diff options
| author | Jules Laplace <julescarbon@gmail.com> | 2019-02-18 13:28:06 +0100 |
|---|---|---|
| committer | Jules Laplace <julescarbon@gmail.com> | 2019-02-18 13:28:06 +0100 |
| commit | b28f65ad5016ba3c3c9f973bd2a64ea3c8a3f84c (patch) | |
| tree | 66cd5ba28f0b3687871d4f7f65078329463fd00c /scraper/s2-final-report.py | |
| parent | 98d7c92f0050ca92ec6b72758a6d5b680f9e4d7f (diff) | |
get institutions from spreadsheet
Diffstat (limited to 'scraper/s2-final-report.py')
| -rw-r--r-- | scraper/s2-final-report.py | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/scraper/s2-final-report.py b/scraper/s2-final-report.py index f08bc748..eff9a736 100644 --- a/scraper/s2-final-report.py +++ b/scraper/s2-final-report.py @@ -12,6 +12,7 @@ DIR_PUBLIC_CITATIONS = "../site/datasets/final" DIR_UNKNOWN_CITATIONS = "../site/datasets/unknown" addresses = AddressBook() +paper_location_lookup = fetch_google_lookup('paper_locations') @click.command() def s2_final_report(): @@ -90,7 +91,7 @@ def process_single_paper(row, paper_id, addresses, aggregate_citations, unknown_ # res['authors'] = ', '.join(paper.authors) # res['citations'] = [] - paper_institutions = load_institutions(paper.paper_id) + paper_institutions = load_institutions(paper.paper_id, paper_location_lookup) paper_address = None for inst in sorted(paper_institutions, key=operator.itemgetter(1)): # print(inst[1]) @@ -119,7 +120,7 @@ def process_single_paper(row, paper_id, addresses, aggregate_citations, unknown_ if citation is None or citation.data is None: print("Citation missing! {}".format(cite['paperId'])) continue - institutions = load_institutions(citationId) + institutions = load_institutions(citationId, paper_location_lookup) geocoded_addresses = [] geocoded_institutions = [] institution = '' |