slstr-lst strip (as opposed to "chips" or "frames") are not detected

[pjhartzell]

Describe the bug

slstr-lst strip geometries are not corrected. In

if item.properties["s3:product_name"] == "slstr-lst" and item.id.endswith("_____"):

item.id.endswith("_____") is never True because the trailing underscores are stripped from the Item ID. Strip data does not have a frame number, so those trailing underscores are inserted as fill values. Brute force is to check for item ID length, something like len(item.id) == 57, which indicates there is no frame number (and, hence, the data is a "strip").

sentinel3/src/stactools/sentinel3/stac.py

Line 342 in 604505a

if item.properties["s3:product_name"] == "slstr-lst" and item.id.endswith("_____"):

To reproduce

Run create_item on a slstr-lst strip archive, e.g., S3A_SL_2_LST____20160518T063623_20160518T081722_20180929T115803_6059_004_177______LR1_R_NT_003.SEN3

Expected behavior
Strips to be detected.

[ircwaves]

Makes sense. Missed it because that combo isn't in the pile of examples. Is there a doc on how to make those data-less netcdf files? I could add the one you referenced.

To use the same logic as before, I pulled the instance_id out of the file name instead. That seems to line up with Sentinel-3 anming conventions.