pyodide | Python ported to WebAssembly

CryptoJS only supports symmetric encryption and therefore not ECDSA. WebCrypto supports ECDSA, but not secp256k1.
WebCrypto has the advantage that it is supported by all major browsers. Since you can use other curves according to your comment, I will describe a solution with a curve supported by WebCrypto.
Otherwise, sjcl would also be an alternative, a pure JavaScript library that supports ECDSA and especially secp256k1, s.here.

WebCrypto is a low level API that provides the functionality you need like key generation, key export and signing. Regarding ECDSA WebCrypto supports the curves P-256 (aka secp256r1), P-384 (aka secp384r1) and p-521 (aka secp521r1). In the following I use P-256.

The following JavaScript code generates a key pair for P-256, exports the public key in X.509/SPKI format, DER encoded (so it can be sent to the Python site), and signs a message:

So, I'm still not sure why directly read_excel or Excelfile functions don't work (especially since they just call openpyxl anyway), but I was able to get it to work by opening the file using openpyxl and then converting it to a dataframe. Working code below:

Short answer is that downloading files with Python currently won't work in Pyodide because http.client, requests etc require POSIX sockets which are not supported in the browser VM.

It's curious that nltk.download doesn't error though -- it should have.

The workaround is to manually download the needed resources, for instance, using the JavaScript fetch API as illustrated in this comment;

You are loading Pyodide multiple times by including the loadPyodide call in the main function, which gives an error for your subsequent main calls. Move loadPyodide outside main should resolve the issue:

You're probably using a newer version of pyodide. Since 0.18.0, you should use loadPyodide with the pyodide base url instead, which gives the pyodide object in the callback:

Assuming you are using Chrome, it's very likely a bug in Chrome 89 - 90, which should be fixed in Chrome 91.

For more details see: pyodide#1507 and bugs.chromium.org#1200031

Basically, you can implement your own function for printing. However, if you want to use exactly print you can override the function from JS:

It's likely a pickle portability issue. Pickles should be portable between architectures¹, here amd64 and wasm32 however they are not portable across package versions. This means that package versions should be identical between the environement where you train your model and where you do the inference (pyodide).

pyodide 0.16.1 includes Python 3.8.2, scipy 0.17.1 and scikit-learn 0.22.2. Which unfortunately means that you will have to build that version of scipy (and possibly numpy) from sources to train the model, since a Python 3.8 binary wheel doesn't exist for such an outdated version of scipy. In the future this should be resolved with pyodide#1293.

The particular error you are getting is likely due to scipy.sparse version mimatch see scipy#6533

¹Though, tree based models in scikit-learn at present are not portable across architectures, and so won't unpickle in pyodide. This is known bug that should be fixed (scikit-learn#19602)

sometimes I get following errors on iOS v13

You should open a Github issue about it. It's likely a bug somewhere. You could try building with TOTAL_MEMORY multiplied by 2, to see if it helps.

How can I build for very minimum usage not for scientific projects to make it run on iOS v12 also?

If you don't load those packages, they have no impact on the main pyodide package, which consists only of CPython + Python/JS type conversion code.