Remove .detect_column_types() from table, make it a documented API

[simonw]

I used it in geojson-to-sqlite here: https://github.com/simonw/geojson-to-sqlite/blob/f10e44264712dd59ae7dfa2e6fd5a904b682fb33/geojson_to_sqlite/utils.py#L45-L50

It would make more sense for this method to live on the Database rather than the Table - or even to exist as a separate utility method entirely.

Then it should be documented.

[simonw]

Here's the current method:

sqlite-utils/sqlite_utils/db.py

Lines 823 to 845 in f728917

	def detect_column_types(self, records):
	all_column_types = {}
	for record in records:
	for key, value in record.items():
	all_column_types.setdefault(key, set()).add(type(value))
	column_types = {}
	for key, types in all_column_types.items():
	if len(types) == 1:
	t = list(types)[0]
	# But if it's list / tuple / dict, use str instead as we
	# will be storing it as JSON in the table
	if t in (list, tuple, dict):
	t = str
	elif {int, bool}.issuperset(types):
	t = int
	elif {int, float, bool}.issuperset(types):
	t = float
	elif {bytes, str}.issuperset(types):
	t = bytes
	else:
	t = str
	column_types[key] = t
	return column_types

If I make it a utility function instead of a class method I could ensure it is directly importable like so:

from sqlite_utils import detect_column_types

[simonw]

Should I keep table.detect_column_types() working so as not to break existing code?

If it was part of the documented API then I wouldn't break that without bumping to 3.x. Since it's undocumented I'm going to make it as a breaking change instead (and bump the geojson-to-sqlite dependency version).

[simonw]

Actually I'll put it in the utils.py module.

[simonw]

While I'm at it I think I'll rename it to suggest_column_types - it's not really detecting them since the input is just a list of dictionaries.