Module notes 

What datui noticed about a dataset while doing what it was already doing. A note never costs its own request or scan (all come from footers the schema and count needed), is never styled as an alarm, and states its basis, so “in 1 of 3 files” is never read as a claim about files not looked at. One claim per note, and one function (out_of) deciding the only ratio any note states.

Structs

Note

One thing datui noticed.

Functions

from_dataset

What the footers said, as notes. Empty when every file agrees, which is the common case and the one where there is nothing to say.

from_the_open

What the open has to say before any footer: which data files this read passed over, and whether a lake table is read as its plain files. Choices, not defects: datui never refuses a read asked for, and says what it did instead.

left_out_note

Rows a filter or sort leaves out because their files do not hold the named column in its read type: the one note about the view. Phrased about the files (a footer fact), not as “the view is N rows shorter”, which an earlier filter may have made false; so it also appears where those rows were already gone.

map_caches

The cache-*.arrow files map() wrote beside a Hugging Face cache’s splits, which the read left out: their columns are the mapping’s, not the split’s.

merged

Every note the panel shows (the open’s, the footers’, the view’s), with the one fact the first two both report said once: a mixed directory’s skipped files appear in the open’s note and the footer walk’s. The open’s sentence is kept; the walk’s is rebuilt without those files, keeping skips only it saw.

no_header

The files a read of several passed over because they hold no header: empty, blank, or nothing but NUL padding.

some_names

names, the first few of them, joined, and an ellipsis for the rest.