A dockerized webservice written in Go that uses wkhtmltopdf to convert HTML into documents (images or pdf files).
HTTP: docker build -t wkhtmltopdf . && docker run --rm -it -p 3000:3000 -e SECURE=false pukkasoftware/wkhtmltopdf
HTTPS: docker build -t wkhtmltopdf . && docker run --rm -it -p 3000:3000 -v $(pwd)/app/ssl:/app/ssl -e AUTHENTICATION=username:password pukkasoftware/wkhtmltopdf
For security purposes, we do not embed the SSL certificates in the image. Instead, mount a volume containing a cert.pem and key.pem to /app/ssl
Optionally enable HTTP basic auth using the AUTHENTICATION environment variable. Failure to auth will result in a 401 error.
The service listens on port 3000 for POST requests on the root path (/). Any other method returns a 405 not allowed status. Any other path returns a 404 not found status.
The body should contain a JSON-encoded object containing the following parameters:
-
output: The type of document to generate, can be either
jpg,pngorpdf. Defauts topdfif not specified. Depending on the output type the appropriate binary is called. -
requests: An array of documents to be converted, each of which has the following parameters:
One of:
- content: The HTML/CSS to be converted.
- url: The URL of the page to fetch and convert.
And:
- options: A list of key-value arguments that are passed on to the appropriate
wkhtmltopdfbinary. Boolean values are interpreted as flag arguments (e.g.:--greyscale). - cookies: A list of key-value arguments that are passed on to the appropriate
wkhtmltopdfbinary as separatecookiearguments.
Example: posting the following JSON:
{
"output": "pdf",
"requests": [
{
"content": "<html><body><h1>Google.com<\/h1><p>The HTML that would have been fetched<\/p><\/body><\/html>",
"options": {
"margin-bottom": "1cm",
"orientation": "Landscape"
},
"cookies": {
"foo": "bar",
"baz": "foo"
}
},
{
"url": "google.com",
"options": {
"margin-bottom": "1cm",
"orientation": "Landscape",
"grayscale": true
},
"cookies": {
"foo": "bar",
"baz": "foo"
}
},
{
"content": "<html><body><h1>Yahoo.com<\/h1><p>The HTML that would have been fetched<\/p><\/body><\/html>",
"options": {
"margin-bottom": "1cm",
"orientation": "Landscape",
"grayscale": true
},
"cookies": {
"foo": "bar",
"baz": "foo"
}
}
]
}
will have the effect of the following command-line being executed on the server:
echo "content" | /usr/local/bin/wkhtmltopdf --margin-bottom 1cm --orientation Landscape --grayscale --cookie foo bar --cookie baz foo - -
/usr/local/bin/wkhtmltopdf --margin-bottom 1cm --orientation Landscape --cookie foo bar --cookie baz foo google.com -
echo "content" | /usr/local/bin/wkhtmltopdf --margin-bottom 1cm --orientation Landscape --grayscale --cookie foo bar --cookie baz foo - -
The - at the end of the command-line is so that the document contents are redirected to stdout so we can in turn redirect it to the web server's response stream.
When using jpg or png output, the set of options you can pass are actually more limited. Please check wkhtmltopdf usage docs or rather just use wkhtmltopdf --help or wkhtmltoimage --help to get help on the available command-line arguments. jpg and png only process the first element of the requests array.
You can also pass in the HTML/CSS directly like so:
{"output": "pdf",
"requests": [{
"content": "<html><body><h1>My Web Page<\/h1><p>The HTML that would have been fetched<\/p><\/body><\/html>",
"options": {
"margin-bottom": "1cm",
"orientation": "Landscape",
"grayscale": true
},
"cookies": {
"foo": "bar",
"baz": "foo"
}
}]
}
There are challenges with converting raw HTML when you have relative links. There are known workarounds that we haven't yet tried (see this stack overflow question for more information).
The server currently does not check if the arguments are valid, just passes them on to the wkhtmltopdf and wkhtmltoimage binaries. If you pass a wrong argument you don't currently get an error, just an empty image or pdf back.
This code is released under the MIT License.