Fix Metaproxy stops logging after check config failed MP-590
[metaproxy-moved-to-github.git] / src / test_filter_rewrite.cpp
index 2e4ae54..110d614 100644 (file)
@@ -1,5 +1,5 @@
 /* This file is part of Metaproxy.
-   Copyright (C) 2005-2013 Index Data
+   Copyright (C) Index Data
 
 Metaproxy is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free
@@ -38,29 +38,11 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 using namespace boost::unit_test;
 namespace mp = metaproxy_1;
-/*
- * The global testconfig is commented out, as it won't even compile
- * on old Centos5 machines
-struct TestConfig {
-    TestConfig()
-    {
-        std::cout << "global setup\n";
-        yaz_log_init_level(YLOG_ALL);
-    }
-    ~TestConfig()
-    {
-        std::cout << "global teardown\n";
-    }
-};
-
-BOOST_GLOBAL_FIXTURE( TestConfig );
-*/
 
 BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 {
     try
     {
-        std::cout << "Running non-xml config test case" << std::endl;
         mp::RouterChain router;
         mp::filter::HttpRewrite fhr;
 
@@ -69,30 +51,44 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "<filter xmlns='http://indexdata.com/metaproxy'\n"
             "        id='rewrite1' type='http_rewrite'>\n"
             " <request verbose=\"1\">\n"
-            "   <rule name=\"null\"/>\n"
-            "   <rule name=\"url\">\n"
+            "  <rule name=\"null\"/>\n"
+            "  <rule name=\"url\">\n"
             "     <rewrite from='"
     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
             "            to='${proto}${host}${path}' />\n"
             "     <rewrite from='(?:Host: )(.*)'\n"
             "            to='Host: ${host}' />\n"
-            "   </rule>\n"
-            "  <within header=\"link\" rule=\"null\"/>\n"
-            "  <within reqline=\"1\" rule=\"url\"/>\n"
+            "  </rule>\n"
+            "  <content type=\"headers\">\n"
+            "    <within header=\"link\" rule=\"null\"/>\n"
+            "    <within reqline=\"1\" rule=\"url\"/>\n"
+            "  </content>\n"
             " </request>\n"
             " <response verbose=\"1\">\n"
-            "   <rule name=\"null\"/>\n"
-            "   <rule name=\"url\">\n"
+            "  <rule name=\"null\"/>\n"
+            "  <rule name=\"cx\">\n"
+            "     <rewrite from='^cx' to='cy'/>\n"
+            "  </rule>\n"
+            "  <rule name=\"url\">\n"
+            "     <rewrite from='foo' to='bar'/>\n"
             "     <rewrite from='"
     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
             "  </rule>\n"
-            "  <within header=\"link\" rule=\"url\"/>\n"
-            "  <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
-            "  <within tag=\"script\" attr=\"#text\" rule=\"url\"/>\n"
-            "  <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
-            "  <within attr=\"href,src\" rule=\"url\"/>\n"
+            "  <content type=\"headers\">\n"
+            "    <within header=\"link\" rule=\"url\"/>\n"
+            "  </content>\n"
+            "  <content type=\"html\" mime=\"text/xml|text/html\">\n"
+            "    <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
+            "    <within tag=\"script\" attr=\"#text\" type=\"quoted-literal\" rule=\"url\"/>\n"
+            "    <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
+            "    <within attr=\"href|src\" rule=\"url,cx\"/>\n"
+            "    <within attr=\"onclick\" type=\"quoted-literal\" rule=\"url\"/>\n"
+            "  </content>\n"
+            "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
+            "    <within rule=\"url\"/>\n"
+            "  </content>\n"
             " </response>\n"
             "</filter>\n"
         ;
@@ -113,15 +109,17 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 
         mp::odr odr;
         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
-        "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
-
+            "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
+        
+        Z_HTTP_Request *hreq = gdu_req->u.HTTP_Request;
+        z_HTTP_header_set(odr, &hreq->headers,
+                          "X-Metaproxy-SkipLink", ".* skiplink.com" );
         pack.request() = gdu_req;
 
         //create the http response
 
         const char *resp_buf =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 441\r\n"
             "Content-Type: text/html\r\n"
             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
             "Link: </dir/file.xml>; rel=relative\r\n"
@@ -133,7 +131,10 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "}"
             "</style>"
             "</head>"
-            "<script>var jslink=\"http://targetsite/webservice.xml\";</script>"
+            "<script>var jslink=\"http://targetsite/webservice.xml\";"
+            "for (i = 0; i<foo; i++) ;\n"
+            "var some=\"foo\"; foo=1;"
+            "</script>"
             "<body>"
             "<p>Welcome to our website. It doesn't make it easy to get pro"
             "xified"
@@ -142,11 +143,14 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "<a target=_blank href=\"http://targetsite/page3.html\">"
             "  Another abs link</a>"
             "<a href=\"/docs/page4.html\" />"
+            "<A href=\"cxcx\" />"
+            "<a HREF=\"cx \" onclick=\"foo(&quot;foo&quot;);\"/>\n"
+            "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
             "</body></html>";
 
         const char *resp_expected =
             "HTTP/1.1 200 OK\r\n"
-            "Content-Length: 522\r\n"
+            "Content-Length: 686\r\n"
             "Content-Type: text/html\r\n"
             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
             "Link: </dir/file.xml>; rel=relative\r\n"
@@ -158,65 +162,213 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
             "}"
             "</style>"
             "</head>"
-            "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";</script>"
+            "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
+            "for (i = 0; i<foo; i++) ;\n"
+            "var some=\"bar\"; foo=1;"
+            "</script>"
             "<body>"
             "<p>Welcome to our website. It doesn't make it easy to get pro"
             "xified"
             "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
             "  An absolute link</a>"
-            "<a target=\"_blank\" href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
+            "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
             "  Another abs link</a>"
             "<a href=\"/docs/page4.html\"/>"
+            "<A href=\"cycx\"/>"
+            "<a HREF=\"cy \" onclick=\"foo(&quot;bar&quot;);\"/>\n"
+            "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
             "</body></html>";
 
-        int r;
         Z_GDU *gdu_res;
-        ODR dec = odr_createmem(ODR_DECODE);
+        mp::odr dec(ODR_DECODE);
+        mp::odr enc(ODR_ENCODE);
         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
-        r = z_GDU(dec, &gdu_res, 0, 0);
+        int r = z_GDU(dec, &gdu_res, 0, 0);
 
         BOOST_CHECK(r);
         if (r)
         {
             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
+
+            pack.response() = gdu_res;
+
+            //feed to the router
+            pack.router(router).move();
+
+            //analyze the response
+            Z_GDU *gdu_res_rew = pack.response().get();
+            BOOST_CHECK(gdu_res_rew);
+            BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
+
+            Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
+            BOOST_CHECK(hres);
+
+            z_GDU(enc, &gdu_res_rew, 0, 0);
+            char *resp_result;
+            int resp_result_len;
+            resp_result = odr_getbuf(enc, &resp_result_len, 0);
+
+            int equal = ((size_t) resp_result_len == strlen(resp_expected))
+                && !memcmp(resp_result, resp_expected, resp_result_len);
+            BOOST_CHECK(equal);
+
+            if (!equal)
+            {
+                //compare buffers
+                std::cout << "Expected result:\n" << resp_expected << "\n";
+                std::cout << "Got result:\n";
+                fflush(stdout);
+                fwrite(resp_result, 1, resp_result_len, stdout);
+                fflush(stdout);
+                std::cout << "\nGot result buf len: " << resp_result_len
+                          << "\n";
+            }
         }
+    }
+    catch (std::exception & e) {
+        std::cout << e.what();
+        std::cout << std::endl;
+        BOOST_CHECK (false);
+    }
+}
+
+BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
+{
+    try
+    {
+        mp::RouterChain router;
+        mp::filter::HttpRewrite fhr;
+
+        std::string xmlconf =
+            "<?xml version='1.0'?>\n"
+            "<filter xmlns='http://indexdata.com/metaproxy'\n"
+            "        id='rewrite1' type='http_rewrite'>\n"
+            " <request verbose=\"1\">\n"
+            "  <rule name=\"null\"/>\n"
+            "  <rule name=\"url\">\n"
+            "     <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
+    "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
+            "            to='${proto}${host}${path}' />\n"
+            "     <rewrite from='(?:Host: )(.*)'\n"
+            "            to='Host: ${host}' />\n"
+            "  </rule>\n"
+            "  <content type=\"headers\">\n"
+            "    <within header=\"link\" rule=\"null\"/>\n"
+            "    <within reqline=\"1\" rule=\"url\"/>\n"
+            "  </content>\n"
+            " </request>\n"
+            " <response verbose=\"1\">\n"
+            "  <rule name=\"null\"/>\n"
+            "  <rule name=\"url\">\n"
+            "     <rewrite from='foo' to='bar'/>\n"
+            "     <rewrite from='^cx' to='cy'/>\n"
+            "     <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
+            "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
+            "  </rule>\n"
+            "  <content type=\"headers\">\n"
+            "    <within header=\"link\" rule=\"url\"/>\n"
+            "  </content>\n"
+            "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
+            "    <within rule=\"url\"/>\n"
+            "  </content>\n"
+            " </response>\n"
+            "</filter>\n"
+        ;
 
-        pack.response() = gdu_res;
+        std::cout << xmlconf;
 
-        //feed to the router
-        pack.router(router).move();
+        // reading and parsing XML conf
+        xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
+        BOOST_CHECK(doc);
+        xmlNode *root_element = xmlDocGetRootElement(doc);
+        fhr.configure(root_element, true, "");
+        xmlFreeDoc(doc);
 
-        //analyze the response
-        Z_GDU *gdu_res_rew = pack.response().get();
-        BOOST_CHECK(gdu_res_rew);
-        BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
+        router.append(fhr);
 
-        Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
-        BOOST_CHECK(hres);
+        // create an http request
+        mp::Package pack;
 
-        //compare buffers
-        std::cout << "Expected result:\n" << resp_expected << std::endl;
+        mp::odr odr;
+        Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
+        "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 
-        ODR enc = odr_createmem(ODR_ENCODE);
-        z_GDU(enc, &gdu_res_rew, 0, 0);
-        char *resp_result;
-        int resp_result_len;
-        resp_result = odr_getbuf(enc, &resp_result_len, 0);
+        pack.request() = gdu_req;
 
-        BOOST_CHECK(resp_result);
-        BOOST_CHECK_EQUAL((size_t) resp_result_len, strlen(resp_expected));
+        //create the http response
 
-        std::cout << "Rewritten result:\n" << std::endl;
-        fflush(stdout);
-        fwrite(resp_result, 1, resp_result_len, stdout);
-        fflush(stdout);
-        std::cout << "\nRewritten result buf len: " << resp_result_len
-            << std::endl;
+        const char *resp_buf =
+            "HTTP/1.1 200 OK\r\n"
+            "Content-Type: application/javascript\r\n"
+            "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
+            "Link: </dir/file.xml>; rel=relative\r\n"
+            "\r\n"
+            "// \"\n"
+            "my.location = 'http://targetsite/images/bg.png';\n"
+            "my.other = \"http://targetsite/images/fg.png\";\n"
+            "my.thrd = \"other\";\n"
+            "// \"http://targetsite/images/bg.png\n";
+
+        const char *resp_expected =
+            "HTTP/1.1 200 OK\r\n"
+            "Content-Length: 195\r\n"
+            "Content-Type: application/javascript\r\n"
+            "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
+            "Link: </dir/file.xml>; rel=relative\r\n"
+            "\r\n"
+            "// \"\n"
+            "my.location = 'http://proxyhost/proxypath/targetsite/images/bg.png';\n"
+            "my.other = \"http://proxyhost/proxypath/targetsite/images/fg.png\";\n"
+            "my.thrd = \"other\";\n"
+            "// \"http://targetsite/images/bg.png\n";
 
-        BOOST_CHECK(memcmp(resp_result, resp_expected, resp_result_len) == 0);
+        Z_GDU *gdu_res;
+        mp::odr dec(ODR_DECODE);
+        mp::odr enc(ODR_ENCODE);
+        odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
+        int r = z_GDU(dec, &gdu_res, 0, 0);
+
+        BOOST_CHECK(r);
+        if (r)
+        {
+            BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 
-        odr_destroy(dec);
-        odr_destroy(enc);
+            pack.response() = gdu_res;
+
+            //feed to the router
+            pack.router(router).move();
+
+            //analyze the response
+            Z_GDU *gdu_res_rew = pack.response().get();
+            BOOST_CHECK(gdu_res_rew);
+            BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
+
+            Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
+            BOOST_CHECK(hres);
+
+            z_GDU(enc, &gdu_res_rew, 0, 0);
+            char *resp_result;
+            int resp_result_len;
+            resp_result = odr_getbuf(enc, &resp_result_len, 0);
+
+            int equal = ((size_t) resp_result_len == strlen(resp_expected))
+                && !memcmp(resp_result, resp_expected, resp_result_len);
+            BOOST_CHECK(equal);
+
+            if (!equal)
+            {
+                //compare buffers
+                std::cout << "Expected result:\n" << resp_expected << "\n";
+                std::cout << "Got result:\n";
+                fflush(stdout);
+                fwrite(resp_result, 1, resp_result_len, stdout);
+                fflush(stdout);
+                std::cout << "\nGot result buf len: " << resp_result_len
+                          << "\n";
+            }
+        }
     }
     catch (std::exception & e) {
         std::cout << e.what();
@@ -225,7 +377,6 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
     }
 }
 
-
 /*
  * Local variables:
  * c-basic-offset: 4