scholar_scrape_example.py

                                                                                                                                                                                                                          
#!pip install -e +git https://github.com/ckreibich/scholar.py                                                                                                                                                                                 
import sys
from scholar_scrape import scholar
import pandas as pd

unicode = str # pylint: disable-msg=W0622                                                                                                                                                                                                     
encode = lambda s: unicode(s) # pylint: disable-msg=C0103                                                                                                                                                                                ''' 


def csv(querier, header=False, sep='|'):
    articles = querier.articles
    results = []
    for art in articles:
        result = art.as_csv(header=header, sep=sep)
        results.append(result)
        print(encode(result))
        header = False
    return results

def citation_export(querier):
    articles = querier.articles
    for art in articles:
        print(art.as_citation() + '\n')
    return articles


def search_author(author):
    # from https://github.com/ckreibich/scholar.py/issues/80                                                                                                                                                                                  

    querier = scholar.ScholarQuerier()
    settings = scholar.ScholarSettings()
    querier.apply_settings(settings)
    query = scholar.SearchScholarQuery()

    query.set_words(str('author:')+author)
    querier.send_query(query)
    results0 = csv(querier)
    results1 = citation_export(querier)
    links = [ a.attrs['url'][0] for a in querier.articles if a.attrs['url'][0] is not None ]
    return results0, results1, links

links, results0, _ = search_author('R Gerkin')
print(links, results0)