Materials Project: get all references for a material

get_references.py

# mktmpenv
# pip install numpy
# pip install scipy
# pip install pymongo
# pip install monty
# pip install -U setuptools
# pip install -e git+git@github:materialsproject/mpworks#egg=mpworks
import os
from pymongo import MongoClient
from monty.serialization import loadfn
from mpworks.snl_utils.snl_mongo import SNLMongoAdapter
sma = SNLMongoAdapter.from_file(
    os.path.join(os.environ['DB_LOC'], 'snl_db.yaml')
)

def get_mp_references(mp_id='mp-568345', db_yaml='materials_db.yaml'):
    """get all references in MP for specific mp_id"""
    # load materials database
    config = loadfn(os.path.join(os.environ['DB_LOC'], db_yaml))
    client = MongoClient(config['host'], config['port'], j=False)
    client[config['db']].authenticate(config['username'], config['password'])
    materials = client[config['db']].materials
    # get snlgroup_id_final for mp_id
    snlgroup_id_final = materials.find_one(
        {'task_id': mp_id}, {'_id': 0, 'snlgroup_id_final': 1}
    )['snlgroup_id_final']
    # get all_snl_ids for snlgroup_id_final
    all_snl_ids = sma.database.snlgroups.find_one(
        {'snlgroup_id': snlgroup_id_final}, {'_id': 0, 'all_snl_ids': 1}
    )['all_snl_ids']
    # get cursor for snls in all_snl_ids
    cursor = sma.database.snl.find(
        {'snl_id': {'$in': all_snl_ids}}, {'_id': 0, 'about.references': 1}
    )
    # compile references and remove empty strings
    references = filter(None, [doc['about']['references'] for doc in cursor])
    # return as single bibtex string (can be parse by user via pybtex)
    # TODO: should we fix double entries and bibtex keys?
    # TODO: should we check validity of bibtex code via pybtex?
    return '\n'.join(references)

if __name__ == '__main__':
    print get_mp_references()