summaryrefslogtreecommitdiff
path: root/scraper/util.py
diff options
context:
space:
mode:
authorjules@lens <julescarbon@gmail.com>2019-03-30 16:02:16 +0100
committerjules@lens <julescarbon@gmail.com>2019-03-30 16:02:16 +0100
commit4772fbdbfe18c5d0e8abd87ac39ebbed5a306b6f (patch)
treef8fee86c1cd08993579ab0353c042557fd0db3c7 /scraper/util.py
parentf99d2d7b13f03e4cc29a8fa9f763dfd5a70102bd (diff)
FIX institution lookup from paper thingie
Diffstat (limited to 'scraper/util.py')
-rw-r--r--scraper/util.py6
1 files changed, 4 insertions, 2 deletions
diff --git a/scraper/util.py b/scraper/util.py
index 65194fd1..05b01fa7 100644
--- a/scraper/util.py
+++ b/scraper/util.py
@@ -308,7 +308,10 @@ class AddressBook (object):
for index, line in enumerate(data):
if line[0] == line[1] or line[0] not in entities:
entities[line[0]] = index
- name = line[1].lower().strip()
+ name = line[0].lower().strip().strip(string.digits)
+ if name not in lookup:
+ lookup[name] = line[0]
+ name = line[1].lower().strip().strip(string.digits)
if name not in lookup:
lookup[name] = line[0]
self.data = data
@@ -347,7 +350,6 @@ def load_institutions(paperId, paper_location_lookup={}):
if paperId in paper_location_lookup:
items = paper_location_lookup[paperId]
result = []
- print(items)
if items['validate_1'] is True or items['validate_1'] == 'TRUE':
result.append(['', items['institution_1'], '', ''])
if items['validate_2'] is True or items['validate_2'] == 'TRUE':