XML-RPC document conversion server
Project description
Install Cloudooo
$ python2.6 setup.py install
- Warnings:
you must have installed setuptools>=0.6c11 in this python.
Install LibreOffice / OpenOffice.org
- Install LibreOffice or OpenOffice.org.
Create Configuration File
The configuration file is used to start the application using paster. $ cp ./cloudooo/samples/samples.conf . # Copy to current folder
- The next step is define some attributes in cloudooo.conf:
working_path - folder to run the application. This folder need be created.
uno_path - folder where UNO library is installed (ex. /opt/libreoffice/basis-link/program/)
soffice_binary_path - folder where soffice.bin is installed (ex. /opt/libreoffice/program/)
Run Application
$ paster serve ./cloudooo.conf
or run as a daemon:
$ paster serve ./cloudoo.conf –daemon
Stop Application
$ kill -1 PASTER_PID
Warning: always use SIGHUP because only with this signal all processes are stopped correctly.
Cloudooo Description
XMLRPC + WSGI will be one bridge for easy access to LibreOffice / OpenOffice.org. This will implement one XMLRPC server into WSGI (Paster).
PyUno is used to connect to LibreOffice / OpenOffice.org stated with open socket. The features will be handled all by pyuno.
Only a process will have access to LibreOffice / OpenOffice.org by time.
All clients receive the same object(proxy) when connects with XMLRPC Server.
Managing LibreOffice / OpenOffice.org process
start ‘soffice.bin’; - Pyuno start ‘soffice.bin’ processes and the communication is through sockets; - ‘soffice.bin’ processes run in brackground;
control ‘soffice.bin’; - If the socket lose the connection, cloudooo kills the process, restartes processes and submit again the file;
XMLRPC Server - XMLRPC + WSGI
- Send document to ‘soffice.bin’ and return the document converted with metadata;
XMLRPC receives a file and connects to ‘soffice.bin’ process by pyuno;
The pyuno opens a new document, write, add metadata and returns the document edited or converted to xmlrpc and it return the document to the user;
When finalize the use of ‘soffice.bin’, should make sure that it was finalized;
Export to another format;
Invite document and return metadata only;
Edit metadata of the document;
- Problems and possible solution
‘soffice.bin’ is stalled; - finalize the process, start ‘soffice.bin’ and submit the document again(without restart the cloudooo);
‘soffice.bin’ is crashed; - finalize the process, verify if all the process was killed, start ‘soffice.bin’ and submit the document again(without restart the cloudooo)
‘soffice.bin’ received the document and stalled; - if ‘soffice.bin’ isn’t responding, kill the process and start
The document that was sent is corrupt; - write in log the error and verify that the process aren’t in memory
1.2.3 (2010-04-01)
Eggify handlers
Configuration file tells to cloudooo which handler must be loaded
Update interface (not implemented yet)
1.2.2 (2010-02-07)
No longer use Xvfb.
1.2.1 (2010-01-21)
Improve ERP5 compatibility.
1.2.0 (2010-01-19)
Support LibreOffice3.3rc3.
1.1.0 (2010-01-13)
Refactor runCloudOOoUnitTest.py to control better the daemon process
Use argparse instead of optparse
Refactor tests to use boolean asserts to validate True or False.
Use python-magic to validate the output documents according to mimetype.
Change folder structure of cloudooo to put handler, mimemapper, helpers related to handler in same folder than handler.
Refactor code to use json instead of jsonpickle.
Add getTableItem, getTableItemList and getTableMatrix for OOGranulate
Add getParagraphItemList and getParagraphItem for OOGranulate
Add getImageItemList and getImage for OOGranulate
Add OdfDocument
Add granulate interface.
1.0.9
use pkg_resources to get helper scripts.
move internal scripts to helper folder.
removed cloudooo as dependency of internal scripts.
modified way that the filters of OpenOffice.org are passed.
1.0.8
Remove all attributes that works with cloudooo script paths.
Use all scripts according to your python eggs.
Fixed problem when a spreadsheet will be converted to html.
1.0.7
Remove entry points, treat those as ordinary files.
Search all script files using pkg_resources.
