X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ftest_filter_rewrite.cpp;h=36e1a15f1dfb938e32cc9e3fdaaedd5b420664f4;hb=6bc88607a5726bcf482ceb23d0cf5e7dd9db8341;hp=40ab6ba0fb0afcec3946da48d45918e8dffded4e;hpb=8181c47896ffd66e31add5fd15236bc5f28a2596;p=metaproxy-moved-to-github.git

diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp
index 40ab6ba..36e1a15 100644
--- a/src/test_filter_rewrite.cpp
+++ b/src/test_filter_rewrite.cpp
@@ -21,271 +21,337 @@ 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/xpressive/xpressive.hpp>
+#include <boost/regex.hpp>
 #include <boost/lexical_cast.hpp>
 
+#include <yaz/log.h>
+
 #define BOOST_AUTO_TEST_MAIN
 #define BOOST_TEST_DYN_LINK
 
 #include <boost/test/auto_unit_test.hpp>
 
 using namespace boost::unit_test;
-using namespace boost::xpressive;
 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;
-          }
-        }
-
-    };
-    void configure(const xmlNode* ptr, bool test_only, const char *path) {};
-
-    void rewrite_req_header(Z_HTTP_Header *header) const
-    {
-        //exec regex against value
-        sregex re = sregex::compile(req_uri_rx);
-        smatch what;
-        std::string hvalue(header->value);
-        std::map<std::string, std::string> vars;
-        if (regex_match(hvalue, what, re))
-        {
-            unsigned i;
-            for (i = 1; i < what.size(); ++i)
-            {
-                //check if the group is named
-                std::map<int, std::string>::const_iterator it
-                    = groups_by_num.find(i);
-                if (it != groups_by_num.end()) 
-                {   //it is
-                    std::string name = it->second;
-                    vars[name] = what[i];
-                }
-            }
-            //rewrite the header according to the recipe
-            std::string rvalue = sub_vars(req_uri_pat, vars);
-            std::cout << "Rewritten '"+hvalue+"' to '"+rvalue+"'\n";
-        }
-        else
-        {
-            std::cout << "No match found in '" + hvalue + "'\n";
-        }
-    };
-
-    static void parse_groups(const std::string & str,
-            std::map<int, std::string> & groups_bynum,
-            std::map<std::string, int> & groups_byname)
+/*
+ * The global testconfig is commented out, as it won't even compile
+ * on old Centos5 machines
+struct TestConfig {
+    TestConfig()   
     {
-       int gnum = 0;
-       bool esc = false;
-       for (int i = 0; i < str.size(); ++i)
-       {
-           if (!esc && str[i] == '\\')
-           {
-               esc = true;
-               continue;
-           }
-           if (!esc && str[i] == '(') //group starts
-           {
-               gnum++;
-               if (i+1 < str.size() && str[i+1] == '?') //group with attrs 
-               {
-                   i++;
-                   if (i+1 < str.size() && str[i+1] == 'P') //optional, python
-                       i++;
-                   if (i+1 < str.size() && str[i+1] == '<') //named
-                   {
-                       i++;
-                       std::string gname;
-                       bool term = false;
-                       while (++i < str.size())
-                       {
-                           if (str[i] == '>') { term = true; break; }
-                           if (!isalnum(str[i])) 
-                               throw mp::filter::FilterException
-                                   ("Only alphanumeric chars allowed, found "
-                                    " in '" 
-                                    + str 
-                                    + "' at " 
-                                    + boost::lexical_cast<std::string>(i)); 
-                           gname += str[i];
-                       }
-                       if (!term)
-                           throw mp::filter::FilterException
-                               ("Unterminated group name '" + gname 
-                                + " in '" + str +"'");
-                      groups_bynum[gnum] = gname;
-                      groups_byname[gname] = gnum;
-                      std::cout << "Found named group '" << gname 
-                          << "' at $" << gnum << std::endl;
-                   }
-               }
-           }
-           esc = false;
-       }
+        std::cout << "global setup\n"; 
+        yaz_log_init_level(YLOG_ALL);
     }
-
-    static std::string sub_vars (const std::string & in, 
-            const std::map<std::string, std::string> & vars)
-    {
-        std::string out;
-        bool esc = false;
-        for (int i = 0; i < in.size(); ++i)
-        {
-            if (!esc && in[i] == '\\')
-            {
-                esc = true;
-                continue;
-            }
-            if (!esc && in[i] == '$') //var
-            {
-                if (i+1 < in.size() && in[i+1] == '{') //ref prefix
-                {
-                    ++i;
-                    std::string name;
-                    bool term = false;
-                    while (++i < in.size()) 
-                    {
-                        if (in[i] == '}') { term = true; break; }
-                        name += in[i];
-                    }
-                    if (!term) throw mp::filter::FilterException
-                        ("Unterminated var ref in '"+in+"' at "
-                         + boost::lexical_cast<std::string>(i));
-                    std::map<std::string, std::string>::const_iterator it
-                        = vars.find(name);
-                    if (it != vars.end())
-                        out += it->second;
-                }
-                else
-                {
-                    throw mp::filter::FilterException
-                        ("Malformed or trimmed var ref in '"
-                         +in+"' at "+boost::lexical_cast<std::string>(i)); 
-                }
-                continue;
-            }
-            //passthru
-            out += in[i];
-            esc = false;
-        }
-        return out;
+    ~TestConfig() 
+    { 
+        std::cout << "global teardown\n"; 
     }
-    
-    void configure(
-            const std::string & req_uri_rx, 
-            const std::string & req_uri_pat, 
-            const std::string & resp_uri_rx, 
-            const std::string & resp_uri_pat) 
-    {
-       this->req_uri_rx = req_uri_rx;
-       this->req_uri_pat = req_uri_pat;
-       //pick up names
-       parse_groups(req_uri_rx, groups_by_num, groups_by_name);
-       this->resp_uri_rx = resp_uri_rx;
-       this->resp_uri_pat = resp_uri_pat;
-    };
-
-private:
-    std::map<std::string, std::string> vars;
-    std::string req_uri_rx;
-    std::string resp_uri_rx;
-    std::string req_uri_pat;
-    std::string resp_uri_pat;
-    std::map<int, std::string> groups_by_num;
-    std::map<std::string, int> groups_by_name;
-
 };
 
+BOOST_GLOBAL_FIXTURE( TestConfig );
+*/
 
 BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
 {
     try
     {
-       FilterHeaderRewrite fhr;
+        std::cout << "Running non-xml config test case" << std::endl;
+        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>\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: ${host}' />\n"
+            "   </rule>\n"
+            "   <within rule=\"url\"/>\n"
+            " </request>\n"
+            " <response>\n"
+            "   <rule name=\"url\">\n"
+            "     <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
+            "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n" 
+            "  </rule>\n"
+            "  <within rule=\"url\"/>\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);
+       
+        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);
+
+        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"
+            "\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\";</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\" />"
+            "</body></html>";
+
+        const char *resp_expected =
+            "HTTP/1.1 200 OK\r\n"
+            "Content-Length: 521\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\";</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\" />"
+            "</body></html>";
+
+        int r;
+        Z_GDU *gdu_res;
+        ODR dec = odr_createmem(ODR_DECODE);
+        odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
+        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);
+
+        //compare buffers
+        std::cout << "Expected result:\n" << resp_expected << std::endl;
+
+        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);
+        
+        BOOST_CHECK(resp_result);
+        BOOST_CHECK_EQUAL((size_t) resp_result_len, strlen(resp_expected));
+
+        std::cout << "Rewriten result:\n" << resp_result << std::endl;
+        std::cout << "Rewriten result buf len: " << resp_result_len 
+            << std::endl;
+
+        BOOST_CHECK(memcmp(resp_result, resp_expected, resp_result_len) == 0);
+
+        odr_destroy(dec);
+        odr_destroy(enc);
     }
-    catch ( ... ) {
+    catch (std::exception & e) {
+        std::cout << e.what();
+        std::cout << std::endl;
         BOOST_CHECK (false);
     }
 }
 
+/*
 BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
 {
     try
     {
+        std::cout << "Running xml config test case" << std::endl;
         mp::RouterChain router;
+        mp::filter::HttpRewrite fhr;
 
-        FilterHeaderRewrite fhr;
-        fhr.configure(
-                ".*?(?P<host>[^:]+):(?P<port>\\d+).*",
-                "http://${host}:${port}/somepath",
-                ".*(localhost).*",
-                "http:://g");
-        mp::filter::HTTPClient hc;
+        std::string xmlconf =
+            "<?xml version='1.0'?>\n"
+            "<filter xmlns='http://indexdata.com/metaproxy'\n"
+            "        id='rewrite1' type='http_rewrite'>\n"
+            " <request>\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: ${host}' />\n" 
+            " </request>\n"
+            " <response>\n"
+            "   <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
+            "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\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);
         
         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);
+        "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 
         pack.request() = gdu_req;
 
+        //create the http response
+
+        const char *resp_buf =
+            "HTTP/1.1 200 OK\r\n"
+            "Content-Length: 50\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\";</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\" />"
+            "</body></html>";
+
+        const char *resp_buf_rew =
+            "HTTP/1.1 200 OK\r\n"
+            "Content-Length: 50\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\";</script>"
+            "<body>"
+            "<p>Welcome to our website. It doesn't make it easy to get pro"
+            "xified"
+            "<a href=\"http://proxyhost/proxypath/targetsite/page.html\">"
+            "  An absolute link</a>"
+            "<a target=_blank href='http://proxyhost/proxypath/targetsite/anotherpage.html\">"
+            "  Another abs link</a>"
+            "<a href=\"/docs/page2.html\" />"
+            "</body></html>";
+
+        int r;
+        Z_GDU *gdu_res;
+        ODR odr2 = odr_createmem(ODR_DECODE);
+        odr_setbuf(odr2, (char *) resp_buf, strlen(resp_buf), 0);
+        r = z_GDU(odr2, &gdu_res, 0, 0);
+
+        BOOST_CHECK(r == 0);
+        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 = pack.response().get();
-        BOOST_CHECK(gdu_res);
-        BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_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->u.HTTP_Response;
+        Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
         BOOST_CHECK(hres);
 
+        //how to compare the buffers:
+
+        odr_destroy(odr2);
     }
     catch (std::exception & e) {
         std::cout << e.what();
+        std::cout << std::endl;
         BOOST_CHECK (false);
     }
 }
+*/
 
 /*
  * Local variables: