Module parse 

Parallel chunk parsing (ADR-0011): record sub-ranges of one chunk fan out across rayon workers, each producing a ParsedBatch; the builder appends the batches in chunk order, so EntryId assignment is deterministic.

Structs

ParsedBatch 🔒

One worker’s output for a record sub-range, in record order.

ParsedMeta 🔒

One entry parsed by a worker; the name lives in its batch’s pools.

RecordArena 🔒

Fixed-size record store for the deferred/extension caches: records live back-to-back in one growable allocation, addressed by slot (ADR-0012).

RecordAttrs 🔒

$STANDARD_INFORMATION + $DATA extract shared by every parse path.

Constants

FILE_ATTRIBUTE_HIDDEN 🔒

FILE_ATTRIBUTE_REPARSE_POINT 🔒

FILE_ATTRIBUTE_SYSTEM 🔒

PARSE_SUB 🔒

Sub-range fed to one parse worker — small enough to spread a 16MiB chunk across cores, large enough to amortize the per-task overhead.

Functions

append_batches 🔒

extract_attrs 🔒

parse_chunk 🔒

Fan a chunk’s record sub-ranges across rayon workers. The returned batches are in sub-range order, so appending them sequentially yields the same EntryId assignment as a fully sequential parse.

parse_subrange 🔒

Validate, fix up and parse every record in bytes (a record-aligned slice whose first byte sits at first_logical in the $MFT stream). Mirrors the sequential loop exactly — same skip conditions, same counts.