This is the final release to gloss-colletion.
Find future updates at https://pypi.org/project/gloss/

gloss-collection

Gloss (short for glossary) is a different kind of dictionary. All values also automatically become keys and so can be looked up in the Gloss to find its corresponding 'key'. This makes it easy to lookup and maintain one-to-one relationships such as a label to a certain magnitude.

As both the key and value are gettable items from the Gloss and are indistinguishable once added, term is the preferred name for items in a Gloss, and a term pair is the 1-1 mapping each term belongs to.

Gloss is a MutableMapping and supports all the same methods that dict does.

Examples

from gloss import Gloss

example = Gloss()
example["stdin"] = 0
example.update({"stdout": 1, "stderr": 2})
print(example)
#Gloss({"stdin": 0, "stdout": 1, "stderr": 2})
print(example[1], "goes to" example["stderr"])
#stdout goes to 2
example[3] = "config.toml"
example[1] = "shell pipe"
print(example)
#Gloss({"stdin": 0, "stderr": 2, "config.toml": 3, "shell pipe": 1})
print([fd_or_desc for fd_or_desc in example])
#["stdin", "stderr", "config.toml", "shell pipe", 0, 1, 2 , 3]

Shouldn't I just use an enum?

Great thought! You can, and often should, use an Enum for these sorts of relationsips. Besides being built in to Python, Enums are probably faster and more space efficient. However, some limitations of Enums solved by Gloss are:

• enum members are static. They are defined all at once in the class and their values cannot change. A Gloss on the other hand can be added to, altered, even have members deleted or popped, all at runtime
• Accessing Enum member namess and their values are differnt operations. Looking up a member by name is done with either dot dereference or getitem; looking up a member by value is done with a call. With a Gloss you don't have to know which side of the mapping your key is on (if there even is a distinction to the mapping), it is all done by getitem
• One side of an enum mapping must be a string. Because member names are attributes they must follow Python identifier naming rules. In a Gloss, all keys may be any hashable object

Shouldn't I just use a dict?

While the classic dict solves some of limitations of using an Enum, primarily being mutable at runtime, looking up any value is O(n) and therefore so is updating, deleting, poping, etc that value. In a Gloss searching for any value is O(1) time, and updating that value (for example chaning the key it maps to) can be done in a single operation.