From aba828abbabcb9a952ad3144d8ae03e7f4e0ea87 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <patrick@lilt.com>
Date: Fri, 21 May 2021 17:49:56 +0200
Subject: tsv-exclude

---
 tsv-exclude | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
 create mode 100755 tsv-exclude

(limited to 'tsv-exclude')

diff --git a/tsv-exclude b/tsv-exclude
new file mode 100755
index 0000000..e951ea1
--- /dev/null
+++ b/tsv-exclude
@@ -0,0 +1,21 @@
+#!/usr/bin/env ruby
+
+require 'zipf'
+require 'set'
+
+to_exclude0 = {}
+to_exclude1 = {}
+f = ReadFile.new ARGV[0]
+
+while line = f.gets
+  p0, p1 = line.strip.split "\t"
+  to_exclude0[p0] = true
+  to_exclude1[p1] = true
+end
+
+while line = STDIN.gets
+  p0, p1 = line.strip.split "\t"
+  if not to_exclude0.has_key? p0 and not to_exclude1.has_key? p1
+    puts line
+  end
+end
-- 
cgit v1.2.3