Building the Right Environment to Support AI, Machine Learning and Deep Learning
The OpenDocument Format (ODF) is an XML-based file format for representing electronic documents such as spreadsheets, charts, presentations and word processing documents. The standard was developed by the OASIS (Organization for the Advancement of Structured Information Standards), and it is free and open format.
The OpenDocument format is used in free software and in proprietary software. Originally, format was implemented by the OpenOffice.org office suite and, with Office 2007 SP2, Microsoft also supports ODF subset.
OpenDocument format supports document representation:
- As a single XML document.
- As a collection of several subdocuments within a package.
Office applications use second approach so we will explain it detail.
Every ODF file is a collection of several subdocuments within a package (ZIP file), each of which stores part of the complete document. Each subdocument stores a particular aspect of the document. For example, one subdocument contains the style information and another subdocument contains the content of the document.
This approach has following benefits:
- You don't need to process entire file in order to extract specific data.
- Images and multimedia are now encoded in native format, not as text streams.
- Files are smaller as a result of compression and native multimedia storage.
There are four subdocuments in the package that contains file's data:
- content.xml - Document content and automatic styles used in the content.
- styles.xml - Styles used in the document content and automatic styles used in the styles themselves.
- meta.xml - Document meta information, such as the author or the time of the last save action.
- settings.xml - Application-specific settings, such as the window size or printer information.
Besides them, in package can be many other subdocuments like document thumbnail, images, etc.
In order to read the data from an ODF file you need to:
- Open package as a ZIP archive.
- Find parts that contain data you want to read.
- Read parts you are interested in.
On the other side, if you want to create a new ODF file, you need to:
- Create/get all necessary parts.
- Package everything into a ZIP file with appropriate extension.
Spreadsheet document files are the subset of ODF files. Spreadsheet files have .ods file extensions.
The content (sheets) is stored in content.xml subdocument.
Picture 1: content.xml subdocument.
As we can see on Picture 1, sheets are stored as xml elements. They contain column and row definitions, rows contain cells and so on... On picture is data from one specific document, but from this we can see basic structure of content.xml file (you can also download full ODF specification).
Our demo is Windows Forms application (Picture 2), written in C# using Visual Studio 2008.
Picture 2: Demo application.
- Create new Spreadsheet document.
- Read existing Spreadsheet document.
- Write created Spreadsheet document.
Creating new document and underlying model of application
Internal, Spreadsheet document is stored as DataSet. Each sheet is represented with DataTable, sheet's row with DataRow, and sheet's column with DataColumn. So, to create a new document we have to create new DataSet, with DataTables. Each DataTable has number of rows and columns that conforms to our needs.
To show data from our DataSet (and to allow editing that data) application dynamically creates tabs with DataGridViews (that are connected to our DataTables).
Through interface, user can read, write, edit data and add new rows to Spreadsheet document.
Zip component and XML parser
Although classes from System.IO.Packaging namespace (.NET 3.0) provides way to read and write ZIP files, they require different format of ZIP file. Because of that, our demo uses open source component called DotNetZip.
Using ZIP component we can extract files, get subdocument, replace (or add) subdocuments that we want and save that file as .ods file (which is ZIP file).
For processing documents we have used XmlDocument because it offers easy way to reach part that we want. Note that, if performance is crucial for you, you should use XmlTextReader and XmlTextWriter. That solution need more work (and code), but provides better performance.
Reading Spreadsheet Document
To read document we follow these steps:
- Extracting .ods file.
- Getting content.xml file (which contains sheets data).
- Creating XmlDocument object from content.xml file.
- Creating DataSet (that represent Spreadsheet file).
- With XmlDocument we select "table:table" elements, and then we create adequate DataTables.
- We parse child's of "table:table" element and fill DataTables with those data.
- At the end, we return DataSet and show it in application's interface.
Although ODF specification provides way to specify default row, column and cell style, implementations have nasty practice (that specially applies for Excel) that they rather write sheet as sheet with maximum number of columns and maximum number of rows, and then they write all cells with their style. So you could see that your sheet has more than 1000 columns (1024 in Calc and 16384 in Excel), and even more rows (and each rows contains the number of cells that is equal to the number of columns), although you only have write data to first few rows/columns.
ODF specification provides a way that you specify some element (like column/row/cell) and then you specify the number of time it repeats. So above behavior doesn't affect the size of the file, but that complicates our implementation.
Because of that, we can't just read the number of columns and add equal number of DataColumns to DataTable (because of performance issues). In this implementation, we rather read cells and, if they have data, we first create rows/columns they belong to, and then we add those cells to the DataTable. So, at the end, we allocate only space that we need to.
Writing Spreadsheet Document
To write document we follow these steps:
- Extracting template.ods file (.ods file that we use as template).
- Getting content.xml file.
- Creating XmlDocument object from content.xml file.
- Erasing all "table:table" elements from the content.xml file.
- Reading data from our DataSet and composing adequate "table:table" elements.
- Adding "table:table" elements to content.xml file.
- Zipping that file as new .ods file.
In this application, as template, we have use empty document. But application can be easy modified to use some other template (so, that you have preserved styles, etc).
You can download the latest version of the demo application (together with the C# source code) from here.
As always in programming, there is more than one method to achieve the same thing.
ODF files are just a collection of xml files, packed in zip files so, any of the vast number of tools for handling zip files and XML data can be used to handle OpenDocument.
As another option, you could use some third party component which has support for ODF format. This will probably cost you some money but has advantage that usually more than one format (for example: GemBox.Spreadsheet reads/writes XLS, XLSX, CSV, HTML and ODS) are supported within the same API, so your application will be able to target different file formats using the same code.