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