fix(rule): fix false-positive result (#3)

azu · web-flow · commit 6a3063c16c34 · 2017-12-24T21:08:13.000+09:00
* fix(rule): fix false-positive result Completely rewrite sentence parser https://github.com/azu/sentence-splitter/releases/tag/3.0.0 fix #2 * chore(npm): remove unused library
diff --git a/package.json b/package.json
@@ -37,9 +37,9 @@
   "devDependencies": {
     "babel-register": "^6.26.0",
     "husky": "^0.14.3",
-    "lint-staged": "^4.3.0",
-    "prettier": "^1.7.4",
-    "textlint-scripts": "^1.3.0"
+    "lint-staged": "^6.0.0",
+    "prettier": "^1.9.2",
+    "textlint-scripts": "^1.4.0"
   },
   "prettier": {
     "printWidth": 120,
@@ -53,8 +53,7 @@
   },
   "dependencies": {
     "en-pos": "^1.0.16",
-    "sentence-splitter": "^2.2.0",
-    "textlint-rule-helper": "^2.0.0",
-    "textlint-util-to-string": "^2.1.1"
+    "sentence-splitter": "^3.0.0",
+    "textlint-rule-helper": "^2.0.0"
   }
 }
diff --git a/src/textlint-rule-en-capitalization.js b/src/textlint-rule-en-capitalization.js
@@ -1,10 +1,14 @@
 // MIT © 2017 azu
 "use strict";
 const { RuleHelper } = require("textlint-rule-helper");
-const StringSource = require("textlint-util-to-string");
-const { split, Syntax: SentenceSyntax } = require("sentence-splitter");
+const { splitAST, Syntax: SentenceSyntax } = require("sentence-splitter");
 import { getPosFromSingleWord, isCapitalized, upperFirstCharacter } from "./captalize";
 
+const REPORT_TYPE = {
+    Heading: "Heading",
+    Paragraph: "Paragraph",
+    List: "List"
+};
 const shouldNotCapitalized = (string, allowWords) => {
     // allow words
     const shouldAllowed = allowWords.some(allowWord => {
@@ -25,24 +29,6 @@ const shouldNotCapitalized = (string, allowWords) => {
     return false;
 };
 
-/**
- * get node at index
- * @param node
- * @param index
- * @returns {undefined|*}
- */
-const getNodeAtIndex = (node, index) => {
-    if (!node || !node.children) {
-        return undefined;
-    }
-    for (let i = 0; i < node.children.length; i++) {
-        const childNode = node.children[i];
-        if (childNode.range[0] <= index && index <= childNode.range[1]) {
-            return childNode;
-        }
-    }
-    return undefined;
-};
 /**
  * @param node
  * @param Syntax
@@ -52,49 +38,46 @@ const getNodeAtIndex = (node, index) => {
  * @param fixer
  * @param {boolean} allowFigures enable figures check
  * @param {string[]} allowWords allow lower-case words
+ * @param {string} reportType REPORT_TYPE
  */
-const checkNode = ({ node, Syntax, getSource, report, RuleError, fixer, allowFigures, allowWords }) => {
-    const source = new StringSource(node);
-    const sourceText = source.toString();
-    const sentences = split(sourceText);
-    sentences.filter(sentence => sentence.type === SentenceSyntax.Sentence).forEach(sentence => {
-        const originalIndex = source.originalIndexFromIndex(sentence.range[0]);
-        let targetNode;
-        if (node.type === Syntax.ListItem) {
-            targetNode = getNodeAtIndex(node.children[0], node.range[0] + originalIndex);
-        } else {
-            targetNode = getNodeAtIndex(node, originalIndex);
-        }
-        if (!targetNode) {
+const checkNode = ({ node, Syntax, getSource, report, RuleError, fixer, allowFigures, allowWords, reportType }) => {
+    const DocumentURL = "https://owl.english.purdue.edu/owl/resource/592/01/";
+    const paragraphNode = splitAST(node);
+    paragraphNode.children.filter(sentence => sentence.type === SentenceSyntax.Sentence).forEach(sentence => {
+        const sentenceFirstNode = sentence.children[0];
+        if (!sentenceFirstNode) {
             return;
         }
-        const DocumentURL = "https://owl.english.purdue.edu/owl/resource/592/01/";
-        // check
-        if (targetNode.type === Syntax.Str) {
-            const text = sentence.value;
-            const firstWord = text.split(" ")[0];
+        // check first word is String
+        if (sentenceFirstNode.type === Syntax.Str) {
+            const text = sentenceFirstNode.value;
+            const firstWord = text.split(/\s/)[0];
             if (isCapitalized(firstWord) || shouldNotCapitalized(firstWord, allowWords)) {
                 return;
             }
-            const index = originalIndex;
+            const index = 0;
             return report(
-                node,
+                sentenceFirstNode,
                 new RuleError(
-                    `Heading: Follow the standard capitalization rules for American English.
+                    `${reportType}: Follow the standard capitalization rules for American English.
 See ${DocumentURL}`,
                     {
                         index: index,
                         fix: fixer.replaceTextRange([index, index + firstWord.length], upperFirstCharacter(firstWord))
                     }
                 )
             );
-        } else if (allowFigures && targetNode.type === Syntax.Image && typeof targetNode.alt === "string") {
-            const text = targetNode.alt;
+        } else if (
+            allowFigures &&
+            sentenceFirstNode.type === Syntax.Image &&
+            typeof sentenceFirstNode.alt === "string"
+        ) {
+            const text = sentenceFirstNode.alt;
             if (isCapitalized(text) || shouldNotCapitalized(text, allowWords)) {
                 return;
             }
             return report(
-                targetNode,
+                sentenceFirstNode,
                 new RuleError(
                     `Image alt: Follow the standard capitalization rules for American English
 See ${DocumentURL}`
@@ -127,7 +110,17 @@ const report = (context, options = {}) => {
             if (!allowHeading) {
                 return;
             }
-            checkNode({ node, Syntax, getSource, report, RuleError, fixer, allowFigures, allowWords });
+            checkNode({
+                node,
+                Syntax,
+                getSource,
+                report,
+                RuleError,
+                fixer,
+                allowFigures,
+                allowWords,
+                reportType: REPORT_TYPE.Heading
+            });
         },
         [Syntax.Paragraph](node) {
             if (helper.isChildNode(node, [Syntax.Link, Syntax.Image, Syntax.BlockQuote, Syntax.Emphasis])) {
@@ -136,13 +129,35 @@ const report = (context, options = {}) => {
             if (helper.isChildNode(node, [Syntax.ListItem])) {
                 return;
             }
-            checkNode({ node, Syntax, getSource, report, RuleError, fixer, allowFigures, allowWords });
+            checkNode({
+                node,
+                Syntax,
+                getSource,
+                report,
+                RuleError,
+                fixer,
+                allowFigures,
+                allowWords,
+                reportType: REPORT_TYPE.Paragraph
+            });
         },
         [Syntax.ListItem](node) {
             if (!allowLists) {
                 return;
             }
-            checkNode({ node, Syntax, getSource, report, RuleError, fixer, allowFigures, allowWords });
+            node.children.forEach(paragraph => {
+                checkNode({
+                    node: paragraph,
+                    Syntax,
+                    getSource,
+                    report,
+                    RuleError,
+                    fixer,
+                    allowFigures,
+                    allowWords,
+                    reportType: REPORT_TYPE.List
+                });
+            });
         }
     };
 };
diff --git a/test/mocha.opts b/test/mocha.opts
@@ -1 +1 @@
---require babel-register
+--require textlint-scripts/register
diff --git a/test/textlint-rule-en-capitalization-test.js b/test/textlint-rule-en-capitalization-test.js
@@ -7,6 +7,8 @@ tester.run("capitalization", rule, {
     valid: [
         "In text, follow the standard capitalization rules for American English. Additionally:",
         "First, sentence should be capital. Second, sentence should be capital.",
+        "`Code`, this is ok.",
+        "This is **good**. But that it not good.",
         "# Capitalization in titles and headings",
         "## Capitalization and colons\n" +
             "Use a lowercase letter to begin the first word of the text immediately following a colon, unless the text is one of the following:",
@@ -28,7 +30,13 @@ tester.run("capitalization", rule, {
             options: {
                 allowWords: ["textlint"]
             }
-        }
+        },
+        // https://github.com/textlint-rule/textlint-rule-en-capitalization/issues/2
+        `# Anatomy of a Package
+
+A minimal npm package should contain metadata in a _package.json_ file and an associated source file (usually _index.js_). In practice, packages contain more than that and you will have at least a license file and the source in various formats.`,
+        // need space after .
+        "This is pass.it's style is bad..."
     ],
     invalid: [
         {
@@ -38,7 +46,7 @@ tester.run("capitalization", rule, {
                 {
                     index: 0,
                     message:
-                        "Heading: Follow the standard capitalization rules for American English.\n" +
+                        "Paragraph: Follow the standard capitalization rules for American English.\n" +
                         "See https://owl.english.purdue.edu/owl/resource/592/01/"
                 }
             ]
@@ -81,11 +89,21 @@ tester.run("capitalization", rule, {
 `,
             errors: [{}, {}, {}, {}, {}]
         },
+        {
+            text: "This is **good**. but that it not good.",
+            errors: [
+                {
+                    index: 18
+                }
+            ]
+        },
         {
             text: "![image](https://exmaple.com) is not capital.",
             errors: [
                 {
-                    index: 0
+                    index: 0,
+                    message: `Image alt: Follow the standard capitalization rules for American English
+See https://owl.english.purdue.edu/owl/resource/592/01/`
                 }
             ]
         }
diff --git a/yarn.lock b/yarn.lock

Original file line number	Diff line number	Diff line change
`@@ -1 +1 @@`
`1`		`---require babel-register`
	`1`	`+--require textlint-scripts/register`
Original file line number	Diff line number	Diff line change
`@@ -7,6 +7,8 @@ tester.run("capitalization", rule, {`
`7`	`7`	`valid: [`
`8`	`8`	`"In text, follow the standard capitalization rules for American English. Additionally:",`
`9`	`9`	`"First, sentence should be capital. Second, sentence should be capital.",`
	`10`	+ "`Code`, this is ok.",
	`11`	`+ "This is good. But that it not good.",`
`10`	`12`	`"# Capitalization in titles and headings",`
`11`	`13`	`"## Capitalization and colons\n" +`
`12`	`14`	`"Use a lowercase letter to begin the first word of the text immediately following a colon, unless the text is one of the following:",`
`@@ -28,7 +30,13 @@ tester.run("capitalization", rule, {`
`28`	`30`	`options: {`
`29`	`31`	`allowWords: ["textlint"]`
`30`	`32`	`}`
`31`		`- }`
	`33`	`+ },`
	`34`	`+ // https://github.com/textlint-rule/textlint-rule-en-capitalization/issues/2`
	`35`	+ `# Anatomy of a Package
	`36`	`+`
	`37`	+A minimal npm package should contain metadata in a _package.json_ file and an associated source file (usually _index.js_). In practice, packages contain more than that and you will have at least a license file and the source in various formats.`,
	`38`	`+ // need space after .`
	`39`	`+ "This is pass.it's style is bad..."`
`32`	`40`	`],`
`33`	`41`	`invalid: [`
`34`	`42`	`{`
`@@ -38,7 +46,7 @@ tester.run("capitalization", rule, {`
`38`	`46`	`{`
`39`	`47`	`index: 0,`
`40`	`48`	`message:`
`41`		`- "Heading: Follow the standard capitalization rules for American English.\n" +`
	`49`	`+ "Paragraph: Follow the standard capitalization rules for American English.\n" +`
`42`	`50`	`"See https://owl.english.purdue.edu/owl/resource/592/01/"`
`43`	`51`	`}`
`44`	`52`	`]`
`@@ -81,11 +89,21 @@ tester.run("capitalization", rule, {`
`81`	`89`	`,
`82`	`90`	`errors: [{}, {}, {}, {}, {}]`
`83`	`91`	`},`
	`92`	`+ {`
	`93`	`+ text: "This is good. but that it not good.",`
	`94`	`+ errors: [`
	`95`	`+ {`
	`96`	`+ index: 18`
	`97`	`+ }`
	`98`	`+ ]`
	`99`	`+ },`
`84`	`100`	`{`
`85`	`101`	`text: "![image](https://exmaple.com) is not capital.",`
`86`	`102`	`errors: [`
`87`	`103`	`{`
`88`		`- index: 0`
	`104`	`+ index: 0,`
	`105`	+ message: `Image alt: Follow the standard capitalization rules for American English
	`106`	+See https://owl.english.purdue.edu/owl/resource/592/01/`
`89`	`107`	`}`
`90`	`108`	`]`
`91`	`109`	`}`