From 704b2fe733ca8a8f5d680c1cf829a124bf3041bd Mon Sep 17 00:00:00 2001 From: Yury Melnichek Date: Mon, 10 Sep 2012 23:54:37 +0200 Subject: [PATCH] [crawler] Add script to do manual renames. --- crawler/wikitravel-manual-rename.py | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) create mode 100755 crawler/wikitravel-manual-rename.py diff --git a/crawler/wikitravel-manual-rename.py b/crawler/wikitravel-manual-rename.py new file mode 100755 index 0000000000..99593637bf --- /dev/null +++ b/crawler/wikitravel-manual-rename.py @@ -0,0 +1,18 @@ +#!/opt/local/bin/python +import hashlib +import json +import os.path +import sys +import string + +for i, line in enumerate(sys.stdin): + (url, title, fileName1) = json.loads(line) + page1 = url[27:] + page2 = page1.replace('(', '%28').replace(')', '%29') + fileName2 = page2.replace('/', '_') + '_' + hashlib.md5(page2).hexdigest()[:8]; + suffix = '.google_geocoded' + if os.path.exists(fileName2 + suffix): + if not os.path.exists(fileName1 + suffix): + cmd = 'mv "%s" "%s"' % (fileName2 + suffix, fileName1 + suffix) + print(cmd) + os.system(cmd)