From 616445e027c203a3caa17d3e0a859f6afe107dbc Mon Sep 17 00:00:00 2001
From: Matthew Honnibal <honnibal@gmail.com>
Date: Sat, 25 Jul 2015 21:12:45 +0200
Subject: [PATCH] * Add simple script to collate frequencies from sorted file

---
 bin/gather_freqs.py | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)
 create mode 100644 bin/gather_freqs.py

diff --git a/bin/gather_freqs.py b/bin/gather_freqs.py
new file mode 100644
index 000000000..f0cbdfa4f
--- /dev/null
+++ b/bin/gather_freqs.py
@@ -0,0 +1,27 @@
+import plac
+
+def main(in_loc, out_loc):
+    out_file = open(out_loc, 'w')
+    this_key = None
+    this_freq = 0
+    df = 0
+    for line in open(in_loc):
+        line = line.strip()
+        if not line:
+            continue
+        freq, key = line.split('\t', 1)
+        freq = int(freq)
+        if this_key is not None and key != this_key:
+            out_file.write('%d\t%d\t%s\n' % (this_freq, df, this_key))
+            this_key = key
+            this_freq = freq
+            df = 1
+        else:
+            this_freq += freq
+            df += 1
+    out_file.write('%d\t%d\t%s\n' % (this_freq, df, this_key))
+    out_file.close()
+
+
+if __name__ == '__main__':
+    plac.call(main)