From a2cf90d9ce532dd03d58a16402faa8924f5492a6 Mon Sep 17 00:00:00 2001
From: Olivier Bertrand <bertrandop@gmail.com>
Date: Fri, 15 Feb 2013 15:53:27 +0100
Subject: Suppress multiple blanks and control characters from the node text
 retrieved by libxml2.

Modified:
libdoc.cpp
---
 storage/connect/libdoc.cpp | 22 +++++++++++++++++++++-
 1 file changed, 21 insertions(+), 1 deletion(-)

(limited to 'storage/connect/libdoc.cpp')

diff --git a/storage/connect/libdoc.cpp b/storage/connect/libdoc.cpp
index 36e920540aa..5768dfa8177 100644
--- a/storage/connect/libdoc.cpp
+++ b/storage/connect/libdoc.cpp
@@ -461,7 +461,27 @@ char *XML2NODE::GetText(char *buf, int len)
     xmlFree(Content);
 
   if ((Content = xmlNodeGetContent(Nodep))) {
-    int rc = ((PXDOC2)Doc)->Decode(Content, buf, len);
+    char *p1, *p2;
+    bool  b = false;
+    int   rc = ((PXDOC2)Doc)->Decode(Content, buf, len);
+
+    // Eliminate extra characters
+    for (p1 = p2 = buf; *p1; p1++)
+      if (strchr(" \t\r\n", *p1)) {
+        if (b) {
+          *p2++ = ' ';
+          b = false;
+          }  // endif b
+
+      } else {
+        *p2++ = *p1;
+        b = true;
+      } // endif p1
+
+    if (*(p2 - 1) == ' ')
+      *(p2 - 1) = 0;
+    else
+      *p2 = 0;
 
     if (trace)
       htrc("GetText buf='%s' len=%d rc=%d\n", buf, len, rc);
-- 
cgit v1.2.1