From f826e90538dc15fbde4152b0e2f7d30fd6e56784 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@cs.cmu.edu>
Date: Mon, 12 Sep 2011 19:22:59 +0100
Subject: source syntax features ~ blunsom emnlp 2008

---
 utils/stringlib.cc | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

(limited to 'utils/stringlib.cc')

diff --git a/utils/stringlib.cc b/utils/stringlib.cc
index 7aaee9f0..ade02ca9 100644
--- a/utils/stringlib.cc
+++ b/utils/stringlib.cc
@@ -32,7 +32,12 @@ void ParseTranslatorInput(const string& line, string* input, string* ref) {
 void ProcessAndStripSGML(string* pline, map<string, string>* out) {
   map<string, string>& meta = *out;
   string& line = *pline;
-  string lline = LowercaseString(line);
+  string lline = *pline;
+  if (lline.find("<SEG")==0 || lline.find("<Seg")==0) {
+    cerr << "Segment tags <seg> must be lowercase!\n";
+    cerr << "  " << *pline << endl;
+    abort();
+  } 
   if (lline.find("<seg")!=0) return;
   size_t close = lline.find(">");
   if (close == string::npos) return; // error
-- 
cgit v1.2.3