Begin PHOENIX converter

.gitignore

@@ -137,6 +137,7 @@ notebooks/
 
 # ignore data files
 data/
+*.parquet.gz
 
 # ignore vscode settings
 .vscode/settings.json

src/gollum/V2/conversion/phoenix_conversion.py

@@ -0,0 +1,32 @@
+from astropy.io import fits
+from numpy import float64
+from pandas import DataFrame
+from pathlib import Path
+from re import search
+from tqdm import tqdm
+
+def main():
+    fluff = 'PHOENIX-ACES-AGSS-COND-2011'
+    p = Path('~/data/PHOENIX').expanduser()
+    wavelength = fits.open(p / f'WAVE_{fluff}.fits')[0].data.astype(float64)
+    p /= fluff
+    fluxes = dict()
+    for f in tqdm(map(lambda x: x.name[:-39], sorted(p.glob('*/*')))):
+        T = int(search(r'\d{5}', f)[0])
+        GZ = search(r'(\d\.\d{2})([+-]\d\.\d)', f)
+        G = float(GZ[1])
+        Z = float(GZ[2])
+        A = float(x[1]) if (x := search(r'Alpha=([+-]\d\.\d{2})', f)) else 0.0
+        z_str = f'Z{Z:+.1f}' if Z else 'Z-0.0'
+        a_str = f'.Alpha={A:+.2f}' if A else ''
+        flux = fits.open(p / f'{z_str}{a_str}/{f}.{fluff}-HiRes.fits')[0].data.astype(float64)
+        fluxes[str((T, G, Z, A))] = flux
+    DataFrame(index=wavelength, data=fluxes, dtype='float64[pyarrow]').rename_axis('wavelength').to_parquet('PHOENIX.parquet.br', compression='brotli')
+
+    #flux = fits.open(p / f'Z-0.0/lte02300+0.50-0.0.{fluff}-HiRes.fits')[0].data.astype(np.float64)
+    #df = pd.DataFrame(index=wavelength, data={'(2300, 0.5, 0.0, 0.0)': flux}).rename_axis('wavelength')
+    #df.to_parquet('test.parquet.gz', compression='gzip')
+
+if __name__ == "__main__":
+    main()
+

src/gollum/V2/phoenix.py

@@ -38,13 +38,4 @@ def __init__(self, teff: int = None, logg: float = None, Z: float = None, alpha:
 
         path = get_key(self._env(), 'PHOENIX_PATH') or 'zenodo link'
         df = read_parquet(path, 'pyarrow', [str(point)], filters=[('wavelength', '>=', extent[0]), ('wavelength', '<=', extent[1])] if extent else None)
-        super(PHOENIXSpectrum, self).__init__(df.index, df[str(point)])
-
-
-import pandas as pd
-#df = pd.DataFrame(index=[1, 2, 3], data={'(6000, 4.5, 0.0, 0.0)': [4, 5, 6]}).rename_axis('wavelength')
-#df.to_parquet('test.parquet.gz', compression='gzip')
-
-x = PHOENIXSpectrum(6000, 4.5, 0, 0)
-print(x)
-
+        super(PHOENIXSpectrum, self).__init__(df.index, df[str(point)])