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