url's scheme extraction was fixed according to rfc3986.

dbms/CMakeLists.txt

@@ -453,6 +453,7 @@ add_library (dbms
 	include/DB/Common/randomSeed.h
 	include/DB/Common/unaligned.h
 	include/DB/Common/ThreadPool.h
+	include/DB/Common/StringView.h
 	include/DB/IO/CompressedStream.h
 	include/DB/IO/ReadBufferFromFileDescriptor.h
 	include/DB/IO/CompressedWriteBuffer.h
@@ -627,6 +628,7 @@ add_library (dbms
 	src/Common/getNumberOfPhysicalCPUCores.cpp
 	src/Common/randomSeed.cpp
 	src/Common/ThreadPool.cpp
+	src/Common/UrlUtils.cpp
 
 	src/Core/Field.cpp
 	src/Core/FieldVisitors.cpp

dbms/include/DB/Common/StringView.h

@@ -0,0 +1,66 @@
+#pragma once
+
+#include <cassert>
+#include <string>
+
+/// A lightweight non-owning read-only view into a subsequence of a string.
+class StringView
+{
+public:
+	inline StringView() noexcept
+		: str(nullptr)
+		, len(0)
+	{
+	}
+
+	constexpr inline StringView(const char* data_, size_t len_) noexcept
+		: str(data_)
+		, len(len_)
+	{
+	}
+
+	inline StringView(const std::string& str) noexcept
+		: str(str.data())
+		, len(str.size())
+	{
+	}
+
+	inline const char* data() const noexcept
+	{
+		return str;
+	}
+
+	inline bool empty() const noexcept
+	{
+		return len == 0;
+	}
+
+	inline bool null() const noexcept
+	{
+		assert(len == 0);
+		return str == nullptr;
+	}
+
+	inline size_t size() const noexcept
+	{
+		return len;
+	}
+
+public:
+	inline operator bool () const noexcept
+	{
+		return !empty();
+	}
+
+private:
+	const char* str;
+	const size_t len;
+};
+
+
+/// It creates StringView from literal constant at compile time.
+template <size_t size>
+constexpr inline StringView MakeStringView(const char (&str)[size])
+{
+    return StringView(str, size - 1);
+}

dbms/include/DB/Common/UrlUtils.h

@@ -0,0 +1,10 @@
+#pragma once
+
+#include <DB/Common/StringView.h>
+
+/** Extracts scheme from given url.
+ *
+ * If there is no valid scheme then
+ * empty StringView will be returned.
+ */
+StringView getUrlScheme(const StringView& url);

dbms/include/DB/Functions/FunctionsURL.h

@@ -3,6 +3,7 @@
 #include <DB/DataTypes/DataTypeString.h>
 #include <DB/Columns/ColumnString.h>
 #include <DB/Columns/ColumnConst.h>
+#include <DB/Common/UrlUtils.h>
 #include <DB/Functions/FunctionsString.h>
 #include <DB/Functions/FunctionsStringSearch.h>
 #include <DB/Functions/FunctionsStringArray.h>
@@ -66,12 +67,10 @@ struct ExtractProtocol
 		res_data = data;
 		res_size = 0;
 
-		Pos pos = data;
+		StringView scheme = getUrlScheme(StringView(data, size));
+		Pos pos = data + scheme.size();
 
-		while (isAlphaNumericASCII(*pos))
-			++pos;
-
-		if (pos == data || pos + 3 >= data + size)
+		if (scheme.empty() || (data + size) - pos < 4)
 			return;
 
 		if (pos[0] == ':')

dbms/src/Common/UrlUtils.cpp

@@ -0,0 +1,24 @@
+#include <DB/Common/StringUtils.h>
+#include <DB/Common/UrlUtils.h>
+
+StringView getUrlScheme(const StringView& url)
+{
+	// scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
+	const char* p = url.data();
+	const char* end = url.data() + url.size();
+
+	if (isAlphaASCII(*p))
+	{
+		for (++p; p < end; ++p)
+		{
+			if (!(isAlphaNumericASCII(*p) || *p == '+' || *p == '-' || *p == '.'))
+			{
+				break;
+			}
+		}
+
+		return StringView(url.data(), p - url.data());
+	}
+
+	return StringView();
+}

dbms/tests/queries/0_stateless/00395_url_functions.reference

@@ -0,0 +1,4 @@
+http
+https
+svn+ssh
+

dbms/tests/queries/0_stateless/00395_url_functions.sql

@@ -0,0 +1,4 @@
+SELECT protocol('http://example.com') AS Scheme;
+SELECT protocol('https://example.com/') AS Scheme;
+SELECT protocol('svn+ssh://example.com?q=hello%20world') AS Scheme;
+SELECT protocol('ftp!://example.com/') AS Scheme;