X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Ftest_filter_rewrite.cpp;h=430e7846dfd2911fdfc3c952923675737579702c;hb=327c0b4339d90ee73a9dac0d692bc94bf7f05ea1;hp=eee048d79b9a2928d4a84ac2427cfbd804b2d6f8;hpb=c106731ba24ab8570ca249809c7aab46364c303a;p=metaproxy-moved-to-github.git

diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp
index eee048d..430e784 100644
--- a/src/test_filter_rewrite.cpp
+++ b/src/test_filter_rewrite.cpp
@@ -21,6 +21,7 @@ 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/package.hpp>
@@ -28,6 +29,8 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <boost/regex.hpp>
 #include <boost/lexical_cast.hpp>
 
+#include <yaz/log.h>
+
 #define BOOST_AUTO_TEST_MAIN
 #define BOOST_TEST_DYN_LINK
 
@@ -36,384 +39,301 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 using namespace boost::unit_test;
 namespace mp = metaproxy_1;
 
-typedef std::pair<std::string, std::string> string_pair;
-typedef std::vector<string_pair> spair_vec;
-typedef spair_vec::iterator spv_iter;
-
-class FilterHeaderRewrite: public mp::filter::Base {
-public:
-    void process(mp::Package & package) const 
+struct TestConfig {
+    TestConfig()   
     {
-        Z_GDU *gdu = package.request().get();
-        //map of request/response vars
-        std::map<std::string, std::string> vars;
-        //we have an http req
-        if (gdu && gdu->which == Z_GDU_HTTP_Request)
-        {
-            Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
-            mp::odr o;
-            std::cout << ">> Request headers" << std::endl;
-            rewrite_reqline(o, hreq, vars);
-            rewrite_headers(o, hreq->headers, vars);
-            rewrite_body(o, &hreq->content_buf, &hreq->content_len, vars);
-            package.request() = gdu;
-        }
-        package.move();
-        gdu = package.response().get();
-        if (gdu && gdu->which == Z_GDU_HTTP_Response)
-        {
-            Z_HTTP_Response *hres = gdu->u.HTTP_Response;
-            std::cout << "Response " << hres->code;
-            std::cout << "<< Respose headers" << std::endl;
-            mp::odr o;
-            rewrite_headers(o, hres->headers, vars);
-            rewrite_body(o, &hres->content_buf, &hres->content_len, vars);
-            package.response() = gdu;
-        }
+        std::cout << "global setup\n"; 
+        yaz_log_init_level(YLOG_ALL);
     }
-
-    void rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq,
-            std::map<std::string, std::string> & vars) const 
-    {
-        //rewrite the request line
-        std::string path;
-        if (strstr(hreq->path, "http://") == hreq->path)
-        {
-            std::cout << "Path in the method line is absolute, " 
-                "possibly a proxy request\n";
-            path += hreq->path;
-        }
-        else
-        {
-            //TODO what about proto
-            path += z_HTTP_header_lookup(hreq->headers, "Host");
-            path += hreq->path; 
-        }
-        std::cout << "Proxy request URL is " << path << std::endl;
-        std::string npath = 
-            test_patterns(vars, path, req_uri_pats, req_groups_bynum);
-        std::cout << "Resp request URL is " << npath << std::endl;
-        if (!npath.empty())
-            hreq->path = odr_strdup(o, npath.c_str());
-    }
- 
-    void rewrite_headers (mp::odr & o, Z_HTTP_Header *headers,
-            std::map<std::string, std::string> & vars) const 
-    {
-        for (Z_HTTP_Header *header = headers;
-                header != 0; 
-                header = header->next) 
-        {
-            std::string sheader(header->name);
-            sheader += ": ";
-            sheader += header->value;
-            std::cout << header->name << ": " << header->value << std::endl;
-            std::string out = test_patterns(vars, 
-                    sheader, 
-                    req_uri_pats, req_groups_bynum);
-            if (!out.empty()) 
-            {
-                size_t pos = out.find(": ");
-                if (pos == std::string::npos)
-                {
-                    std::cout << "Header malformed during rewrite, ignoring";
-                    continue;
-                }
-                header->name = odr_strdup(o, out.substr(0, pos).c_str());
-                header->value = odr_strdup(o, out.substr(pos+2, 
-                            std::string::npos).c_str());
-            }
-        }
+    ~TestConfig() 
+    { 
+        std::cout << "global teardown\n"; 
     }
+};
+
+BOOST_GLOBAL_FIXTURE( TestConfig );
 
-    void rewrite_body (mp::odr & o, char **content_buf, int *content_len,
-            std::map<std::string, std::string> & vars) const 
+BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
+{
+    try
     {
-        if (*content_buf)
-        {
-            std::string body(*content_buf);
-            std::string nbody = 
-                test_patterns(vars, body, req_uri_pats, req_groups_bynum);
-            if (!nbody.empty())
-            {
-                *content_buf = odr_strdup(o, nbody.c_str());
-                *content_len = nbody.size();
-            }
-        }
-    }
+        std::cout << "Running non-xml config test case" << std::endl;
+        mp::RouterChain router;
+        mp::filter::HttpRewrite fhr;
+        
+        //configure the filter
+        mp::filter::HttpRewrite::spair_vec vec_req;
+        vec_req.push_back(std::make_pair(
+        "(?<proto>http\\:\\/\\/s?)(?<pxhost>[^\\/?#]+)\\/(?<pxpath>[^\\/]+)"
+        "\\/(?<host>[^\\/]+)(?<path>.*)",
+        "${proto}${host}${path}"
+        ));
+        vec_req.push_back(std::make_pair(
+        "(?:Host\\: )(.*)",
+        "Host: ${host}"
+        ));
 
+        mp::filter::HttpRewrite::spair_vec vec_res;
+        vec_res.push_back(std::make_pair(
+        "(?<proto>http\\:\\/\\/s?)(?<host>[^\\/?# \"'>]+)\\/(?<path>[^ \"'>]+)",
+        "${proto}${pxhost}/${pxpath}/${host}/${path}"
+        ));
+        
+        fhr.configure(vec_req, vec_res);
+        
+        router.append(fhr);
 
-    void configure(const xmlNode* ptr, bool test_only, const char *path) {};
+        // create an http request
+        mp::Package pack;
 
-    /**
-     * Tests pattern from the vector in order and executes recipe on
-       the first match.
-     */
-    const std::string test_patterns(
-            std::map<std::string, std::string> & vars,
-            const std::string & txt, 
-            const spair_vec & uri_pats,
-            const std::vector<std::map<int, std::string> > & groups_bynum_vec)
-        const
-    {
-        for (int i = 0; i < uri_pats.size(); i++) 
-        {
-            std::string out = search_replace(vars, txt, 
-                    uri_pats[i].first, uri_pats[i].second,
-                    groups_bynum_vec[i]);
-            if (!out.empty()) return out;
-        }
-        return "";
-    }
+        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;
 
-    const std::string search_replace(
-            std::map<std::string, std::string> & vars,
-            const std::string & txt,
-            const std::string & uri_re,
-            const std::string & uri_pat,
-            const std::map<int, std::string> & groups_bynum) const
-    {
-        //exec regex against value
-        boost::regex re(uri_re);
-        boost::smatch what;
-        std::string::const_iterator start, end;
-        start = txt.begin();
-        end = txt.end();
-        std::string out;
-        while (regex_search(start, end, what, re)) //find next full match
+        //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)
         {
-            unsigned i;
-            for (i = 1; i < what.size(); ++i)
-            {
-                //check if the group is named
-                std::map<int, std::string>::const_iterator it
-                    = groups_bynum.find(i);
-                if (it != groups_bynum.end()) 
-                {   //it is
-                    std::string name = it->second;
-                    if (!what[i].str().empty())
-                        vars[name] = what[i];
-                }
-
-            }
-            //prepare replacement string
-            std::string rvalue = sub_vars(uri_pat, vars);
-            //rewrite value
-            std::string rhvalue = what.prefix().str() 
-                + rvalue + what.suffix().str();
-            std::cout << "! Rewritten '"+what.str(0)+"' to '"+rvalue+"'\n";
-            out += rhvalue;
-            start = what[0].second; //move search forward
+            BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
         }
-        return out;
-    }
 
-    static void parse_groups(
-            const spair_vec & uri_pats,
-            std::vector<std::map<int, std::string> > & groups_bynum_vec)
-    {
-        for (int h = 0; h < uri_pats.size(); h++) 
-        {
-            int gnum = 0;
-            bool esc = false;
-            //regex is first, subpat is second
-            std::string str = uri_pats[h].first;
-            //for each pair we have an indexing map
-            std::map<int, std::string> groups_bynum;
-            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] == ':') //non-capturing
-                        {
-                            if (gnum > 0) gnum--;
-                            i++;
-                            continue;
-                        }
-                        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;
-                            std::cout << "Found named group '" << gname 
-                                << "' at $" << gnum << std::endl;
-                        }
-                    }
-                }
-                esc = false;
-            }
-            groups_bynum_vec.push_back(groups_bynum);
-        }
-    }
+        pack.response() = gdu_res;
 
-    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;
-    }
-    
-    void configure(
-            const spair_vec req_uri_pats,
-            const spair_vec res_uri_pats)
-    {
-       //TODO should we really copy them out?
-       this->req_uri_pats = req_uri_pats;
-       this->res_uri_pats = res_uri_pats;
-       //pick up names
-       parse_groups(req_uri_pats, req_groups_bynum);
-       parse_groups(res_uri_pats, res_groups_bynum);
-    };
-
-private:
-    spair_vec req_uri_pats;
-    spair_vec res_uri_pats;
-    std::vector<std::map<int, std::string> > req_groups_bynum;
-    std::vector<std::map<int, std::string> > res_groups_bynum;
+        //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;
 
-BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
-{
-    try
-    {
-       FilterHeaderRewrite fhr;
+        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(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;
-
-        FilterHeaderRewrite fhr;
-        
-        spair_vec vec_req;
-        vec_req.push_back(std::make_pair(
-        "(?<proto>http\\:\\/\\/s?)(?<pxhost>[^\\/?#]+)\\/(?<pxpath>[^\\/]+)"
-        "\\/(?<host>[^\\/]+)(?<path>.*)",
-        "${proto}${host}${path}"
-        ));
-        vec_req.push_back(std::make_pair(
-        "(?:Host\\: )(.*)",
-        "Host: localhost"
-        ));
-
-        spair_vec vec_res;
-        vec_res.push_back(std::make_pair(
-        "(?<proto>http\\:\\/\\/s?)(?<host>[^\\/?#]+)\\/(?<path>[^ >]+)",
-        "http://${pxhost}/${pxpath}/${host}/${path}"
-        ));
-        
-        fhr.configure(vec_req, vec_res);
-
-        mp::filter::HTTPClient hc;
+        mp::filter::HttpRewrite fhr;
+
+        std::string xmlconf =
+            "<?xml version='1.0'?>\n"
+            "<filter xmlns='http://indexdata.com/metaproxy'\n"
+            "        id='rewrite1' type='http_rewrite'>\n"
+            " <request>\n"
+            "   <rewrite from='"
+    "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
+    "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
+            "            to='${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://proxyhost/proxypath/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: