feat(ja-space-between-half-and-full-width): add allows option (#64)

azu · web-flow · commit 76da13295e7a · 2024-04-20T20:10:19.000+09:00
* feat(ja-space-between-half-and-full-width): add `allows` option

* test: add test case
diff --git a/packages/textlint-rule-ja-space-between-half-and-full-width/README.md b/packages/textlint-rule-ja-space-between-half-and-full-width/README.md
@@ -49,13 +49,17 @@ textlint --rule ja-space-between-half-and-full-width README.md
       - 対象としたい物のみ指定する
       - 例えば、数値と句読点（、。）を例外として扱いたい場合は以下
         - `["alphabets"]`
-- （非推奨）`exceptPunctuation`: `boolean`
-    - デフォルト: `true`
-    - 句読点（、。）を例外として扱うかどうか
-    - 代わりに `space` オプションを用いて `["alphabets", "numbers"]` と指定する
 - `lintStyledNode`: `boolean`
     - デフォルト: `false`
     - プレーンテキスト以外(リンクや画像のキャプションなど)を lint の対象とするかどうか (プレーンテキストの判断基準は [textlint/textlint-rule-helper: This is helper library for creating textlint rule](https://github.com/textlint/textlint-rule-helper#rulehelperisplainstrnodenode-boolean) を参照してください)
+- `allows: string[]`
+    - デフォルト: `[]`
+    - 例外として扱う文字列の配列
+    - [RegExp-like String](https://github.com/textlint/regexp-string-matcher?tab=readme-ov-file#regexp-like-string)も指定可能
+- （非推奨）`exceptPunctuation`: `boolean`
+  - デフォルト: `true`
+  - 句読点（、。）を例外として扱うかどうか
+  - 代わりに `space` オプションを用いて `["alphabets", "numbers"]` と指定する
 
 ```json
 {
@@ -83,6 +87,13 @@ textlint --rule ja-space-between-half-and-full-width README.md
             space: []
         }
 
+スペースは必須だが、`Eコーマス`だけはスペースなしを許可する。
+
+        text: "例外的にEコーマスはスペースなしでも通す",
+        options: {
+            space: "always",
+            allows: ["Eコーマス"]
+        }
 
 ## Changelog
 
diff --git a/packages/textlint-rule-ja-space-between-half-and-full-width/package.json b/packages/textlint-rule-ja-space-between-half-and-full-width/package.json
@@ -29,6 +29,7 @@
     "textlintrule"
   ],
   "devDependencies": {
+    "@textlint/regexp-string-matcher": "^2.0.2",
     "textlint-scripts": "^13.3.3"
   },
   "dependencies": {
diff --git a/packages/textlint-rule-ja-space-between-half-and-full-width/src/index.js b/packages/textlint-rule-ja-space-between-half-and-full-width/src/index.js
@@ -4,68 +4,79 @@ const assert = require("assert");
 /*
  全角文字と半角文字の間にスペースを入れるかどうか
  */
-import {RuleHelper} from "textlint-rule-helper";
-import {matchCaptureGroupAll} from "match-index";
+import { RuleHelper } from "textlint-rule-helper";
+import { matchCaptureGroupAll } from "match-index";
+import { matchPatterns } from "@textlint/regexp-string-matcher";
+
 const PunctuationRegExp = /[。、]/;
 const ZenRegExpStr = '[、。]|[\u3400-\u4DBF\u4E00-\u9FFF\uF900-\uFAFF]|[\uD840-\uD87F][\uDC00-\uDFFF]|[ぁ-んァ-ヶ]';
 const defaultSpaceOptions = {
     alphabets: false,
     numbers: false,
-    punctuation: false
+    punctuation: false,
 };
 const defaultOptions = {
     // プレーンテキスト以外を対象とするか See https://github.com/textlint/textlint-rule-helper#rulehelperisplainstrnodenode-boolean
     lintStyledNode: false,
+    /**
+     * 例外として無視する文字列
+     * RegExp-like Stringの配列を指定
+     * https://github.com/textlint/regexp-string-matcher?tab=readme-ov-file#regexp-like-string
+     */
+    allows: []
 };
+
 function reporter(context, options = {}) {
-   /**
-    * 入力された `space` オプションを内部処理用に成形する
-    * @param {string|Array|undefined} opt `space` オプションのインプット
-    * @param {boolean|undefined} exceptPunctuation `exceptPunctuation` オプションのインプット
-    * @returns {Object}
-    */
+    /**
+     * 入力された `space` オプションを内部処理用に成形する
+     * @param {string|Array|undefined} opt `space` オプションのインプット
+     * @param {boolean|undefined} exceptPunctuation `exceptPunctuation` オプションのインプット
+     * @returns {Object}
+     */
     const parseSpaceOption = (opt, exceptPunctuation) => {
-      if (typeof opt === 'string') {
-        assert(opt === "always" || opt === "never", `"space" options should be "always", "never" or an array.`);
-
-        if (opt === "always") {
-          if (exceptPunctuation === false) {
-            return {...defaultSpaceOptions, alphabets: true, numbers: true, punctuation: true};
-          } else {
-            return {...defaultSpaceOptions, alphabets: true, numbers: true};
-          }
-        } else if (opt === "never") {
-          if (exceptPunctuation === false) {
-            return {...defaultSpaceOptions, punctuation: true};
-          } else {
-            return defaultSpaceOptions;
-          }
+        if (typeof opt === 'string') {
+            assert(opt === "always" || opt === "never", `"space" options should be "always", "never" or an array.`);
+            
+            if (opt === "always") {
+                if (exceptPunctuation === false) {
+                    return { ...defaultSpaceOptions, alphabets: true, numbers: true, punctuation: true };
+                } else {
+                    return { ...defaultSpaceOptions, alphabets: true, numbers: true };
+                }
+            } else if (opt === "never") {
+                if (exceptPunctuation === false) {
+                    return { ...defaultSpaceOptions, punctuation: true };
+                } else {
+                    return defaultSpaceOptions;
+                }
+            }
+        } else if (Array.isArray(opt)) {
+            assert(
+                opt.every((v) => Object.keys(defaultSpaceOptions).includes(v)),
+                `Only "alphabets", "numbers", "punctuation" can be included in the array.`
+            );
+            const userOptions = Object.fromEntries(opt.map(key => [key, true]));
+            return { ...defaultSpaceOptions, ...userOptions };
         }
-      } else if (Array.isArray(opt)) {
-        assert(
-          opt.every((v) => Object.keys(defaultSpaceOptions).includes(v)),
-          `Only "alphabets", "numbers", "punctuation" can be included in the array.`
-        );
-        const userOptions = Object.fromEntries(opt.map(key => [key, true]));
-        return {...defaultSpaceOptions, ...userOptions};
-      }
-
-      return defaultSpaceOptions;
+        
+        return defaultSpaceOptions;
     }
-
-    const {Syntax, RuleError, report, fixer, getSource} = context;
+    
+    const { Syntax, RuleError, report, fixer, getSource } = context;
     const helper = new RuleHelper();
     const spaceOption = parseSpaceOption(options.space, options.exceptPunctuation);
     const lintStyledNode = options.lintStyledNode !== undefined
         ? options.lintStyledNode
         : defaultOptions.lintStyledNode;
+    const allows = options.allows !== undefined ? options.allows : defaultOptions.allows;
     /**
      * `text`を対象に例外オプションを取り除くfilter関数を返す
      * @param {string} text テスト対象のテキスト全体
      * @param {number} padding +1 or -1
      * @returns {function(*, *)}
      */
     const createFilter = (text, padding) => {
+        const allowedPatterns = allows.length > 0 ? matchPatterns(text, allows) : [];
         /**
          * `PunctuationRegExp`で指定された例外を取り除く
          * @param {Object} match
@@ -79,15 +90,22 @@ function reporter(context, options = {}) {
             if (!spaceOption.punctuation && PunctuationRegExp.test(targetChar)) {
                 return false;
             }
-            return true;
+            const isAllowed = allowedPatterns.some((allow) => {
+                // start ... end
+                if (allow.startIndex <= match.index && match.index <= allow.endIndex) {
+                    return true;
+                }
+                return false
+            })
+            return !isAllowed;
         }
     };
     // Never: アルファベットと全角の間はスペースを入れない
     const noSpaceBetween = (node, text) => {
         const betweenHanAndZen = matchCaptureGroupAll(text, new RegExp(`[A-Za-z0-9]([ 　])(?:${ZenRegExpStr})`));
         const betweenZenAndHan = matchCaptureGroupAll(text, new RegExp(`(?:${ZenRegExpStr})([ 　])[A-Za-z0-9]`));
         const reportMatch = (match) => {
-            const {index} = match;
+            const { index } = match;
             report(node, new RuleError("原則として、全角文字と半角文字の間にスペースを入れません。", {
                 index: match.index,
                 fix: fixer.replaceTextRange([index, index + 1], "")
@@ -96,37 +114,37 @@ function reporter(context, options = {}) {
         betweenHanAndZen.filter(createFilter(text, 1)).forEach(reportMatch);
         betweenZenAndHan.filter(createFilter(text, -1)).forEach(reportMatch);
     };
-
+    
     // Always: アルファベットと全角の間はスペースを入れる
     const needSpaceBetween = (node, text, options) => {
-       /**
-        * オプションを元に正規表現オプジェクトを生成する
-        * @param {Array} opt `space` オプション
-        * @param {boolean} btwHanAndZen=true 半角全角の間か全角半角の間か
-        * @returns {Object}
-        */
+        /**
+         * オプションを元に正規表現オプジェクトを生成する
+         * @param {Array} opt `space` オプション
+         * @param {boolean} btwHanAndZen=true 半角全角の間か全角半角の間か
+         * @returns {Object}
+         */
         const generateRegExp = (opt, btwHanAndZen = true) => {
-          const alphabets = opt.alphabets ? 'A-Za-z' : '';
-          const numbers = opt.numbers ? '0-9' : '';
-
-          let expStr;
-          if (btwHanAndZen) {
-            expStr = `([${alphabets}${numbers}])(?:${ZenRegExpStr})`;
-          } else {
-            expStr = `(${ZenRegExpStr})[${alphabets}${numbers}]`;
-          }
-
-          return new RegExp(expStr);
+            const alphabets = opt.alphabets ? 'A-Za-z' : '';
+            const numbers = opt.numbers ? '0-9' : '';
+            
+            let expStr;
+            if (btwHanAndZen) {
+                expStr = `([${alphabets}${numbers}])(?:${ZenRegExpStr})`;
+            } else {
+                expStr = `(${ZenRegExpStr})[${alphabets}${numbers}]`;
+            }
+            
+            return new RegExp(expStr);
         };
-
+        
         const betweenHanAndZenRegExp = generateRegExp(options);
         const betweenZenAndHanRegExp = generateRegExp(options, false);
         const errorMsg = '原則として、全角文字と半角文字の間にスペースを入れます。';
-
+        
         const betweenHanAndZen = matchCaptureGroupAll(text, betweenHanAndZenRegExp);
         const betweenZenAndHan = matchCaptureGroupAll(text, betweenZenAndHanRegExp);
         const reportMatch = (match) => {
-            const {index} = match;
+            const { index } = match;
             report(node, new RuleError(errorMsg, {
                 index: match.index,
                 fix: fixer.replaceTextRange([index + 1, index + 1], " ")
@@ -136,12 +154,12 @@ function reporter(context, options = {}) {
         betweenZenAndHan.filter(createFilter(text, 0)).forEach(reportMatch);
     };
     return {
-        [Syntax.Str](node){
+        [Syntax.Str](node) {
             if (!lintStyledNode && !helper.isPlainStrNode(node)) {
                 return;
             }
             const text = getSource(node);
-
+            
             const noSpace = (key) => key === 'punctuation' ? true : !spaceOption[key];
             if (Object.keys(spaceOption).every(noSpace)) {
                 noSpaceBetween(node, text);
@@ -151,6 +169,7 @@ function reporter(context, options = {}) {
         }
     }
 }
+
 module.exports = {
     linter: reporter,
     fixer: reporter
diff --git a/packages/textlint-rule-ja-space-between-half-and-full-width/test/index-test.js b/packages/textlint-rule-ja-space-between-half-and-full-width/test/index-test.js
@@ -125,7 +125,35 @@ Pull Request、コミットのやりかたなどが書かれています。`,
             options: {
                 space: ["alphabets", "punctuation"]
             }
-       }
+       },
+        // allows,
+        {
+            text: "Eコーマス",
+            options: {
+                space: "always",
+                allows: [
+                    "Eコーマス"
+                ]
+            }
+        },
+        {
+            text: "これは A言語、B言語、C言語です。",
+            options: {
+                space: "always",
+                allows: [
+                    "/(\\w)言語/"
+                ]
+            }
+        },
+        {
+            text: "E コーマス",
+            options: {
+                space: "never",
+                allows: [
+                    "E コーマス"
+                ]
+            }
+        },
     ],
     invalid: [
         {
diff --git a/yarn.lock b/yarn.lock
@@ -1659,6 +1659,16 @@
   resolved "https://registry.yarnpkg.com/@textlint/module-interop/-/module-interop-13.3.3.tgz#645b47b9e951030b2d656e2c9266b5587de2a17b"
   integrity sha512-CwfVpRGAxbkhGY9vLLU06Q/dy/RMNnyzbmt6IS2WIyxqxvGaF7QZtFYpKEEm63aemVyUvzQ7WM3yVOoUg6P92w==
 
+"@textlint/regexp-string-matcher@^2.0.2":
+  version "2.0.2"
+  resolved "https://registry.yarnpkg.com/@textlint/regexp-string-matcher/-/regexp-string-matcher-2.0.2.tgz#cef4d8353dac624086069e290d9631ca285df34d"
+  integrity sha512-OXLD9XRxMhd3S0LWuPHpiARQOI7z9tCOs0FsynccW2lmyZzHHFJ9/eR6kuK9xF459Qf+740qI5h+/0cx+NljzA==
+  dependencies:
+    escape-string-regexp "^4.0.0"
+    lodash.sortby "^4.7.0"
+    lodash.uniq "^4.5.0"
+    lodash.uniqwith "^4.5.0"
+
 "@textlint/runtime-helper@^0.16.0":
   version "0.16.0"
   resolved "https://registry.yarnpkg.com/@textlint/runtime-helper/-/runtime-helper-0.16.0.tgz#b59967ac861cc873bf3e9cd69739f9bf098a2534"
@@ -4656,11 +4666,26 @@ lodash.ismatch@^4.4.0:
   resolved "https://registry.yarnpkg.com/lodash.ismatch/-/lodash.ismatch-4.4.0.tgz#756cb5150ca3ba6f11085a78849645f188f85f37"
   integrity sha512-fPMfXjGQEV9Xsq/8MTSgUf255gawYRbjwMyDbcvDhXgV7enSZA0hynz6vMPnpAb5iONEzBHBPsT+0zes5Z301g==
 
+lodash.sortby@^4.7.0:
+  version "4.7.0"
+  resolved "https://registry.yarnpkg.com/lodash.sortby/-/lodash.sortby-4.7.0.tgz#edd14c824e2cc9c1e0b0a1b42bb5210516a42438"
+  integrity sha512-HDWXG8isMntAyRF5vZ7xKuEvOhT4AhlRt/3czTSjvGUxjYCBVRQY48ViDHyfYz9VIoBkW4TMGQNapx+l3RUwdA==
+
 lodash.truncate@^4.4.2:
   version "4.4.2"
   resolved "https://registry.yarnpkg.com/lodash.truncate/-/lodash.truncate-4.4.2.tgz#5a350da0b1113b837ecfffd5812cbe58d6eae193"
   integrity sha512-jttmRe7bRse52OsWIMDLaXxWqRAmtIUccAQ3garviCqJjafXOfNMO0yMfNpdD6zbGaTU0P5Nz7e7gAT6cKmJRw==
 
+lodash.uniq@^4.5.0:
+  version "4.5.0"
+  resolved "https://registry.yarnpkg.com/lodash.uniq/-/lodash.uniq-4.5.0.tgz#d0225373aeb652adc1bc82e4945339a842754773"
+  integrity sha512-xfBaXQd9ryd9dlSDvnvI0lvxfLJlYAZzXomUYzLKtUeOQvOP5piqAWuGtrhWeqaXK9hhoM/iyJc5AV+XfsX3HQ==
+
+lodash.uniqwith@^4.5.0:
+  version "4.5.0"
+  resolved "https://registry.yarnpkg.com/lodash.uniqwith/-/lodash.uniqwith-4.5.0.tgz#7a0cbf65f43b5928625a9d4d0dc54b18cadc7ef3"
+  integrity sha512-7lYL8bLopMoy4CTICbxygAUq6CdRJ36vFc80DucPueUee+d5NBRxz3FdT9Pes/HEx5mPoT9jwnsEJWz1N7uq7Q==
+
 lodash@^4.17.21:
   version "4.17.21"
   resolved "https://registry.yarnpkg.com/lodash/-/lodash-4.17.21.tgz#679591c564c3bffaae8454cf0b3df370c3d6911c"
