1 files changed, 24 insertions, 0 deletions
diff --git a/sql/backfill/petabox_transform.py b/sql/backfill/petabox_transform.py
new file mode 100755
index 0000000..b638911
--- /dev/null
+++ b/sql/backfill/petabox_transform.py
@@ -0,0 +1,24 @@
+#!/usr/bin/env python3
+
+import json, sys, os
+
+for l in sys.stdin.readlines():
+    l = l.strip()
+    if not l:
+        continue
+    r = json.loads(l)
+    if not r['sha1']:
+        continue
+    sha1hex = r['sha1']
+    for url in r['urls']:
+        u = url['url']
+        if not '//archive.org/' in u:
+            continue
+        u = u.split('/')
+        if u[2] == 'web.archive.org':
+            continue
+        #print(u)
+        assert u[2] == 'archive.org' and u[3] in ('download', 'serve')
+        item = u[4]
+        path = '/'.join(u[5:])
+        print("\t".join([item, path, sha1hex]))