summaryrefslogtreecommitdiff
path: root/scraper/util.py
diff options
context:
space:
mode:
authoradamhrv <adam@ahprojects.com>2019-03-08 21:56:41 +0100
committeradamhrv <adam@ahprojects.com>2019-03-08 21:56:41 +0100
commit69c4157f599846173362d2a450f58b5f58c0d8d8 (patch)
tree342e4e295820f6daaa130032fdb10bdf4ee1f2b7 /scraper/util.py
parent5bcbb24791062b6524c06e2bf2bd4a61965ba6e8 (diff)
parent377279bd1fb43d8752f7a664fe0f5c2ae20c0bc3 (diff)
Merge branch 'master' of github.com:adamhrv/megapixels_dev
Diffstat (limited to 'scraper/util.py')
-rw-r--r--scraper/util.py4
1 files changed, 2 insertions, 2 deletions
diff --git a/scraper/util.py b/scraper/util.py
index fdbc0534..830dbe8b 100644
--- a/scraper/util.py
+++ b/scraper/util.py
@@ -386,10 +386,10 @@ def parallelize(func, rows):
with Pool(processes=processCount) as pool:
pool.starmap(func, rows, chunksize)
-def fetch_paper(s2, paper_id):
+def fetch_paper(s2, paper_id, freshen=False):
os.makedirs('./datasets/s2/papers/{}/{}'.format(paper_id[0:2], paper_id), exist_ok=True)
paper_fn = './datasets/s2/papers/{}/{}/paper.json'.format(paper_id[0:2], paper_id)
- if os.path.exists(paper_fn):
+ if os.path.exists(paper_fn) and not freshen:
return read_json(paper_fn)
print(paper_id)
paper = s2.paper(paper_id)