viewtsv is a great example of a minimal vdtui application. This is an extremely functional utility in 25 lines of code. Here it is in its entirety, with line by line annotations:
VisiData 1.0 requires Python 3.4+.
import sys import vdtui
vdtui.py into an application verbatim. It's freely usable under the MIT license, and then it can just be imported like this and used without adding any external dependencies to the project.
class TsvSheet(vdtui.Sheet): rowtype = 'rows' # rowdef: tuple of values
All tabular data sheets inherit from
Sheet. The rowtype is displayed on the right status bar. The
rowdef comment declares the structure of every row object, and should be present for every Sheet.
columns = [vdtui.ColumnItem('tsv', 0)]
An initial column. Generally the class-level
columns is set to the actual columns of the sheet, but in this case, the columns aren't known until the source is loaded. (See the
reload() function below where they are set from the contents of the first row.) This line is not strictly necessary but makes loading feel a bit more responsive.
@async marks the function to spawn a new thread whenever it is called.
reload() function collects data from the source and puts it into
rows. It's called once automatically when first pushed, and manually with
self.rows = 
rows is a list of Python objects. The row definition ('rowdef') for the TsvSheet is a tuple of values, with each position corresponding to one column.
with open(self.source, encoding=vdtui.options.encoding) as fp:
source is the filename, which has been passed to the constructor (see the last line with
for line in fp: line = line[:-1] if line: self.rows.append(line.split('\t'))
For each line, strip the included newline character, and filter out any blank lines. Add each split tuple to
self.columns = [ vdtui.ColumnItem(colname, i) for i, colname in enumerate(self.rows) ]
The actual columns are set from the first (header) row.
ColumnItem is a builtin, which creates a column to use getitem/setitem with the given key/index.
self.rows = self.rows[1:]
The header row is removed from the list of rows. (Column names are displayed on the first row anyway).
vdtui.run(*(TsvSheet(fn, source=fn) for fn in sys.argv[1:]))
run(*sheets) is the toplevel entry point for vdtui.