Food menu parsing stuff assumes data is for dining commons

[simon-andrews]

Dining commons have really nice menus that give you all sorts of nutritional information and ingredients and stuff. We take advantage of this to give users a whole bunch of useful information:

umass-toolkit/umass_toolkit/dining_utils.py

Lines 35 to 59 in 3a78af4

	def category_html_to_dict(html_string, meal, category):
	soup = BeautifulSoup(html_string, 'html.parser')
	items = soup.find_all('a', href='#inline')
	ret = []
	for item in items:
	dish = {}
	dish['category-name'] = category
	dish['meal-name'] = meal
	for attribute in item.attrs.keys():
	if attribute.startswith('data-') and not attribute.endswith('dv'):
	attribute_name = attribute[5:]
	data = item.attrs[attribute]
	if attribute_name == 'calories' or attribute_name == 'calories-from-fat':
	data = int(data) if data else None
	elif attribute_name == 'clean-diet-str':
	data = data.split(', ')
	attribute_name = 'diets'
	elif attribute_name in ['allergens', 'ingredient-list']:
	data = parse_list(data)
	elif attribute_name in ['cholesterol', 'sodium', 'dietary-fiber', 'protein', 'sat-fat', 'sugars',
	'total-carb', 'total-fat', 'trans-fat']:
	data = ureg.Quantity(data) if data else None
	dish[attribute_name] = data
	ret.append(dish)
	return ret

Unfortunately, non-DC locations do not do this, and basically just have plain text menus. We should handle this.

I'm not sure what I think is the best way to handle this is. My current idea is maybe a get_dc_menu function, separate from get_menu. This could be confusing though because getting DC menus is probably going to be the more useful and more commonly used feature, so I don't like making it the "special case."

I dunno, thoughts?

[mattrossman]

I think separate functions is good. I was toying with the idea of get_menu having a different response depending on the ID it receives (list for DC locations, string for non-DC locations) but I think a predictable return type for each function is important.

We should probably have a location_is_dc function to help determine if an ID corresponds to a DC or non-DC location and use this to raise errors accordingly.

So users would write something like:

def my_dc_function(id):
  try:
    menu = get_dc_menu(id)
    do_something(menu)
  except ValueError:
    print('ID does not correspond to a DC location')

As for the names, I agree that get_menu could be too vague since it only applies to non-DC locations.
Some potential ideas:

• get_dc_menu and get_non_dc_menu
• get_parsed_menu and get_raw_menu