uio.tasks.reconfirming_stellar_parameters

  1"""
  2Reconfirming stellar parameters, enriching original datasets
  3with additional data from various data sources.
  4"""
  5
  6import pandas
  7import numpy
  8# import json
  9
 10from typing import Optional, List, Dict
 11
 12from ..utility.files import pickle
 13from ..utility.databases import tap
 14from ..utility.databases import simbad
 15
 16
 17def lookForParametersInGaia(
 18    originalTable: pandas.DataFrame,
 19    adqlTable: str,
 20    adqlParameters: List[str],
 21    simbadIDversion: Optional[str] = None
 22) -> pandas.DataFrame:
 23    """
 24    Looking for specified parameters in GAIA database:
 25
 26    1. Opens a pickle file with original [Pandas](https://pandas.pydata.org) table;
 27    2. Extracts unique list of star names;
 28    3. Gets their GAIA IDs from Simbad database;
 29    4. Queries GAIA database for given parameters;
 30    5. Adds found parameters to the original table as new columns.
 31
 32    Example:
 33
 34    ``` py
 35    from uio.utility.files import pickle
 36    from uio.tasks import reconfirming_stellar_parameters
 37
 38    originalTable = pickle.openPickleAsPandasTable("./data/systems-528n.pkl")
 39    tbl = reconfirming_stellar_parameters.lookForParametersInGaia(
 40        originalTable,
 41        "gaiadr3.astrophysical_parameters",
 42        [
 43            "age_flame",
 44            "logg_gspphot",
 45            "mass_flame",
 46            "mh_gspphot",
 47            "mh_gspspec",
 48            "radius_flame",
 49            "radius_gspphot",
 50            "teff_esphs",
 51            "teff_espucd",
 52            "teff_gspphot",
 53            "teff_gspspec",
 54            "teff_msc1",
 55            "ew_espels_halpha",
 56            "ew_espels_halpha_model"
 57        ],
 58        "dr3"
 59    )
 60    ```
 61
 62    You might need to provide `simbadIDversion` parameter (*the `dr3` value
 63    here*) if SIMBAD (`uio.utility.databases.simbad.findIdentificatorFromAnotherCatalogue`)
 64    returns IDs like `DR3 2135237601028549888` and you need to get exactly
 65    the DR3 ones.
 66
 67    As a result, your original table `tbl` will be enriched with additional
 68    columns according to the list of provided astrophysical parameters.
 69    """
 70
 71    starNames = originalTable["star_name"].unique()
 72
 73    print("\nGetting GAIA IDs from SIMBAD...\n")
 74
 75    stars: Dict[str, Optional[str]] = {}
 76    for star in starNames:
 77        oid = simbad.findIdentificatorFromAnotherCatalogue(star, "gaia", simbadIDversion)
 78        if oid is None:
 79            print(f"- [WARNING] did not GAIA ID for [{star}]")
 80        else:
 81            print(f"- found GAIA ID for [{star}]: {oid}")
 82            stars[star] = oid
 83
 84    # print(json.dumps(stars, indent=4))
 85
 86    print("\nLooking for parameters in GAIA...\n")
 87
 88    for parameter in adqlParameters:
 89        originalTable[parameter] = numpy.array(numpy.NaN, dtype=float)
 90
 91    foundCnt = 0
 92    for star in stars:
 93        gaiaID = stars[star]
 94        print(f"- {star} | {gaiaID}...")
 95        resultsGAIA = tap.queryService(
 96            tap.getServiceEndpoint("gaia"),
 97            " ".join((
 98                f"SELECT {', '.join(adqlParameters)}",
 99                f"FROM {adqlTable}",
100                f"WHERE source_id = {gaiaID}"
101            ))
102        )
103        if resultsGAIA is None:
104            print(f"- [WARNING] did not found anything in GAIA for [{gaiaID}]")
105        else:
106            tbl = resultsGAIA.to_table().to_pandas()
107            foundCnt += 1
108            if len(tbl) > 1:
109                print(
110                    " ".join((
111                        "- [WARNING] GAIA has more than one record",
112                        f"for ID [{gaiaID}], will take only the first one"
113                    ))
114                )
115            # add found values to the new columns in the original table
116            for parameter in adqlParameters:
117                originalTable.loc[
118                    originalTable["star_name"] == star,
119                    parameter
120                ] = tbl.head(1)[parameter][0]
121
122    print(f"\nFound parameters for {foundCnt}/{len(stars)} stars\n")
123
124    return originalTable