summaryrefslogtreecommitdiff
path: root/scraper/s2-merge-csv.py
diff options
context:
space:
mode:
authorAdam Harvey <adam@ahprojects.com>2018-12-23 01:37:03 +0100
committerAdam Harvey <adam@ahprojects.com>2018-12-23 01:37:03 +0100
commit4452e02e8b04f3476273574a875bb60cfbb4568b (patch)
tree3ffa44f9621b736250a8b94da14a187dc785c2fe /scraper/s2-merge-csv.py
parent2a65f7a157bd4bace970cef73529867b0e0a374d (diff)
parent5340bee951c18910fd764241945f1f136b5a22b4 (diff)
.
Diffstat (limited to 'scraper/s2-merge-csv.py')
-rw-r--r--scraper/s2-merge-csv.py28
1 files changed, 28 insertions, 0 deletions
diff --git a/scraper/s2-merge-csv.py b/scraper/s2-merge-csv.py
new file mode 100644
index 00000000..301f47ae
--- /dev/null
+++ b/scraper/s2-merge-csv.py
@@ -0,0 +1,28 @@
+import os
+import glob
+import time
+import simplejson as json
+import click
+import operator
+from util import *
+import random
+
+@click.command()
+@click.option('--path', '-d', default='report/institutions_geocoded', help='Path to CSVs')
+def s2_merge_csv(path):
+ print(path)
+ lookup = {}
+ for fn in glob.iglob('{}/*.csv'.format(path)):
+ rows = read_csv(fn, keys=False)
+ for row in rows:
+ key = row[0]
+ if key not in lookup:
+ lookup[key] = row
+ keys = sorted(lookup.keys())
+ deduped = []
+ for key in keys:
+ deduped.append(lookup[key])
+ write_csv('{}.csv'.format(path), keys=None, rows=deduped)
+
+if __name__ == '__main__':
+ s2_merge_csv()