Markdown

Crate carries two text modules that work together: crate.text.markdown turns a block-structured JSON document into Markdown source, and crate.text.markdownVibe turns Markdown source into HTML.

crate.text.markdown

Editor-style rich-text fields are usually stored as a list of typed blocks rather than as prose. toMarkdown flattens such a document:

import crate.text.markdown;

auto article = `{
  "blocks": [
    { "type": "header",    "data": { "text": "Release notes" } },
    { "type": "paragraph", "data": { "text": "Bulk operations landed." } }
  ]
}`.parseJsonString;

auto source = article.toMarkdown;

Block type	Rendered as
header	## <text>
paragraph	<text>

Every block is followed by a blank line, and the result is stripped. A document with no blocks array, or one whose blocks are all of unrecognized types, produces an empty string — unknown types are skipped rather than rejected, so a newer editor’s document degrades instead of failing.

toHeader and toParagraph are exposed if you need to render a single block.

crate.text.markdownVibe

A Markdown-to-HTML filter, extended past vanilla Markdown with tables, attributes, figures, and a document outline.

import crate.text.markdownVibe;

auto html = filterMarkdown(source, MarkdownFlags.backtickCodeBlocks | MarkdownFlags.tables);

filterMarkdown has four forms: return a string or write into an output range dst, each taking either a MarkdownFlags value or a MarkdownSettings object.

MarkdownSettings

Field	Default	Purpose
flags	vanillaMarkdown	Which extensions the parser recognizes
headingBaseLevel	1	Level the first heading renders at, so embedded content can sit under a page title
urlFilter	null	string delegate(string url, bool isImage) — rewrite every link and image target
allowedURISchemas	["http", "https", "ftp", "mailto"]	Schemas permitted in link and image targets

The schema allow-list is a security control, not a convenience: a javascript: URI in a link is blocked and escaped rather than emitted.

MarkdownFlags

Flag	Effect
none / vanillaMarkdown	Standard Markdown only
keepLineBreaks	Single newlines become <br>; suits plain-text sources such as email
backtickCodeBlocks	Fenced code blocks
noInlineHtml	Embedded HTML is escaped instead of passed through
tables	Markdown Extra / GitHub table syntax
attributes	{ … } after a link or image adds HTML attributes to the element
figures	- %%% list syntax produces <figure>, - ### produces <figcaption>
forumDefault	keepLineBreaks, backtickCodeBlocks, noInlineHtml and tables combined

Flags are a bit set — combine them with |.

Document Outline

getMarkdownOutline returns the heading tree without rendering the document, which is what a table-of-contents component needs:

Section[] outline = getMarkdownOutline(source);

struct Section {
  size_t headingLevel;
  string caption;
  string anchor;
  Section[] subSections;
}

anchor is the slug generated for the heading — lowercased, with dashes for spaces — and matches the id filterMarkdown puts on the rendered heading, so outline entries link to the right place.

Limits

The parser is not fully CommonMark compliant. vanillaMarkdown is the intent — standard Markdown — rather than a conformance claim.