In general, the conversion process works by applying rules that convert incoming data from one converter state to another. Whether or not a given rule is applied depends on conditions that can include the input document type (the type-in for the data), and logical conditions such as URL pattern matches, data content, and so on. The conversion process searches for the first rule that matches the incoming data, and then applies the action associated with that rule. The output of that action can then be further processed by other rules, depending on its type-out value and whether other rules exist that are triggered by that value.
The next few sections explain the state of a converter at any point, the internals of the conversion process, and the interaction between crawling, indexing, index updates, and cached content.