src/test_filter_rewrite.cpp

   1 /* This file is part of Metaproxy.
   2    Copyright (C) Index Data
   3
   4 Metaproxy is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Metaproxy is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17 */
  18
  19 #include "config.hpp"
  20 #include <iostream>
  21 #include <stdexcept>
  22
  23 #include "filter_http_client.hpp"
  24 #include "filter_http_rewrite.hpp"
  25 #include <metaproxy/util.hpp>
  26 #include <metaproxy/router_chain.hpp>
  27 #include <metaproxy/package.hpp>
  28
  29 #include <boost/regex.hpp>
  30 #include <boost/lexical_cast.hpp>
  31
  32 #include <yaz/log.h>
  33
  34 #define BOOST_AUTO_TEST_MAIN
  35 #define BOOST_TEST_DYN_LINK
  36
  37 #include <boost/test/auto_unit_test.hpp>
  38
  39 using namespace boost::unit_test;
  40 namespace mp = metaproxy_1;
  41
  42 BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
  43 {
  44     try
  45     {
  46         mp::RouterChain router;
  47         mp::filter::HttpRewrite fhr;
  48
  49         std::string xmlconf =
  50             "<?xml version='1.0'?>\n"
  51             "<filter xmlns='http://indexdata.com/metaproxy'\n"
  52             "        id='rewrite1' type='http_rewrite'>\n"
  53             " <request verbose=\"1\">\n"
  54             "  <rule name=\"null\"/>\n"
  55             "  <rule name=\"url\">\n"
  56             "     <rewrite from='"
  57     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
  58     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
  59             "            to='${proto}${host}${path}' />\n"
  60             "     <rewrite from='(?:Host: )(.*)'\n"
  61             "            to='Host: ${host}' />\n"
  62             "  </rule>\n"
  63             "  <content type=\"headers\">\n"
  64             "    <within header=\"link\" rule=\"null\"/>\n"
  65             "    <within reqline=\"1\" rule=\"url\"/>\n"
  66             "  </content>\n"
  67             " </request>\n"
  68             " <response verbose=\"1\">\n"
  69             "  <rule name=\"null\"/>\n"
  70             "  <rule name=\"cx\">\n"
  71             "     <rewrite from='^cx' to='cy'/>\n"
  72             "  </rule>\n"
  73             "  <rule name=\"url\">\n"
  74             "     <rewrite from='foo' to='bar'/>\n"
  75             "     <rewrite from='"
  76     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
  77             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
  78             "  </rule>\n"
  79             "  <content type=\"headers\">\n"
  80             "    <within header=\"link\" rule=\"url\"/>\n"
  81             "  </content>\n"
  82             "  <content type=\"html\" mime=\"text/xml|text/html\">\n"
  83             "    <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
  84             "    <within tag=\"script\" attr=\"#text\" type=\"quoted-literal\" rule=\"url\"/>\n"
  85             "    <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
  86             "    <within attr=\"href|src\" rule=\"url,cx\"/>\n"
  87             "    <within attr=\"onclick\" type=\"quoted-literal\" rule=\"url\"/>\n"
  88             "  </content>\n"
  89             "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
  90             "    <within rule=\"url\"/>\n"
  91             "  </content>\n"
  92             " </response>\n"
  93             "</filter>\n"
  94         ;
  95
  96         std::cout << xmlconf;
  97
  98         // reading and parsing XML conf
  99         xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
 100         BOOST_CHECK(doc);
 101         xmlNode *root_element = xmlDocGetRootElement(doc);
 102         fhr.configure(root_element, true, "");
 103         xmlFreeDoc(doc);
 104
 105         router.append(fhr);
 106
 107         // create an http request
 108         mp::Package pack;
 109
 110         mp::odr odr;
 111         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
 112             "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 113
 114         Z_HTTP_Request *hreq = gdu_req->u.HTTP_Request;
 115         z_HTTP_header_set(odr, &hreq->headers,
 116                           "X-Metaproxy-SkipLink", ".* skiplink.com" );
 117         pack.request() = gdu_req;
 118
 119         //create the http response
 120
 121         const char *resp_buf =
 122             "HTTP/1.1 200 OK\r\n"
 123             "Content-Type: text/html\r\n"
 124             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 125             "Link: </dir/file.xml>; rel=relative\r\n"
 126             "\r\n"
 127             "<html><head><title>Hello proxy!</title>"
 128             "<style>"
 129             "body {"
 130             "  background-image:url('http://targetsite/images/bg.png');"
 131             "}"
 132             "</style>"
 133             "</head>"
 134             "<script>var jslink=\"http://targetsite/webservice.xml\";"
 135             "for (i = 0; i<foo; i++) ;\n"
 136             "var some=\"foo\"; foo=1;"
 137             "</script>"
 138             "<body>"
 139             "<p>Welcome to our website. It doesn't make it easy to get pro"
 140             "xified"
 141             "<a href=\"http://targetsite/page2.html\">"
 142             "  An absolute link</a>"
 143             "<a target=_blank href=\"http://targetsite/page3.html\">"
 144             "  Another abs link</a>"
 145             "<a href=\"/docs/page4.html\" />"
 146             "<A href=\"cxcx\" />"
 147             "<a HREF=\"cx \" onclick=\"foo(&quot;foo&quot;);\"/>\n"
 148             "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
 149             "</body></html>";
 150
 151         const char *resp_expected =
 152             "HTTP/1.1 200 OK\r\n"
 153             "Content-Length: 686\r\n"
 154             "Content-Type: text/html\r\n"
 155             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 156             "Link: </dir/file.xml>; rel=relative\r\n"
 157             "\r\n"
 158             "<html><head><title>Hello proxy!</title>"
 159             "<style>"
 160             "body {"
 161             "  background-image:url('http://proxyhost/proxypath/targetsite/images/bg.png');"
 162             "}"
 163             "</style>"
 164             "</head>"
 165             "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
 166             "for (i = 0; i<foo; i++) ;\n"
 167             "var some=\"bar\"; foo=1;"
 168             "</script>"
 169             "<body>"
 170             "<p>Welcome to our website. It doesn't make it easy to get pro"
 171             "xified"
 172             "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
 173             "  An absolute link</a>"
 174             "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
 175             "  Another abs link</a>"
 176             "<a href=\"/docs/page4.html\"/>"
 177             "<A href=\"cycx\"/>"
 178             "<a HREF=\"cy \" onclick=\"foo(&quot;bar&quot;);\"/>\n"
 179             "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
 180             "</body></html>";
 181
 182         Z_GDU *gdu_res;
 183         mp::odr dec(ODR_DECODE);
 184         mp::odr enc(ODR_ENCODE);
 185         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 186         int r = z_GDU(dec, &gdu_res, 0, 0);
 187
 188         BOOST_CHECK(r);
 189         if (r)
 190         {
 191             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 192
 193             pack.response() = gdu_res;
 194
 195             //feed to the router
 196             pack.router(router).move();
 197
 198             //analyze the response
 199             Z_GDU *gdu_res_rew = pack.response().get();
 200             BOOST_CHECK(gdu_res_rew);
 201             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 202
 203             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 204             BOOST_CHECK(hres);
 205
 206             z_GDU(enc, &gdu_res_rew, 0, 0);
 207             char *resp_result;
 208             int resp_result_len;
 209             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 210
 211             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 212                 && !memcmp(resp_result, resp_expected, resp_result_len);
 213             BOOST_CHECK(equal);
 214
 215             if (!equal)
 216             {
 217                 //compare buffers
 218                 std::cout << "Expected result:\n" << resp_expected << "\n";
 219                 std::cout << "Got result:\n";
 220                 fflush(stdout);
 221                 fwrite(resp_result, 1, resp_result_len, stdout);
 222                 fflush(stdout);
 223                 std::cout << "\nGot result buf len: " << resp_result_len
 224                           << "\n";
 225             }
 226         }
 227     }
 228     catch (std::exception & e) {
 229         std::cout << e.what();
 230         std::cout << std::endl;
 231         BOOST_CHECK (false);
 232     }
 233 }
 234
 235 BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
 236 {
 237     try
 238     {
 239         mp::RouterChain router;
 240         mp::filter::HttpRewrite fhr;
 241
 242         std::string xmlconf =
 243             "<?xml version='1.0'?>\n"
 244             "<filter xmlns='http://indexdata.com/metaproxy'\n"
 245             "        id='rewrite1' type='http_rewrite'>\n"
 246             " <request verbose=\"1\">\n"
 247             "  <rule name=\"null\"/>\n"
 248             "  <rule name=\"url\">\n"
 249             "     <rewrite from='"
 250     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
 251     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
 252             "            to='${proto}${host}${path}' />\n"
 253             "     <rewrite from='(?:Host: )(.*)'\n"
 254             "            to='Host: ${host}' />\n"
 255             "  </rule>\n"
 256             "  <content type=\"headers\">\n"
 257             "    <within header=\"link\" rule=\"null\"/>\n"
 258             "    <within reqline=\"1\" rule=\"url\"/>\n"
 259             "  </content>\n"
 260             " </request>\n"
 261             " <response verbose=\"1\">\n"
 262             "  <rule name=\"null\"/>\n"
 263             "  <rule name=\"url\">\n"
 264             "     <rewrite from='foo' to='bar'/>\n"
 265             "     <rewrite from='^cx' to='cy'/>\n"
 266             "     <rewrite from='"
 267     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
 268             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
 269             "  </rule>\n"
 270             "  <content type=\"headers\">\n"
 271             "    <within header=\"link\" rule=\"url\"/>\n"
 272             "  </content>\n"
 273             "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
 274             "    <within rule=\"url\"/>\n"
 275             "  </content>\n"
 276             " </response>\n"
 277             "</filter>\n"
 278         ;
 279
 280         std::cout << xmlconf;
 281
 282         // reading and parsing XML conf
 283         xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
 284         BOOST_CHECK(doc);
 285         xmlNode *root_element = xmlDocGetRootElement(doc);
 286         fhr.configure(root_element, true, "");
 287         xmlFreeDoc(doc);
 288
 289         router.append(fhr);
 290
 291         // create an http request
 292         mp::Package pack;
 293
 294         mp::odr odr;
 295         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
 296         "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 297
 298         pack.request() = gdu_req;
 299
 300         //create the http response
 301
 302         const char *resp_buf =
 303             "HTTP/1.1 200 OK\r\n"
 304             "Content-Type: application/javascript\r\n"
 305             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 306             "Link: </dir/file.xml>; rel=relative\r\n"
 307             "\r\n"
 308             "// \"\n"
 309             "my.location = 'http://targetsite/images/bg.png';\n"
 310             "my.other = \"http://targetsite/images/fg.png\";\n"
 311             "my.thrd = \"other\";\n"
 312             "// \"http://targetsite/images/bg.png\n";
 313
 314         const char *resp_expected =
 315             "HTTP/1.1 200 OK\r\n"
 316             "Content-Length: 195\r\n"
 317             "Content-Type: application/javascript\r\n"
 318             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 319             "Link: </dir/file.xml>; rel=relative\r\n"
 320             "\r\n"
 321             "// \"\n"
 322             "my.location = 'http://proxyhost/proxypath/targetsite/images/bg.png';\n"
 323             "my.other = \"http://proxyhost/proxypath/targetsite/images/fg.png\";\n"
 324             "my.thrd = \"other\";\n"
 325             "// \"http://targetsite/images/bg.png\n";
 326
 327         Z_GDU *gdu_res;
 328         mp::odr dec(ODR_DECODE);
 329         mp::odr enc(ODR_ENCODE);
 330         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 331         int r = z_GDU(dec, &gdu_res, 0, 0);
 332
 333         BOOST_CHECK(r);
 334         if (r)
 335         {
 336             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 337
 338             pack.response() = gdu_res;
 339
 340             //feed to the router
 341             pack.router(router).move();
 342
 343             //analyze the response
 344             Z_GDU *gdu_res_rew = pack.response().get();
 345             BOOST_CHECK(gdu_res_rew);
 346             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 347
 348             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 349             BOOST_CHECK(hres);
 350
 351             z_GDU(enc, &gdu_res_rew, 0, 0);
 352             char *resp_result;
 353             int resp_result_len;
 354             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 355
 356             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 357                 && !memcmp(resp_result, resp_expected, resp_result_len);
 358             BOOST_CHECK(equal);
 359
 360             if (!equal)
 361             {
 362                 //compare buffers
 363                 std::cout << "Expected result:\n" << resp_expected << "\n";
 364                 std::cout << "Got result:\n";
 365                 fflush(stdout);
 366                 fwrite(resp_result, 1, resp_result_len, stdout);
 367                 fflush(stdout);
 368                 std::cout << "\nGot result buf len: " << resp_result_len
 369                           << "\n";
 370             }
 371         }
 372     }
 373     catch (std::exception & e) {
 374         std::cout << e.what();
 375         std::cout << std::endl;
 376         BOOST_CHECK (false);
 377     }
 378 }
 379
 380 /*
 381  * Local variables:
 382  * c-basic-offset: 4
 383  * c-file-style: "Stroustrup"
 384  * indent-tabs-mode: nil
 385  * End:
 386  * vim: shiftwidth=4 tabstop=8 expandtab
 387  */
 388