From 8adea055298189643a3c7a76e2d529f536a94e11 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <p@simianer.de>
Date: Tue, 23 Apr 2019 13:55:47 +0200
Subject: init

---
 stats.rb | 38 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 38 insertions(+)
 create mode 100755 stats.rb

(limited to 'stats.rb')

diff --git a/stats.rb b/stats.rb
new file mode 100755
index 0000000..390cbd9
--- /dev/null
+++ b/stats.rb
@@ -0,0 +1,38 @@
+#!/usr/bin/env ruby
+
+require_relative 'util'
+
+def setup reference_file, stopwords_file
+  references = ReadFile.readlines_strip reference_file
+  stopwords  = read_stopwords_file stopwords_file
+
+  return references, stopwords
+end
+
+def stats references, stopwords
+  references.each { |r|
+    types, uniq_types = get_types r, stopwords
+
+    counts = []
+    uniq_types.each { |t|
+      counts <<  types.count(t)
+    }
+    if counts.size > 0
+      puts counts.inject(:+) / counts.size.to_f
+    end
+  }
+end
+
+def main
+  config = Trollop::options do
+    opt :references, "File with references, truecased and tokenized", :type => :string, :short => "-r", :required => true
+    opt :stopwords, "File with stopwords, one per line",              :type => :string, :short => "-s", :required => true
+  end
+
+  references, stopwords = setup config[:references], config[:stopwords]
+
+  stats references, stopwords
+end
+
+main
+
-- 
cgit v1.2.3