mdb.validate

Integrity checks.

Every check here holds in both reference samples, which is what makes them useful: a deviation is a corruption signal, not a quirk. The most valuable is heap tiling – in a healthy file the region in front of the TOC is 100% covered by TOC-referenced values with no gaps and no overlaps, so a gap means data nothing points at and an overlap means two objects claiming the same bytes.

>>> for finding in mdb.validate(f):
...     print(finding)

class Finding(severity, check, message, detail=None)

Bases: NamedTuple

One validation result.

severity: str

Alias for field number 0

check: str

Alias for field number 1

message: str

Alias for field number 2

detail: Optional[object]

Alias for field number 3

property ok: bool

validate(f, deep=True)

Run every check and return the findings, worst first.

deep=False skips the whole-file passes (heap tiling, reference resolution) for a fast structural check on a large database.

Return type:

List[Finding]

heap_coverage(container)

Tile the value heap and report (regions, gaps, overlaps).

Geometry properties (the TOC object, the container itself) are excluded: they describe file regions rather than heap allocations.

Return type:

tuple[List[Region], List[tuple], List[tuple]]