create netloc function

src/Functions/URL/netloc.cpp

@@ -0,0 +1,17 @@
+#include <Functions/FunctionFactory.h>
+#include <Functions/FunctionStringToString.h>
+#include "netloc.h"
+
+namespace DB
+{
+
+struct NameNetloc { static constexpr auto name = "netloc"; };
+using FunctionNetloc = FunctionStringToString<ExtractSubstringImpl<ExtractNetloc>, NameNetloc>;
+
+void registerFunctionNetloc(FunctionFactory & factory)
+{
+    factory.registerFunction<FunctionNetloc>();
+}
+
+}
+

src/Functions/URL/netloc.h

@@ -0,0 +1,42 @@
+#pragma once
+
+#include "FunctionsURL.h"
+#include <common/find_symbols.h>
+
+
+namespace DB
+{
+
+struct ExtractNetloc
+{
+    static size_t getReserveLengthForElement() { return 10; }
+
+    static void execute(Pos data, size_t size, Pos & res_data, size_t & res_size)
+    {
+        res_data = data;
+        res_size = size;
+
+        Pos pos = data;
+        Pos end = pos + size;
+
+        pos = find_first_symbols<'/'>(pos, end);
+        if (end == pos)
+            return;
+
+        /// Note that strings are zero-terminated.
+        bool has_subsequent_slash = pos[1] == '/';
+        if (!has_subsequent_slash)
+            return;
+        res_data = pos + 2;
+        res_size = end - res_data;
+
+        /// Search for next slash.
+        pos = find_first_symbols<'/', '?'>(pos + 2, end);
+        if (end == pos)
+            return;
+        res_size = pos - res_data;
+    }
+};
+
+}
+

src/Functions/URL/registerFunctionsURL.cpp

@@ -26,6 +26,7 @@ void registerFunctionCutFragment(FunctionFactory & factory);
 void registerFunctionCutQueryStringAndFragment(FunctionFactory & factory);
 void registerFunctionCutURLParameter(FunctionFactory & factory);
 void registerFunctionDecodeURLComponent(FunctionFactory & factory);
+void registerFunctionNetloc(FunctionFactory & factory);
 
 void registerFunctionsURL(FunctionFactory & factory)
 {
@@ -52,6 +53,7 @@ void registerFunctionsURL(FunctionFactory & factory)
     registerFunctionCutQueryStringAndFragment(factory);
     registerFunctionCutURLParameter(factory);
     registerFunctionDecodeURLComponent(factory);
+    registerFunctionNetloc(factory);
 }
 
 }

tests/queries/0_stateless/00398_url_functions.reference

@@ -16,6 +16,17 @@ www.example.com
 example.com
 example.com
 example.com
+====NETLOC====
+paul@www.example.com:80
+127.0.0.1:443
+127.0.0.1:443
+example.ru
+example.ru
+paul:zozo@example.ru
+paul:zozo@example.ru
+www.example.com
+www.example.com
+example.com
 ====DOMAIN====
 com
 

tests/queries/0_stateless/00398_url_functions.sql

@@ -18,6 +18,17 @@ SELECT domain('example.com') as Host;
 SELECT domainWithoutWWW('//paul@www.example.com') AS Host;
 SELECT domainWithoutWWW('http://paul@www.example.com:80/') AS Host;
 
+SELECT '====NETLOC====';
+SELECT netloc('http://paul@www.example.com:80/') AS Netloc;
+SELECT netloc('http://127.0.0.1:443/') AS Netloc;
+SELECT netloc('http://127.0.0.1:443') AS Netloc;
+SELECT netloc('svn+ssh://example.ru/?q=hello%20world') AS Netloc;
+SELECT netloc('svn+ssh://example.ru/?q=hello%20world') AS Netloc;
+SELECT netloc('svn+ssh://paul:zozo@example.ru/?q=hello%20world') AS Netloc;
+SELECT netloc('svn+ssh://paul:zozo@example.ru/?q=hello%20world') AS Netloc;
+SELECT netloc('//www.example.com') AS Netloc;
+SELECT netloc('www.example.com') as Netloc;
+SELECT netloc('example.com') as Netloc;
 
 SELECT '====DOMAIN====';
 SELECT topLevelDomain('http://paul@www.example.com:80/') AS Domain;