summaryrefslogtreecommitdiff
path: root/scraper/s2-fetch-pdf.py
diff options
context:
space:
mode:
Diffstat (limited to 'scraper/s2-fetch-pdf.py')
-rw-r--r--scraper/s2-fetch-pdf.py5
1 files changed, 2 insertions, 3 deletions
diff --git a/scraper/s2-fetch-pdf.py b/scraper/s2-fetch-pdf.py
index 72ca4ca8..61574b90 100644
--- a/scraper/s2-fetch-pdf.py
+++ b/scraper/s2-fetch-pdf.py
@@ -25,13 +25,12 @@ def fetch_pdf(paper_id, url):
pdf_fn = make_pdf_fn(paper_id)
txt_fn = make_txt_fn(paper_id)
if os.path.exists(pdf_fn) or os.path.exists(txt_fn):
- # return read_json(pdf_fn)
- return
+ return None
size = s2.fetch_file(url, pdf_fn)
if size is None:
print("{} empty?".format(paper_id))
return None
- print("{} {} kb".format(paper_id, int(size / 1024)))
+ print("{} {} kb {}".format(paper_id, int(size / 1024), url))
def make_pdf_path(paper_id):
return './datasets/s2/pdf/{}/{}'.format(paper_id[0:2], paper_id)