Support time points as numeric indexes

[moznuy]

This is a rough draft

What's is implemented:

• datetime.datetime works as numeric index.
• datetime.date works as numeric index.
• datetime.time works as numeric index. (very sketchy, see below)
• Add test for all above for:
  • Saving to Redis, checking if is valid.
  • Finding by less operator(any comparison should work) in sorted fashion.
  • Saving no-UTC and checking that restored version(UTC) is equal to pre-saving no-UTC value.

What can be added / improved:

• datetime.time custom parsing implementation
• Our type for datetime.time because user will have to write parsing(validation) themselves otherwise(see post_model_time in tests).
• Support for datetime.timedelta (I skipped it for now).
• Support for Model.find(Model.field < "Thu Sep 25 10:36:28 2003") or ISO format, or etc...
• Documentation.
• More tests.
• MORE TESTS(Seriously, I don't think I understand time zones 100%).

Questions:

• How to store timezone information as numeric field? Right now I just convert everything to UTC before saving. (This needs to be checked)
• What to do if user did not provide timezone?
• Do we want to support Model.find(Model.field < "Thu Sep 25 10:36:28 2003") or "2003-09-25T10:49:41.5-03:00" or any other string in expressions. (Because pydantic supports only ISO format(I think)). We could use dateutil. Or make it optional dependency.
• Precision. To what precision should we save time-points. I implemented milliseconds because:
  • It should be enough
  • pydantic parsing support parsing datetime from integers ms out of the box.
  • Later I checked redis-om-node, and there they support precision to seconds(I'm not sure): https://github.com/redis/redis-om-node#searching-on-dates
• Cross type expression(Model.datetime < date/time). Should we support this? Or catch this and raise exception?
• Will Redis at some point add "Timepoint" index field type(like NUMERIC)? (with timezone support). Because that would simplify things.
• I probably forgot thing or 10, will add later

Problems:

• datetime.time: pydantic support parsing only as:

  time, existing time object
  str, following formats work:
   HH:MM[:SS[.ffffff]][Z or [±]HH[:]MM]]]

  so no int parsing , very custom logic(probably wrong)

Sketchy things:

• datetime.time custom parsing implementation
• I added this to resolve_value. Maybe there is better approach.

  elif field_type is RediSearchFieldTypes.NUMERIC:
  +     value = jsonable_encoder(value)

  https://github.com/redis/redis-om-python/pull/240/files#diff-ed42e1a9e452289cee8007ec1dc71beb5e883a54f7f58db63320901b9e4baaf3R546

Fixes #199

[codecov-commenter]

Codecov Report

Merging #240 (97c617c) into main (c2339b3) will increase coverage by 0.36%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #240      +/-   ##
==========================================
+ Coverage   77.25%   77.61%   +0.36%     
==========================================
  Files          12       12              
  Lines        1165     1184      +19     
==========================================
+ Hits          900      919      +19     
  Misses        265      265              

Flag	Coverage Δ
unit	77.61% <100.00%> (+0.36%)	:arrow_up:

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
aredis_om/model/encoders.py	59.77% <100.00%> (+8.38%)	:arrow_up:
aredis_om/model/model.py	86.33% <100.00%> (+0.06%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update c2339b3...97c617c. Read the comment docs.

[moznuy]

Any input will be appreciated