From 13c9f64b23f2610a233eb1ea778fda05329e9237 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <simianer@cl.uni-heidelberg.de>
Date: Tue, 4 Mar 2014 21:41:48 +0100
Subject: data and scripts

---
 scripts/geoquery/extract.rb | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100755 scripts/geoquery/extract.rb

(limited to 'scripts/geoquery/extract.rb')

diff --git a/scripts/geoquery/extract.rb b/scripts/geoquery/extract.rb
new file mode 100755
index 0000000..ea6e864
--- /dev/null
+++ b/scripts/geoquery/extract.rb
@@ -0,0 +1,33 @@
+#!/usr/bin/env ruby
+
+require 'nlp_ruby'
+require 'xmlsimple'
+
+
+def extract fn='./corpus.xml', lang='en', ids
+  doc = XmlSimple.xml_in(fn)
+  doc['example'].each { |example|
+    next if (!ids.include? example['id']) && ids.size>0
+    if lang == 'funql' || lang == 'geo-prolog'
+      puts example['mrl'][0]['content'].to_s.strip
+    else
+      example['nl'].each { |nl|
+        if nl['lang'] == lang
+          puts nl['content']
+        else
+          next
+        end
+      }
+    end
+  }
+end
+
+def main
+  ids = []
+  ids = ReadFile.readlines_strip ARGV[2]
+  extract ARGV[0], ARGV[1], ids
+end
+
+
+main
+
-- 
cgit v1.2.3