7 |
19 |
20 |
21 |
--------------------------------------------------------------------------------
/src/tests/prepareMessage/trimmed-reply-bug.output-message.html:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
9 |
10 |
11 | 8 | Hello 9 |
10 |
11 | 
Cheers!
Jonathan 17 |
18 | Cheers!
Jonathan 17 |
Hi Mette,
12 |13 | I have attached our contracts for the USD 500.000 investment regarding BoostVC. Please let me know if thats what 14 | is needed. 15 |
16 |Kind Regards
17 |Henrique
18 |--
I use Tempo to improve my focus
2 |
3 |
4 |
23 |
--------------------------------------------------------------------------------
/src/tests/prepareMessage/attached-image.output-complete.html:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
10 |
11 |
12 | 4 |
5 |
22 |
6 | On Tue, Dec 31, 2019 at 12:08 AM Nicolas Gaborit <hello@soreine.dev> wrote:
8 |
9 | 8 |
13 |21 |14 |20 |15 |19 |16 | Replied message 17 |
18 |
13 |
25 |
26 |
27 |
--------------------------------------------------------------------------------
/src/tests/prepareMessage/attached-image.output-message.html:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
10 |
11 |
12 | 14 | Hello 15 |
16 |
17 |
Cheers!
Jonathan 23 |
24 | Cheers!
Jonathan 23 |
13 |
25 |
26 |
27 |
--------------------------------------------------------------------------------
/tsconfig.json:
--------------------------------------------------------------------------------
1 | {
2 | "include": ["src/**/*"],
3 | "exclude": [
4 | "lib/",
5 | "old/",
6 | "benchmarks/",
7 | "node_modules",
8 | "src/tests/",
9 | "**/*.test.ts"
10 | ],
11 | "compilerOptions": {
12 | "outDir": "lib/",
13 | "rootDir": "src/",
14 | "incremental": true,
15 | "target": "es5",
16 | "module": "commonjs",
17 | "lib": [],
18 | "allowJs": true,
19 | "skipLibCheck": true,
20 | "isolatedModules": true,
21 | "strict": true,
22 | /* Allow use of generators */
23 | "downlevelIteration": true,
24 | "moduleResolution": "node",
25 | "typeRoots": ["./typings", "node_modules/@types"],
26 | "allowSyntheticDefaultImports": true,
27 | "esModuleInterop": true,
28 | "forceConsistentCasingInFileNames": true
29 | }
30 | }
31 |
--------------------------------------------------------------------------------
/src/tests/prepareMessage/github-emoji.input.html:
--------------------------------------------------------------------------------
1 |
2 |
3 | 14 | Hello 15 |
16 |
17 |
Cheers!
Jonathan 23 |
24 | Cheers!
Jonathan 23 |
@alexandersandberg approved this pull request.
4 | 5 |Nice job!
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub, or unsubscribe.
@alexandersandberg approved this pull request.
14 | 15 |
16 | Nice job!
17 |
27 | —
You are receiving this because you authored the thread.
Reply to this email directly,
28 | view it on GitHub, or
29 | unsubscribe.
30 |
@alexandersandberg approved this pull request.
14 | 15 |
16 | Nice job!
17 |
27 | —
You are receiving this because you authored the thread.
Reply to this email directly,
28 | view it on GitHub, or
29 | unsubscribe.
30 |
12 |
13 |
35 |
36 |
37 |
--------------------------------------------------------------------------------
/src/tests/prepareMessage/no-empty-message.output-message.html:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
9 |
10 |
11 | 13 |
14 |
34 |
15 | On Tue, Dec 31, 2019 at 12:08 AM Nicolas Gaborit <hello@soreine.dev> wrote:
20 |
21 | 20 |
25 |33 |26 |32 |27 |31 |28 | Replied message 29 |
30 |
12 |
13 |
35 |
36 |
37 |
--------------------------------------------------------------------------------
/benchmarks/message-splitter.ts:
--------------------------------------------------------------------------------
1 | /*
2 | These benchmarks measures the performance of the message-splitter module
3 | */
4 |
5 | import cheerio from 'cheerio';
6 | import { createSuite } from './utils';
7 | import prepareMessage, { linkify } from '../src';
8 | import EMAILS from './emails';
9 | import removeQuotations from '../src/removeQuotations';
10 |
11 | const suite = createSuite();
12 |
13 | // Measure against a real-world, HTML-heavy, marketing email
14 | suite
15 | .add('removeQuotations # Marketing email', () => {
16 | removeQuotations(cheerio.load(EMAILS.MARKETING));
17 | })
18 | .add('linkify # Marketing email', () => {
19 | linkify(EMAILS.MARKETING);
20 | })
21 | .add('prepareMessage # Marketing email', () => {
22 | prepareMessage(EMAILS.MARKETING);
23 | });
24 |
25 | // Using a linear scale of input complexity, we can see if the time complexity is linear.
26 | suite
27 | .add('prepareMessage # Size 1', () => {
28 | prepareMessage(EMAILS.BASIC);
29 | })
30 | .add('prepareMessage # Size 2', () => {
31 | prepareMessage(EMAILS.BASIC_REPLIED_X1);
32 | })
33 | .add('prepareMessage # Size 3', () => {
34 | prepareMessage(EMAILS.BASIC_REPLIED_X2);
35 | });
36 |
37 | suite.run();
38 |
--------------------------------------------------------------------------------
/src/tests/utils.ts:
--------------------------------------------------------------------------------
1 | import fs from 'fs';
2 | import path from 'path';
3 | import prettier from 'prettier';
4 | import expect from 'expect';
5 |
6 | function formatHtml(html: string): string {
7 | return prettier.format(html, {
8 | parser: 'html',
9 | endOfLine: 'lf',
10 | printWidth: 120,
11 | });
12 | }
13 | /**
14 | * Expect two HTMLs to be identical, disregarding formatting differences
15 | */
16 | function expectHtml(actual: string, expected: string) {
17 | // Use prettier to avoid formatting discrepencies
18 | actual = formatHtml(actual);
19 | expected = formatHtml(expected);
20 |
21 | // console.log(actual);
22 | // console.log(expected);
23 | expect(actual).toBe(expected);
24 | }
25 |
26 | function readFile(...paths: string[]): string {
27 | return fs.readFileSync(path.join(...paths)).toString();
28 | }
29 |
30 | function fileExists(...paths: string[]): boolean {
31 | return fs.existsSync(path.join(...paths));
32 | }
33 |
34 | function readFileIfExists(...paths: string[]): string | null {
35 | if (!fileExists(...paths)) {
36 | return null;
37 | } else {
38 | return readFile(...paths);
39 | }
40 | }
41 |
42 | export { expectHtml, formatHtml, readFile, fileExists, readFileIfExists };
43 |
--------------------------------------------------------------------------------
/.circleci/config.yml:
--------------------------------------------------------------------------------
1 | version: 2.1
2 | orbs:
3 | node: circleci/node@1.1.6
4 | commands:
5 | yarn_install:
6 | steps:
7 | - restore_cache:
8 | name: Restore Yarn Package Cache
9 | keys:
10 | - yarn-packages-{{ checksum "yarn.lock" }}
11 | - run:
12 | name: Install Dependencies
13 | command: yarn install --frozen-lockfile
14 | - save_cache:
15 | name: Save Yarn Package Cache
16 | key: yarn-packages-{{ checksum "yarn.lock" }}
17 | paths:
18 | - ~/.cache/yarn
19 |
20 | jobs:
21 | build:
22 | executor:
23 | name: node/default
24 | steps:
25 | - checkout
26 | - yarn_install
27 | - run:
28 | name: Type check and lint
29 | command: yarn lint
30 | - run:
31 | name: Test
32 | command: yarn test
33 | - run:
34 | name: Build
35 | command: yarn build
36 | workflows:
37 | test-and-build:
38 | jobs:
39 | - build
40 |
--------------------------------------------------------------------------------
/.eslintrc.js:
--------------------------------------------------------------------------------
1 | module.exports = {
2 | env: {
3 | browser: true,
4 | commonjs: true,
5 | es6: true,
6 | node: true,
7 | },
8 | extends: [
9 | 'plugin:@typescript-eslint/recommended', // Base rules configuration from Eslint, adapted to Typescript
10 | 'prettier/@typescript-eslint', // disable ESLint rules from @typescript-eslint/eslint-plugin that would conflict with prettier
11 | 'plugin:prettier/recommended', // Enables eslint-plugin-prettier and eslint-config-prettier. This will display prettier errors as ESLint errors,, so that they can be fixed automatically with `eslint --fix`. Make sure this is always the last configuration in the extends array.
12 | ],
13 | parser: '@typescript-eslint/parser',
14 | parserOptions: {
15 | ecmaVersion: 2018,
16 | sourceType: 'module',
17 | },
18 | plugins: ['@typescript-eslint', 'prettier'],
19 | rules: {
20 | indent: ['off', 'tab', { SwitchCase: 1 }],
21 | 'linebreak-style': ['error', 'unix'],
22 | 'no-useless-escape': [0, 'unix'],
23 | semi: ['error', 'always'],
24 | '@typescript-eslint/no-use-before-define': 'off',
25 | '@typescript-eslint/explicit-function-return-type': 'off',
26 | '@typescript-eslint/no-empty-function': 'off',
27 | '@typescript-eslint/no-unused-vars': 'error',
28 | },
29 | ignorePatterns: ['./old/', './node_modules/'],
30 | };
31 |
--------------------------------------------------------------------------------
/src/enhanceLinks.ts:
--------------------------------------------------------------------------------
1 | import { validate as isValidEmail } from 'email-validator';
2 | import isValidDomain from 'is-valid-domain';
3 |
4 | export const hasProtocol = (link: string) => {
5 | const lowerCaseLink = link.toLowerCase();
6 | return Boolean(lowerCaseLink.match(/^[a-z][a-z0-9+-.]*:/i));
7 | };
8 |
9 | /**
10 | * Scans the document for links that don't start
11 | * with a protocol and adds one automatically.
12 | */
13 | const addHttpsToRelativeLinks = ($: CheerioStatic) => {
14 | $('a').each((_, link) => {
15 | const href = link.attribs.href;
16 |
17 | if (href && !hasProtocol(href)) {
18 | if (isValidDomain(href.split('/')[0])) {
19 | // Valid domains are prepended with https://
20 | link.attribs.href = `https://${href}`;
21 | } else if (isValidEmail(href)) {
22 | // Valid email addresses get an extra mailto:
23 | link.attribs.href = `mailto:${href}`;
24 | }
25 | }
26 | });
27 | };
28 |
29 | const addTitlesToLinks = ($: CheerioStatic) => {
30 | $('a').each((_, link) => {
31 | if (link.attribs.title) {
32 | link.attribs.title += ` (${link.attribs.href})`;
33 | } else {
34 | link.attribs.title = link.attribs.href;
35 | }
36 | });
37 | };
38 |
39 | export const enhanceLinks = ($: CheerioStatic) => {
40 | addHttpsToRelativeLinks($);
41 | addTitlesToLinks($);
42 | };
43 |
--------------------------------------------------------------------------------
/benchmarks/libraries-output.txt:
--------------------------------------------------------------------------------
1 | Parse # JSDom # Marketing email
2 | 53.3 ms
3 | 18.76 ops/sec
4 | (39 runs sampled)
5 | Relative Margin of Error: ±5.76%
6 |
7 | Parse # Cheerio # Marketing email
8 | 1.808 ms
9 | 553.1 ops/sec
10 | (92 runs sampled)
11 | Relative Margin of Error: ±0.57%
12 |
13 | Parse # XmlDom # Marketing email
14 | 2.665 ms
15 | 375.3 ops/sec
16 | (90 runs sampled)
17 | Relative Margin of Error: ±0.63%
18 |
19 | Sanitizing # Marketing # DOMPurify
20 | 89.84 ms
21 | 11.13 ops/sec
22 | (32 runs sampled)
23 | Relative Margin of Error: ±2.51%
24 |
25 | Sanitizing # Marketing # sanitize-html
26 | 2.831 ms
27 | 353.2 ops/sec
28 | (87 runs sampled)
29 | Relative Margin of Error: ±1.94%
30 |
31 | Quotation # TalonJS # Marketing email
32 | 184.5 ms
33 | 5.419 ops/sec
34 | (18 runs sampled)
35 | Relative Margin of Error: ±2.01%
36 |
37 | Quotation # Planer with JSDom # Marketing email
38 | 1,096 ms
39 | 0.9127 ops/sec
40 | (7 runs sampled)
41 | Relative Margin of Error: ±4.44%
42 |
43 | Auto links # Marketing # autolink
44 | 6.578 ms
45 | 152 ops/sec
46 | (86 runs sampled)
47 | Relative Margin of Error: ±0.63%
48 |
49 | Auto links # Marketing # linkify
50 | 14.82 ms
51 | 67.47 ops/sec
52 | (70 runs sampled)
53 | Relative Margin of Error: ±0.70%
54 |
55 | Done in 62.65s.
--------------------------------------------------------------------------------
/src/linkify.ts:
--------------------------------------------------------------------------------
1 | import Autolinker from 'autolinker';
2 |
3 | const GH_EMOJI_URL = 'github.githubassets.com/images/icons/emoji';
4 |
5 | /**
6 | * Wrap text links in anchor tags
7 | */
8 | function linkify(inputHtml: string): string {
9 | const headOffset = inputHtml.indexOf('');
10 |
11 | return Autolinker.link(inputHtml, {
12 | urls: {
13 | schemeMatches: true,
14 | wwwMatches: true,
15 | tldMatches: true,
16 | },
17 | email: true,
18 | phone: true,
19 | mention: false,
20 | hashtag: false,
21 |
22 | stripPrefix: false,
23 | stripTrailingSlash: false,
24 | newWindow: true,
25 |
26 | className: '',
27 |
28 | replaceFn: function(match) {
29 | if (match.getType() === 'url') {
30 | // Don't autolink filenames
31 | // https://github.com/gregjacobs/Autolinker.js/issues/270#issuecomment-498878987
32 | const previousChar = inputHtml.charAt(match.getOffset() - 1);
33 | if (previousChar === '/') {
34 | return false; // don't autolink this match
35 | }
36 |
37 | // Don't autolink URLs for GitHub's Emoji
38 | if (match.getAnchorHref().includes(GH_EMOJI_URL)) {
39 | return false;
40 | }
41 |
42 | // Ignore URLs in head
43 | if (match.getOffset() < headOffset) {
44 | return false;
45 | }
46 |
47 | // To avoid tabnabbing
48 | const tag = match.buildTag();
49 | tag.setAttr('rel', 'noopener noreferrer');
50 | }
51 | },
52 | });
53 | }
54 |
55 | export default linkify;
56 |
--------------------------------------------------------------------------------
/src/fixBrokenHtml.ts:
--------------------------------------------------------------------------------
1 | import * as htmlparser2 from 'htmlparser2';
2 |
3 | /**
4 | * Fix various problems in input HTML before it can be parsed by cheerio.
5 | */
6 | function fixBrokenHtml(inputHtml: string): string {
7 | const fixedHead = fixBrokenHead(inputHtml);
8 |
9 | return fixedHead;
10 | }
11 |
12 | function fixBrokenHead(inputHtml: string): string {
13 | const encounteredTags = new Set13 |
14 |
34 |
15 | On Tue, Dec 31, 2019 at 12:08 AM Nicolas Gaborit <hello@soreine.dev> wrote:
20 |
21 | 20 |
25 |33 |26 |32 |27 |31 |28 | Replied message 29 |
30 |
21 |
39 |
40 |
41 |
--------------------------------------------------------------------------------
/src/removeTrailingWhitespaces.ts:
--------------------------------------------------------------------------------
1 | import {
2 | getTopLevelElement,
3 | isRootElement,
4 | isTextualElement,
5 | isEmptyish,
6 | } from './cheerio-utils';
7 |
8 | /**
9 | * Remove trailing whitespaces in given element, using given cheerio context.
10 | * Returns true if the element was empty and removed completely
11 | */
12 | function removeTrailingWhitespaces(
13 | $: CheerioStatic,
14 | el: CheerioElement = getTopLevelElement($)
15 | ): boolean {
16 | const hasChildren = el.childNodes && el.childNodes.length > 0;
17 | const isText = el.type === 'text';
18 | const isComment = el.type === 'comment';
19 | const isTextual = isTextualElement(el);
20 |
21 | if (isComment) {
22 | // Remove it
23 | $(el).remove();
24 | return true;
25 | } else if (isText) {
26 | if (isEmptyish(el)) {
27 | $(el).remove();
28 | // The element was removed completely
29 | return true;
30 | } else {
31 | const trimmed = (el.data as string).trimRight();
32 | $(el).replaceWith(trimmed);
33 | // We're done trimming
34 | return false;
35 | }
36 | } else if (!isTextual) {
37 | // Contains content other than text, we stop trimming here
38 | return false;
39 | } else if (hasChildren) {
40 | // Textual element with children
41 |
42 | // Trim last child
43 | const wasEmpty = removeTrailingWhitespaces($, el.lastChild);
44 | if (wasEmpty) {
45 | // Continue trimming this element
46 | return removeTrailingWhitespaces($, el);
47 | } else {
48 | // The last element was trimmed as much as possible.
49 | // We stop here
50 | return false;
51 | }
52 | } else if (isRootElement(el)) {
53 | // Stop here
54 | return false;
55 | } else {
56 | // Empty textual element, we can remove it.
57 | $(el).remove();
58 | return true;
59 | }
60 | }
61 |
62 | export default removeTrailingWhitespaces;
63 |
--------------------------------------------------------------------------------
/benchmarks/emails/basic-lorem-gmail.html:
--------------------------------------------------------------------------------
1 |
22 |
38 |
23 | Here is the body of the message, with a text
24 | link.com
25 |
26 | 
29 |
30 | An hoc usque quaque, aliter in vita? Terram, mihi crede, ea lanx et maria deprimet. Duo Reges:
31 | constructio interrete. Id est enim, de quo quaerimus. Parvi enim primo ortu sic iacent, tamquam omnino
32 | sine animo sint. Quis est tam dissimile homini. Claudii libidini, qui tum erat summo ne imperio, dederetur.
33 | Beatus sibi videtur esse moriens.
36 |
2 |
40 |
--------------------------------------------------------------------------------
/src/tests/appendStyle.test.ts:
--------------------------------------------------------------------------------
1 | import cheerio from 'cheerio';
2 | import appendStyle from '../appendStyle';
3 | import { expectHtml } from './utils';
4 |
5 | describe('appendStyle', () => {
6 | it('should add style', () => {
7 | const email = `
8 |
3 |
39 | 4 | Lorem ipsum dolor sit amet, consectetur adipiscing elit. 5 | An hoc usque quaque, aliter in vita? Terram, mihi crede, ea 6 | lanx et maria deprimet. Duo Reges: constructio interrete. 7 | Id est enim, de quo quaerimus. Parvi enim primo ortu sic 8 | iacent, tamquam omnino sine animo sint. Quis est tam dissimile 9 | homini. Claudii libidini, qui tum erat summo ne imperio, dederetur. 10 | Beatus sibi videtur esse moriens. 13 |
14 | 15 |16 | Hoc dixerit potius Ennius: Nimium boni est, cui nihil est mali. 17 | Egone quaeris, inquit, quid sentiam? Si mala non sunt, iacet omnis 18 | ratio Peripateticorum. Itaque rursus eadem ratione, qua sum paulo 19 | ante usus, haerebitis. Paulum, cum regem Persem captum adduceret, 20 | eodem flumine invectio? Mihi enim erit isdem istis fortasse iam 21 | utendum. Ab hoc autem quaedam non melius quam veteres, quaedam 22 | omnino relicta. Tamen a proposito, inquam, aberramus. 23 |
24 | 25 |26 | Sic, et quidem diligentius saepiusque ista loquemur inter nos 27 | agemusque communiter. 28 | Quid sequatur, quid repugnet, vident. 31 | Serpere anguiculos, nare anaticulas, evolare merulas, cornibus uti 32 | videmus boves, nepas aculeis. Ego vero volo in virtute vim esse quam 33 | maximam; Aufert enim sensus actionemque tollit omnem. Qua tu etiam 34 | inprudens utebare non numquam. Apud ceteros autem philosophos, qui 35 | quaesivit aliquid, tacet; Totum autem id externum est, et quod 36 | externum, id in casu est. 37 |
38 |Hello
9 | `;
10 |
11 | const $ = cheerio.load(email);
12 | appendStyle(
13 | $,
14 | `
15 | p {
16 | background: red;
17 | }
18 |
19 | .title {
20 | color: black;
21 | }
22 | `
23 | );
24 |
25 | const actual = $.html();
26 |
27 | const expected = `
28 |
29 |
30 |
39 |
40 |
41 | Hello
42 |
43 |
44 | `;
45 |
46 | expectHtml(actual, expected);
47 | });
48 |
49 | it('should append style after existing ones', () => {
50 | const email = `
51 |
52 |
53 |
58 |
59 |
60 | Hello
61 |
62 |
63 | `;
64 |
65 | const $ = cheerio.load(email);
66 | appendStyle(
67 | $,
68 | `
69 | .title {
70 | color: black;
71 | }
72 | `
73 | );
74 |
75 | const actual = $.html();
76 |
77 | const expected = `
78 |
79 |
80 |
85 |
90 |
91 |
92 | Hello
93 |
94 |
95 | `;
96 |
97 | expectHtml(actual, expected);
98 | });
99 | });
100 |
--------------------------------------------------------------------------------
/src/blockRemoteContent/index.ts:
--------------------------------------------------------------------------------
1 | import cheerio from 'cheerio';
2 | import blockRemoteContentInAttributes from './blockRemoteContentInAttributes';
3 | import blockRemoteContentInStyle from './blockRemoteContentInStyle';
4 |
5 | // This is a 1x100 transparent PNG used to replace images
6 | // Generated using http://png-pixel.com/
7 | // Note: using a 1x1 square results in large square empty
8 | // spaces in many e-mails, because only the width is
9 | // defined in the HTML; and the height gets scaled
10 | // proportionally. Thus the 1x100 ratio instead
11 | // https://github.com/mailpile/Mailpile/blob/babc3e5c3e7dfa3326998d1628ffad5b0bbd27f5/shared-data/default-theme/html/jsapi/message/html-sandbox.js#L43-L47
12 | const TRANSPARENT_1X100_URL =
13 | 'data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=';
14 |
15 | // What to replace remote URLs with
16 | export type ReplacementOptions = {
17 | image?: string;
18 | other?: string;
19 | };
20 |
21 | /**
22 | * Replace all URLs that could be automatically fetched when displaying the HTML. These can be used for tracking, or can consume bandwidth.
23 | */
24 | function blockRemoteContentCheerio(
25 | $: CheerioStatic,
26 | replacements: ReplacementOptions = {}
27 | ) {
28 | const { image = TRANSPARENT_1X100_URL, other = '#' } = replacements;
29 |
30 | // Block remote URLs in style tags
31 | blockRemoteContentInStyle($, image);
32 | // Block remote URLs in tags attributes
33 | blockRemoteContentInAttributes($, { image, other });
34 | }
35 |
36 | /**
37 | * Same as blockRemoteContentCheerio, but to be used as a standalone.
38 | */
39 | function blockRemoteContent(
40 | html: string,
41 | replacements: ReplacementOptions = {}
42 | ): string {
43 | const $ = cheerio.load(html);
44 | blockRemoteContentCheerio($, replacements);
45 | return $.html();
46 | }
47 |
48 | export { blockRemoteContentCheerio, blockRemoteContent };
49 |
--------------------------------------------------------------------------------
/package.json:
--------------------------------------------------------------------------------
1 | {
2 | "name": "tempo-email-parser",
3 | "version": "0.7.7",
4 | "description": "Processes HTML email for display. Extracts quotations and more.",
5 | "main": "./lib/index.js",
6 | "types": "./lib/index.d.ts",
7 | "scripts": {
8 | "prepublishOnly": "yarn run build",
9 | "build": "tsc -D",
10 | "test": "jest",
11 | "generate:fixtures": "npx ts-node ./src/tests/prepareMessage/generateFixtureOutputs.ts",
12 | "lint": "tsc --noEmit && eslint . --ext .js,.jsx,.ts,.tsx",
13 | "benchmarks": "npx ts-node --transpile-only ./benchmarks/message-splitter.ts",
14 | "benchmarks:libraries": "npx ts-node --transpile-only ./benchmarks/libraries.ts"
15 | },
16 | "repository": {
17 | "type": "git",
18 | "url": "git+https://github.com/yourtempo/tempo-email-parser.git"
19 | },
20 | "author": "",
21 | "license": "ISC",
22 | "bugs": {
23 | "url": "https://github.com/yourtempo/tempo-email-parser/issues"
24 | },
25 | "homepage": "https://github.com/yourtempo/tempo-email-parser#readme",
26 | "devDependencies": {
27 | "@types/benchmark": "^1.0.31",
28 | "@types/dompurify": "^2.0.1",
29 | "@types/jest": "^26.0.22",
30 | "@types/jsdom": "^12.2.4",
31 | "@types/linkifyjs": "^2.1.3",
32 | "@types/prettier": "^1.19.0",
33 | "@types/sanitize-html": "^1.20.2",
34 | "@types/xmldom": "^0.1.29",
35 | "@typescript-eslint/eslint-plugin": "^2.14.0",
36 | "@typescript-eslint/parser": "^2.14.0",
37 | "benchmark": "^2.1.4",
38 | "dompurify": "^2.0.7",
39 | "eslint": "^6.8.0",
40 | "eslint-config-prettier": "^6.9.0",
41 | "eslint-plugin-prettier": "^3.1.2",
42 | "jest": "^26.6.3",
43 | "jsdom": "^15.2.1",
44 | "linkifyjs": "^2.1.8",
45 | "planer": "^1.1.1",
46 | "prettier": "^1.19.1",
47 | "sanitize-html": "^1.20.1",
48 | "talonjs": "^1.0.24",
49 | "ts-jest": "^26.5.4",
50 | "typescript": "^4.2.3",
51 | "xmldom": "^0.2.1"
52 | },
53 | "dependencies": {
54 | "@types/cheerio": "^0.22.15",
55 | "autolinker": "^3.11.1",
56 | "cheerio": "^1.0.0-rc.3",
57 | "email-validator": "^2.0.4",
58 | "htmlparser2": "^6.0.1",
59 | "is-valid-domain": "^0.0.20",
60 | "regx": "^1.0.4"
61 | }
62 | }
63 |
--------------------------------------------------------------------------------
/benchmarks/libraries.ts:
--------------------------------------------------------------------------------
1 | /*
2 | These benchmarks compare different libraries (not all are used by this project).
3 | This helps estimate the cost of using any of them.
4 | */
5 |
6 | import { JSDOM } from 'jsdom';
7 | import sanitizeHtml from 'sanitize-html';
8 | import * as htmlparser2 from 'htmlparser2';
9 | import createDOMPurify from 'dompurify';
10 | import cheerio from 'cheerio';
11 | import Talon from 'talonjs';
12 | import XmlDom from 'xmldom';
13 | import planer from 'planer';
14 | import Autolinker from 'autolinker';
15 | import linkifyHtml from 'linkifyjs/html';
16 |
17 | import { createSuite } from './utils';
18 | import EMAILS from './emails';
19 |
20 | const suite = createSuite();
21 |
22 | // Test parsing a real-world, HTML-heavy, marketing email
23 | suite
24 | .add('Parse # JSDom # Marketing email', () => {
25 | new JSDOM(EMAILS.MARKETING);
26 | })
27 | .add('Parse # Cheerio # Marketing email', () => {
28 | cheerio.load(EMAILS.MARKETING);
29 | })
30 | .add('Parse # HtmlParser2 # Marketing email', () => {
31 | htmlparser2.parseDOM(EMAILS.MARKETING);
32 | })
33 | .add('Parse # XmlDom # Marketing email', () => {
34 | const parser = new XmlDom.DOMParser();
35 | parser.parseFromString(EMAILS.MARKETING);
36 | });
37 |
38 | const window = new JSDOM().window;
39 | const DOMPurify = createDOMPurify(window);
40 |
41 | suite
42 | .add('Sanitizing # Marketing # DOMPurify', () => {
43 | DOMPurify.sanitize(EMAILS.MARKETING, { WHOLE_DOCUMENT: true });
44 | })
45 | .add('Sanitizing # Marketing # sanitize-html', () => {
46 | sanitizeHtml(EMAILS.MARKETING);
47 | });
48 |
49 | suite
50 | .add('Quotation # TalonJS # Marketing email', () => {
51 | Talon.quotations.extractFromHtml(EMAILS.MARKETING);
52 | })
53 | .add('Quotation # Planer with JSDom # Marketing email', () => {
54 | const document = new JSDOM().window.document;
55 | planer.extractFrom(EMAILS.MARKETING, 'text/html', document);
56 | });
57 |
58 | suite
59 | .add('Auto links # Marketing # autolink', () => {
60 | Autolinker.link(EMAILS.MARKETING);
61 | })
62 | .add('Auto links # Marketing # linkify', () => {
63 | linkifyHtml(EMAILS.MARKETING);
64 | });
65 |
66 | suite.run();
67 |
--------------------------------------------------------------------------------
/src/tests/walkBackwards.test.ts:
--------------------------------------------------------------------------------
1 | import cheerio from 'cheerio';
2 | import expect from 'expect';
3 | import walkBackwards from '../walkBackwards';
4 | import { getTopLevelElement } from '../cheerio-utils';
5 |
6 | function printEl(el: CheerioElement): string {
7 | if (el.type === 'text') {
8 | return (el.data as string).trim();
9 | } else {
10 | return el.tagName;
11 | }
12 | }
13 |
14 | describe('walkBackwards', () => {
15 | it('should walk depth-first, in reverse order', () => {
16 | const $ = cheerio.load(`
17 |
18 |
19 |
20 |
18 |
47 |
67 |
--
68 |
75 |
76 |
19 |
20 | 
28 |
29 |
46 |
30 | Here is the body of the message, with a text link.com
31 |
32 |
33 |
34 | An hoc usque quaque, aliter in vita? Terram, mihi crede, ea lanx et maria deprimet. Duo
35 | Reges: constructio interrete. Id est enim, de quo quaerimus. Parvi enim primo ortu sic
36 | iacent, tamquam omnino sine animo sint. Quis est tam dissimile homini. Claudii libidini, qui tum
37 | erat summo ne imperio, dederetur.
38 | Beatus sibi videtur esse moriens.
39 |
42 |
43 |
44 |
47 |
48 |
66 |
49 | On Tue, Dec 31, 2019 at 12:08 AM Nicolas Gaborit <hello@soreine.dev> wrote:
52 |
53 | 52 |
57 |65 |58 |64 |59 |63 |60 | This is the replied message 61 |
62 |
67 |
--
68 |
69 |
74 |
70 |
73 | Nicolas Gaborit (Soreine)
71 | Web Developper
72 |