resources
 
 The Hunspell source for this dictionary can be found:

diff --git a/dictionaries/es_ES/Spanish.trie.gz b/dictionaries/es_ES/Spanish.trie.gz
diff --git a/dictionaries/es_ES/Spanish.txt b/dictionaries/es_ES/Spanish.txt
diff --git a/dictionaries/es_ES/checksum.txt b/dictionaries/es_ES/checksum.txt
@@ -0,0 +1,6 @@
+f8ffd462c0991b024b9b3fb58ded6f588099f255  ./Spanish.trie.gz
+e9852fedec06e16eadf561a89fcfc4352800f320  ./source-dependencies.txt
+30b5a9cd705505c82b9d576840d32cf6c83c352f  ./source-files.txt
+942c6fb2ca5fd83d0eeee1fdf50e81b2e766b60f  ./src/additional_words.txt
+50acae0932a1576ccb76393db2f7cd495e53e376  ./src/hunspell/index.aff
+8f93fdb38bd02bfbfab931916773cc048cadd84d  ./src/hunspell/index.dic
diff --git a/dictionaries/es_ES/cspell-ext.json b/dictionaries/es_ES/cspell-ext.json
@@ -8,7 +8,29 @@
         {
             "name": "es-es",
             "path": "./Spanish.trie.gz",
-            "description": "Spanish Dictionary (Spain)"
+            "description": "Spanish Dictionary (Spain)",
+            "dictionaryInformation": {
+                "locale": "es-ES",
+                "alphabet": "a-zA-ZóíjáéñúüÓÍÁÉÑÚÜ",
+                "suggestionEditCosts": [
+                    {
+                        "map": "MAP eéèëêEÉ|aáàAÁ|iíìïîIÍ|oóòöôOÓ|uüúùûUÚÜ",
+                        "replace": 1
+                    },
+                    {
+                        "map": "(ás)(az)|(cc)(x)|(és)(ez)|(güe)(hue)|(güi)(hui)",
+                        "replace": 75
+                    },
+                    {
+                        "map": "(ís)(iz)|(ío)(ido)|(ke)(que)|(ki)(qui)",
+                        "replace": 75
+                    },
+                    {
+                        "map": "(ll)(y)|(mb)(nv)|(seci)(cesi)|(vámonos)(vayámonos)",
+                        "replace": 75
+                    }
+                ]
+            }
         }
     ],
     "dictionaries": [],

diff --git a/dictionaries/es_ES/cspell.config.yaml b/dictionaries/es_ES/cspell.config.yaml
@@ -0,0 +1,31 @@
+---
+id: "es-es"
+version: "0.2"
+readonly: true
+name: "Spanish Dictionary (Spain)"
+description: "Spanish Dictionary (Spain)"
+dictionaryDefinitions:
+  - name: "es-es"
+    path: "./Spanish.trie.gz"
+    description: "Spanish Dictionary (Spain)"
+    dictionaryInformation:
+      locale: "es-ES"
+      # cspell:ignore ZóíjáéñúüÓÍÁÉÑÚÜ
+      alphabet: a-zA-ZóíjáéñúüÓÍÁÉÑÚÜ
+      suggestionEditCosts:
+        # cspell:ignore eéèëê iíìïî oóòöô uüúùû
+        - map: MAP eéèëêEÉ|aáàAÁ|iíìïîIÍ|oóòöôOÓ|uüúùûUÚÜ
+          replace: 1
+        - map: (ás)(az)|(cc)(x)|(és)(ez)|(güe)(hue)|(güi)(hui)
+          replace: 75
+        - map: (ís)(iz)|(ío)(ido)|(ke)(que)|(ki)(qui)
+          replace: 75
+        # cspell:ignore seci cesi vámonos vayámonos
+        - map: (ll)(y)|(mb)(nv)|(seci)(cesi)|(vámonos)(vayámonos)
+          replace: 75
+dictionaries: []
+languageSettings:
+  - languageId: "*"
+    locale: "es, es_ES"
+    caseSensitive: true
+    dictionaries: ["es-es"]
diff --git a/dictionaries/es_ES/cspell.json b/dictionaries/es_ES/cspell.json
@@ -1,16 +1,8 @@
 {
-    "files": [
-        "**/*.{md,txt}"
-    ],
-    "ignorePaths": [
-        "*.aff",
-        "*.dic",
-        "cspell-ext.json"
-    ],
+    "files": ["**/*.{md,txt}"],
+    "ignorePaths": ["cspell-ext.json", "src/hunspell"],
     "language": "en,es",
-    "import": [
-        "./cspell-ext.json"
-    ],
+    "import": ["./cspell-ext.json"],
     "caseSensitive": true,
     "ignoreWords": [
         "Alexandro",
@@ -25,6 +17,7 @@
         "Garrone",
         "Gelbort",
         "gmail",
+        "hunspell",
         "HYPH",
         "KNTRO",
         "LGPL",

diff --git a/dictionaries/es_ES/package.json b/dictionaries/es_ES/package.json
@@ -11,8 +11,20 @@
     "./cspell-ext.json": "./cspell-ext.json"
   },
   "scripts": {
-    "build": "cspell-tools-cli compile --trie3 -x compound \"src/hunspell/Spanish.dic\" -o .",
-    "test": "hunspell-reader words -n 1000 \"src/hunspell/Spanish.dic\" | cspell -v -c ./cspell-ext.json --local=es --languageId=* stdin",
+    "#build": "cspell-tools-cli compile --trie3 -x compound \"src/hunspell/Spanish.dic\" -o .",
+    "#test": "hunspell-reader words -n 1000 \"src/hunspell/Spanish.dic\" | cspell -v -c ./cspell-ext.json --local=es --languageId=* stdin",
+    "#prepublishOnly": "echo pre-publish",
+    "cspell-yaml-to-json": "yarn -s yaml2json -p cspell.config.yaml > cspell-ext.json && yarn lint",
+    "build": "yarn run compile",
+    "sync": "yarn cpy \"../../node_modules/dictionary-es/**\" src/hunspell",
+    "checksum": "shasum -c checksum.txt",
+    "compile": "cat source-files.txt | xargs cspell-tools-cli compile --trie3 -x compound --merge Spanish -o . && yarn run gen-checksum",
+    "conditional-build": "yarn run sync && yarn run --silent checksum || yarn run build",
+    "gen-checksum": "cat source-files.txt source-dependencies.txt | sort -u | xargs shasum > checksum.txt",
+    "lint": "yarn prettier -w \"*.{yaml,json}\"",
+    "test-dict": "hunspell-reader words -n 1000 \"src/hunspell/index.dic\" | cspell -v -c ./cspell-ext.json --local=es --languageId=* stdin",
+    "test-text": "cspell",
+    "test": "yarn run test-dict && yarn run test-text",
     "prepublishOnly": "echo pre-publish"
   },
   "repository": {
@@ -27,14 +39,17 @@
     "spelling"
   ],
   "author": "Jason Dent",
-  "license": "MIT",
+  "license": "GPL-3.0-or-later",
   "bugs": {
     "url": "https://github.com/streetsidesoftware/cspell-dicts/issues"
   },
   "homepage": "https://github.com/streetsidesoftware/cspell-dicts/blob/main/dictionaries/es_ES#readme",
   "files": [
     "Spanish.trie.gz",
     "cspell-ext.json",
-    "Spanish.txt"
-  ]
+    "src/hunspell/license"
+  ],
+  "devDependencies": {
+    "dictionary-es": "^3.2.0"
+  }
 }
diff --git a/dictionaries/es_ES/source-dependencies.txt b/dictionaries/es_ES/source-dependencies.txt
@@ -0,0 +1,5 @@
+./Spanish.trie.gz
+./source-dependencies.txt
+./source-files.txt
+./src/additional_words.txt
+./src/hunspell/index.aff
diff --git a/dictionaries/es_ES/source-files.txt b/dictionaries/es_ES/source-files.txt
@@ -0,0 +1,2 @@
+./src/additional_words.txt
+./src/hunspell/index.dic
diff --git a/dictionaries/es_ES/src/additional_words.txt b/dictionaries/es_ES/src/additional_words.txt
@@ -0,0 +1,2 @@
+# When adding words to the CSpell Spanish dictionary via a pull request,
+# Please add the words to this file, one term per line.