@tryghost/html-to-plaintext 1.0.1

package/README.md

@@ -0,0 +1,21 @@
+# Html To Plaintext
+
+
+## Usage
+
+
+## Develop
+
+This is a monorepo package.
+
+Follow the instructions for the top-level repo.
+1. `git clone` this repo & `cd` into it as usual
+2. Run `yarn` to install top-level dependencies.
+
+
+
+## Test
+
+- `yarn lint` run just eslint
+- `yarn test` run lint and tests
+

package/index.js

@@ -0,0 +1 @@
+module.exports = require('./lib/html-to-plaintext');

package/lib/html-to-plaintext.js

@@ -0,0 +1,120 @@
+const _ = require('lodash');
+
+const mergeSettings = (extraSettings) => {
+    return _.mergeWith({}, baseSettings, extraSettings, function customizer(objValue, srcValue) {
+        if (_.isArray(objValue)) {
+            return objValue.concat(srcValue);
+        }
+    });
+};
+
+const baseSettings = {
+    wordwrap: false,
+    preserveNewlines: true,
+
+    // equiv returnDomByDefault: true,
+    baseElements: {returnDomByDefault: true},
+    selectors: [
+        // Ignore images, equiv ignoreImage: true
+        {selector: 'img', format: 'skip'},
+
+        // disable uppercase headings, equiv uppercaseHeadings: false
+        {selector: 'h1', options: {uppercase: false}},
+        {selector: 'h2', options: {uppercase: false}},
+        {selector: 'h3', options: {uppercase: false}},
+        {selector: 'h4', options: {uppercase: false}},
+        {selector: 'h5', options: {uppercase: false}},
+        {selector: 'h6', options: {uppercase: false}},
+        {selector: 'table', options: {uppercaseHeaderCells: false}},
+
+        // Backwards compatibility with html-to-text 5.1.1
+        {selector: 'div', format: 'inline'}
+    ]
+};
+
+let excerptConverter;
+let emailConverter;
+let commentConverter;
+let commentSnippetConverter;
+
+const loadConverters = () => {
+    if (excerptConverter && emailConverter) {
+        return;
+    }
+
+    const {compile} = require('html-to-text');
+
+    const excerptSettings = mergeSettings({
+        selectors: [
+            {selector: 'a', options: {ignoreHref: true}},
+            {selector: 'figcaption', format: 'skip'},
+            // Strip inline and bottom footnotes
+            {selector: 'a[rel=footnote]', format: 'skip'},
+            {selector: 'div.footnotes', format: 'skip'},
+            // Don't output hrs
+            {selector: 'hr', format: 'skip'},
+            // Don't output > in blockquotes
+            {selector: 'blockquote', format: 'block'},
+            // Don't include signup cards in excerpts
+            {selector: '.kg-signup-card', format: 'skip'}
+        ]
+    });
+
+    const emailSettings = mergeSettings({
+        selectors: [
+            // equiv hideLinkHrefIfSameAsText: true
+            {selector: 'a', options: {hideLinkHrefIfSameAsText: true}},
+            // Don't include html .preheader in email
+            {selector: '.preheader', format: 'skip'}
+        ]
+    });
+
+    const commentSettings = mergeSettings({
+        preserveNewlines: false,
+        selectors: [
+            // equiv hideLinkHrefIfSameAsText: true
+            {selector: 'a', options: {hideLinkHrefIfSameAsText: true}},
+            // No space between <p> tags. An empty <p> is needed
+            {selector: 'p', options: {leadingLineBreaks: 1, trailingLineBreaks: 1}}
+        ]
+    });
+
+    const commentSnippetSettings = mergeSettings({
+        preserveNewlines: false,
+        ignoreHref: true,
+        selectors: [
+            {selector: 'blockquote', format: 'skip'}
+        ]
+    });
+
+    excerptConverter = compile(excerptSettings);
+    emailConverter = compile(emailSettings);
+    commentConverter = compile(commentSettings);
+    commentSnippetConverter = compile(commentSnippetSettings);
+};
+
+module.exports.excerpt = (html) => {
+    loadConverters();
+
+    return excerptConverter(html);
+};
+
+module.exports.email = (html) => {
+    loadConverters();
+
+    return emailConverter(html);
+};
+
+module.exports.comment = (html) => {
+    loadConverters();
+
+    return commentConverter(html);
+};
+
+module.exports.commentSnippet = (html) => {
+    loadConverters();
+
+    return commentSnippetConverter(html)
+        .replace(/\n/g, ' ')
+        .replace(/\s+/g, ' ');
+};

package/package.json

@@ -0,0 +1,28 @@
+{
+  "name": "@tryghost/html-to-plaintext",
+  "version": "1.0.1",
+  "repository": "https://github.com/TryGhost/Ghost/tree/main/packages/html-to-plaintext",
+  "author": "Ghost Foundation",
+  "private": false,
+  "main": "index.js",
+  "files": [
+    "index.js",
+    "lib"
+  ],
+  "scripts": {
+    "dev": "echo \"Implement me!\"",
+    "test:unit": "NODE_ENV=testing c8 --all --check-coverage --reporter text --reporter cobertura -- mocha --reporter dot './test/**/*.test.js'",
+    "test": "yarn test:unit",
+    "lint:code": "eslint *.js lib/ --ext .js --cache",
+    "lint": "yarn lint:code && yarn lint:test",
+    "lint:test": "eslint -c test/.eslintrc.js test/ --ext .js --cache"
+  },
+  "devDependencies": {
+    "c8": "8.0.1",
+    "mocha": "10.8.2"
+  },
+  "dependencies": {
+    "html-to-text": "8.2.1",
+    "lodash": "4.17.21"
+  }
+}