summaryrefslogtreecommitdiff
path: root/scraper/s2-search.py
diff options
context:
space:
mode:
Diffstat (limited to 'scraper/s2-search.py')
-rw-r--r--scraper/s2-search.py6
1 files changed, 4 insertions, 2 deletions
diff --git a/scraper/s2-search.py b/scraper/s2-search.py
index ddecf2f9..169a8d19 100644
--- a/scraper/s2-search.py
+++ b/scraper/s2-search.py
@@ -7,7 +7,7 @@ import random
import re
import simplejson as json
import click
-from s2 import SemanticScholarAPI
+from s2 import SemanticScholarAPI, fetch_paper
from util import *
'''
@@ -33,6 +33,8 @@ def fetch_entries(index):
key = line[0]
name = line[1]
title = re.sub(r'[^-0-9a-zA-Z ]+', '', line[2])
+ if len(title) < 2:
+ continue
dump_fn = './datasets/s2/dumps/{}.json'.format(key)
entry_fn = './datasets/s2/entries/{}.json'.format(key)
result = None
@@ -49,7 +51,7 @@ def fetch_entries(index):
write_json(entry_fn, result)
if result:
paper_id = result['id']
- paper = fetch_paper(paper_id)
+ paper = fetch_paper(s2, paper_id)
citation_lookup.append([key, name, title, paper_id])
write_csv("datasets/citation_lookup.csv", keys=['key', 'name', 'title', 'paper_id'], rows=citation_lookup)