not able to install geopy package

[preetiverma525]

Hi

i need to install geopy package as i want to find the latitude and longitude of location in pyodide but unable to install
error no module find geopy

i already mention geopy in html page as well
pyodide.loadPackage(['matplotlib','pandas','micropip','geopy']).then(() => {
pyodide.runPython(pythonText);

     });

please help

[rth]

geopy is not included pyodide directly. You need to use micropip to install it from PyPi https://pyodide.readthedocs.io/en/latest/pypi.html (that should work as geopy has a pure python wheel, with a filename ending in none-any.whl). You also need to do this for all of its dependencies.

Don't hesitate to open issues or preferably post stack overflow questions with the pyodide tag if you have other questions.

[yxdragon]

many package is released as .gz (not whl), So I must download the source, and rebuild it as whl, and upload to my cloud disk.

could micropip support .gz directly?

[yxdragon]

some request: hdf and netcdf is two important science format, could pyodide include h5py and netcdf4?

[rth]

could micropip support .gz directly?

No unfortunately it's technically very difficult / not possible cf #1501 (comment). See also pypa/packaging-problems#496 discussion for a way to improve this situation in the ecosystem.

hdf and netcdf is two important science format, could pyodide include h5py and netcdf4?

Please open a separate issue. Assuming it's not to hard to build it should be possible. You can also use the zarr format as a more modern alternative (which is already packaged)

[yxdragon]

yes, zarr is more modern.
but for geography and meteorology, history data is very important, and most of history data is in hdf and netcdf.

[yxdragon]

@rth another question about hdf/netcdf/zarr. One file may contain many date, many bands. When we read it from native, the reader hold a pointer to the file. And read data we need just in time. But in pyodide, a web environment, is the "just in time read" possible? I worry about browser would load the whole file as memory stream when we upload the file. (hdf/nc/zarr file may be very large)

[rth]

Replied in #1582