namedtuple-table

Problem

• You want to make a "sample table" config file (e.g. for Snakemake), so that various system-specific attributes can be accessed via an index.
• You want to store it as a human-readable tab-separated text file.
• You don't want to install Pandas.

Solution

• NamedTupleTable represents tabular data as a mapping between some index column and rows of some NamedTuple.

  my_table["label_1"] -> ThisTableNamedTuple
  

• To index on a different column, produce a new table by calling .with_index("new_index"). Values of the new index must be unique in every row.

• Tables are immutable and hashable, so should play nicely with caching, filters etc. We could add a "select" method etc. but it should be straightforward to do this stuff with Python's functional programming features.

Drawbacks

• This is not designed to scale; in the intended use-case the table size is modest and you are doing somewhat expensive things with the data. If you need performance/scale, consider Pandas or a database interface like dataset.

Usage example

Store data in a tab-separated variable file. The first non-comment line must be a set of column headers. Other lines can be commented out with # or !

# dogs.tsv
ref	name		collar	age
1	Bertie		red	4
2	Geoff		blue	2
!3	Bandit		none	40
4	Gertrude	blue	5

and load with

>>> from namedtuple_table import NamedTupleTable
>>> from pathlib import Path

>>> dogs = NamedTupleTable.from_tsv(Path("dogs.tsv"))

Now you have a dict-like Mapping of data rows represented as NamedTuple objects. All data is loaded as strings, so you might need to cast back and forth to int.

>>> print(dogs)
NamedTupleTable (3 items, index = ref)

>>> for i in range(5):
...     if str(i) in dogs:
...         print(dogs[str(i)])
...
TableRow(ref='1', name='Bertie', collar='red', age='4')
TableRow(ref='2', name='Geoff', collar='blue', age='2')
TableRow(ref='4', name='Gertrude', collar='blue', age='5')

>>> dogs['4'].collar
'blue'

To use a different index column, get a new table with the .index_by method.

>>> dogs_by_name = dogs.with_index("name")
>>> dogs_by_name["Geoff"]
TableRow(ref='2', name='Geoff', collar='blue', age='2')