Time slots iterators and helper functions

[iblislin]

• I need a iterator to create continuous time slots in a given interval of a day.
  • e.g. start from 2021/1/1 to 2021/1/10, in each day, I need to create 15m time slots from am 8:45 to pm 1:45.
• Helper functions:
  • Given a timestamp, find the time slots which the timestamp belongs to.

[anurag-mds]

Hi @iblislin! I've been contributing to
TimeSeries.jl and I'm interested in working
on this issue.

Could you clarify a few things before I start:

1. Is this issue still relevant?
2. What would the expected API look like?
3. Are there any existing packages or
   functions I should build on top of?

Happy to take this on if it's still needed!

[iblislin]

well, I think it still relevant, but not sure about there is some re-usable codebase about the resampling function recently added or not.

About the API, maybe a factory function to create iterator object that obey the iterator protocol in internal implementation, then return the TimeArray for the public API.

maybe you could put your idea about the API design first here.

[anurag-mds]

Thanks! Here's a rough API idea:
time_slots(start, stop, interval; from=Time(8,45), to=Time(13,45))
returning a TimeArray. Does that direction make sense?

[iblislin]

Thanks! Here's a rough API idea:
time_slots(start, stop, interval; from=Time(8,45), to=Time(13,45))
returning a TimeArray. Does that direction make sense?

oh, I just got an idea:
filter(∈(Time(8,45):Time(13,45)), DateTime(2026, 1, 1):Minute(15):DateTime(2026, 12, 31))

Do this style make sense to you?

Also, it' fine for me to provide several styles of API.

[anurag-mds]

That's a clean approach! So the idea would be to make Time ranges work with ∈ and step through DateTime ranges essentially making the filtering composable with Julia's existing iteration protocol rather than a custom API. I'll prototype this and share what I come up with.

[anurag-mds]

julia> using Dates

julia> result = filter(∈(Time(8,45):Minute(15):Time(13,45)),
                       DateTime(2026,1,1):Minute(15):DateTime(2026,1,10))
ERROR: promotion of types Time and DateTime failed to change any arguments
Stacktrace:
 [1] error(::String, ::String, ::String)
   @ Base .\error.jl:54
 [2] sametype_error(input::Tuple{Time, DateTime})
   @ Base .\promotion.jl:428
 [3] not_sametype(x::Tuple{Time, DateTime}, y::Tuple{Time, DateTime})
   @ Base .\promotion.jl:422
 [4] promote
   @ .\promotion.jl:405 [inlined]
 [5] ==(x::Time, y::DateTime)
   @ Dates C:\Users\RadhaKunjBalram\.julia\juliaup\julia-1.12.5+0.x64.w64.mingw32\share\julia\stdlib\v1.12\Dates\src\types.jl:475
 [6] in
   @ .\operators.jl:1362 [inlined]
 [7] Fix
   @ .\operators.jl:1194 [inlined]
 [8] filter(f::Base.Fix2{typeof(in), StepRange{Time, Minute}}, a::StepRange{DateTime, Minute})
   @ Base .\array.jl:2952
 [9] top-level scope
   @ REPL[5]:1

julia> 

Prototyped the core logic. The key insight is that ∈ fails when comparing Time and DateTime directly due to type promotion, but extracting Time(dt) from each DateTime solves it cleanly:

juliatimestamps = filter(dt -> Time(dt) ∈ Time(8,45):Minute(15):Time(13,45),
                    DateTime(2026,1,1):Minute(15):DateTime(2026,1,10))
ta = TimeArray(timestamps, ones(length(timestamps)))
# 189×1 TimeArray{Float64,1,DateTime,...}

Planning to wrap this as:

juliatime_slots(start, stop, interval; from=Time(8,45), to=Time(13,45))

returning a TimeArray. Does this direction look good to proceed with a PR?

               _
   _       _ _(_)_     |  Documentation: https://docs.julialang.org
  (_)     | (_) (_)    |
   _ _   _| |_  __ _   |  Type "?" for help, "]?" for Pkg help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 1.12.5 (2026-02-09)
 _/ |\__'_|_|_|\__'_|  |  Official https://julialang.org release
|__/                   |

julia> using TimeSeries

julia> 

julia> timestamps = filter(dt -> Time(dt) ∈ Time(8,45):Minute(15):Time(13,45),
                           DateTime(2026,1,1):Minute(15):DateTime(2026,1,10))
189-element Vector{DateTime}:
 2026-01-01T08:45:00
 2026-01-01T09:00:00
 2026-01-01T09:15:00
 2026-01-01T09:30:00
 2026-01-01T09:45:00
 2026-01-01T10:00:00
 2026-01-01T10:15:00
 2026-01-01T10:30:00
 2026-01-01T10:45:00
 2026-01-01T11:00:00
 2026-01-01T11:15:00
 2026-01-01T11:30:00
 2026-01-01T11:45:00
 2026-01-01T12:00:00
 2026-01-01T12:15:00
 2026-01-01T12:30:00
 2026-01-01T12:45:00
 2026-01-01T13:00:00
 2026-01-01T13:15:00
 2026-01-01T13:30:00
 2026-01-01T13:45:00
 ⋮
 2026-01-09T09:00:00
 2026-01-09T09:15:00
 2026-01-09T09:30:00
 2026-01-09T09:45:00
 2026-01-09T10:00:00
 2026-01-09T10:15:00
 2026-01-09T10:30:00
 2026-01-09T10:45:00
 2026-01-09T11:00:00
 2026-01-09T11:15:00
 2026-01-09T11:30:00
 2026-01-09T11:45:00
 2026-01-09T12:00:00
 2026-01-09T12:15:00
 2026-01-09T12:30:00
 2026-01-09T12:45:00
 2026-01-09T13:00:00
 2026-01-09T13:15:00
 2026-01-09T13:30:00
 2026-01-09T13:45:00

julia> 

julia> ta = TimeArray(timestamps, ones(length(timestamps)))
189×1 TimeArray{Float64, 1, DateTime, Vector{Float64}} 2026-01-01T08:45:00 to 2026-01-09T13:45:00
┌─────────────────────┬─────┐
│                     │ A   │
├─────────────────────┼─────┤
│ 2026-01-01T08:45:00 │ 1.0 │
│ 2026-01-01T09:00:00 │ 1.0 │
│ 2026-01-01T09:15:00 │ 1.0 │
│ 2026-01-01T09:30:00 │ 1.0 │
│ 2026-01-01T09:45:00 │ 1.0 │
│ 2026-01-01T10:00:00 │ 1.0 │
│ 2026-01-01T10:15:00 │ 1.0 │
│ 2026-01-01T10:30:00 │ 1.0 │
│ 2026-01-01T10:45:00 │ 1.0 │
│ 2026-01-01T11:00:00 │ 1.0 │
│ 2026-01-01T11:15:00 │ 1.0 │
│ 2026-01-01T11:30:00 │ 1.0 │
│ 2026-01-01T11:45:00 │ 1.0 │
│ 2026-01-01T12:00:00 │ 1.0 │
│ 2026-01-01T12:15:00 │ 1.0 │
│ 2026-01-01T12:30:00 │ 1.0 │
│ 2026-01-01T12:45:00 │ 1.0 │
│ 2026-01-01T13:00:00 │ 1.0 │
│ 2026-01-01T13:15:00 │ 1.0 │
│          ⋮          │  ⋮  │
│ 2026-01-09T09:30:00 │ 1.0 │
│ 2026-01-09T09:45:00 │ 1.0 │
│ 2026-01-09T10:00:00 │ 1.0 │
│ 2026-01-09T10:15:00 │ 1.0 │
│ 2026-01-09T10:30:00 │ 1.0 │
│ 2026-01-09T10:45:00 │ 1.0 │
│ 2026-01-09T11:00:00 │ 1.0 │
│ 2026-01-09T11:15:00 │ 1.0 │
│ 2026-01-09T11:30:00 │ 1.0 │
│ 2026-01-09T11:45:00 │ 1.0 │
│ 2026-01-09T12:00:00 │ 1.0 │
│ 2026-01-09T12:15:00 │ 1.0 │
│ 2026-01-09T12:30:00 │ 1.0 │
│ 2026-01-09T12:45:00 │ 1.0 │
│ 2026-01-09T13:00:00 │ 1.0 │
│ 2026-01-09T13:15:00 │ 1.0 │
│ 2026-01-09T13:30:00 │ 1.0 │
│ 2026-01-09T13:45:00 │ 1.0 │
└─────────────────────┴─────┘
             152 rows omitted

julia> 

[iblislin]

ah, fine

Maybe we could have following public APIs:

time_slots(start, stop, interval; from=Time(8,45), to=Time(13,45)) # could returns a lazy iterator that do not materialized the vector at first?

time_slots(range_object; from=..., to=...)

when(ta, time_slots(; from=..., to=...)) # this returns a new TimeArray

If you have more ideas, fine to put them here.

[anurag-mds]

Thanks!,
I'll implement all three. The when variant is especially useful for filtering existing TimeArrays by time-of-day very composable with real data. Will start with basic time_slots and share a draft PR soon