host extraction functionality was moved to UrlUtils

dbms/include/DB/Common/StringView.h

@@ -1,11 +1,17 @@
 #pragma once
 
 #include <cassert>
+#include <stdexcept>
 #include <string>
 
 /// A lightweight non-owning read-only view into a subsequence of a string.
 class StringView
 {
+public:
+	using size_type = size_t;
+
+	static constexpr size_type npos = size_type(-1);
+
 public:
 	inline StringView() noexcept
 		: str(nullptr)
@@ -41,11 +47,27 @@ public:
 		return str == nullptr;
 	}
 
-	inline size_t size() const noexcept
+	inline size_type size() const noexcept
 	{
 		return len;
 	}
 
+public:
+	/**
+	 * Returns a substring [pos, pos + count).
+	 * If the requested substring extends past the end of the string,
+	 * or if count == npos, the returned substring is [pos, size()).
+	 */
+	StringView substr(size_type pos, size_type count = npos) const
+	{
+		if (pos >= len)
+			throw std::out_of_range("pos must be less than len");
+		if (pos + count >= len || count == npos)
+			return StringView(str + pos, len - pos);
+		else
+			return StringView(str + pos, count);
+	}
+
 public:
 	inline operator bool () const noexcept
 	{
@@ -54,7 +76,7 @@ public:
 
 private:
 	const char* str;
-	const size_t len;
+	size_t len;
 };
 
 

dbms/include/DB/Common/UrlUtils.h

@@ -2,9 +2,9 @@
 
 #include <DB/Common/StringView.h>
 
-/** Extracts scheme from given url.
- *
- * If there is no valid scheme then
- * empty StringView will be returned.
- */
+/// Extracts scheme from given url.
 StringView getUrlScheme(const StringView& url);
+
+
+/// Extracts host from given url.
+StringView getUrlHost(const StringView& url);

dbms/include/DB/Functions/FunctionsURL.h

@@ -85,33 +85,21 @@ struct ExtractDomain
 
 	static void execute(Pos data, size_t size, Pos & res_data, size_t & res_size)
 	{
-		res_data = data;
-		res_size = 0;
+		StringView host = getUrlHost(StringView(data, size));
 
-		Pos pos = data;
-		Pos end = pos + size;
+		if (host.empty())
+		{
+			res_data = data;
+			res_size = 0;
+		}
+		else
+		{
+			if (without_www && host.size() > 4 && !strncmp(host.data(), "www.", 4))
+				host = host.substr(4);
 
-		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 (without_www && pos + 4 < end && !strncmp(pos, "www.", 4))
-			pos += 4;
-
-		Pos domain_begin = pos;
-
-		while (pos < end && *pos != '/' && *pos != ':' && *pos != '?' && *pos != '#')
-			++pos;
-
-		if (pos == domain_begin)
-			return;
-
-		res_data = domain_begin;
-		res_size = pos - domain_begin;
+			res_data = host.data();
+			res_size = host.size();
+		}
 	}
 };
 

dbms/src/Common/UrlUtils.cpp

@@ -22,3 +22,36 @@ StringView getUrlScheme(const StringView& url)
 
 	return StringView();
 }
+
+
+StringView getUrlHost(const StringView& url)
+{
+	StringView scheme = getUrlScheme(url);
+	const char* p = url.data() + scheme.size();
+	const char* end = url.data() + url.size();
+
+	// Colon must follows after scheme.
+	if (p == end || *p != ':')
+		return StringView();
+	// Authority component must starts with "//".
+	if (end - p < 2 || (p[1] != '/' || p[2] != '/'))
+		return StringView();
+	else
+		p += 3;
+
+	const char* st = p;
+
+	for (; p < end; ++p)
+	{
+		if (*p == '@')
+		{
+			st = p + 1;
+		}
+		else if (*p == ':' || *p == '/' || *p == '?' || *p == '#')
+		{
+			break;
+		}
+	}
+
+	return (p == st) ? StringView() : StringView(st, p - st);
+}

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

@@ -2,3 +2,7 @@ http
 https
 svn+ssh
 
+www.example.com
+
+www.example.com
+example.com

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

@@ -2,3 +2,8 @@ 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;
+
+SELECT domain('http://paul@www.example.com:80/') AS Host;
+SELECT domain('http:/paul/example/com') AS Host;
+SELECT domain('http://www.example.com?q=4') AS Host;
+SELECT domainWithoutWWW('http://paul@www.example.com:80/') AS Host;