html-writer's Introduction

html-writer

A Racket library for converting X-expressions to strings of HTML with content-aware line wrapping and indentation.

Expected benefits of this library over xexpr->string, xexpr->html and display-xml/content:

Wraps and indents based on the tag type according to HTML5 standard (block/flow/inline) rather than based on scanning a tag's contents
Allows you to set the column width of the output
Never inserts a line break where doing so would introduce whitespace in the HTML output
Attribute values are never broken across lines
Does not modify content of <script> and <style> tags
Unicode-aware: wraps based on grapheme count vs. character count (depends on unicode-breaks)
Outputs boolean attributes the HTML5 way (<option selected> rather than <option selected="selected">)
Outputs self-closing tags the HTML5 way (<meta charset="UTF-8">, note > to close rather than />)

Still in early stages, comments and PRs welcome.

See the tests at the end of main.rkt to get an idea of what it can do so far. The tests that are there are passing at this point. Lots to do though:

Add many more tests
Add lookahead (or deferred output) for smarter wrapping (see this test)
Defer whitespace to avoid trailing spaces on lines (update tests first)
Compare classification of “block” and “flow” tags against the HTML5 standard
Thoroughly compare output against HTML Tidy
Add option to self-close tags the XHTML way (/>)
Split tests into their own files
Documentation

Questions to resolve

How important is the Unicode/grapheme thing and what is the performance cost of using in-words from unicode-breaks, as opposed to simply letting the wrapping be a little bit wrong when multi-byte graphemes are present?
How much logging/debugging instrumentation should be left in? (Probably none except for errors, but see next question)
What should happen when an X-expression's structure is not valid HTML, such as a <div> inside a <p>? Currently it just logs an error, but maybe it should throw an exception?
Support CDATA/PCDATA? (Probably not?) If not, then the things we are converting maybe aren’t technically X-expressions, what exactly are they and how should they be validated?

html-writer's People

Contributors

Stargazers

Watchers

html-writer's Issues

Whitespace between words is not always preserved

(display (xexpr->html5 '(p "What if there are linebreaks\nin the input?") #:wrap-at 20))

Produces:

<p>What if there 
are linebreaksin 
the input?</p>

but should produce

<p>What if there 
are linebreaks in 
the input?</p>

Input with spaces causes inconsistent indentation

When I have an xexpr? with some whitespace it causes the indentation to be wonky.

Input

(html
 ()
 "\n  "
 (head () "\n    " (title () "Minimal Post") "\n  ")
 "\n  "
 (body
  ()
  "\n    "
  (div
   ((class "article"))
   (article
    (h1 "Minimal Post")
    (p
     "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Maecenas nisi libero,\nscelerisque vel nulla eu, tristique porta metus.")))
  "\n  ")
 "\n")

Output

<html>
    <head>
        <title>Minimal Post</title>
  
  </head>
    <body>
        <div class="article">
      <article>
        <h1>Minimal Post</h1>
        <p>Lorem ipsum dolor sit amet, consectetur adipiscing elit. Maecenas 
        nisi libero,scelerisque vel nulla eu, tristique porta metus.</p>
      </article>
    </div>
  
  </body>

</html>

Recommend Projects