src/filter_http_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 <metaproxy/filter.hpp>
  21 #include <metaproxy/package.hpp>
  22 #include <metaproxy/util.hpp>
  23 #include "filter_http_rewrite.hpp"
  24
  25 #include <yaz/zgdu.h>
  26 #include <yaz/log.h>
  27
  28 #include <boost/regex.hpp>
  29 #include <boost/lexical_cast.hpp>
  30
  31 #include <list>
  32 #include <map>
  33
  34 #if HAVE_SYS_TYPES_H
  35 #include <sys/types.h>
  36 #endif
  37
  38 namespace mp = metaproxy_1;
  39 namespace yf = mp::filter;
  40
  41 yf::HttpRewrite::HttpRewrite()
  42 {
  43 }
  44
  45 yf::HttpRewrite::~HttpRewrite()
  46 {
  47 }
  48
  49 void yf::HttpRewrite::process(mp::Package & package) const
  50 {
  51     yaz_log(YLOG_DEBUG, "HttpRewrite begins....");
  52     Z_GDU *gdu = package.request().get();
  53     //map of request/response vars
  54     std::map<std::string, std::string> vars;
  55     //we have an http req
  56     if (gdu && gdu->which == Z_GDU_HTTP_Request)
  57     {
  58         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
  59         mp::odr o;
  60         rewrite_reqline(o, hreq, vars);
  61         yaz_log(YLOG_DEBUG, ">> Request headers");
  62         rewrite_headers(o, hreq->headers, vars);
  63         rewrite_body(o, &hreq->content_buf, &hreq->content_len, vars);
  64         package.request() = gdu;
  65     }
  66     package.move();
  67     gdu = package.response().get();
  68     if (gdu && gdu->which == Z_GDU_HTTP_Response)
  69     {
  70         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
  71         yaz_log(YLOG_DEBUG, "Response code %d", hres->code);
  72         mp::odr o;
  73         yaz_log(YLOG_DEBUG, "<< Respose headers");
  74         rewrite_headers(o, hres->headers, vars);
  75         rewrite_body(o, &hres->content_buf, &hres->content_len, vars);
  76         package.response() = gdu;
  77     }
  78 }
  79
  80 void yf::HttpRewrite::rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq,
  81         std::map<std::string, std::string> & vars) const
  82 {
  83     //rewrite the request line
  84     std::string path;
  85     if (strstr(hreq->path, "http://") == hreq->path)
  86     {
  87         yaz_log(YLOG_DEBUG, "Path in the method line is absolute, "
  88             "possibly a proxy request");
  89         path += hreq->path;
  90     }
  91     else
  92     {
  93         //TODO what about proto
  94         path += z_HTTP_header_lookup(hreq->headers, "Host");
  95         path += hreq->path;
  96     }
  97     yaz_log(YLOG_DEBUG, "Proxy request URL is %s", path.c_str());
  98     std::string npath =
  99         test_patterns(vars, path, req_uri_pats, req_groups_bynum);
 100     if (!npath.empty())
 101     {
 102         yaz_log(YLOG_DEBUG, "Rewritten request URL is %s", npath.c_str());
 103         hreq->path = odr_strdup(o, npath.c_str());
 104     }
 105 }
 106
 107 void yf::HttpRewrite::rewrite_headers (mp::odr & o, Z_HTTP_Header *headers,
 108         std::map<std::string, std::string> & vars) const
 109 {
 110     for (Z_HTTP_Header *header = headers;
 111             header != 0;
 112             header = header->next)
 113     {
 114         std::string sheader(header->name);
 115         sheader += ": ";
 116         sheader += header->value;
 117         yaz_log(YLOG_DEBUG, "%s: %s", header->name, header->value);
 118         std::string out = test_patterns(vars,
 119                 sheader,
 120                 req_uri_pats, req_groups_bynum);
 121         if (!out.empty())
 122         {
 123             size_t pos = out.find(": ");
 124             if (pos == std::string::npos)
 125             {
 126                 yaz_log(YLOG_DEBUG, "Header malformed during rewrite, ignoring");
 127                 continue;
 128             }
 129             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 130             header->value = odr_strdup(o, out.substr(pos+2,
 131                         std::string::npos).c_str());
 132         }
 133     }
 134 }
 135
 136 void yf::HttpRewrite::rewrite_body (mp::odr & o, char **content_buf, int *content_len,
 137         std::map<std::string, std::string> & vars) const
 138 {
 139     if (*content_buf)
 140     {
 141         std::string body(*content_buf);
 142         std::string nbody =
 143             test_patterns(vars, body, req_uri_pats, req_groups_bynum);
 144         if (!nbody.empty())
 145         {
 146             *content_buf = odr_strdup(o, nbody.c_str());
 147             *content_len = nbody.size();
 148         }
 149     }
 150 }
 151
 152 /**
 153  * Tests pattern from the vector in order and executes recipe on
 154  the first match.
 155  */
 156 const std::string yf::HttpRewrite::test_patterns(
 157         std::map<std::string, std::string> & vars,
 158         const std::string & txt,
 159         const spair_vec & uri_pats,
 160         const std::vector<std::map<int, std::string> > & groups_bynum_vec)
 161     const
 162 {
 163     for (unsigned i = 0; i < uri_pats.size(); i++)
 164     {
 165         std::string out = search_replace(vars, txt,
 166                 uri_pats[i].first, uri_pats[i].second,
 167                 groups_bynum_vec[i]);
 168         if (!out.empty()) return out;
 169     }
 170     return "";
 171 }
 172
 173
 174 const std::string yf::HttpRewrite::search_replace(
 175         std::map<std::string, std::string> & vars,
 176         const std::string & txt,
 177         const std::string & uri_re,
 178         const std::string & uri_pat,
 179         const std::map<int, std::string> & groups_bynum) const
 180 {
 181     //exec regex against value
 182     boost::regex re(uri_re);
 183     boost::smatch what;
 184     std::string::const_iterator start, end;
 185     start = txt.begin();
 186     end = txt.end();
 187     std::string out;
 188     while (regex_search(start, end, what, re)) //find next full match
 189     {
 190         unsigned i;
 191         for (i = 1; i < what.size(); ++i)
 192         {
 193             //check if the group is named
 194             std::map<int, std::string>::const_iterator it
 195                 = groups_bynum.find(i);
 196             if (it != groups_bynum.end())
 197             {   //it is
 198                 std::string name = it->second;
 199                 if (!what[i].str().empty())
 200                     vars[name] = what[i];
 201             }
 202
 203         }
 204         //prepare replacement string
 205         std::string rvalue = sub_vars(uri_pat, vars);
 206         //rewrite value
 207         std::string rhvalue = what.prefix().str()
 208             + rvalue + what.suffix().str();
 209         yaz_log(YLOG_DEBUG, "! Rewritten '%s' to '%s'",
 210                 what.str(0).c_str(), rvalue.c_str());
 211         out += rhvalue;
 212         start = what[0].second; //move search forward
 213     }
 214     return out;
 215 }
 216
 217 void yf::HttpRewrite::parse_groups(
 218         const spair_vec & uri_pats,
 219         std::vector<std::map<int, std::string> > & groups_bynum_vec)
 220 {
 221     for (unsigned h = 0; h < uri_pats.size(); h++)
 222     {
 223         int gnum = 0;
 224         bool esc = false;
 225         //regex is first, subpat is second
 226         std::string str = uri_pats[h].first;
 227         //for each pair we have an indexing map
 228         std::map<int, std::string> groups_bynum;
 229         for (unsigned i = 0; i < str.size(); ++i)
 230         {
 231             if (!esc && str[i] == '\\')
 232             {
 233                 esc = true;
 234                 continue;
 235             }
 236             if (!esc && str[i] == '(') //group starts
 237             {
 238                 gnum++;
 239                 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 240                 {
 241                     i++;
 242                     if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 243                     {
 244                         if (gnum > 0) gnum--;
 245                         i++;
 246                         continue;
 247                     }
 248                     if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 249                         i++;
 250                     if (i+1 < str.size() && str[i+1] == '<') //named
 251                     {
 252                         i++;
 253                         std::string gname;
 254                         bool term = false;
 255                         while (++i < str.size())
 256                         {
 257                             if (str[i] == '>') { term = true; break; }
 258                             if (!isalnum(str[i]))
 259                                 throw mp::filter::FilterException
 260                                     ("Only alphanumeric chars allowed, found "
 261                                      " in '"
 262                                      + str
 263                                      + "' at "
 264                                      + boost::lexical_cast<std::string>(i));
 265                             gname += str[i];
 266                         }
 267                         if (!term)
 268                             throw mp::filter::FilterException
 269                                 ("Unterminated group name '" + gname
 270                                  + " in '" + str +"'");
 271                         groups_bynum[gnum] = gname;
 272                         yaz_log(YLOG_DEBUG, "Found named group '%s' at $%d",
 273                                 gname.c_str(), gnum);
 274                     }
 275                 }
 276             }
 277             esc = false;
 278         }
 279         groups_bynum_vec.push_back(groups_bynum);
 280     }
 281 }
 282
 283 std::string yf::HttpRewrite::sub_vars (const std::string & in,
 284         const std::map<std::string, std::string> & vars)
 285 {
 286     std::string out;
 287     bool esc = false;
 288     for (unsigned i = 0; i < in.size(); ++i)
 289     {
 290         if (!esc && in[i] == '\\')
 291         {
 292             esc = true;
 293             continue;
 294         }
 295         if (!esc && in[i] == '$') //var
 296         {
 297             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 298             {
 299                 ++i;
 300                 std::string name;
 301                 bool term = false;
 302                 while (++i < in.size())
 303                 {
 304                     if (in[i] == '}') { term = true; break; }
 305                     name += in[i];
 306                 }
 307                 if (!term) throw mp::filter::FilterException
 308                     ("Unterminated var ref in '"+in+"' at "
 309                      + boost::lexical_cast<std::string>(i));
 310                 std::map<std::string, std::string>::const_iterator it
 311                     = vars.find(name);
 312                 if (it != vars.end())
 313                 {
 314                     out += it->second;
 315                 }
 316             }
 317             else
 318             {
 319                 throw mp::filter::FilterException
 320                     ("Malformed or trimmed var ref in '"
 321                      +in+"' at "+boost::lexical_cast<std::string>(i));
 322             }
 323             continue;
 324         }
 325         //passthru
 326         out += in[i];
 327         esc = false;
 328     }
 329     return out;
 330 }
 331
 332 void yf::HttpRewrite::configure(
 333         const spair_vec req_uri_pats,
 334         const spair_vec res_uri_pats)
 335 {
 336     //TODO should we really copy them out?
 337     this->req_uri_pats = req_uri_pats;
 338     this->res_uri_pats = res_uri_pats;
 339     //pick up names
 340     parse_groups(req_uri_pats, req_groups_bynum);
 341     parse_groups(res_uri_pats, res_groups_bynum);
 342 }
 343
 344
 345 static void configure_rules(const xmlNode *ptr, yf::HttpRewrite::spair_vec & dest)
 346 {
 347     for (ptr = ptr->children; ptr; ptr = ptr->next)
 348     {
 349         if (ptr->type != XML_ELEMENT_NODE)
 350             continue;
 351         else if (!strcmp((const char *) ptr->name, "rewrite"))
 352         {
 353             std::string from, to;
 354             const struct _xmlAttr *attr;
 355             for (attr = ptr->properties; attr; attr = attr->next)
 356             {
 357                 if (!strcmp((const char *) attr->name,  "from"))
 358                     from = mp::xml::get_text(attr->children);
 359                 else if (!strcmp((const char *) attr->name,  "to"))
 360                     to = mp::xml::get_text(attr->children);
 361                 else
 362                     throw mp::filter::FilterException
 363                         ("Bad attribute "
 364                          + std::string((const char *) attr->name)
 365                          + " in rewrite section of http_rewrite");
 366             }
 367             yaz_log(YLOG_DEBUG, "Found rewrite rule from '%s' to '%s'",
 368                     from.c_str(), to.c_str());
 369             if (!from.empty())
 370                 dest.push_back(std::make_pair(from, to));
 371         }
 372         else
 373         {
 374             throw mp::filter::FilterException
 375                 ("Bad element o"
 376                  + std::string((const char *) ptr->name)
 377                  + " in http_rewrite1 filter");
 378         }
 379     }
 380 }
 381
 382 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 383         const char *path)
 384 {
 385     spair_vec req_uri_pats;
 386     spair_vec res_uri_pats;
 387     for (ptr = ptr->children; ptr; ptr = ptr->next)
 388     {
 389         if (ptr->type != XML_ELEMENT_NODE)
 390             continue;
 391         else if (!strcmp((const char *) ptr->name, "request"))
 392         {
 393             configure_rules(ptr, req_uri_pats);
 394         }
 395         else if (!strcmp((const char *) ptr->name, "response"))
 396         {
 397             configure_rules(ptr, res_uri_pats);
 398         }
 399         else
 400         {
 401             throw mp::filter::FilterException
 402                 ("Bad element "
 403                  + std::string((const char *) ptr->name)
 404                  + " in http_rewrite1 filter");
 405         }
 406     }
 407     configure(req_uri_pats, res_uri_pats);
 408 }
 409
 410 static mp::filter::Base* filter_creator()
 411 {
 412     return new mp::filter::HttpRewrite;
 413 }
 414
 415 extern "C" {
 416     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 417         0,
 418         "http_rewrite",
 419         filter_creator
 420     };
 421 }
 422
 423
 424 /*
 425  * Local variables:
 426  * c-basic-offset: 4
 427  * c-file-style: "Stroustrup"
 428  * indent-tabs-mode: nil
 429  * End:
 430  * vim: shiftwidth=4 tabstop=8 expandtab
 431  */
 432