cache behaves incorrectly, ignores period parameter

[verbal]

I think I found 2 problems with the cache

1. cache will return any data thats less than the end_date even if its period parameter asked for something else.

>>> from tools import api
>>> [(m.report_period, m.period) for m in api.get_financial_metrics('TSLA', end_date='2025-04-03', period='annual')]
[('2024-12-31', 'annual'), ('2023-12-31', 'annual'), ('2022-12-31', 'annual'), ('2021-12-31', 'annual'), ('2020-12-31', 'annual'), ('2019-12-31', 'annual'), ('2018-12-31', 'annual'), ('2017-12-31', 'annual'), ('2016-12-31', 'annual'), ('2015-12-31', 'annual')]
>>> [(m.report_period, m.period) for m in api.get_financial_metrics('TSLA', end_date='2025-04-03', period='quarterly')]
[('2024-12-31', 'annual'), ('2023-12-31', 'annual'), ('2022-12-31', 'annual'), ('2021-12-31', 'annual'), ('2020-12-31', 'annual'), ('2019-12-31', 'annual'), ('2018-12-31', 'annual'), ('2017-12-31', 'annual'), ('2016-12-31', 'annual'), ('2015-12-31', 'annual')]

Since the first call asked for the period to be annual, thats the data stored in the cache. If I execute the call again asking for quarterly results, it will just get the annual results from the cache and return those instead.

2. cache only merges on report_period, ignoring differences between ttm and quarterly results

If the first issue was fixed by knowing that different keys (report_period) will be returned by using period annual and quarterly/ttm, it will expose this second issue. For some api calls such as tools.api.get_financial_metrics, the call supports annual, ttm, and quarterly values for the period param. If you choose ttm or quarterly, the report_period will be the same, but the values will be different, so they need to be cached separately.

>>> [(m.report_period, m.revenue_growth) for m in api.get_financial_metrics('TSLA', end_date='2025-04-03', period='quarterly')]
[('2024-12-31', 0.020848224922563736), ('2024-09-30', -0.012470588235294117), ('2024-06-30', 0.1971268954509178), ('2024-03-31', -0.15361385941908054), ('2023-12-31', 0.07781584582441113), ('2023-09-30', -0.06326473302041963), ('2023-06-30', 0.06849843542372155), ('2023-03-31', -0.0406694629492557), ('2022-12-31', 0.13349491936235666), ('2022-09-30', 0.2669186252509744)]

>>> [(m.report_period, m.revenue_growth) for m in api.get_financial_metrics('TSLA', end_date='2025-04-03', period='ttm')]
[('2024-12-31', 0.0055584148224395264), ('2024-09-30', 0.019219874525273295), ('2024-06-30', 0.006047812549474906), ('2024-03-31', -0.020956258460521012), ('2023-12-31', 0.008850756849172262), ('2023-09-30', 0.020164206406602287), ('2023-06-30', 0.09290405067705004), ('2023-03-31', 0.05613660356976259), ('2022-12-31', 0.0881476831011314), ('2022-09-30', 0.11459667093469911)]

I think the solution is to take the period param into account when caching results.