ChatGPT解决这个技术问题 Extra ChatGPT

Is there a command line utility for rendering GitHub flavored Markdown?

I'm wondering if there is a command line utility for taking a GitHub flavored Markdown file and rendering it to HTML.

I'm using a GitHub wiki to create website content. I've cloned the repository on my server and would then like to process it into regular HTML. It's important to me that what appears on GitHub is exactly how it should look for my website. I'd also really like to use the fenced blocks with ~~~, so I'd rather not use standard Markdown syntax only.

I've looked a bit into the JavaScript live preview thinking I could hook it into Node.js, but they say it is deprecated. I've looked at the redcarpet repository, but it doesn't look like it has a command line interface.

I rolled my own solution, however, since no solution here is clearly better than the others, I'll leave the question without a selected answer.

RE: UPDATE: If I added an --out argument to grip to render to an HTML file instead of the browser, would that be acceptable?
@Joe please do add that option!
@McLeopold @bguiz just deployed the --export option, which renders GFM and its styles to a single file. Does this answer the question?
@McLeopold, Joe kind of knocked this out of the park by creating a simple reusable solution, might want to give him the answer.
Github itself uses Sundown. See stackoverflow.com/a/7694931/632951

P
Peter Mortensen

I wrote a small CLI in Python and added GFM support. It's called Grip (Github Readme Instant Preview).

Install it with:

$ pip install grip

And to use it, simply:

$ grip

Then visit localhost:5000 to view the readme.md file at that location.

You can also specify your own file:

$ grip CHANGES.md

And change port:

$ grip 8080

And of course, specifically render GitHub-Flavored Markdown, optionally with repository context:

$ grip --gfm --context=username/repo issue.md

Notable features:

Renders pages to appear exactly like on GitHub

Fenced blocks

Python API

Navigate between linked files (thanks, vladwing!) added in 2.0

Export to a single file (thanks, iliggio!) added in 2.0

New: Read from stdin and export to stdout added in 3.0

Hope this helps someone here. Check it out.


Works really well and you can't beat ease of install for Pythonistas!
This should be a first hit for "github markdown preview." Everything else is complicated, doesn't work, or doesn't do all the GitHub features. grip works right out of the box.
@Houdini That issue is out of date. Flask does support 3.3, see flask.pocoo.org/docs/python3. Here's a more recent Github thread on the topic github.com/mitsuhiko/flask/issues/587. If there's another dependency that needs updated, feel free to open an issue or a pull request.
It should be noted that this package requires an active internet connection and your github authentication credentials (provided at command line) if you do more than 60 refreshes per hour.
As mentioned earlier, I don't think this is a particularly great solution because all it does it goes off to Github and gets Github to render your Markdown. It requires a working internet connection with access to Github, and if Github dies then this tool stops working. I'd rather have a completely offline solution.
D
Dale Wijnand

I've not found a quick and easy method for GitHub-flavoured Markdown, but I have found a slightly more generic version - Pandoc. It converts from/to a number of formats, including Markdown, Rest, HTML and others.

I've also developed a Makefile to convert all .md files to .html (in large part to the example at Writing, Markdown and Pandoc):

# 'Makefile'
MARKDOWN = pandoc --from gfm --to html --standalone
all: $(patsubst %.md,%.html,$(wildcard *.md)) Makefile

clean:
    rm -f $(patsubst %.md,%.html,$(wildcard *.md))
    rm -f *.bak *~

%.html: %.md
    $(MARKDOWN) $< --output $@

i've been using "watch pandoc ..." to continuously convert a markdown file to html, and the chrome "live reload" extension to get real time "stay where i'm scrolled too" functionality with this, and it works great. chrome.google.com/webstore/detail/livereload/…
Pandoc reads GFM fine but it doesn't generate the same HTML GitHub does -- for instance, if you have a multi-line <pre/> tag in your GFM source, Pandoc will put <br/> tags in for the line breaks in it, while GitHub's renderer, though it strips leading whitespace, seems to otherwise leave the content alone.
How does one go about getting nice styling on the resulting HTML? My output is still rendered with Times New Roman, for example.
Pandoc install instructions are here. On macOS: brew install pandoc
neither the gfm nor the markdown_github input formats correctly render things like code blocks.
G
Gringo Suave
pip3 install --user markdown
python3 -m markdown readme.md > readme.html

It doesn't handle GitHub extensions, but it is better than nothing. I believe you can extend the module to handle the GitHub additions.


/usr/bin/python: markdown is a package and cannot be directly executed
Simple HTML output with no fancy tags.
Strictly speaking, this is "John Gruber's Markdown", not GitHub flavored Markdown.
J
James Lim

Maybe this might help:

gem install github-markdown

No documentation exists, but I got it from the gollum documentation. Looking at rubydoc.info, it looks like you can use:

require 'github/markdown'  
puts GitHub::Markdown.render_gfm('your markdown string')

in your Ruby code. You can wrap that easily in a script to turn it into a command line utility:

#!/usr/bin/env ruby

# render.rb
require 'github/markdown'

puts GitHub::Markdown.render_gfm File.read(ARGV[0])

Execute it with ./render.rb path/to/my/markdown/file.md. Note that this is not safe for use in production without sanitization.


Thanks Jim, due to virtually non-existent examples, I was stuck at the require step (replacing dash with slash made it).. ;)
This is great if you're already using NPM. I had to use it, on account of DOxygen causing me constant problems with specifically github-flavored markdown + exporting to HTML.
is this running locally or is it sending data out to GitHub API?
t
toraritte

To read a README.md file in the terminal I use:

pandoc README.md | lynx -stdin

Pandoc outputs it in HTML format, which Lynx renders in your terminal.

It works great: It fills my terminal, shortcuts are shown below, I can scroll through, and the links work! There is only one font size though, but the colors + indentation + alignment make up for that.

Installation:

apt: sudo apt-get install pandoc lynx

nix: nix-shell -p pandoc lynx


The question is specifically about command-line usage. Before writing your own ruby script (or egad node server), give this a shot.
Exactly this works inside your terminal. Or if your favorite (desktop?) browser can access that folder use pandoc readme.md -o readme.md.html and open the resulting file.
@baerry-staes Yes, sorry, I hope it was clear that yours was my favored answer.
@JustinMiddleton yes i got that, thank you. My comment was just to add some extra info for desktop users.. i figured someone someday reading this might find it useful.
I've tried about 5-6 other console md readers and this has by far been the best solution. I just added the most basic function to my config to make it a little quicker to use. function md { pandoc $@ | lynx -stdin }
y
youurayy

Probably not what you want, but since you mentioned Node.js: I could not find a good tool to preview GitHub Flavored Markdown documentation on my local drive before committing them to GitHub, so today I created one, based on Node.js: https://github.com/ypocat/gfms

So perhaps you can reuse the showdown.js from it for your Wiki, if your question is still actual. If not, maybe other people facing the same problem as I did will find (just as I did) this question and this answer to it.


Well done, man. Saves me from having to resort to Ruby or Python when I'm writing a node application, which is great.
P
Peter Mortensen

GitHub has a Markdown API you can use.


jq --slurp --raw-input '{"text": "\(.)", "mode": "markdown"}' < README.md | curl --data @- https://api.github.com/markdown > README.html
@VebjornLjosa * that * or grip... You chose. :P
grip is amazing. But, unfortunately, due to its name it is not possible to find it easily if you forget how it is named. (not available via MacPorts either).
P
Peter Mortensen

Use marked. It supports GitHub Flavored Markdown, can be used as a Node.js module and from the command line.

An example would be:

$ marked -o hello.html
hello world
^D
$ cat hello.html
<p>hello world</p>

I've noticed that this doesn't support features like syntax highlighting for code blocks and newer features like checklists. But hey it gets most of the way!
Very nice, the only thing I'm missing is some borders for the tables. Well, at least I can render them at all, this is pretty much exactly what I need. Pipe in the GFM, pipe out HTML :)
I'm confused about this example, what is ^D?
M
Mateusz Piotrowski

I created a tool similar to Atom's Preview functionality, but as a standalone application. Not sure if this is what you're looking for, but it might be helpful. -- https://github.com/yoshuawuyts/vmd

https://i.stack.imgur.com/eciJE.png


You guys still supporting this? I tried to install with NPM today, but no dice. >downloading electron-v0.36.9-win32-x64.zip >Error: self signed certificate
Yeah, we are! What version of npm / node did you run this on? - feel free to open up an issue on GH and we'll take a look at this. Thanks!
P
Peter Mortensen

This is mostly a follow-on to @barry-staes's answer for using Pandoc. Homebrew has it as well, if you're on a Mac:

brew install pandoc

Pandoc supports GFM as an input format via the markdown_github name.

Output to file

cat foo.md | pandoc -f markdown_github > foo.html

Open in Lynx

cat foo.md | pandoc -f markdown_github | lynx -stdin # To open in Lynx

Open in the default browser on OS X

cat foo.md | pandoc -f markdown_github > foo.html && open foo.html # To open in the default browser on OS X`

TextMate Integration

You can always pipe the current selection or current document to one of the above, as most editors allow you to do. You can also easily configure the environment so that pandoc replaces the default Markdown processor used by the Markdown bundle.

First, create a shell script with the following contents (I'll call it ghmarkdown):

#!/bin/bash
# Note included, optional --email-obfuscation arg
pandoc -f markdown_github --email-obfuscation=references

You can then set the TM_MARKDOWN variable (in Preferences→Variables) to /path/to/ghmarkdown, and it will replace the default Markdown processor.


M
Martin Eden

pandoc with browser works well for me.

Usage: cat README.md | pandoc -f markdown_github | browser

Installation (Assuming you are using Mac OSX):

$ brew install pandoc

$ brew install browser

Or on Debian/Ubuntu: apt-get install pandoc browser


apt-get isntall pandoc will do, no need to use insecure, local stuff like brew.
@DominikGeorge there is a typo, it is install, not isntall
@DominikGeorge there's no apt-get on macOS.
n
notpeter

Building on this comment I wrote a one-liner to hit the Github Markdown API using curl and jq.

Paste this bash function onto the command line or into your ~/.bash_profile:

mdsee(){ 
    HTMLFILE="$(mktemp -u).html"
    cat "$1" | \
      jq --slurp --raw-input '{"text": "\(.)", "mode": "markdown"}' | \
      curl -s --data @- https://api.github.com/markdown > "$HTMLFILE"
    echo $HTMLFILE
    open "$HTMLFILE"
}

And then to see the rendered HTML in-browser run:

mdsee readme.md

Replace open "$HTMLFILE" with lynx "$HTMLFILE" if you need a pure terminal solution.


A
Asme Just

I use Pandoc with the option --from=gfm for GitHub Flavored Markdown like this:

$ pandoc my_file.md   --from=gfm -t html -o my_file.html

With pandoc 1.16.0.2 on linux mint 18 I get an error: pandoc: Unknown reader: gfm. Going to 2.2.1 fixes this.
I tested it with Pandoc 2.1.2.
how to install 2.2.1 on ubuntu?
@AlexanderMills Did you try sudo apt install pandoc ?
C
Community

Also see https://softwareengineering.stackexchange.com/a/128721/24257.

If you're interested in how we [Github] render Markdown files, you might want to check out Redcarpet, our Ruby interface to the Sundown library.

Ruby-script, which use Redcarpet, will be "command line utility", if you'll have local Ruby


Haven't ideas - I don't write Ruby and I didn't read Redcarpet sources
@LazyBadger, Sundown is the actual parser (written in C). Redcarpet is not needed.
This is arguably the best solution, but you don't give actual instructions on what to do. So after installing the gem gem install redcarpet, suppose we're in a directory containing README.md what next?
P
Peter Mortensen

My final solution was to use Python Markdown. I rolled my own extension that fixed the fence blocks.


P
Peter Mortensen

There is a really nice and simple tool for browsing GFM Markdown documents:

GFMS - Github Flavored Markdown Server

It's simple and lightweight (no configuration needed) HTTP server you can start in any directory containing markdown files to browse them.

Features:

Full GFM Markdown support

Source code syntax highlighting

Browsing files and directories

Nice looking output (and configurable CSS stylesheets)

Export to PDF


P
Peter Mortensen

GitHub has (since) developed a nice modular text editor called Atom (based on Chromium and uses Node.js modules for packages).

A default preinstalled package Markdown Preview lets you display your preview in a separate tab using Ctrl + Shift + M.

I haven't tested its full syntax, but since it's coming from GitHub, I'd be highly surprised if the preview's syntax was different from theirs (fenced blocks using ~~~ work).

Now, while it's not technically command-line based, it uses Node.js and outputs to a DOM-based renderer, which might help anyone trying to render GitHub syntax-based HTML on a Node.js-based webserver, or just edit her/his README.md offline.


In Atom you can install a package called gfm-pdf (atom.io/packages/gfm-pdf), which exports your markdown document to a HTML and/or PDF document. The library wkhtmltopdf is required.
T
Tivie

Late addition but showdownjs has a CLI tool you can use to parse MD to HTML.


e
eyllanesc

I managed to use a one-line Ruby script for that purpose (although it had to go in a separate file). First, run these commands once on each client machine you'll be pushing docs from:

gem install github-markup
gem install commonmarker

Next, install this script in your client image, and call it render-readme-for-javadoc.rb:

require 'github/markup'

puts GitHub::Markup.render_s(GitHub::Markups::MARKUP_MARKDOWN, File.read('README.md'))

Finally, invoke it like this:

ruby ./render-readme-for-javadoc.rb >> project/src/main/javadoc/overview.html

ETA: This won't help you with StackOverflow-flavor Markdown, which seems to be failing on this answer.


I think this is the "closest to source" answer from all of them since these tools are the ones used by github.
S
Sandeep

Improving upon @barry-stae's solution. Stick this snippet in ~/.bashrc

function mdviewer(){
  pandoc $* | lynx -stdin
}

Then we can quickly view the file from the command-line. Also works nicely over SSH/Telnet sessions.

mdviewer README.md

P
Peter Mortensen

I found a website that will do this for you: http://tmpvar.com/markdown.html. Paste in your Markdown, and it'll display it for you. It seems to work just fine!

However, it doesn't seem to handle the syntax highlighting option for code; that is, the ~~~ruby feature doesn't work. It just prints 'ruby'.


tmpvar doesn't seem to do GFM version enhancements like tables
P
Peter Mortensen

A 'quick-and-dirty' approach is to download the wiki HTML pages using the wget utility, instead of cloning it. For example, this is how I downloaded the Hystrix wiki from GitHub (I'm using Ubuntu Linux):

 $ wget -e robots=off -nH -E -H -k -K -p https://github.com/Netflix/Hystrix/wiki
 $ wget -e robots=off -nH -E -H -k -K -I "Netflix/Hystrix/wiki" -r -l 1 https://github.com/Netflix/Hystrix/wiki

The first call will download the wiki entry page and all its dependencies. The second one will call all sub-pages on it. You can browse now the wiki by opening Netflix/Hystrix/wiki.1.html.

Note that both calls to wget are necessary. If you just run the second one then you will miss some dependencies required to show the pages properly.


d
daBertl

Improving upon @barry-stae and @Sandeep answers for regular users of elinks you would add the following to .bashrc:

function mdviewer() {
  pandoc $* | elinks --force-html
}

Don't forget to install pandoc (and elinks).


C
Community

Based on Jim Lim's answer, I installed the GitHub Markdown gem. That included a script called gfm that takes a filename on the command line and writes the equivalent HTML to standard output. I modified that slightly to save the file to disk and then to open the standard browser with launchy:

#!/usr/bin/env ruby

HELP = <<-help
  Usage: gfm [--readme | --plaintext] [<file>]
  Convert a GitHub-Flavored Markdown file to HTML and write to standard output.
  With no <file> or when <file> is '-', read Markdown source text from standard input.
  With `--readme`, the files are parsed like README.md files in GitHub.com. By default,
  the files are parsed with all the GFM extensions.
help

if ARGV.include?('--help')
  puts HELP
  exit 0
end

root = File.expand_path('../../', __FILE__)
$:.unshift File.expand_path('lib', root)

require 'github/markdown'
require 'tempfile'
require 'launchy'

mode = :gfm
mode = :markdown if ARGV.delete('--readme')
mode = :plaintext if ARGV.delete('--plaintext')

outputFilePath = File.join(Dir.tmpdir, File.basename(ARGF.path))  + ".html"

File.open(outputFilePath, "w") do |outputFile |
    outputFile.write(GitHub::Markdown.to_html(ARGF.read, mode))
end

outputFileUri = 'file:///' + outputFilePath

Launchy.open(outputFileUri)

I tried this with fenced blocks for Ruby and Cucumber. While the fences (ruby, cucumber, etc.) appear to be recognized as fences (because they're rendered in fixed width text), there is no syntax highlighting. Any idea why?
M
Mohnish

Another option is AllMark - the markdown server.
Docker images available for ready-to-go setup.

$ allmark serve .

Note: It recursively scans directories to serve website from markdown files. So for faster processing of single file, move it to a separate directory.


P
Peter Mortensen

I recently made what you want, because I was in need to generate documentation from Markdown files and the GitHub style is pretty nice. Try it. It is written in Node.js.

gfm


I know, the module has not been udated in 9 months, why do you downvote an old post?