yvictor / polars_ta_extension Goto Github PK

Currently period accepts only i32 inputs. It should support f32 type. The original talib / talib python have no such restriction.

Environment

• Operating System: macOS (Apple M1)
• Python Version: Python 3.12.3
• pip Version: pip 24.0

Description

I am trying to install polars-talib==0.1.2 on my M1 Mac but am encountering an error indicating that there are no matching distributions found. This suggests that there might not be pre-built wheels available for ARM architecture.

Steps to Reproduce

1. Run pip install polars-talib==0.1.2
2. Encounter error messages indicating incompatibility.

× No solution found when resolving dependencies: ╰─▶ Because polars-talib==0.1.2 is unusable because no wheels are available with a matching Python implementation and you require polars-talib==0.1.2, we can conclude that the requirements are unsatisfiable.

Possible Solution

It would be highly beneficial for Mac users with ARM architecture if ARM-compatible wheels could be made available.

Additional Context

Really appreciate your help in making this library accessible on different architectures.

ta-lib-python is supposed to already natively support polars dataframes?
TA-Lib/ta-lib-python#471

Just out of interest: What is the difference to this library?

So, in my project I did cargo add talib-sys
The build fails with this error:

   Compiling talib-sys v0.1.0
error: could not find native static library `ta_lib`, perhaps an -L flag is missing?

error: could not compile `talib-sys` (lib) due to 1 previous error

I figured that the solution is to find out the location. In my case:
~/.cargo/registry/src/index.crates.io-6f17d22bba15001f/talib-sys-0.1.0

Then, i cloned this repo into /tmp, built it myself and did:
cp /tmp/polars_ta_extension/talib-sys/dependencies ~/.cargo/registry/src/index.crates.io-6f17d22bba15001f/talib-sys-0.1.0

That allowed me to build properly.
Maybe the cargo version is not up to date?

using this script (based on the example [here]:

df_data = (df_data
.with_columns(pl.col("LDPS_w").ta.wma(lookback_period).over(["ticker", "timeframe"]).alias("LDPS_w")))

I get this error:

"polars.exceptions.ComputeError: the plugin failed with message: Could not compute indicator, err: TA_OUT_OF_RANGE_END_INDEX"

could you please help me understand what I should be doing?

apologies, my mistake, works fine

Just a warning I got. May dig into and make PR myself but just making note