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