X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ftest_filter_rewrite.cpp;h=110d6144037ec4d8d29d7e92f4fb2990c22fd67f;hb=586d78659d671683f33ec55f4a7d32b28e345ccd;hp=be6a0718f89f0019b1b743fc736b9fb5cf15cb13;hpb=a627adf593e208501bb62986ae0d1a99a8f5fa4b;p=metaproxy-moved-to-github.git

diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp
index be6a071..110d614 100644
--- a/src/test_filter_rewrite.cpp
+++ b/src/test_filter_rewrite.cpp
@@ -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
@@ -21,13 +21,15 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <stdexcept>
 
 #include "filter_http_client.hpp"
+#include "filter_http_rewrite.hpp"
 #include <metaproxy/util.hpp>
-#include "router_chain.hpp"
+#include <metaproxy/router_chain.hpp>
 #include <metaproxy/package.hpp>
 
-#define BOOST_REGEX_MATCH_EXTRA
-
 #include <boost/regex.hpp>
+#include <boost/lexical_cast.hpp>
+
+#include <yaz/log.h>
 
 #define BOOST_AUTO_TEST_MAIN
 #define BOOST_TEST_DYN_LINK
@@ -37,99 +39,195 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 using namespace boost::unit_test;
 namespace mp = metaproxy_1;
 
-class FilterHeaderRewrite: public mp::filter::Base {
-public:
-    void process(mp::Package & package) const {
-        Z_GDU *gdu = package.request().get();
-        //we have an http req
-        if (gdu && gdu->which == Z_GDU_HTTP_Request)
-        {
-          std::cout << "Request headers" << std::endl;
-          Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
-          //dump req headers
-          for (Z_HTTP_Header *header = hreq->headers;
-              header != 0; 
-              header = header->next) 
-          {
-            std::cout << header->name << ": " << header->value << std::endl;
-            rewrite_req_header(header);
-          }
-        }
-        package.move();
-        gdu = package.response().get();
-        if (gdu && gdu->which == Z_GDU_HTTP_Response)
-        {
-          std::cout << "Respose headers" << std::endl;
-          Z_HTTP_Response *hr = gdu->u.HTTP_Response;
-          //dump resp headers
-          for (Z_HTTP_Header *header = hr->headers;
-              header != 0; 
-              header = header->next) 
-          {
-            std::cout << header->name << ": " << header->value << std::endl;
-          }
-        }
+BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
+{
+    try
+    {
+        mp::RouterChain router;
+        mp::filter::HttpRewrite fhr;
 
-    };
-    void configure(const xmlNode* ptr, bool test_only, const char *path) {};
+        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: ${host}' />\n"
+            "  </rule>\n"
+            "  <content type=\"headers\">\n"
+            "    <within header=\"link\" rule=\"null\"/>\n"
+            "    <within reqline=\"1\" rule=\"url\"/>\n"
+            "  </content>\n"
+            " </request>\n"
+            " <response verbose=\"1\">\n"
+            "  <rule name=\"null\"/>\n"
+            "  <rule name=\"cx\">\n"
+            "     <rewrite from='^cx' to='cy'/>\n"
+            "  </rule>\n"
+            "  <rule name=\"url\">\n"
+            "     <rewrite from='foo' to='bar'/>\n"
+            "     <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
+            "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
+            "  </rule>\n"
+            "  <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"
+        ;
 
-    void rewrite_req_header(Z_HTTP_Header *header) const
-    {
-        //exec regex against value
-        boost::regex e(req_uri_pat, boost::regex::perl);
-        boost::smatch what;
-        std::string hvalue(header->value);
-        if(boost::regex_match(hvalue, what, e, boost::match_extra))
-        {
-            unsigned i, j;
-            std::cout << "** Match found **\n   Sub-Expressions:\n";
-            for(i = 0; i < what.size(); ++i)
-                std::cout << "      $" << i << " = \"" << what[i] << "\"\n";
-            std::cout << "   Captures:\n";
-            for(i = 0; i < what.size(); ++i)
-            {
-                std::cout << "      $" << i << " = {";
-                for(j = 0; j < what.captures(i).size(); ++j)
-                {
-                    if(j)
-                        std::cout << ", ";
-                    else
-                        std::cout << " ";
-                    std::cout << "\"" << what.captures(i)[j] << "\"";
-                }
-                std::cout << " }\n";
-            }
-        }
-        else
+        std::cout << xmlconf;
+
+        // 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);
+
+        router.append(fhr);
+
+        // create an http request
+        mp::Package pack;
+
+        mp::odr odr;
+        Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
+            "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-Type: text/html\r\n"
+            "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
+            "Link: </dir/file.xml>; rel=relative\r\n"
+            "\r\n"
+            "<html><head><title>Hello proxy!</title>"
+            "<style>"
+            "body {"
+            "  background-image:url('http://targetsite/images/bg.png');"
+            "}"
+            "</style>"
+            "</head>"
+            "<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"
+            "<a href=\"http://targetsite/page2.html\">"
+            "  An absolute link</a>"
+            "<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: 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"
+            "\r\n"
+            "<html><head><title>Hello proxy!</title>"
+            "<style>"
+            "body {"
+            "  background-image:url('http://proxyhost/proxypath/targetsite/images/bg.png');"
+            "}"
+            "</style>"
+            "</head>"
+            "<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\">"
+            "  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>";
+
+        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)
         {
-            std::cout << "** No Match found **\n";
-        }
-        //iteratate over named groups
-        //set the captured values in the map
-        //rewrite the header according to the hardcoded recipe
-    };
-    
-    void configure(const std::string & req_uri_pat, 
-            const std::string & resp_uri_pat) 
-    {
-       this->req_uri_pat = req_uri_pat;
-       this->resp_uri_pat = resp_uri_pat;
-    };
+            BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 
-private:
-    std::map<std::string, std::string> vars;
-    std::string req_uri_pat;
-    std::string resp_uri_pat;
-};
+            pack.response() = gdu_res;
 
+            //feed to the router
+            pack.router(router).move();
 
-BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
-{
-    try
-    {
-       FilterHeaderRewrite fhr;
+            //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 ( ... ) {
+    catch (std::exception & e) {
+        std::cout << e.what();
+        std::cout << std::endl;
         BOOST_CHECK (false);
     }
 }
@@ -139,36 +237,142 @@ 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"
+        ;
+
+        std::cout << xmlconf;
+
+        // 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);
 
-        FilterHeaderRewrite fhr;
-        fhr.configure(".*(localhost).*", ".*(localhost).*");
-        mp::filter::HTTPClient hc;
-        
         router.append(fhr);
-        router.append(hc);
 
         // create an http request
         mp::Package pack;
 
         mp::odr odr;
-        Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr, 
-            "http://localhost:80/~jakub/targetsite.php", 0, 1);
+        Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
+        "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 
         pack.request() = gdu_req;
 
-        //feed to the router
-        pack.router(router).move();
+        //create the http response
 
-        //analyze the response
-        Z_GDU *gdu_res = pack.response().get();
-        BOOST_CHECK(gdu_res);
-        BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
-        
-        Z_HTTP_Response *hres = gdu_res->u.HTTP_Response;
-        BOOST_CHECK(hres);
+        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";
+
+        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);
+
+            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 ( ... ) {
+    catch (std::exception & e) {
+        std::cout << e.what();
+        std::cout << std::endl;
         BOOST_CHECK (false);
     }
 }