two small issues with the docker instructions

[abate]

Hello devs :)

I encountered two minor problem in the Dockerfile instruction provided :

• the first is easily fix: docker build -t alwrity -f "Getting Started/Option_3_Docker_Install/Dockerfile" . the path of the docker file changed

The second here, I was not able to solve it. I tried to change the base image ( upgrade to 3.12-slim ) and change the requirements.txt file, but I didn't manage.

Not sure what is wrong. There must be some requirements that keep pulling in the wrong version of panda-ta

ImportError: cannot import name 'NaN' from 'numpy' (/usr/local/lib/python3.12/site-packages/numpy/__init__.py)
Traceback:

File "/app/alwrity.py", line 81, in <module>
    from lib.utils.ui_setup import setup_ui, setup_alwrity_ui
File "/app/lib/utils/ui_setup.py", line 10, in <module>
    from lib.utils.alwrity_utils import ai_social_writer
File "/app/lib/utils/alwrity_utils.py", line 11, in <module>
    from lib.ai_writers.ai_finance_report_generator.ai_financial_dashboard import get_dashboard
File "/app/lib/ai_writers/ai_finance_report_generator/ai_financial_dashboard.py", line 22, in <module>
    from ...ai_web_researcher.finance_data_researcher import get_finance_data, get_fin_options_data
File "/app/lib/ai_web_researcher/finance_data_researcher.py", line 4, in <module>
    import pandas_ta as ta
File "/usr/local/lib/python3.12/site-packages/pandas_ta/__init__.py", line 116, in <module>
    from pandas_ta.core import *
File "/usr/local/lib/python3.12/site-packages/pandas_ta/core.py", line 18, in <module>
    from pandas_ta.momentum import *
File "/usr/local/lib/python3.12/site-packages/pandas_ta/momentum/__init__.py", line 34, in <module>
    from .squeeze_pro import squeeze_pro
File "/usr/local/lib/python3.12/site-packages/pandas_ta/momentum/squeeze_pro.py", line 2, in <module>
    from numpy import NaN as npNaN

It would also be great if you can also release the app as a docker image on dockerhub or similar services, and maybe
provide a ready to use docker-compose file for a quick local installation and testing.

regards

[uniqueumesh]

Hello @abate please give us some time to look into this issue. We will update you ASAP.

Regards

[AJaySi]

@abate @uniqueumesh
I am looking into it and will update and fix it by tomorrow.
Not easily reproducible at my end, but I can acknowledge the bug as relevant.

[AJaySi]

1. [requirements.txt] specifies:pandas-ta>=0.3.14b0
2. [numpy>=1.24.0]
3. The Dockerfile uses Python 3.12 for build, but Python 3.11-slim for runtime (this can cause subtle issues).
4. The error is due to an incompatibility between recent numpy versions (1.24+) and older pandas-ta versions, which try to import NaN from numpy (removed in numpy 1.24).

---

Solution:

Pin pandas-ta to a version compatible with numpy 1.24+ (e.g., pandas-ta>=0.3.14b0,!=0.3.14b0,>=0.3.15), or downgrade numpy to <1.24.
Also, use the same Python version in both build and runtime stages in Dockerfile.

---

requirements.txt: pandas-ta is now pinned to >=0.3.15 for compatibility with numpy >=1.24.0.
Dockerfile: Both build and runtime stages now use Python 3.12 for consistency.
docker-compose.yml: Added for easy local installation and testing. You can now run the app with:

cd "Getting Started/Option_3_Docker_Install"
docker-compose up --build

For Docker Hub release, you can build and push the image with:

docker build -t yourdockerhubusername/alwrity:latest -f "Getting Started/Option_3_Docker_Install/Dockerfile" .
docker push yourdockerhubusername/alwrity:latest