Dumbo is a webserver that provides minimal REST API to convert various text formats to HTML or LaTeX with Pandoc.
Features:
- Convert any Pandoc-supported markup format to HTML or LaTeX
- Convert embedded LaTeX and AsciiMath to SVG using DVISVGM
You can compile and run the server locally with stack build.
Available options:
Dumboest markdown converter
Usage: Dumbo.EXE [--latex STRING] [--dvisvgm STRING] --cacheDir STRING
--tmpDir STRING --port INT
Available options:
-h,--help Show this help text
--latex STRING Absolute path to latex binary
--dvisvgm STRING Absolute path to dvisvgm binary
--cacheDir STRING Directory for cached math
--tmpDir STRING Directory to store temporary files
--port INT Service port number
Note: The tool requires TeX and DVISVGM installed for all features to be available.
A prebuilt Docker container with all necessary dependencies is available via timimages/dumbo container.
Usage example:
docker run --rm -p 5000:5000 --tmpfs /docker_cache --tmpfs /dumbo_tmp timimages/dumbo:latestThe command above will start the server on localhost:5000
The main input can be defined as the following TypeScript schema:
type InputElement<T> = {
// Main content
content: T,
// Format of the input content as a valid Pandoc input type
// If not specified, assumes input is Markdown
inputFormat: string | undefined,
// How to process LaTex math.
// mathjax (default) -> Let Pandoc process math and output MathJax-compatible output
// svg -> Render math using proper LaTeX and DVISVGM
// undefined -> defaults to mathjax
mathOption: "svg" | "mathjax" | undefined,
// If mathOption is set to "svg", this allows to specify a custom LaTeX preamble
mathPreamble: string | undefined,
// If true, applies Pandoc's smart punctuation extension (Ext_smart)
smartPunct: bool | undefined,
};Input: JSON-encoded text paragraphs as InputElement<string[]>
Output: JSON-encoded resulting HTML snippets as string[]
Description: Converts input text (assumes Markdown by default) to HTML.
Examples:
@ curl --data '{"content":["# Hello, world!"]}' localhost:5000/markdown
["<h1 id=\"hello-world\">Hello, world!</h1>"]@ curl --data '{"content":["Math: $1+1$"]}' localhost:5000/markdown
["<p>Math: <span class=\"math inline\">\\(1+1\\)</span></p>"]
@ curl --data '{"content":["Math: $1+1$"], "mathOption": "svg"}' localhost:5000/markdown
["<p>Math: <span class=\"mathp inline\"><img style=\"width:2.77670em; vertical-align:-0.15963em\" src=\"data:image/svg+xml;...\" title=\"1+1\"></span></p>"]Input: JSON-encoded text paragraphs as InputElement<string[]>
Output: JSON-encoded resulting LaTeX snippets as string
Description: Converts input text (assumes Markdown by default) to LaTeX.
Examples:
@ curl --data '{"content":["# Hello, world!"]}' localhost:5000/latex
["\\hypertarget{hello-world}{%\n\\section{Hello, world!}\\label{hello-world}}"]@ curl --data '{"content":["Math: $1+1$"]}' localhost:5000/latex
["Math: \\(1+1\\)"]
# Note that mathOption: svg does not have any effect on LaTeX rendering
@ curl --data '{"content":["Math: $1+1$"], "mathOption": "svg"}' localhost:5000/latex
["Math: \\(1+1\\)"]Input: JSON-encoded data as InputElement<T> where T is any JSON-encodable object or InputElement<T> itself.
Output: T with specific fields converted to HTML
Description: Walks an aribtrary JSON-encoded object and converts any keys/values to HTML if they contain one of the following prefixes (\n is newline):
md:- Parses the key/value as Markdownam:- Parse sthe key/value as AsciiMath
Note: Note that T can be InputElement<T> itself. In that case, any options specified in it will be applied to the any child values. You can use this approach to override any rendering option for each possible inner object.
Examples:
Processes an object {"a": "md:# Hello, world", "b": ["*foo*", "md:**bar**"]} and converts all values prefixed with md: to HTML:
@ curl --data '{"content":[{"a": "md:# Hello, world", "b": ["*foo*", "md:**bar**"]}]}' localhost:5000/mdkeys
[{"a":"<h1 id=\"hello-world\">Hello, world</h1>","b":["*foo*","<strong>bar</strong>"]}]Recursively specify the content with different input options:
@ curl --data '{"content": [{"content": "md:$1+1$"}, {"content": "md:$1+1$", "mathOption": "svg"}]}' localhost:5000/mdk
eys
["<span class=\"math inline\">\\(1+1\\)</span>","<span class=\"mathp inline\"><img style=\"width:2.77670em; vertical-align:-0.15963em\" src=\"data:image/svg+xml;...\" title=\"1+1\"></span>"]Input: JSON-encoded data as InputElement<T> where T is any JSON-encodable object or InputElement<T> itself.
Output: T with specific fields converted to LaTeX
Description: Walks an aribtrary JSON-encoded object and converts any keys/values to HTML if they contain one of the following prefixes (\n is newline):
md:- Parses the key/value as Markdownam:- Parse sthe key/value as AsciiMath
Note: Note that T can be InputElement<T> itself. In that case, any options specified in it will be applied to the any child values. You can use this approach to override any rendering option for each possible inner object.
Examples:
Processes an object {"a": "md:# Hello, world", "b": ["*foo*", "md:**bar**"]} and converts all values prefixed with md: to HTML:
@ curl --data '{"content":[{"a": "md:# Hello, world", "b": ["*foo*", "md:**bar**"]}]}' localhost:5000/latexkeys
[{"a":"\\hypertarget{hello-world}{%\n\\section{Hello, world}\\label{hello-world}}","b":["*foo*","\\textbf{bar}"]}]Dumbo was created and primarily developed by Ville Tirronen (@aleator) with some minor tweaks added by Mika Lehtinen (@Smibu).
Dumbo is licensed under GPLv3.