citation_grapher.py

import requests
from edit_queue import EditQueue
from site_credentials import *

class CitationGrapher:
    def __init__(self, eq):
        self.eq = eq

    def get_entitydata(self, manifest):
        url = 'https://www.wikidata.org/w/api.php?format=json&action=wbgetentities&ids='
        for wikidata_id in manifest.keys():
            url += wikidata_id + '|'
        url = url[:-1]  # remove trailing pipe

        try:
            r = requests.get(url).json()['entities']
        except:
           return

        for wikidata_id, blob in r.items():
            # (pmcid, cites, retrieve_date)
            yield wikidata_id, blob, manifest[wikidata_id][0], manifest[wikidata_id][1], manifest[wikidata_id][2]

    def process_manifest(self, manifest):
        # Screen against current values on Wikidata
        for relevant_item, blob, relevant_external_id, raw_cites, retrieve_date in self.get_entitydata(manifest):

            cites = []

            # Creating a convenient data object for keeping track of existing
            # 'cites' claims and their references on a given Wikidata item
            references = {}
            if 'claims' in blob:
                if 'P2860' in blob['claims']:
                    extant_claims = blob['claims']['P2860']
                    for claim in extant_claims:
                        if 'mainsnak' in claim:
                            if 'datavalue' in claim['mainsnak']:
                                extant_cited_item = claim['mainsnak']['datavalue']['value']['id']
                                    
                                if 'references' in claim:
                                    for reference in claim['references']:
                                        snaks = {}
                                        for prop_nr, values in reference['snaks'].items():
                                            snaks[prop_nr] = [v['datavalue'] for v in values]

                                        references[extant_cited_item] = snaks

            # Don't generate a statement if the statement already exists and
            # features a fully filled out citation. *Do* otherwise generate
            # a statement, even if the statement exists, if it has a lousy,
            # not-filled-out citation.
            for cited_item in raw_cites:
                generate_statement = True
                if cited_item in references:
                    if 'P248' in references[cited_item] \
                    and 'P813' in references[cited_item] \
                    and 'P854' in references[cited_item]:
                        for x in references[cited_item]['P248']:
                            # stated in: PubMed Central or Crossref
                            if x['value']['id'] in ['Q5188229', 'Q229883']:
                                generate_statement = False

                if generate_statement is True:
                    cites.append(cited_item)

            if len(cites) > 0:
                self.eq.post(relevant_item, relevant_external_id, retrieve_date, cites)