query_glottolog.py

from pyglottolog.api import Glottolog

# 'full' is a pandas dataframe with glottocodes

api = Glottolog('/Users/macbook/tmp/glottolog')

gltc_temp = {}
gltc_err = set()

for i in range(full.shape[0]):
    r = full.iloc[i,:]
    gltc = r['Borrowing language (glottocode)']
    if gltc in gltc_temp or gltc in gltc_err:
        continue
    else:
        lang = api.languoid(gltc)
        if lang is None:
            gltc_err.add(gltc)
            print('Check gltc: %s' % gltc)
        else:
            try:
                macroarea = api.languoid(gltc).macroareas[0].name
            except:
                macroarea = ''
            try:
                family = lang.family.name
            except:
                family = ''
            gltc_temp[gltc] = {
                'name': lang.name,
                'family': family,
                'lat': lang.latitude,
                'lon': lang.longitude,
                'macroarea': macroarea
            }

for i in range(full.shape[0]):
    r = full.iloc[i,:]
    gltc = r['Borrowing language (glottocode)']
    if gltc in gltc_err or len(gltc) < 3:
        continue
    r['Borrowing language (name)'] = gltc_temp[gltc]['name']
    r['Borrowing language family'] = gltc_temp[gltc]['family']
    r['Borrowing language latitude'] = gltc_temp[gltc]['lat']
    r['Borrowing language longitude'] = gltc_temp[gltc]['lon']
    r['Macroregion'] = gltc_temp[gltc]['macroarea']

full.to_csv('full_df.csv', sep = '\t', index = False)