A Webservice to analyze the content of PDF Documents using a HTTP API. This is a simple HTTP wrapper around ad-freiburg/pdfact and builds the container image directly from their source.
Start the service:
> docker run -p 80:80 dnlbauer/pdfact-service
[2022-11-27 12:36:38 +0000] [1] [INFO] Starting gunicorn 20.1.0
[2022-11-27 12:36:38 +0000] [1] [INFO] Listening at: http://0.0.0.0:80 (1)
[2022-11-27 12:36:38 +0000] [1] [INFO] Using worker: gthread
[2022-11-27 12:36:38 +0000] [7] [INFO] Booting worker with pid: 7
PDFs can be POST
ed to /analyze
as multipart file request. The response will contain the output of pdfact
. The response format can be specified using the correct MIME Accept
header; pdfact
van provide json, xml and plain text as output format.
> curl -H "Accept: application/json" -F [email protected] localhost:80/analyze
{"paragraphs": [
{"paragraph": {
"role": "page-header",
"positions": [{
"minY": 642.6,
"minX": 210.3,
"maxY": 652.6,
"maxX": 401.6,
"page": 1
}],
"text": "This is a test PDF document."
}},
{"paragraph": {
"role": "body",
"positions": [{
"minY": 628.5,
"minX": 91.2,
"maxY": 636.6,
"maxX": 520.8,
"page": 1
}],
"text": "If you can read this, you are lucky."
}}
]}
Supported cli arguments (--units
, --roles
) can be supplied as http parameters. Example:
> curl ... localhost:80/analyze?roles=body&units=words
All credits go to the Algorithms and Data Structures Group from University of Freiburg for ad-freiburg/pdfact.
Published under CC0. Do whatever you want :-)
(but also check the license of pdfact if you are going to use the image as is).