diff options
Diffstat (limited to 'scraper/s2-raw-papers.py')
| -rw-r--r-- | scraper/s2-raw-papers.py | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/scraper/s2-raw-papers.py b/scraper/s2-raw-papers.py index 612c8099..67ff0b65 100644 --- a/scraper/s2-raw-papers.py +++ b/scraper/s2-raw-papers.py @@ -18,10 +18,10 @@ def fetch_raw_papers(fn): lines = read_csv(fn, keys=False) parallelize(fetch_raw_paper, lines) -def fetch_raw_paper(paper_id): +def fetch_raw_paper(paper_id, freshen=False): os.makedirs(make_raw_paper_path(paper_id), exist_ok=True) paper_fn = make_raw_paper_fn(paper_id) - if os.path.exists(paper_fn): + if os.path.exists(paper_fn) and not freshen: paper = read_json(paper_fn) else: paper = s2.raw_paper(paper_id) @@ -33,7 +33,7 @@ def fetch_raw_paper(paper_id): if 'responseType' in paper and paper['responseType'] == 'CANONICAL': write_json(paper_fn, paper) - paper = s2.raw_paper(data['canonicalId']) + paper = s2.raw_paper(paper['canonicalId']) paper_fn = make_raw_paper_fn(paper_id) return paper |
