Search code examples
javascriptregexstringmatchmarkdown

Regex - extract all headers from markdown string


I am using gray-matter in order to parse .MD files from the file system into a string. The result the parser produces is a string like this:

\n# Clean-er ReactJS Code - Conditional Rendering\n\n## TL;DR\n\nMove render conditions into appropriately named variables. Abstract the condition logic into a function. This makes the render function code a lot easier to understand, refactor, reuse, test, and think about.\n\n## Introduction\n\nConditional rendering is when a logical operator determines what will be rendered. The following code is from the examples in the official ReactJS documentation. It is one of the simplest examples of conditional rendering that I can think of.\n\n

I am now trying to write a regular expression that would extract all the heading text from the string. Headers in markdown start with a # (there can be from 1-6), and in my case always end with a new line.

I've tried using the following regular expression but calling it on my test string returns nothing:

const testString = "\n# Clean-er ReactJS Code - Conditional Rendering\n\n## TL;DR\n\nMove render conditions into appropriately named variables. Abstract the condition logic into a function. This makes the render function code a lot easier to understand, refactor, reuse, test, and think about.\n\n## Introduction\n\nConditional rendering is when a logical operator determines what will be rendered. The following code is from the examples in the official ReactJS documentation. It is one of the simplest examples of conditional rendering that I can think of.\n\n"
const HEADING_R = /(?<!#)#{1,6} (.*?)(\\r(?:\\n)?|\\n)/gm;
const headings = HEADING_R.exec(content);

console.log('headings: ', headings);

This console logs headings as null (no matches found). The result that I am looking for would be: ["# Clean-er ReactJS Code - Conditional Rendering", "## TL;DR", "## Introduction"].

I believe the regular expression is wrong, but have no idea why.


Solution

  • /#{1,6}.+(?=\n)/g

    • #{1,6} ... matches the character # at least once or as sequence of maximum 6 equal characters.

    • .+ matches any character (except for line terminators) at least once and as many times as possible (greedy)

    • does so until the positive lookahead (?=\n) matches ...

      • which is ... \n ... a newline / line-feed.
    • uses the global modifier which does match everything.

    Edit

    Having mentioned

    "matches any character (except for line terminators)"

    thus a regex like ... /#{1,6}.+/g ... should already do the job (no need for a positive lookahead) for the OP's use case which is ...

    "Headers in markdown start with a # (there can be from 1-6), and in my case always end with a new line."

    The result that I am looking for would be: ["# Clean-er ReactJS Code - Conditional Rendering", "## TL;DR", "## Introduction"].

    const testString = `\n# Clean-er ReactJS Code - Conditional Rendering\n\n## TL;DR\n\nMove render conditions into appropriately named variables. Abstract the condition logic into a function. This makes the render function code a lot easier to understand, refactor, reuse, test, and think about.\n\n## Introduction\n\nConditional rendering is when a logical operator determines what will be rendered. The following code is from the examples in the official ReactJS documentation. It is one of the simplest examples of conditional rendering that I can think of.\n\n`;
    
    // see...[https://regex101.com/r/n6XQub/2]
    const regXHeader = /#{1,6}.+/g
    
    console.log(
      testString.match(regXHeader)
    );
    .as-console-wrapper { min-height: 100%!important; top: 0; }

    Bonus

    Refactoring the above regex into e.g. /(?<flag>#{1,6})\s+(?<content>.+)/g by utilizing named capturing groups alongside with matchAll and a mapping task, one could achieve a result like computed by the next provided example code ...

    const testString = `\n# Clean-er ReactJS Code - Conditional Rendering\n\n## TL;DR\n\nMove render conditions into appropriately named variables. Abstract the condition logic into a function. This makes the render function code a lot easier to understand, refactor, reuse, test, and think about.\n\n## Introduction\n\nConditional rendering is when a logical operator determines what will be rendered. The following code is from the examples in the official ReactJS documentation. It is one of the simplest examples of conditional rendering that I can think of.\n\n`;
    
    // see...[https://regex101.com/r/n6XQub/4]
    const regXHeader = /(?<flag>#{1,6})\s+(?<content>.+)/g
    
    console.log(
      Array
        .from(
          testString.matchAll(regXHeader)
        )
        .map(({ groups: { flag, content } }) => ({
          heading: `h${ flag.length }`,
          content,
        }))
    );
    .as-console-wrapper { min-height: 100%!important; top: 0; }