From d637e517919852217705a34d1355020a97af2f6a Mon Sep 17 00:00:00 2001
From: Patrick Simianer <simianer@cl.uni-heidelberg.de>
Date: Fri, 11 Apr 2014 11:55:24 +0200
Subject: hashing, helper scripts

---
 hopefear.rb                            |  1 +
 lampion.rb                             |  3 ++-
 scripts/geoquery/sigf_format.rb        | 20 ++++++++++++++++++++
 scripts/geoquery/test-nof-old-crawl.sh |  5 +++++
 4 files changed, 28 insertions(+), 1 deletion(-)
 create mode 100755 scripts/geoquery/sigf_format.rb
 create mode 100755 scripts/geoquery/test-nof-old-crawl.sh

diff --git a/hopefear.rb b/hopefear.rb
index ef1fd43..49ef593 100644
--- a/hopefear.rb
+++ b/hopefear.rb
@@ -115,6 +115,7 @@ def gethopefear_fear_no_exec_hope_exec_skip kbest, feedback, gold, max
   return hope, fear, skip, type1, type2
 end
 
+# new variant w/ "real" reference
 def gethopefear_only_exec kbest, feedback, gold, max, own_reference=nil
   hope = fear = nil; hope_idx = 0; new_reference = nil
   type1 = type2 = false
diff --git a/lampion.rb b/lampion.rb
index ab8861f..ab527d0 100755
--- a/lampion.rb
+++ b/lampion.rb
@@ -4,6 +4,7 @@ require 'nlp_ruby'
 require 'trollop'
 require 'tempfile'
 require 'memcached'
+require 'digest'
 require_relative './hopefear'
 
 
@@ -11,7 +12,7 @@ def exec natural_language_string, reference_output, no_output=false
   mrl = output = feedback = nil
   # this may cause collisions, but there are not so many German words that
   # could have different Umlauts at the same position, e.g. Häuser => H?user
-  key_prefix = natural_language_string.encode('ASCII', :invalid => :replace, :undef => :replace, :replace => '?').gsub(/ /,'_')
+  key_prefix = Digest::SHA1.hexdigest(natural_language_string.encode('ASCII', :invalid => :replace, :undef => :replace, :replace => '?').gsub(/ /,'_'))
   begin
     mrl      = $cache.get key_prefix+'__MRL'
     output   = $cache.get key_prefix+'__OUTPUT'
diff --git a/scripts/geoquery/sigf_format.rb b/scripts/geoquery/sigf_format.rb
new file mode 100755
index 0000000..9d0125b
--- /dev/null
+++ b/scripts/geoquery/sigf_format.rb
@@ -0,0 +1,20 @@
+#!/usr/bin/env ruby
+
+require 'nlp_ruby'
+
+
+gold = ReadFile.readlines_strip ARGV[0]
+i = -1
+while line = STDIN.gets
+  i += 1
+  line.strip!
+  a = [0, 0, 1]
+  if line==gold[i]
+    a[0] = 1
+    a[1] = 1
+  elsif line!=''
+    a[1] = 1 
+  end
+  puts a.join " "
+end
+
diff --git a/scripts/geoquery/test-nof-old-crawl.sh b/scripts/geoquery/test-nof-old-crawl.sh
new file mode 100755
index 0000000..79c35a8
--- /dev/null
+++ b/scripts/geoquery/test-nof-old-crawl.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+/workspace/grounded/lampion/scripts/geoquery/translate.sh $1 $2 < /workspace/grounded/lampion/proper/d/split880.test-nof-old-crawl.in | tee $2.transl | /workspace/grounded/lampion/scripts/geoquery/semparse.rb $3 | tee $2.parsed | /workspace/grounded/lampion/scripts/geoquery/query.rb $3 > $2.output
+/workspace/grounded/lampion/scripts/geoquery/eval.rb /workspace/grounded/lampion/proper/d/split880.test-nof.gold < $2.output > $2.result
+
-- 
cgit v1.2.3