From 97e0bc2e76ee25633944de3d62eaa00380fcb6c4 Mon Sep 17 00:00:00 2001
From: Dafydd Harries <daf@rhydd.org>
Date: Mon, 18 Mar 2013 00:28:40 -0400
Subject: add tool for generating summary of licenses

---
 license_summary.py | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
 create mode 100644 license_summary.py

diff --git a/license_summary.py b/license_summary.py
new file mode 100644
index 0000000..7e7899c
--- /dev/null
+++ b/license_summary.py
@@ -0,0 +1,21 @@
+# summary of the License: fields found in Files: clauses
+
+import pandas as pd
+
+store = pd.HDFStore('cp.h5')
+cpf = store['cp_files']
+
+def histogram(values):
+    hist = {}
+
+    for v in values:
+        v_ = v.lower()
+        hist[v_] = hist.get(v_, 0) + 1
+
+    return hist
+
+licenses = histogram(cpf['_license'])
+
+for (k, v) in sorted(licenses.iteritems(), key=lambda x: x[1], reverse=True):
+    print '%-40s %6d' % (k.encode('utf8'), v)
+
-- 
cgit v1.2.3