On the SFTP, in adamsampson/Faults/issue809, there's the first few frames of Easy Rider (LD10005) side 2.
Decoding this causes FieldPAL.calc_burstmedian to return NaN. This appears to be because the disc's mastered a bit hot - in the first field, all the lines trip the test for excessive burst amplitude in get_burstlevel.
The NaN eventually gets written to burstMedianIre in the JSON file... but Python's json.dump writes it as NaN, which isn't permitted by the JSON spec.
These two commits pass allow_nan=False to json.dump, and fix the test in calc_burstmedian so it returns 0 when no burst is found. It might be worth looking at the get_burstlevel heuristic to see if there's a better way of handling discs like this, though?
On the SFTP, in
adamsampson/Faults/issue809, there's the first few frames of Easy Rider (LD10005) side 2.Decoding this causes
FieldPAL.calc_burstmedianto return NaN. This appears to be because the disc's mastered a bit hot - in the first field, all the lines trip the test for excessive burst amplitude inget_burstlevel.The NaN eventually gets written to
burstMedianIrein the JSON file... but Python'sjson.dumpwrites it asNaN, which isn't permitted by the JSON spec.These two commits pass
allow_nan=Falsetojson.dump, and fix the test incalc_burstmedianso it returns 0 when no burst is found. It might be worth looking at theget_burstlevelheuristic to see if there's a better way of handling discs like this, though?