PegDF uses "peg" grammar for pdf file structure parsing. The parser walks through a file one object/xref/etc. a time, until reached EOF. This grammar is oblivious of pdf's cross reference structure, instead just store each object in memory, in a b-plus tree.

External logics captures the "Catalog", and from there recovers the whole content tree.

By doing this way, a single pass walking replaces the necessities of re-contruct document based on cross reference table.