Source code for indra.literature.crossref_client

from __future__ import absolute_import, print_function, unicode_literals
from builtins import dict, str
import os
import logging
import requests
from indra.config import get_config
from indra.literature import pubmed_client
# Python3
    from functools import lru_cache
# Python2
except ImportError:
    from functools32 import lru_cache

logger = logging.getLogger(__name__)

crossref_url = ''
crossref_search_url = ''

def get_api_key():
    return get_config('CROSSREF_CLICKTHROUGH_KEY')

[docs]@lru_cache(maxsize=100) def get_metadata(doi): """Returns the metadata of an article given its DOI from CrossRef as a JSON dict""" url = crossref_url + 'works/' + doi res = requests.get(url) if res.status_code != 200:'Could not get CrossRef metadata for DOI %s, code %d' % (doi, res.status_code)) return None raw_message = res.json() metadata = raw_message.get('message') return metadata
def get_publisher(doi): metadata = get_metadata(doi) if metadata is None: return None publisher = metadata.get('publisher') return publisher def get_url(doi): metadata = get_metadata(doi) if metadata is None: return None url = metadata.get('URL') return url def get_license_links(doi): metadata = get_metadata(doi) if metadata is None: return None licenses = metadata.get('license') if licenses is None: return None urls = [l.get('URL') for l in licenses] return urls
[docs]def doi_query(pmid, search_limit=10): """Get the DOI for a PMID by matching CrossRef and Pubmed metadata. Searches CrossRef using the article title and then accepts search hits only if they have a matching journal ISSN and page number with what is obtained from the Pubmed database. """ # Get article metadata from PubMed pubmed_meta_dict = pubmed_client.get_metadata_for_ids( [pmid], get_issns_from_nlm=True) if pubmed_meta_dict is None or pubmed_meta_dict.get(pmid) is None: logger.warning('No metadata found in Pubmed for PMID%s' % pmid) return None # The test above ensures we've got this now pubmed_meta = pubmed_meta_dict[pmid] # Check if we already got a DOI from Pubmed itself! if pubmed_meta.get('doi'): return pubmed_meta.get('doi') # Check for the title, which we'll need for the CrossRef search pm_article_title = pubmed_meta.get('title') if pm_article_title is None: logger.warning('No article title found in Pubmed for PMID%s' % pmid) return None # Get the ISSN list pm_issn_list = pubmed_meta.get('issn_list') if not pm_issn_list: logger.warning('No ISSNs found in Pubmed for PMID%s' % pmid) return None # Get the page number pm_page = pubmed_meta.get('page') if not pm_page: logger.debug('No page number found in Pubmed for PMID%s' % pmid) return None # Now query CrossRef using the title we've got url = crossref_search_url params = {'q': pm_article_title, 'sort': 'score'} try: res = requests.get(crossref_search_url, params) except requests.exceptions.ConnectionError as e: logger.error('CrossRef service could not be reached.') logger.error(e) return None except Exception as e: logger.error('Error accessing CrossRef service: %s' % str(e)) return None if res.status_code != 200:'PMID%s: no search results from CrossRef, code %d' % (pmid, res.status_code)) return None # This is to handle a special case where we get a 200 status code but # an error message in res.text that precludes json loading try: raw_message = res.json() except ValueError: return None mapped_doi = None # Iterate over the search results, looking up XREF metadata for result_ix, result in enumerate(raw_message): if result_ix > search_limit:'PMID%s: No match found within first %s results, ' 'giving up!' % (pmid, search_limit)) break xref_doi = result['doi'] # Get the XREF metadata using the DOI xref_meta = get_metadata(xref_doi) if xref_meta is None: continue xref_issn_list = xref_meta.get('ISSN') xref_page = xref_meta.get('page') # If there's no ISSN info for this article, skip to the next result if not xref_issn_list: logger.debug('No ISSN found for DOI %s, skipping' % xref_doi) continue # If there's no page info for this article, skip to the next result if not xref_page: logger.debug('No page number found for DOI %s, skipping' % xref_doi) continue # Now check for an ISSN match by looking for the set intersection # between the Pubmed ISSN list and the CrossRef ISSN list. matching_issns = set(pm_issn_list).intersection(set(xref_issn_list)) # Before comparing page numbers, regularize the page numbers a bit. # Note that we only compare the first page number, since frequently # the final page number will simply be missing in one of the data # sources. We also canonicalize page numbers of the form '14E' to # 'E14' (which is the format used by Pubmed). pm_start_page = pm_page.split('-')[0].upper() xr_start_page = xref_page.split('-')[0].upper() if xr_start_page.endswith('E'): xr_start_page = 'E' + xr_start_page[:-1] # Now compare the ISSN list and page numbers if matching_issns and pm_start_page == xr_start_page: # We found a match! mapped_doi = xref_doi break # Otherwise, keep looking through the results... # Return a DOI, or None if we didn't find one that met our matching # criteria return mapped_doi