From 219de205e39e508825f6cfe7ee1f110d98a321c8 Mon Sep 17 00:00:00 2001
From: zhang2014 <coswde@gmail.com>
Date: Tue, 20 Feb 2018 09:34:50 +0800
Subject: [PATCH] ISSUES-995 fix cut www

---
 dbms/src/Functions/FunctionsURL.h | 43 +++++++++++++++++++++++--------
 1 file changed, 32 insertions(+), 11 deletions(-)

diff --git a/dbms/src/Functions/FunctionsURL.h b/dbms/src/Functions/FunctionsURL.h
index 67b2df2fc30..0614f66a809 100644
--- a/dbms/src/Functions/FunctionsURL.h
+++ b/dbms/src/Functions/FunctionsURL.h
@@ -396,18 +396,39 @@ struct ExtractWWW
         Pos pos = data;
         Pos end = pos + size;
 
-        Pos tmp;
-        size_t protocol_length;
-        ExtractProtocol::execute(data, size, tmp, protocol_length);
-        pos += protocol_length + 3;
-
-        if (pos >= end || pos[-1] != '/' || pos[-2] != '/')
-            return;
-
-        if (pos + 4 < end && !strncmp(pos, "www.", 4))
+        if (nullptr != (pos = strchr(pos, '/')))
         {
-            res_data = pos;
-            res_size = 4;
+            if (pos != data)
+            {
+                Pos tmp;
+                size_t protocol_length;
+                ExtractProtocol::execute(data, size, tmp, protocol_length);
+
+                if (pos != data + protocol_length + 1)
+                    return;
+            }
+
+            if (end - pos < 2 || *(pos++) != '/' || *(pos++) != '/')
+                return;
+
+            const char *st = pos;
+            for (; pos < end; ++pos)
+            {
+                if (*pos == '@')
+                {
+                    st = pos + 1;
+                } else if (*pos == ':' || *pos == '/' || *pos == '?' || *pos == '#')
+                {
+                    break;
+                }
+            }
+
+
+            if (st + 4 < end && !strncmp(st, "www.", 4))
+            {
+                res_data = st;
+                res_size = 4;
+            }
         }
     }
 };