Stream Analyzers

4.3

At the heart of Strigi Desktop Search are the stream analyzers.The system of having multiple analyzers indexing the same file at the same time is one of the reasons Strigi is be so fast. Also, Strigi has the ability to load analyzers at run time, allowing external packages to provide their own analyzers.

The two main types of analyzer are through analyzers (implemented by subclassing Strigi::StreamThroughAnalyzer) and end analyzers (subclasses of Strigi::StreamEndAnalyzer). For each file, one end analyzer is run together with all the through analyzers.

The end analyzer drives the process, reading data from the file. The through analyzers can either ignore this process entirely, or pick up the data as it is read and act on it. The mechanisms used to do this are described below in the section about through analyzers.

Also covered in that section are several convenience interfaces for analyzing XML files and text files.

End Analyzers

An end analyzer has two important functions. Firstly, since only one end analyzer can be run on any file, Strigi needs a way of working out which one to choose. This is done, in the first instance, by StreamEndAnalyzer::checkHeader().