The OpenOffice.org XML project is looking for contributions in the areas listed below. If you are interested in working in one of these areas, please contact us at email@example.com.
XML Based Filters
The following suggestions are related to the framwork for XML based filters.
Detection of the file type
Every time a document is being imported into OpenOffice.org the so called filter detection chooses between the existing filter implementations to find the one matching the type of document that shall be loaded. For filters that load XML documents using the filter framework, the filter detection only searches for a string with the first 1000 characters of a document. This string can be specified in the XML Filter Settings dialog.
A better solution, that we would like to have instead, would be to check the name and or XML namespace of the root node of the XML document.
Access to images
Currently XML based filters have only access to the BASE64 encoded data of embedded images an objects. Having also access to the binary data (for instance by unpacking the zip file of an OpenDocument file) or having a simple way to convert the BASE64 data into a binary file would simplify the developmenmt of filters.
Currently error logging is comparable weak for the XML based / XSLT filters. The only way to enable logging is to set a Java environment variable (e.g. -DXSLTransformer.statsfile=/usr/local/offices/xslt_debug.txt) in the Office options for Java.
A few new features are imaginable, such as:
- Customizing the filter logging via GUI
- Usage of defined log level (analog to Java Logging)
- GUI flag for a cumulative log file (instead replacing log for every transformation)
Currently, no XML validation takes place during the import or export of documents. Validation is only possible from a test dialog. This test dialog might be dropped in favor of an external development tools and/or (optional) validation during runtime. To achieve this, it is not sufficient to reuse the existing functionality, because it supports only DTDs, but no Relax-NG or other XML schema langauges like Schematron.