Add integration tests validating all regex patterns against real markdown content

.eslintignore

@@ -3,5 +3,4 @@ dist
 rollup.config.js
 babel.config.js
 dist/*
-package-lock.json
-yarn.lock
+*.lock

.npmignore

@@ -17,4 +17,4 @@ renovate.json
 examples
 CONTRIBUTING.md
 CHANGELOG.md
-yarn.lock
+*.lock

src/index.js

@@ -1,44 +1,24 @@
-import {
+export {
   REGEXP_HEADER,
+  REGEXP_H2,
+  REGEXP_H3,
   REGEXP_IMAGE,
   REGEXP_LINK,
   REGEXP_STRONG,
   REGEXP_DEL,
   REGEXP_Q,
   REGEXP_CODE,
-  // -----------
   REGEXP_BLOCKQUOTE,
   REGEXP_HR,
   REGEXP_PARAGRAPH,
-  // -----------
   REGEXP_BR,
   REGEXP_EMPTY_BLOCKQUOTE,
   REGEXP_EM,
-} from "./tags";
-import {
-  REGEXP_UL_LIST,
-  REGEXP_OL_LIST,
-  REGEXP_EMPTY_UL,
-  REGEXP_EMPTY_OL,
-} from "./list";
-
+} from "./tags/index";
 export {
-  REGEXP_HEADER,
-  REGEXP_IMAGE,
-  REGEXP_LINK,
-  REGEXP_STRONG,
-  REGEXP_DEL,
-  REGEXP_Q,
-  REGEXP_CODE,
   REGEXP_UL_LIST,
   REGEXP_OL_LIST,
-  REGEXP_BLOCKQUOTE,
-  REGEXP_HR,
-  REGEXP_PARAGRAPH,
   REGEXP_EMPTY_UL,
   REGEXP_EMPTY_OL,
-  REGEXP_BR,
-  REGEXP_EMPTY_BLOCKQUOTE,
-  REGEXP_EM,
-};
+} from "./lists/index";
 

src/lists/empty-ol.js

@@ -0,0 +1,3 @@
+const REGEXP_EMPTY_OL = /<\/ol>\s?<ol>/g;
+
+export { REGEXP_EMPTY_OL };

src/lists/empty-ul.js

@@ -0,0 +1,3 @@
+const REGEXP_EMPTY_UL = /<\/ul>\s?<ul>/g;
+
+export { REGEXP_EMPTY_UL };

src/lists/index.js

@@ -0,0 +1,4 @@
+export { REGEXP_UL_LIST } from "./ul";
+export { REGEXP_OL_LIST } from "./ol";
+export { REGEXP_EMPTY_UL } from "./empty-ul";
+export { REGEXP_EMPTY_OL } from "./empty-ol";

src/lists/ol.js

@@ -0,0 +1,9 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_OL_LIST = new RegExp(`${newLine}[0-9]+\\.(.*)`, "g");
+
+export { REGEXP_OL_LIST };

src/lists/ul.js

@@ -0,0 +1,12 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_UL_LIST = new RegExp(
+  `${newLine}(((\\s{4})?\\*(.*?)${newLine}){1,})`,
+  "g"
+);
+
+export { REGEXP_UL_LIST };

src/tags/blockquote.js

@@ -0,0 +1,9 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_BLOCKQUOTE = new RegExp(`${newLine}(>|\\>)(.*)`, "g");
+
+export { REGEXP_BLOCKQUOTE };

src/tags/br.js

@@ -0,0 +1,9 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_BR = new RegExp(`((${newLine}){2,})`, "g");
+
+export { REGEXP_BR };

src/tags/code.js

@@ -0,0 +1,3 @@
+const REGEXP_CODE = /`(.*?)`/g;
+
+export { REGEXP_CODE };

src/tags/del.js

@@ -0,0 +1,3 @@
+const REGEXP_DEL = /\~\~(.*?)\~\~/g;
+
+export { REGEXP_DEL };

src/tags/em.js

@@ -0,0 +1,3 @@
+const REGEXP_EM = /(\s|>)(\*|_)(.*?)\2(\s|<)/g;
+
+export { REGEXP_EM };

src/tags/empty-blockquote.js

@@ -0,0 +1,3 @@
+const REGEXP_EMPTY_BLOCKQUOTE = /<\/blockquote><blockquote>/g;
+
+export { REGEXP_EMPTY_BLOCKQUOTE };

src/tags/header.js

@@ -0,0 +1,12 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_HEADER = new RegExp(`${newLine}(#+)(.*)`, "g");
+
+const REGEXP_H2 = /^## (.*$)/gim;
+const REGEXP_H3 = /^### (.*$)/gim;
+
+export { REGEXP_HEADER, REGEXP_H2, REGEXP_H3 };

src/tags/hr.js

@@ -0,0 +1,9 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_HR = new RegExp(`${newLine}-{5,}`, "g");
+
+export { REGEXP_HR };

src/tags/image.js

@@ -0,0 +1,3 @@
+const REGEXP_IMAGE = /!\[([^\[]+)\]\(([^\)]+)\)/g;
+
+export { REGEXP_IMAGE };

src/tags/index.js

@@ -0,0 +1,13 @@
+export { REGEXP_HEADER, REGEXP_H2, REGEXP_H3 } from "./header";
+export { REGEXP_BLOCKQUOTE } from "./blockquote";
+export { REGEXP_CODE } from "./code";
+export { REGEXP_LINK } from "./link";
+export { REGEXP_IMAGE } from "./image";
+export { REGEXP_STRONG } from "./strong";
+export { REGEXP_DEL } from "./del";
+export { REGEXP_EM } from "./em";
+export { REGEXP_HR } from "./hr";
+export { REGEXP_BR } from "./br";
+export { REGEXP_PARAGRAPH } from "./paragraph";
+export { REGEXP_Q } from "./quote";
+export { REGEXP_EMPTY_BLOCKQUOTE } from "./empty-blockquote";

src/tags/link.js

@@ -0,0 +1,3 @@
+const REGEXP_LINK = /\[([^\[]+)\]\(([^\)]+)\)/g;
+
+export { REGEXP_LINK };

src/tags/paragraph.js

@@ -0,0 +1,9 @@
+import os from "os";
+
+const platform = os.platform();
+
+const newLine = platform === "win32" ? "\r\n" : "\n";
+
+const REGEXP_PARAGRAPH = new RegExp(`${newLine}(.+?)${newLine}`, "g");
+
+export { REGEXP_PARAGRAPH };

src/tags/quote.js

@@ -0,0 +1,3 @@
+const REGEXP_Q = /\:\"(.*?)\"\:/g;
+
+export { REGEXP_Q };

src/tags/strong.js

@@ -0,0 +1,3 @@
+const REGEXP_STRONG = /(\*\*|__)(.*?)(\*?)\1/g;
+
+export { REGEXP_STRONG };

tests/fixtures/edge-cases.js

@@ -0,0 +1,14 @@
+'use strict';
+
+const EDGE_CASES = {
+  empty: '',
+  whitespaceOnly: '   \n   \t   ',
+  nestedMarkdown: '**bold with _italic_ inside**',
+  multiplePatterns: '\n# Header\n\n> blockquote\n\n**bold** and _italic_\n',
+  specialChars: '!@#$%^&*()[]{}',
+  unicode: '# Title with émojis 🚀',
+  longText: 'a'.repeat(1000),
+  mixedNewlines: '\n\n\nTriple newline',
+};
+
+module.exports = { EDGE_CASES };

tests/fixtures/header-samples.js

@@ -0,0 +1,42 @@
+'use strict';
+
+const HEADER_SAMPLES = {
+  valid: [
+    '\n# Header 1',
+    '\n## Header 2',
+    '\n### Header 3',
+    '\n#### Header 4',
+    '\n# Combatting Increased Cybersecurity Gaps During The Pandemic',
+  ],
+  invalid: [
+    'No header here',
+    'Just plain text',
+    '',
+  ],
+};
+
+const H2_SAMPLES = {
+  valid: [
+    '## Header 2',
+    '## Another Section',
+  ],
+  invalid: [
+    '# Header 1',
+    '### Header 3',
+    'plain text',
+  ],
+};
+
+const H3_SAMPLES = {
+  valid: [
+    '### Header 3',
+    '### Deep Section',
+  ],
+  invalid: [
+    '## Header 2',
+    '# Header 1',
+    'plain text',
+  ],
+};
+
+module.exports = { HEADER_SAMPLES, H2_SAMPLES, H3_SAMPLES };

tests/fixtures/list-samples.js

@@ -0,0 +1,52 @@
+'use strict';
+
+const LIST_SAMPLES = {
+  ul: {
+    valid: [
+      '\n* Item 1\n* Item 2\n',
+      '\n* First item\n',
+      '\n    * Nested item\n',
+    ],
+    invalid: [
+      'Item 1\nItem 2',
+      '1. Ordered item',
+      'plain text',
+    ],
+  },
+  ol: {
+    valid: [
+      '\n1. First item',
+      '\n2. Second item',
+      '\n10. Tenth item',
+    ],
+    invalid: [
+      '* Unordered item',
+      'plain text',
+      '',
+    ],
+  },
+  emptyUl: {
+    valid: [
+      '</ul><ul>',
+      '</ul> <ul>',
+    ],
+    invalid: [
+      '</ol><ol>',
+      'plain text',
+      '* item',
+    ],
+  },
+  emptyOl: {
+    valid: [
+      '</ol><ol>',
+      '</ol> <ol>',
+    ],
+    invalid: [
+      '</ul><ul>',
+      'plain text',
+      '1. item',
+    ],
+  },
+};
+
+module.exports = { LIST_SAMPLES };

tests/fixtures/markdown-samples.js

@@ -0,0 +1,44 @@
+'use strict';
+
+const MARKDOWN_SAMPLES = {
+  header: '\n# Title\n\nSome paragraph text here.\n\n## Section\n\nMore text.\n',
+  image: '![alt text](https://example.com/image.png)',
+  link: 'Check out [this link](https://example.com) for more.',
+  strong: 'This is **bold text** in a sentence.',
+  del: 'This is ~~strikethrough~~ text.',
+  code: 'Use `inline code` like this.',
+  blockquote: '\n> This is a blockquote.\n',
+  hr: '\n-----\n',
+  paragraph: '\nThis is a paragraph.\n',
+  br: '\nFirst paragraph.\n\nSecond paragraph.\n',
+  em: ' _italic text_ here',
+  quote: ':"This is a quote":',
+  emptyBlockquote: '</blockquote><blockquote>',
+  ulList: '\n* Item one\n* Item two\n',
+  olList: '\n1. First item\n2. Second item\n',
+};
+
+const FULL_MARKDOWN = `
+# Main Title
+
+This is a paragraph with **bold**, _italic_, and \`code\`.
+
+> This is a blockquote
+
+## Section
+
+Here is a [link](https://example.com) and an ![image](https://example.com/img.png).
+
+* Unordered item 1
+* Unordered item 2
+
+1. Ordered item 1
+2. Ordered item 2
+
+-----
+
+~~strikethrough~~ and :"quoted":
+
+`;
+
+module.exports = { MARKDOWN_SAMPLES, FULL_MARKDOWN };

tests/helpers/file-loader.js

@@ -0,0 +1,10 @@
+'use strict';
+
+const fs = require('fs');
+const path = require('path');
+
+function loadMarkdownFile(filePath) {
+  return fs.readFileSync(path.resolve(filePath), 'utf-8');
+}
+
+module.exports = { loadMarkdownFile };

tests/image/image.test.js

@@ -6,7 +6,7 @@ const { REGEXP_IMAGE } = require('../../src/index');
 describe('RegEx: REGEXP_IMAGE', () => {
 
     test('should match the expected ...', () => {
-      let result = REGEXP_IMAGE.test("[alt_text](https://github.com/atherdon/newsletters/blob/master/archive/img/memes/october/4/4.jpg "image_tooltip")");
+      let result = REGEXP_IMAGE.test('![alt_text](https://github.com/atherdon/newsletters/blob/master/archive/img/memes/october/4/4.jpg)');
       expect(result).toBe(true);
     });