[Test] Add one more test for urls parsing

1 files changed, 123 insertions, 118 deletions

diff --git a/test/lua/unit/url.lua b/test/lua/unit/url.lua
index cd23afd7c..e2daa7771 100644
--- a/test/lua/unit/url.lua
+++ b/test/lua/unit/url.lua
@@ -9,7 +9,7 @@ context("URL check functions", function()
   local test_helper = require("rspamd_test_helper")
   local ffi = require("ffi")
 
-  ffi.cdef[[
+  ffi.cdef [[
   void rspamd_normalize_path_inplace(char *path, size_t len, size_t *nlen);
   ]]
 
@@ -18,25 +18,25 @@ context("URL check functions", function()
   local pool = mpool.create()
 
   local cases = {
-    {"test.com", {"test.com", nil}},
-    {" test.com", {"test.com", nil}},
-    {"<test.com> text", {"test.com", nil}},
-    {"test.com. text", {"test.com", nil}},
-    {"mailto:A.User@example.com text", {"example.com", "A.User"}},
-    {"http://Тест.Рф:18 text", {"тест.рф", nil}},
-    {"http://user:password@тест2.РФ:18 text", {"тест2.рф", "user"}},
-    {"somebody@example.com", {"example.com", "somebody"}},
-    {"https://127.0.0.1/abc text", {"127.0.0.1", nil}},
-    {"https:\\\\127.0.0.1/abc text", {"127.0.0.1", nil}},
-    {"https:\\\\127.0.0.1", {"127.0.0.1", nil}},
-    {"https://127.0.0.1 text", {"127.0.0.1", nil}},
-    {"https://[::1]:1", {"::1", nil}},
-    {"https://user:password@[::1]:1", {"::1", nil}},
-    {"https://user:password@[::1]", {"::1", nil}},
-    {"https://user:password@[::1]/1", {"::1", nil}},
+    { "test.com", { "test.com", nil } },
+    { " test.com", { "test.com", nil } },
+    { "<test.com> text", { "test.com", nil } },
+    { "test.com. text", { "test.com", nil } },
+    { "mailto:A.User@example.com text", { "example.com", "A.User" } },
+    { "http://Тест.Рф:18 text", { "тест.рф", nil } },
+    { "http://user:password@тест2.РФ:18 text", { "тест2.рф", "user" } },
+    { "somebody@example.com", { "example.com", "somebody" } },
+    { "https://127.0.0.1/abc text", { "127.0.0.1", nil } },
+    { "https:\\\\127.0.0.1/abc text", { "127.0.0.1", nil } },
+    { "https:\\\\127.0.0.1", { "127.0.0.1", nil } },
+    { "https://127.0.0.1 text", { "127.0.0.1", nil } },
+    { "https://[::1]:1", { "::1", nil } },
+    { "https://user:password@[::1]:1", { "::1", nil } },
+    { "https://user:password@[::1]", { "::1", nil } },
+    { "https://user:password@[::1]/1", { "::1", nil } },
   }
 
-  for i,c in ipairs(cases) do
+  for i, c in ipairs(cases) do
     local res = url.create(pool, c[1])
 
     test("Extract urls from text" .. i, function()
@@ -46,107 +46,112 @@ context("URL check functions", function()
       --print(s)
       assert_not_nil(t, "cannot convert to table " .. c[1])
       assert_equal(c[2][1], t['host'],
-              logger.slog('expected host "%s", but got "%s" in url %s => %s',
+          logger.slog('expected host "%s", but got "%s" in url %s => %s',
               c[2][1], t['host'], c[1], t))
 
       if c[2][2] then
         assert_equal(c[2][1], t['host'],
-                logger.slog('expected user "%s", but got "%s" in url %s => %s',
-                        c[2][1], t['host'], c[1], t))
+            logger.slog('expected user "%s", but got "%s" in url %s => %s',
+                c[2][1], t['host'], c[1], t))
       end
     end)
   end
 
   cases = {
-    {[[http://example.net/path/]], true, {
+    { [[http://example.net/path/]], true, {
       host = 'example.net', path = 'path/'
-    }},
-    {'http://example.net/hello%20world.php?arg=x#fragment', true, {
+    } },
+    { 'http://example.net/hello%20world.php?arg=x#fragment', true, {
       host = 'example.net', fragment = 'fragment', query = 'arg=x',
       path = 'hello world.php',
-    }},
-    {'http://example.net/?arg=%23#fragment', true, {
+    } },
+    { 'http://example.net/?arg=%23#fragment', true, {
       host = 'example.net', fragment = 'fragment', query = 'arg=#',
-    }},
-    {"http:/\\[::eeee:192.168.0.1]/#test", true, {
+    } },
+    { "http:/\\[::eeee:192.168.0.1]/#test", true, {
       host = '::eeee:c0a8:1', fragment = 'test'
-    }},
-    {"http:/\\[::eeee:192.168.0.1]#test", true, {
+    } },
+    { "http:/\\[::eeee:192.168.0.1]#test", true, {
       host = '::eeee:c0a8:1', fragment = 'test'
-    }},
-    {"http:/\\[::eeee:192.168.0.1]?test", true, {
+    } },
+    { "http:/\\[::eeee:192.168.0.1]?test", true, {
       host = '::eeee:c0a8:1', query = 'test'
-    }},
-    {"http:\\\\%30%78%63%30%2e%30%32%35%30.01", true, { --0xc0.0250.01
+    } },
+    { "http:\\\\%30%78%63%30%2e%30%32%35%30.01", true, { --0xc0.0250.01
       host = '192.168.0.1',
-    }},
-    {"http:/\\www.google.com/foo?bar=baz#", true, {
+    } },
+    { "http:/\\www.google.com/foo?bar=baz#", true, {
       host = 'www.google.com', path = 'foo', query = 'bar=baz', tld = 'google.com'
-    }},
-    {"http://[www.google.com]/", true, {
+    } },
+    { "http://[www.google.com]/", true, {
       host = 'www.google.com',
-    }},
-    {"<test.com", true, {
+    } },
+    { "<test.com", true, {
       host = 'test.com', tld = 'test.com',
-    }},
-    {"test.com>", false},
-    {",test.com text", false},
-    {"ht\ttp:@www.google.com:80/;p?#", false},
-    {"http://user:pass@/", false},
-    {"http://foo:-80/", false},
-    {"http:////////user:@google.com:99?foo", true, {
+    } },
+    { "test.com>", false },
+    { ",test.com text", false },
+    { "ht\ttp:@www.google.com:80/;p?#", false },
+    { "http://user:pass@/", false },
+    { "http://foo:-80/", false },
+    { "http:////////user:@google.com:99?foo", true, {
       host = 'google.com', user = 'user', port = 99, query = 'foo'
-    }},
-    {"http://%25DOMAIN:foobar@foodomain.com/", true, {
+    } },
+    { "http://%25DOMAIN:foobar@foodomain.com/", true, {
       host = 'foodomain.com', user = '%25DOMAIN'
-    }},
-    {"http://0.0xFFFFFF", true, {
+    } },
+    { "http://0.0xFFFFFF", true, {
       host = '0.255.255.255'
-    }},
+    } },
     --{"http:/\\030052000001", true, {
     --  host = '192.168.0.1'
     --}},
-    {"http:\\/0xc0.052000001", true, {
+    { "http:\\/0xc0.052000001", true, {
       host = '192.168.0.1'
-    }},
-    {"http://192.168.0.1.?foo", true, {
+    } },
+    { "http://192.168.0.1.?foo", true, {
       host = '192.168.0.1', query = 'foo',
-    }},
-    {"http://twitter.com#test", true, {
+    } },
+    { "http://twitter.com#test", true, {
       host = 'twitter.com', fragment = 'test'
-    }},
-    {"http:www.twitter.com#test", true, {
+    } },
+    { "http:www.twitter.com#test", true, {
       host = 'www.twitter.com', fragment = 'test'
-    }},
-    {"http://example。com#test", true, {
+    } },
+    { "http://example。com#test", true, {
       host = 'example.com', fragment = 'test'
-    }},
-    {"http://hoho.example。com#test", true, {
+    } },
+    { "http://hoho.example。com#test", true, {
       host = 'hoho.example.com', fragment = 'test'
-    }},
-    {"http://hoho。example。com#test", true, {
+    } },
+    { "http://hoho。example。com#test", true, {
       host = 'hoho.example.com', fragment = 'test'
-    }},
-    {"http://hoho．example。com#test", true, {
+    } },
+    { "http://hoho．example。com#test", true, {
       host = 'hoho.example.com', fragment = 'test'
-    }},
-    {"http://hehe｡example。com#test", true, {
+    } },
+    { "http://hehe｡example。com#test", true, {
       host = 'hehe.example.com', fragment = 'test'
-    }},
-    {"http:////$%^&****((@example.org//#f@f", true, {
+    } },
+    { "http:////$%^&****((@example.org//#f@f", true, {
       user = '$%^&****((', host = 'example.org', fragment = 'f@f'
-    }},
-    {"http://@@example.com", true, {
+    } },
+    { "http://@@example.com", true, {
       user = "@", host = "example.com"
-    }},
-    {"https://example.com\\_Resources\\ClientImages\\UserData?ol\\o#ololo\\", true, {
+    } },
+    { "https://example.com\\_Resources\\ClientImages\\UserData?ol\\o#ololo\\", true, {
       host = "example.com", path = "_Resources\\ClientImages\\UserData",
       query = "ol\\o", fragment = "ololo\\",
-    }},
+    } },
+    {
+      "http://0x3f8f29a4/pro/au.html", true, {
+      host = "63.143.41.164",
+      path = "pro/au.html",
+    } },
   }
 
   -- Some cases from https://code.google.com/p/google-url/source/browse/trunk/src/url_canon_unittest.cc
-  for i,c in ipairs(cases) do
+  for i, c in ipairs(cases) do
     local res = url.create(pool, c[1])
 
     test("Parse url: " .. c[1], function()
@@ -155,12 +160,12 @@ context("URL check functions", function()
 
         local uf = res:to_table()
 
-        for k,v in pairs(c[3]) do
+        for k, v in pairs(c[3]) do
           assert_not_nil(uf[k], k .. ' is missing in url, must be ' .. v)
-          assert_equal(uf[k], v, logger.slog('expected "%s", for %s, but got "%s" in url %s => %s',
-                v, k, uf[k], c[1], uf))
+          assert_equal(uf[k], v, logger.slog('expected " %s ", for %s, but got " %s " in url %s => %s',
+              v, k, uf[k], c[1], uf))
         end
-        for k,v in pairs(uf) do
+        for k, v in pairs(uf) do
           if k ~= 'url' and k ~= 'protocol' and k ~= 'tld' then
             assert_not_nil(c[3][k], k .. ' should be absent but it is ' .. v .. ' in: ' .. c[1])
           end
@@ -172,32 +177,32 @@ context("URL check functions", function()
   end
 
   cases = {
-    {"/././foo", "/foo"},
-    {"/a/b/c/./../../g", "/a/g"},
-    {"/./.foo", "/.foo"},
-    {"/foo/.", "/foo/"},
-    {"/foo/./", "/foo/"},
-    {"/foo/bar/..", "/foo"},
-    {"/foo/bar/../", "/foo/"},
-    {"/foo/..bar", "/foo/..bar"},
-    {"/foo/bar/../ton", "/foo/ton"},
-    {"/foo/bar/../ton/../../a", "/a"},
-    {"/foo/../../..", "/"},
-    {"/foo/../../../ton", "/ton"},
-    {"////../..", "/"},
-    {"./", ""},
-    {"/./", "/"},
-    {"/./././././././", "/"},
-    {"/", "/"},
-    {"/a/b", "/a/b"},
-    {"/a/b/", "/a/b/"},
-    {"..", "/"},
-    {"/../", "/"},
-    {"../", "/"},
-    {"///foo", "/foo"},
+    { "/././foo", "/foo" },
+    { "/a/b/c/./../../g", "/a/g" },
+    { "/./.foo", "/.foo" },
+    { "/foo/.", "/foo/" },
+    { "/foo/./", "/foo/" },
+    { "/foo/bar/..", "/foo" },
+    { "/foo/bar/../", "/foo/" },
+    { "/foo/..bar", "/foo/..bar" },
+    { "/foo/bar/../ton", "/foo/ton" },
+    { "/foo/bar/../ton/../../a", "/a" },
+    { "/foo/../../..", "/" },
+    { "/foo/../../../ton", "/ton" },
+    { "////../..", "/" },
+    { "./", "" },
+    { "/./", "/" },
+    { "/./././././././", "/" },
+    { "/", "/" },
+    { "/a/b", "/a/b" },
+    { "/a/b/", "/a/b/" },
+    { "..", "/" },
+    { "/../", "/" },
+    { "../", "/" },
+    { "///foo", "/foo" },
   }
 
-  for i,v in ipairs(cases) do
+  for i, v in ipairs(cases) do
     test(string.format("Normalize paths '%s'", v[1]), function()
       local buf = ffi.new("uint8_t[?]", #v[1])
       local sizbuf = ffi.new("size_t[1]")
@@ -209,28 +214,28 @@ context("URL check functions", function()
   end
 
   cases = {
-    {'example.com', 'example.com'},
-    {'baz.example.com', 'baz.example.com'},
-    {'3.baz.example.com', 'baz.example.com'},
-    {'bar.example.com', 'example.com'},
-    {'foo.example.com', 'foo.example.com'},
-    {'3.foo.example.com', '3.foo.example.com'},
-    {'foo.com', 'foo.com'},
-    {'bar.foo.com', 'foo.com'},
+    { 'example.com', 'example.com' },
+    { 'baz.example.com', 'baz.example.com' },
+    { '3.baz.example.com', 'baz.example.com' },
+    { 'bar.example.com', 'example.com' },
+    { 'foo.example.com', 'foo.example.com' },
+    { '3.foo.example.com', '3.foo.example.com' },
+    { 'foo.com', 'foo.com' },
+    { 'bar.foo.com', 'foo.com' },
   }
 
   local excl_rules1 = {
-      'example.com',
-      '*.foo.example.com',
-      '!bar.example.com'
+    'example.com',
+    '*.foo.example.com',
+    '!bar.example.com'
   }
 
   local comp_rules = lua_urls_compose.inject_composition_rules(rspamd_config, excl_rules1)
 
-  for _,v in ipairs(cases) do
+  for _, v in ipairs(cases) do
     test("URL composition " .. v[1], function()
       local u = url.create(pool, v[1])
-      assert_not_nil(u, "we are able to parse url: " .. v[1])
+      assert_not_nil(u, "we are able to parse url:" .. v[1])
       local res = comp_rules:process_url(nil, u:get_tld(), u:get_host())
       assert_equal(v[2], res, 'expected ' .. v[2] .. ' but got ' .. res .. ' in url ' .. v[1])
     end)