getSAOB.py

import requests
file = open("saob2020.csv", "r")

def createQSurl(SAOBid):
#    https: // quickstatements.toolforge.org /  # /v1=L349183%7CP8478%7C%22A_0787-0228.n0A3%22
    urlbase = "https://quickstatements.toolforge.org/#/v1=LEXEME"
    urlP8478 = f"%7CP8478%7C%22{SAOBid.strip()}%22"
    return urlbase + urlP8478


def main():
    print("start")
    for x in file:
        word = x.split(",")[1]
        wordtype = x.split(",")[2]
        wordSAOB = x.split(",")[4]
        SAOBid = wordSAOB.replace("https://svenska.se/saob/?id=","").replace("&pz=5","")
        # https://svenska.se/saob/?id=A_0001-0018.TWNA&pz=5
        if wordtype == "subst":
            print(wordtype, word, wordSAOB.strip())
            print ("\t",createQSurl(SAOBid))
            params = dict (
                action='wbsearchentities',
                format='json',
                language='sv',
                uselang='sv',
                type='lexeme',
                search=word
                )
            response = requests.get('https://www.wikidata.org/w/api.php?', params).json()
            for l in response.get('search'):
                print(l)
                language = l['match']['language']
                if language == "sv":
                    id = l['id']
                    url = "http:" + l['url']
                    label = l['label']
                    print("\t",id,url,label,language,"\n")


if __name__ == '__main__':
    main()