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         std::cout << "Running non-xml config test case" << std::endl;
  47         mp::RouterChain router;
  48         mp::filter::HttpRewrite fhr;
  49
  50         std::string xmlconf =
  51             "<?xml version='1.0'?>\n"
  52             "<filter xmlns='http://indexdata.com/metaproxy'\n"
  53             "        id='rewrite1' type='http_rewrite'>\n"
  54             " <request verbose=\"1\">\n"
  55             "  <rule name=\"null\"/>\n"
  56             "  <rule name=\"url\">\n"
  57             "     <rewrite from='"
  58     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
  59     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
  60             "            to='${proto}${host}${path}' />\n"
  61             "     <rewrite from='(?:Host: )(.*)'\n"
  62             "            to='Host: ${host}' />\n"
  63             "  </rule>\n"
  64             "  <content type=\"headers\">\n"
  65             "    <within header=\"link\" rule=\"null\"/>\n"
  66             "    <within reqline=\"1\" rule=\"url\"/>\n"
  67             "  </content>\n"
  68             " </request>\n"
  69             " <response verbose=\"1\">\n"
  70             "  <rule name=\"null\"/>\n"
  71             "  <rule name=\"url\">\n"
  72             "     <rewrite from='foo' to='bar'/>\n"
  73             "     <rewrite from='^cx' to='cy'/>\n"
  74             "     <rewrite from='"
  75     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
  76             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
  77             "  </rule>\n"
  78             "  <content type=\"headers\">\n"
  79             "    <within header=\"link\" rule=\"url\"/>\n"
  80             "  </content>\n"
  81             "  <content type=\"html\" mime=\"text/xml|text/html\">\n"
  82             "    <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
  83             "    <within tag=\"script\" attr=\"#text\" type=\"quoted-literal\" rule=\"url\"/>\n"
  84             "    <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
  85             "    <within attr=\"href,src\" rule=\"url\"/>\n"
  86             "  </content>\n"
  87             "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
  88             "    <within rule=\"url\"/>\n"
  89             "  </content>\n"
  90             " </response>\n"
  91             "</filter>\n"
  92         ;
  93
  94         std::cout << xmlconf;
  95
  96         // reading and parsing XML conf
  97         xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
  98         BOOST_CHECK(doc);
  99         xmlNode *root_element = xmlDocGetRootElement(doc);
 100         fhr.configure(root_element, true, "");
 101         xmlFreeDoc(doc);
 102
 103         router.append(fhr);
 104
 105         // create an http request
 106         mp::Package pack;
 107
 108         mp::odr odr;
 109         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
 110         "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 111
 112         pack.request() = gdu_req;
 113
 114         //create the http response
 115
 116         const char *resp_buf =
 117             "HTTP/1.1 200 OK\r\n"
 118             "Content-Type: text/html\r\n"
 119             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 120             "Link: </dir/file.xml>; rel=relative\r\n"
 121             "\r\n"
 122             "<html><head><title>Hello proxy!</title>"
 123             "<style>"
 124             "body {"
 125             "  background-image:url('http://targetsite/images/bg.png');"
 126             "}"
 127             "</style>"
 128             "</head>"
 129             "<script>var jslink=\"http://targetsite/webservice.xml\";"
 130             "var some=\"foo\"; foo=1;"
 131             "</script>"
 132             "<body>"
 133             "<p>Welcome to our website. It doesn't make it easy to get pro"
 134             "xified"
 135             "<a href=\"http://targetsite/page2.html\">"
 136             "  An absolute link</a>"
 137             "<a target=_blank href=\"http://targetsite/page3.html\">"
 138             "  Another abs link</a>"
 139             "<a href=\"/docs/page4.html\" />"
 140             "<a href=\"cxcx\" />"
 141             "<a href=\"cx \" />"
 142             "</body></html>";
 143
 144         const char *resp_expected =
 145             "HTTP/1.1 200 OK\r\n"
 146             "Content-Length: 573\r\n"
 147             "Content-Type: text/html\r\n"
 148             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 149             "Link: </dir/file.xml>; rel=relative\r\n"
 150             "\r\n"
 151             "<html><head><title>Hello proxy!</title>"
 152             "<style>"
 153             "body {"
 154             "  background-image:url('http://proxyhost/proxypath/targetsite/images/bg.png');"
 155             "}"
 156             "</style>"
 157             "</head>"
 158             "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
 159             "var some=\"bar\"; foo=1;"
 160             "</script>"
 161             "<body>"
 162             "<p>Welcome to our website. It doesn't make it easy to get pro"
 163             "xified"
 164             "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
 165             "  An absolute link</a>"
 166             "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
 167             "  Another abs link</a>"
 168             "<a href=\"/docs/page4.html\"/>"
 169             "<a href=\"cycx\"/>"
 170             "<a href=\"cy \"/>"
 171             "</body></html>";
 172
 173         Z_GDU *gdu_res;
 174         mp::odr dec(ODR_DECODE);
 175         mp::odr enc(ODR_ENCODE);
 176         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 177         int r = z_GDU(dec, &gdu_res, 0, 0);
 178
 179         BOOST_CHECK(r);
 180         if (r)
 181         {
 182             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 183
 184             pack.response() = gdu_res;
 185
 186             //feed to the router
 187             pack.router(router).move();
 188
 189             //analyze the response
 190             Z_GDU *gdu_res_rew = pack.response().get();
 191             BOOST_CHECK(gdu_res_rew);
 192             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 193
 194             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 195             BOOST_CHECK(hres);
 196
 197             z_GDU(enc, &gdu_res_rew, 0, 0);
 198             char *resp_result;
 199             int resp_result_len;
 200             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 201
 202             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 203                 && !memcmp(resp_result, resp_expected, resp_result_len);
 204             BOOST_CHECK(equal);
 205
 206             if (!equal)
 207             {
 208                 //compare buffers
 209                 std::cout << "Expected result:\n" << resp_expected << "\n";
 210                 std::cout << "Got result:\n" << "\n";
 211                 fflush(stdout);
 212                 fwrite(resp_result, 1, resp_result_len, stdout);
 213                 fflush(stdout);
 214                 std::cout << "\nGot result buf len: " << resp_result_len
 215                           << "\n";
 216             }
 217         }
 218     }
 219     catch (std::exception & e) {
 220         std::cout << e.what();
 221         std::cout << std::endl;
 222         BOOST_CHECK (false);
 223     }
 224 }
 225
 226
 227 BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 )
 228 {
 229     try
 230     {
 231         std::cout << "Running non-xml config test case" << std::endl;
 232         mp::RouterChain router;
 233         mp::filter::HttpRewrite fhr;
 234
 235         std::string xmlconf =
 236             "<?xml version='1.0'?>\n"
 237             "<filter xmlns='http://indexdata.com/metaproxy'\n"
 238             "        id='rewrite1' type='http_rewrite'>\n"
 239             " <request verbose=\"1\">\n"
 240             "  <rule name=\"null\"/>\n"
 241             "  <rule name=\"url\">\n"
 242             "     <rewrite from='"
 243     "(?&lt;proto>https?://)(?&lt;pxhost>[^ /?#]+)/(?&lt;pxpath>[^ /]+)"
 244     "/(?&lt;host>[^ /]+)(?&lt;path>[^ ]*)'\n"
 245             "            to='${proto}${host}${path}' />\n"
 246             "     <rewrite from='(?:Host: )(.*)'\n"
 247             "            to='Host: ${host}' />\n"
 248             "  </rule>\n"
 249             "  <content type=\"headers\">\n"
 250             "    <within header=\"link\" rule=\"null\"/>\n"
 251             "    <within reqline=\"1\" rule=\"url\"/>\n"
 252             "  </content>\n"
 253             " </request>\n"
 254             " <response verbose=\"1\">\n"
 255             "  <rule name=\"null\"/>\n"
 256             "  <rule name=\"url\">\n"
 257             "     <rewrite from='foo' to='bar'/>\n"
 258             "     <rewrite from='^cx' to='cy'/>\n"
 259             "     <rewrite from='"
 260     "(?&lt;proto>https?://)(?&lt;host>[^/?# &quot;&apos;>]+)/(?&lt;path>[^  &quot;&apos;>]+)'\n"
 261             "            to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
 262             "  </rule>\n"
 263             "  <content type=\"headers\">\n"
 264             "    <within header=\"link\" rule=\"url\"/>\n"
 265             "  </content>\n"
 266             "  <content type=\"html\" mime=\"text/xml|text/html\">\n"
 267             "    <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
 268             "    <within tag=\"script\" attr=\"#text\" rule=\"url\"/>\n"
 269             "    <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
 270             "    <within attr=\"href,src\" rule=\"url\"/>\n"
 271             "  </content>\n"
 272             "  <content type=\"quoted-literal\" mime=\".*javascript\">\n"
 273             "    <within rule=\"url\"/>\n"
 274             "  </content>\n"
 275             " </response>\n"
 276             "</filter>\n"
 277         ;
 278
 279         std::cout << xmlconf;
 280
 281         // reading and parsing XML conf
 282         xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
 283         BOOST_CHECK(doc);
 284         xmlNode *root_element = xmlDocGetRootElement(doc);
 285         fhr.configure(root_element, true, "");
 286         xmlFreeDoc(doc);
 287
 288         router.append(fhr);
 289
 290         // create an http request
 291         mp::Package pack;
 292
 293         mp::odr odr;
 294         Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
 295         "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
 296
 297         pack.request() = gdu_req;
 298
 299         //create the http response
 300
 301         const char *resp_buf =
 302             "HTTP/1.1 200 OK\r\n"
 303             "Content-Type: application/javascript\r\n"
 304             "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
 305             "Link: </dir/file.xml>; rel=relative\r\n"
 306             "\r\n"
 307             "// \"\n"
 308             "my.location = 'http://targetsite/images/bg.png';\n"
 309             "my.other = \"http://targetsite/images/fg.png\";\n"
 310             "my.thrd = \"other\";\n"
 311             "// \"http://targetsite/images/bg.png\n";
 312
 313         const char *resp_expected =
 314             "HTTP/1.1 200 OK\r\n"
 315             "Content-Length: 195\r\n"
 316             "Content-Type: application/javascript\r\n"
 317             "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
 318             "Link: </dir/file.xml>; rel=relative\r\n"
 319             "\r\n"
 320             "// \"\n"
 321             "my.location = 'http://proxyhost/proxypath/targetsite/images/bg.png';\n"
 322             "my.other = \"http://proxyhost/proxypath/targetsite/images/fg.png\";\n"
 323             "my.thrd = \"other\";\n"
 324             "// \"http://targetsite/images/bg.png\n";
 325
 326         Z_GDU *gdu_res;
 327         mp::odr dec(ODR_DECODE);
 328         mp::odr enc(ODR_ENCODE);
 329         odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
 330         int r = z_GDU(dec, &gdu_res, 0, 0);
 331
 332         BOOST_CHECK(r);
 333         if (r)
 334         {
 335             BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
 336
 337             pack.response() = gdu_res;
 338
 339             //feed to the router
 340             pack.router(router).move();
 341
 342             //analyze the response
 343             Z_GDU *gdu_res_rew = pack.response().get();
 344             BOOST_CHECK(gdu_res_rew);
 345             BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
 346
 347             Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
 348             BOOST_CHECK(hres);
 349
 350             z_GDU(enc, &gdu_res_rew, 0, 0);
 351             char *resp_result;
 352             int resp_result_len;
 353             resp_result = odr_getbuf(enc, &resp_result_len, 0);
 354
 355             int equal = ((size_t) resp_result_len == strlen(resp_expected))
 356                 && !memcmp(resp_result, resp_expected, resp_result_len);
 357             BOOST_CHECK(equal);
 358
 359             if (!equal)
 360             {
 361                 //compare buffers
 362                 std::cout << "Expected result:\n" << resp_expected << "\n";
 363                 std::cout << "Got result:\n" << "\n";
 364                 fflush(stdout);
 365                 fwrite(resp_result, 1, resp_result_len, stdout);
 366                 fflush(stdout);
 367                 std::cout << "\nGot result buf len: " << resp_result_len
 368                           << "\n";
 369             }
 370         }
 371     }
 372     catch (std::exception & e) {
 373         std::cout << e.what();
 374         std::cout << std::endl;
 375         BOOST_CHECK (false);
 376     }
 377 }
 378
 379 /*
 380  * Local variables:
 381  * c-basic-offset: 4
 382  * c-file-style: "Stroustrup"
 383  * indent-tabs-mode: nil
 384  * End:
 385  * vim: shiftwidth=4 tabstop=8 expandtab
 386  */
 387