pp.server 1.0.6.1

pp.server - Produce & Publish Server

pp.server is a Pyramid based server implementation and implements the server side functionality of the Produce & Publish platform. It is known as the Produce & Publish Server.

The Produce & Publish Server provided web service APIs for converting HTML/XML + assets to PDF using one of the following external PDF converters:

• PrinceXML (www.princexml.com, commercial)

• PDFreactor (www.realobjects.com, commercial)

• PhantomJS (free, unsupported)

• Speedata Publisher (www.speedata.de, open-source, experimental support)

• WKHTMLTOPDF (www.wkhtmltopdf.org, open-source, experimental support)

• Vivliostyle Formatter (www.vivliostyle.com, commercial, experimental support)

• Antennahouse 6.2 (www.antennahouse.com, commercial)

• Weasyprint (free, unsupported)

In addition there is experimental support for generating EPUB documents using Calibre (www.calibre.org, open-source).

In addition the Produce & Publish server provides a simple conversion API for converting format A to B (as supported through LibreOffice or OpenOffice). The conversion is build on top of unoconv.

The web service provides both synchronous and asynchronous operations.

Requirements

• Python 3.5 or higher, no support for Python 2.x

• the external binaries

  • PrinceXML: prince,

  • PDFreactor up to version 7: pdfreactor,

  • PDFreactor version 8 or higher: pdfreactor.py,

  • Unoconv: unoconv

  • Speedata Publisher: sp

  • Calibre: ebook-convert

  • WKHTMLTOPDF: wkhtmltopdf

  • Vivliostyle: vivliostyle-formatter

  • Weasyprint: weasyprint

  • Antennahouse: run.sh

  must be in the $PATH. Please refer to the installation documentation of the individual products.

Installation

• create an virtualenv environment (Python 2.7 (or Python 3.4)) - either within your current (empty) directory or by letting virtualenv create one for you. (easy_install virtualenv if virtualenv is not available on your system):

  virtualenv --no-site-packages .

  or:

  virtualenv --no-site-packages pp.server

• install the Produce & Publish server:

  bin/pip install pp.server

• create a server.ini configuration file (and change it according to your needs):

  [DEFAULT]
  debug = true
  
  [app:main]
  use = egg:pp.server
  reload_templates = true
  debug_authorization = false
  debug_notfound = false
  
  [server:main]
  use = egg:waitress#main
  host = 127.0.0.1
  port = 6543

• create a circus.ini configuration file (and change it according to your needs):

  [watcher:pserve]
  cmd = bin/pserve server.ini
  
  [env:pserve]
  PATH = $PATH

• start the server (in foreground):

  bin/pserve server.ini

• or start it in background:

  bin/circusd circus.ini  --daemon

Converter requirements

For the PDF conversion the related converter binaries or scripts must be included in the $PATH of your server.

• prince for PrinceXML

• pdfreactor for PDFreactor 7

• pdfreactor8 for PDFreactor 8 or higher

• phantomjs for PhantomJS

• wkhtmltopdf for WKHTMLToPDF

• ebook-convert for Calibre

• sp for the Speedata Publisher

• vivliostyle for the Vivliostyle Formatter

• weasyprint for Weasyprint

• antennahouse for the Antennahouse

API documentation

All API methods are available through a REST api following API URL endpoint:

http://host:port/api/1/<command>

With the default server configuration this translates to:

http://localhost:6543/api/1/pdf

or

http://localhost:6543/api/1/unoconv

PDF conversion API

Remember that all converters use HTML or XML as input for the conversion. All input data (HTML/XML, images, stylesheets, fonts etc.) must be stored in ZIP archive. The filename of the content must be named index.html.

You have to POST the data to the

http://host:port/api/1/pdf

with the following parameters:

• file - the ZIP archive (multi/part encoding)

• converter - a string that determines the the PDF converter to be used (either princexml, pdfreactor, phantomjs, vivliostyle, or calibre for generating EPUB content)

• async - asynchronous (“1”) or synchronous conversion (“0”, default)

• cmd_options - an optional string of command line parameters added as given to the calls of the externals converters

Returns:

The API returns its result as JSON structure with the following key-value pairs:

• status - either OK or ERROR

• data- the generated PDF file encoded as base64 encoded byte string

• output - the conversion transcript (output of the converter run)

Unoconv conversion API

The unoconv web service wraps the OpenOffice/LibreOffice server mode in order to perform document conversion (mainly used in the Produce & Publish world for convertering DOC(X) documents to HTML/XML).

Remember that all converters use HTML or XML as input for the conversion. All input data (HTML/XML, images, stylesheets, fonts etc.) must be stored in ZIP archive. The filename of the content must be named index.html.

You have to POST the data to the

http://host:port/api/1/unoconv

with the following parameters:

• file - the source files (multi/part encoding)

• async - asynchronous (“1”) or synchronous conversion (“0”, default)

• cmd_options - an optional string of command line parameters added as given to the unoconv calls

Returns:

The API returns its result as JSON structure with the following key-value pairs:

• status - either OK or ERROR

• data - the converted output files as ZIP archive (e.g. a DOCX file containing images will be converted to a HTML file plus the list of extract image files)

• output - the conversion transcript (output of the converter run)

Asynchronous operations

If you set async to ‘1’ in the API calls above then both calls will return a JSON datastructure like

{‘job_id’: <some id>}

The job_id can be used to poll the Produce &amp; Publish Server in order to retrieve the conversion result asynchronously.

The poll API is provided through the URL

http://host:port/api/1/poll/<job_id>

If the conversion is still pending the API will return a JSON document

{‘done’: False}

If the conversion has finished then a PDF/Unoconv specific return JSON document will be return (same format as for the synchronous API calls). In addition the key-value pair {‘done’: True} will be included with the JSOn reply.

Introspection API methods

Produce & Publish server version:

http://host:port/api/version

returns:

{“version”: “0.3.2”, “module”: “pp.server”}

Installed/available converters:

http://host:port/api/converters

returns:

{“unoconv”: true, “pdfreactor”: true, “phantomjs”: false, “calibre”: true, “princexml”: true}

Versions of installed converter:

http://host:port/api/converter-versions

returns:

{‘princexml’: ‘Version x.y’, ‘pdfreactor: ‘Version a.b.c’, …}

Other API methods

Cleanup of the queue directory (removes conversion data older than one day)

http://host:port/api/cleanup

returns:

{“directories_removed”: 22}

Authorization support

The pp.server implementation provides a simple and optional authorization mechanism by accepting a pp-token header from the client. In order to enable the authorization support on the server side you need to configure the authenticator method and the authorization token in your .ini file:

[app:main]
use = egg:pp.server
...
pp.authenticator = token_auth
pp.authentication_token = 12345

The token_auth string refers to a method in pp.server.authorization which is a simple authorization method (for the beta phase) supporting only one token for now. The token is configured through the pp.authentication_token value.

Any client sending a HTTP request to the pp.server server instance is required to send a HTTP header for authorization (if enabled on the server):

pp-token: <value of token>
pp-token: 12345

Advanced installation issues

Installation of PDFreactor using zc.buildout

• https://bitbucket.org/ajung/pp.server/raw/master/pdfreactor.cfg

Installation of PrinceXML using zc.buildout

• https://bitbucket.org/ajung/pp.server/raw/master/princexml.cfg

Production setup

pserve and celeryd can be started automatically and controlled using Circus. Look into the following buildout configuration

• https://bitbucket.org/ajung/pp.server/raw/master/circus-app.ini

Source code

https://bitbucket.org/ajung/pp.server

Bug tracker

https://bitbucket.org/ajung/pp.server/issues

Support

Support for Produce & Publish Server is currently only available on a project basis.

License

pp.server is published under the GNU Public License V2 (GPL 2).

Contact

ZOPYX

Hundskapfklinge 33

D-72074 Tuebingen, Germany

[e-mail]

www.zopyx.com

www.produce-and-publish.info

1.0.6 (2017/08/08)

• display available converters during startup

1.0.5 (2017/04/06)

• support for WeasyPrint

1.0.3 (2017/02/13)

• calling princexml with explicit option -o (Prince 11 compatibility)

1.0.0 (2017/01/08)

• first official release

• requires Python 3.5 or higher

0.7.13 (2016/08/30)

• added circus dependency in order to provide daemonizing support out of the box

0.7.12 (2016/08/29)

• fix for PDFreactor 8

0.7.11 (2016/06/01)

• updated to Pyramid 1.7

0.7.7 (2016/01/24)

• updated support for latest Vivliostyle formatter

• added support for Antennahouse Formatter

0.7.6 (2015/11/30)

• support for PDFreactor 8

0.7.5 (2015/11/18)

• fixed race condition while creating directories

0.7.4 (2015/11/14)

• support for nested uploaded ZIP files

0.7.3 (2015/11/14)

• support for Vivliostyle Formatter

0.7.2 (2015/04/20)

• merged https://bitbucket.org/ajung/pp.server/pull-request/1/ (improper check for wkhtmltopdf)

• merged https://bitbucket.org/ajung/pp.server/pull-request/2/ (fix for async operations)

0.7.1 (2015/03/13)

• unicode fix in runcmd()

0.7.0 (2015/02/15)

• 0.6.x was completely badly packaged

• changed repo structure

0.6.1 (2015/02/02)

• add /api/converter-versions to webservice API

0.6.0 (2015/01/26)

• dropped Python 2.X support, Python 3.3 or higher is now a mandatory requirement

0.5.5 (2015/01/23)

• UTF8 handling fix

0.5.3 (2014/11/23)

• support for WKHTMLTOPDF

0.5.2 (2014/11/19)

• support for Speedata Publisher

0.5.1 (2014/10/12)

• improved error handling

0.5.0 (2014/10/12)

• official Python 3.3/3.4 support

0.4.7 (25.09.2014)

• fixed documentation bug

0.4.6 (22.08.2014)

• removed PDFreactor –addlog option

0.4.5 (22.08.2014)

• added supplementary commandline options to pdfreactor commandline call

0.4.4 (24.01.2014)

• minor typos fixed

0.4.3 (20.01.2014)

• implemented automatic queue cleanup after one day

0.4.2 (18.01.2014)

• URL fix in index.pt related to virtual hosting

0.4.1 (13.01.2014)

• show Python version and converters on index.pt

• authorization support added

0.4.0 (17.10.2013)

• Python 3.3 support

• Pyramid 1.5 support

0.3.5 (05.10.2013)

• added ‘cmd_options’ to pdf and unoconv API methods for specifying arbitary command line parameters for the external converters

0.3.4 (05.10.2013)

• added ‘cleanup’ API

0.3.3 (05.10.2013)

• added ‘version’ and ‘converter’ API methods

0.3.2 (04.10.2013)

• added support EPUB conversion using Calibre

0.3.1 (03.10.2013)

• updated documentation

0.3.0 (14.07.2013)

• unoconv conversion now returns a ZIP archive (e.g. a HTML file + extracted images)

0.2.7 (06.07.2013)

• added support for Phantom.js converter

0.2.5 (05.07.2013)

• better detecting of prince and pdfreactor binaries

0.2.2 (05.07.2013)

• updated the documentation

• minor cleanup

0.2.1 (04.07.2013)

• re-added poll API

0.2.0 (03.07.2013)

• converted XML-RPC api to REST api

0.1.9 (01.07.2013)

• monkeypatch pyramid_xmlrpc.parse_xmlrpc_request in order to by-pass its stupid DOS request body check

0.1.7 (29.06.2013)

• more tests

• fixes

• updated documentation

0.1.5 (27.06.2013)

• test for synchronous operations

• fixes

0.1.0 (24.06.2013)

• initial release