Refactored html-to-plaintext to have two functions

refs: https://github.com/TryGhost/Team/issues/1609

- Split html-to-plaintext into a function for excerpts and emails
- Updated all usages so they use the correct function
- There's currently no difference between the two

core/server/api/canary/utils/serializers/output/utils/post-gating.js

@@ -15,7 +15,7 @@ const forPost = (attrs, frame) => {
 
         if (paywallIndex !== -1) {
             attrs.html = attrs.html.slice(0, paywallIndex);
-            attrs.plaintext = htmlToPlaintext(attrs.html);
+            attrs.plaintext = htmlToPlaintext.excerpt(attrs.html);
 
             if (!attrs.custom_excerpt && attrs.excerpt) {
                 attrs.excerpt = attrs.plaintext.substring(0, 500);

core/server/data/migrations/versions/4.0/23-regenerate-posts-html.js

@@ -46,7 +46,7 @@ module.exports = createIrreversibleMigration(async (knex) => {
             };
 
             if (html !== post.html || !post.plaintext) {
-                const plaintext = htmlToPlaintext(html);
+                const plaintext = htmlToPlaintext.excerpt(html);
 
                 if (plaintext !== post.plaintext) {
                     updatedAttrs.plaintext = plaintext;

core/server/data/migrations/versions/4.9/05-fix-missed-mobiledoc-url-transforms.js

@@ -65,7 +65,7 @@ module.exports = createTransactionalMigration(
                     continue;
                 }
 
-                const plaintext = htmlToPlaintext(html);
+                const plaintext = htmlToPlaintext.excerpt(html);
 
                 await knex('posts')
                     .transacting(trx)

core/server/data/migrations/versions/5.0/2022-05-21-00-00-regenerate-posts-html.js

@@ -50,7 +50,7 @@ module.exports = createIrreversibleMigration(async (knex) => {
             };
 
             if (html !== post.html || !post.plaintext) {
-                const plaintext = htmlToPlaintext(html);
+                const plaintext = htmlToPlaintext.excerpt(html);
 
                 if (plaintext !== post.plaintext) {
                     updatedAttrs.plaintext = plaintext;

core/server/models/post.js

@@ -626,7 +626,7 @@ Post = ghostBookshelf.Model.extend({
             if (this.get('html') === null) {
                 plaintext = null;
             } else {
-                plaintext = htmlToPlaintext(this.get('html'));
+                plaintext = htmlToPlaintext.excerpt(this.get('html'));
             }
 
             // CASE: html is e.g. <p></p>

core/server/services/mega/post-email-serializer.js

@@ -248,7 +248,7 @@ const serialize = async (postModel, newsletter, options = {isBrowserPreview: fal
     `).remove();
     post.html = _cheerio('body').html();
 
-    post.plaintext = htmlToPlaintext(post.html);
+    post.plaintext = htmlToPlaintext.email(post.html);
 
     // Outlook will render feature images at full-size breaking the layout.
     // Content images fix this by rendering max 600px images - do the same for feature image here
@@ -321,7 +321,7 @@ function renderEmailForSegment(email, memberSegment) {
     });
 
     result.html = formatHtmlForEmail($.html());
-    result.plaintext = htmlToPlaintext(result.html);
+    result.plaintext = htmlToPlaintext.email(result.html);
 
     return result;
 }

core/shared/html-to-plaintext.js

@@ -1,28 +1,52 @@
-module.exports = function htmlToPlaintext(html) {
-    const {convert} = require('html-to-text');
+const baseSettings = {
+    wordwrap: false,
+    preserveNewlines: true,
 
-    return convert(html, {
-        wordwrap: false,
-        preserveNewlines: true,
+    // equiv returnDomByDefault: true,
+    baseElements: {returnDomByDefault: true},
+    selectors: [
+        // Ignore images, equiv ignoreImage: true
+        {selector: 'img', format: 'skip'},
 
-        // 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}},
-            // equiv hideLinkHrefIfSameAsText: true
-            {selector: 'a', options: {hideLinkHrefIfSameAsText: true}},
+        // 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'}
-        ]
-    });
+        // equiv hideLinkHrefIfSameAsText: true
+        {selector: 'a', options: {hideLinkHrefIfSameAsText: true}},
+
+        // Backwards compatibility with html-to-text 5.1.1
+        {selector: 'div', format: 'inline'}
+    ]
+};
+
+let excerptConverter;
+let emailConverter;
+
+const loadConverters = () => {
+    if (excerptConverter && emailConverter) {
+        return;
+    }
+
+    const {compile} = require('html-to-text');
+
+    excerptConverter = compile(baseSettings);
+    emailConverter = compile(baseSettings);
+};
+
+module.exports.excerpt = (html) => {
+    loadConverters();
+
+    return excerptConverter(html);
+};
+
+module.exports.email = (html) => {
+    loadConverters();
+
+    return emailConverter(html);
 };

test/unit/shared/html-to-plaintext.test.js

@@ -0,0 +1,40 @@
+const assert = require('assert');
+const htmlToPlaintext = require('../../../core/shared/html-to-plaintext');
+
+describe('Html to Plaintext', function () {
+    function getEmailandExcert(input) {
+        const excerpt = htmlToPlaintext.excerpt(input);
+        const email = htmlToPlaintext.email(input);
+
+        return {email, excerpt};
+    }
+
+    describe('excerpt vs email behaviour', function () {
+        it('example case with img & link', function () {
+            const input = '<p>Some thing <a href="https://google.com">Google</a> once told me.</p><img src="https://hotlink.com" alt="An important image"><p>And <strong>another</strong> thing.</p>';
+
+            const {excerpt, email} = getEmailandExcert(input);
+
+            assert.equal(excerpt, 'Some thing Google [https://google.com] once told me.\n\nAnd another thing.');
+            assert.equal(email, 'Some thing Google [https://google.com] once told me.\n\nAnd another thing.');
+        });
+
+        it('example case with figure + figcaption', function () {
+            const input = '<figcaption>A snippet from a post template</figcaption></figure><p>See? Not that scary! But still completely optional. </p>';
+
+            const {excerpt, email} = getEmailandExcert(input);
+
+            assert.equal(excerpt, 'A snippet from a post template\n\nSee? Not that scary! But still completely optional.');
+            assert.equal(email, 'A snippet from a post template\n\nSee? Not that scary! But still completely optional.');
+        });
+
+        it('example case with figure + figcaption inside a link', function () {
+            const input = '<a href="https://mysite.com"><figcaption>A snippet from a post template</figcaption></figure></a><p>See? Not that scary! But still completely optional. </p>';
+
+            const {excerpt, email} = getEmailandExcert(input);
+
+            assert.equal(excerpt, 'A snippet from a post template [https://mysite.com]\n\nSee? Not that scary! But still completely optional.');
+            assert.equal(email, 'A snippet from a post template [https://mysite.com]\n\nSee? Not that scary! But still completely optional.');
+        });
+    });
+});