summaryrefslogtreecommitdiff
path: root/s2-search.py
diff options
context:
space:
mode:
authoradamhrv <adam@ahprojects.com>2018-11-04 21:54:00 +0100
committeradamhrv <adam@ahprojects.com>2018-11-04 21:54:00 +0100
commit9bcba0d02aafb34a5a9ca3db2f894f1fc95401c0 (patch)
tree3dcaf94563498c15b56d51efc62750d0be72e01a /s2-search.py
parentef45f3c93ffd39b57ee56db74a95f9d2dae074a8 (diff)
parent0dc3e40434c23e4d48119465f39b03bf35fb56bd (diff)
.
Diffstat (limited to 's2-search.py')
-rw-r--r--s2-search.py46
1 files changed, 46 insertions, 0 deletions
diff --git a/s2-search.py b/s2-search.py
new file mode 100644
index 00000000..b9e8db2d
--- /dev/null
+++ b/s2-search.py
@@ -0,0 +1,46 @@
+import os
+import sys
+import csv
+import subprocess
+import time
+import random
+import re
+import simplejson as json
+import click
+from s2 import SemanticScholarAPI
+from util import *
+
+'''
+s2 search API format:
+results
+matchedAuthors
+matchedPresentations
+query
+querySuggestions
+results
+stats
+totalPages
+totalResults
+'''
+
+@click.command()
+@click.option('--index', '-n', default=0, help='Index of CSV.')
+def fetch_entries(index):
+ keys, lines = read_citation_list(index)
+ s2 = SemanticScholarAPI()
+ for line in lines:
+ label = line[0]
+ title = re.sub(r'[^-0-9a-zA-Z ]+', '', line[1])
+ entry_fn = './datasets/s2/entries/{}.json'.format(title)
+ if not os.path.exists(entry_fn):
+ results = s2.search(title)
+ write_json(dump_fn, results)
+ if len(results['results']) == 0:
+ print("No results for {}".format(title))
+ else:
+ print(title)
+ write_json(entry_fn, results['results'][0])
+ time.sleep(random.randint(10, 20))
+
+if __name__ == '__main__':
+ fetch_entries()