src/test_filter_rewrite.cpp

   1 /* This file is part of Metaproxy.
   2    Copyright (C) 2005-2013 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         pack.request() = gdu_req;
 115
 116         //create the http response
 117
 118         const char *resp_buf =
 119             "HTTP/1.1 200 OK\r\n"
 120             "Content-Type: text/html\r\n"
 121             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 122             "Link: </dir/file.xml>; rel=relative\r\n"
 123             "\r\n"
 124             "<html><head><title>Hello proxy!</title>"
 125             "<style>"
 126             "body {"
 127             "  background-image:url('http://targetsite/images/bg.png');"
 128             "}"
 129             "</style>"
 130             "</head>"
 131             "<script>var jslink=\"http://targetsite/webservice.xml\";"
 132             "for (i = 0; i<foo; i++) ;\n"
 133             "var some=\"foo\"; foo=1;"
 134             "</script>"
 135             "<body>"
 136             "<p>Welcome to our website. It doesn't make it easy to get pro"
 137             "xified"
 138             "<a href=\"http://targetsite/page2.html\">"
 139             "  An absolute link</a>"
 140             "<a target=_blank href=\"http://targetsite/page3.html\">"
 141             "  Another abs link</a>"
 142             "<a href=\"/docs/page4.html\" />"
 143             "<A href=\"cxcx\" />"
 144             "<a HREF=\"cx \" onclick=\"foo(&quot;foo&quot;);\"/>"
 145             "</body></html>";
 146
 147         const char *resp_expected =
 148             "HTTP/1.1 200 OK\r\n"
 149             "Content-Length: 631\r\n"
 150             "Content-Type: text/html\r\n"
 151             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 152             "Link: </dir/file.xml>; rel=relative\r\n"
 153             "\r\n"
 154             "<html><head><title>Hello proxy!</title>"
 155             "<style>"
 156             "body {"
 157             "  background-image:url('http://proxyhost/proxypath/targetsite/images/bg.png');"
 158             "}"
 159             "</style>"
 160             "</head>"
 161             "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
 162             "for (i = 0; i<foo; i++) ;\n"
 163             "var some=\"bar\"; foo=1;"
 164             "</script>"
 165             "<body>"
 166             "<p>Welcome to our website. It doesn't make it easy to get pro"
 167             "xified"
 168             "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
 169             "  An absolute link</a>"
 170             "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
 171             "  Another abs link</a>"
 172             "<a href=\"/docs/page4.html\"/>"
 173             "<A href=\"cycx\"/>"
 174             "<a HREF=\"cy \" onclick=\"foo(&quot;bar&quot;);\"/>"
 175             "</body></html>";
 176
 177         Z_GDU *gdu_res;
 178         mp::odr dec(ODR_DECODE);
 179         mp::odr enc(ODR_ENCODE);
 180         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 181         int r = z_GDU(dec, &gdu_res, 0, 0);
 182
 183         BOOST_CHECK(r);
 184         if (r)
 185         {
 186             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 187
 188             pack.response() = gdu_res;
 189
 190             //feed to the router
 191             pack.router(router).move();
 192
 193             //analyze the response
 194             Z_GDU *gdu_res_rew = pack.response().get();
 195             BOOST_CHECK(gdu_res_rew);
 196             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 197
 198             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 199             BOOST_CHECK(hres);
 200
 201             z_GDU(enc, &gdu_res_rew, 0, 0);
 202             char *resp_result;
 203             int resp_result_len;
 204             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 205
 206             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 207                 && !memcmp(resp_result, resp_expected, resp_result_len);
 208             BOOST_CHECK(equal);
 209
 210             if (!equal)
 211             {
 212                 //compare buffers
 213                 std::cout << "Expected result:\n" << resp_expected << "\n";
 214                 std::cout << "Got result:\n";
 215                 fflush(stdout);
 216                 fwrite(resp_result, 1, resp_result_len, stdout);
 217                 fflush(stdout);
 218                 std::cout << "\nGot result buf len: " << resp_result_len
 219                           << "\n";
 220             }
 221         }
 222     }
 223     catch (std::exception & e) {
 224         std::cout << e.what();
 225         std::cout << std::endl;
 226         BOOST_CHECK (false);
 227     }
 228 }
 229
 230 BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
 231 {
 232     try
 233     {
 234         mp::RouterChain router;
 235         mp::filter::HttpRewrite fhr;
 236
 237         std::string xmlconf =
 238             "<?xml version='1.0'?>\n"
 239             "<filter xmlns='http://indexdata.com/metaproxy'\n"
 240             "        id='rewrite1' type='http_rewrite'>\n"
 241             " <request verbose=\"1\">\n"
 242             "  <rule name=\"null\"/>\n"
 243             "  <rule name=\"url\">\n"
 244             "     <rewrite from='"
 245     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
 246     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
 247             "            to='${proto}${host}${path}' />\n"
 248             "     <rewrite from='(?:Host: )(.*)'\n"
 249             "            to='Host: ${host}' />\n"
 250             "  </rule>\n"
 251             "  <content type=\"headers\">\n"
 252             "    <within header=\"link\" rule=\"null\"/>\n"
 253             "    <within reqline=\"1\" rule=\"url\"/>\n"
 254             "  </content>\n"
 255             " </request>\n"
 256             " <response verbose=\"1\">\n"
 257             "  <rule name=\"null\"/>\n"
 258             "  <rule name=\"url\">\n"
 259             "     <rewrite from='foo' to='bar'/>\n"
 260             "     <rewrite from='^cx' to='cy'/>\n"
 261             "     <rewrite from='"
 262     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
 263             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
 264             "  </rule>\n"
 265             "  <content type=\"headers\">\n"
 266             "    <within header=\"link\" rule=\"url\"/>\n"
 267             "  </content>\n"
 268             "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
 269             "    <within rule=\"url\"/>\n"
 270             "  </content>\n"
 271             " </response>\n"
 272             "</filter>\n"
 273         ;
 274
 275         std::cout << xmlconf;
 276
 277         // reading and parsing XML conf
 278         xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
 279         BOOST_CHECK(doc);
 280         xmlNode *root_element = xmlDocGetRootElement(doc);
 281         fhr.configure(root_element, true, "");
 282         xmlFreeDoc(doc);
 283
 284         router.append(fhr);
 285
 286         // create an http request
 287         mp::Package pack;
 288
 289         mp::odr odr;
 290         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
 291         "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 292
 293         pack.request() = gdu_req;
 294
 295         //create the http response
 296
 297         const char *resp_buf =
 298             "HTTP/1.1 200 OK\r\n"
 299             "Content-Type: application/javascript\r\n"
 300             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 301             "Link: </dir/file.xml>; rel=relative\r\n"
 302             "\r\n"
 303             "// \"\n"
 304             "my.location = 'http://targetsite/images/bg.png';\n"
 305             "my.other = \"http://targetsite/images/fg.png\";\n"
 306             "my.thrd = \"other\";\n"
 307             "// \"http://targetsite/images/bg.png\n";
 308
 309         const char *resp_expected =
 310             "HTTP/1.1 200 OK\r\n"
 311             "Content-Length: 195\r\n"
 312             "Content-Type: application/javascript\r\n"
 313             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 314             "Link: </dir/file.xml>; rel=relative\r\n"
 315             "\r\n"
 316             "// \"\n"
 317             "my.location = 'http://proxyhost/proxypath/targetsite/images/bg.png';\n"
 318             "my.other = \"http://proxyhost/proxypath/targetsite/images/fg.png\";\n"
 319             "my.thrd = \"other\";\n"
 320             "// \"http://targetsite/images/bg.png\n";
 321
 322         Z_GDU *gdu_res;
 323         mp::odr dec(ODR_DECODE);
 324         mp::odr enc(ODR_ENCODE);
 325         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 326         int r = z_GDU(dec, &gdu_res, 0, 0);
 327
 328         BOOST_CHECK(r);
 329         if (r)
 330         {
 331             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 332
 333             pack.response() = gdu_res;
 334
 335             //feed to the router
 336             pack.router(router).move();
 337
 338             //analyze the response
 339             Z_GDU *gdu_res_rew = pack.response().get();
 340             BOOST_CHECK(gdu_res_rew);
 341             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 342
 343             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 344             BOOST_CHECK(hres);
 345
 346             z_GDU(enc, &gdu_res_rew, 0, 0);
 347             char *resp_result;
 348             int resp_result_len;
 349             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 350
 351             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 352                 && !memcmp(resp_result, resp_expected, resp_result_len);
 353             BOOST_CHECK(equal);
 354
 355             if (!equal)
 356             {
 357                 //compare buffers
 358                 std::cout << "Expected result:\n" << resp_expected << "\n";
 359                 std::cout << "Got result:\n";
 360                 fflush(stdout);
 361                 fwrite(resp_result, 1, resp_result_len, stdout);
 362                 fflush(stdout);
 363                 std::cout << "\nGot result buf len: " << resp_result_len
 364                           << "\n";
 365             }
 366         }
 367     }
 368     catch (std::exception & e) {
 369         std::cout << e.what();
 370         std::cout << std::endl;
 371         BOOST_CHECK (false);
 372     }
 373 }
 374
 375 /*
 376  * Local variables:
 377  * c-basic-offset: 4
 378  * c-file-style: "Stroustrup"
 379  * indent-tabs-mode: nil
 380  * End:
 381  * vim: shiftwidth=4 tabstop=8 expandtab
 382  */
 383