I think it operates as Sax parsing (to oppose to Dom parsing, among others).
Let's expand on that. Parsing means taking a string, and finding out patterns, components. Parsing CSV data means to find field values (and headers) out of a continuous string. Parsing XML means to find tag names, attributes and their values and text data. Etc.
The Sax way is to read the string (file) linearly, keeping little in memory, emitting events on each milestone: a new tag, a new attribute (and its value), a text area, etc. The Sax parser keep very little context, perhaps just enough to know the nesting level.
The Dom way actually uses a Sax parser, but builds a model in memory, a tree reflecting the nesting of elements, the dependencies (attributes related to a tag), etc.
It uses lot of memory, but allows transformations of the tree, navigation, etc. Web browsers build a Dom tree of the HTML pages.