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_LOG, "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_LOG, ">> 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_LOG, "Response code %d", hres->code);
  72         mp::odr o;
  73         yaz_log(YLOG_LOG, "<< 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_LOG, "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 += "http://";
  95         path += z_HTTP_header_lookup(hreq->headers, "Host");
  96         path += hreq->path;
  97     }
  98     yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
  99     std::string npath =
 100         test_patterns(vars, path, req_uri_pats, req_groups_bynum);
 101     if (!npath.empty())
 102     {
 103         yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 104         hreq->path = odr_strdup(o, npath.c_str());
 105     }
 106 }
 107
 108 void yf::HttpRewrite::rewrite_headers (mp::odr & o, Z_HTTP_Header *headers,
 109         std::map<std::string, std::string> & vars) const
 110 {
 111     for (Z_HTTP_Header *header = headers;
 112             header != 0;
 113             header = header->next)
 114     {
 115         std::string sheader(header->name);
 116         sheader += ": ";
 117         sheader += header->value;
 118         yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
 119         std::string out = test_patterns(vars,
 120                 sheader,
 121                 req_uri_pats, req_groups_bynum);
 122         if (!out.empty())
 123         {
 124             size_t pos = out.find(": ");
 125             if (pos == std::string::npos)
 126             {
 127                 yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 128                 continue;
 129             }
 130             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 131             header->value = odr_strdup(o, out.substr(pos+2,
 132                         std::string::npos).c_str());
 133         }
 134     }
 135 }
 136
 137 void yf::HttpRewrite::rewrite_body (mp::odr & o, char **content_buf, int *content_len,
 138         std::map<std::string, std::string> & vars) const
 139 {
 140     if (*content_buf)
 141     {
 142         std::string body(*content_buf);
 143         std::string nbody =
 144             test_patterns(vars, body, req_uri_pats, req_groups_bynum);
 145         if (!nbody.empty())
 146         {
 147             *content_buf = odr_strdup(o, nbody.c_str());
 148             *content_len = nbody.size();
 149         }
 150     }
 151 }
 152
 153 /**
 154  * Tests pattern from the vector in order and executes recipe on
 155  the first match.
 156  */
 157 const std::string yf::HttpRewrite::test_patterns(
 158         std::map<std::string, std::string> & vars,
 159         const std::string & txt,
 160         const spair_vec & uri_pats,
 161         const std::vector<std::map<int, std::string> > & groups_bynum_vec)
 162     const
 163 {
 164     for (unsigned i = 0; i < uri_pats.size(); i++)
 165     {
 166         std::string out = search_replace(vars, txt,
 167                 uri_pats[i].first, uri_pats[i].second,
 168                 groups_bynum_vec[i]);
 169         if (!out.empty()) return out;
 170     }
 171     return "";
 172 }
 173
 174
 175 const std::string yf::HttpRewrite::search_replace(
 176         std::map<std::string, std::string> & vars,
 177         const std::string & txt,
 178         const std::string & uri_re,
 179         const std::string & uri_pat,
 180         const std::map<int, std::string> & groups_bynum) const
 181 {
 182     //exec regex against value
 183     boost::regex re(uri_re);
 184     boost::smatch what;
 185     std::string::const_iterator start, end;
 186     start = txt.begin();
 187     end = txt.end();
 188     std::string out;
 189     while (regex_search(start, end, what, re)) //find next full match
 190     {
 191         unsigned i;
 192         for (i = 1; i < what.size(); ++i)
 193         {
 194             //check if the group is named
 195             std::map<int, std::string>::const_iterator it
 196                 = groups_bynum.find(i);
 197             if (it != groups_bynum.end())
 198             {   //it is
 199                 std::string name = it->second;
 200                 if (!what[i].str().empty())
 201                     vars[name] = what[i];
 202             }
 203
 204         }
 205         //prepare replacement string
 206         std::string rvalue = sub_vars(uri_pat, vars);
 207         //rewrite value
 208         std::string rhvalue = what.prefix().str()
 209             + rvalue + what.suffix().str();
 210         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 211                 what.str(0).c_str(), rvalue.c_str());
 212         out += rhvalue;
 213         start = what[0].second; //move search forward
 214     }
 215     return out;
 216 }
 217
 218 void yf::HttpRewrite::parse_groups(
 219         const spair_vec & uri_pats,
 220         std::vector<std::map<int, std::string> > & groups_bynum_vec)
 221 {
 222     for (unsigned h = 0; h < uri_pats.size(); h++)
 223     {
 224         int gnum = 0;
 225         bool esc = false;
 226         //regex is first, subpat is second
 227         std::string str = uri_pats[h].first;
 228         //for each pair we have an indexing map
 229         std::map<int, std::string> groups_bynum;
 230         for (unsigned i = 0; i < str.size(); ++i)
 231         {
 232             if (!esc && str[i] == '\\')
 233             {
 234                 esc = true;
 235                 continue;
 236             }
 237             if (!esc && str[i] == '(') //group starts
 238             {
 239                 gnum++;
 240                 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 241                 {
 242                     i++;
 243                     if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 244                     {
 245                         if (gnum > 0) gnum--;
 246                         i++;
 247                         continue;
 248                     }
 249                     if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 250                         i++;
 251                     if (i+1 < str.size() && str[i+1] == '<') //named
 252                     {
 253                         i++;
 254                         std::string gname;
 255                         bool term = false;
 256                         while (++i < str.size())
 257                         {
 258                             if (str[i] == '>') { term = true; break; }
 259                             if (!isalnum(str[i]))
 260                                 throw mp::filter::FilterException
 261                                     ("Only alphanumeric chars allowed, found "
 262                                      " in '"
 263                                      + str
 264                                      + "' at "
 265                                      + boost::lexical_cast<std::string>(i));
 266                             gname += str[i];
 267                         }
 268                         if (!term)
 269                             throw mp::filter::FilterException
 270                                 ("Unterminated group name '" + gname
 271                                  + " in '" + str +"'");
 272                         groups_bynum[gnum] = gname;
 273                         yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 274                                 gname.c_str(), gnum);
 275                     }
 276                 }
 277             }
 278             esc = false;
 279         }
 280         groups_bynum_vec.push_back(groups_bynum);
 281     }
 282 }
 283
 284 std::string yf::HttpRewrite::sub_vars (const std::string & in,
 285         const std::map<std::string, std::string> & vars)
 286 {
 287     std::string out;
 288     bool esc = false;
 289     for (unsigned i = 0; i < in.size(); ++i)
 290     {
 291         if (!esc && in[i] == '\\')
 292         {
 293             esc = true;
 294             continue;
 295         }
 296         if (!esc && in[i] == '$') //var
 297         {
 298             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 299             {
 300                 ++i;
 301                 std::string name;
 302                 bool term = false;
 303                 while (++i < in.size())
 304                 {
 305                     if (in[i] == '}') { term = true; break; }
 306                     name += in[i];
 307                 }
 308                 if (!term) throw mp::filter::FilterException
 309                     ("Unterminated var ref in '"+in+"' at "
 310                      + boost::lexical_cast<std::string>(i));
 311                 std::map<std::string, std::string>::const_iterator it
 312                     = vars.find(name);
 313                 if (it != vars.end())
 314                 {
 315                     out += it->second;
 316                 }
 317             }
 318             else
 319             {
 320                 throw mp::filter::FilterException
 321                     ("Malformed or trimmed var ref in '"
 322                      +in+"' at "+boost::lexical_cast<std::string>(i));
 323             }
 324             continue;
 325         }
 326         //passthru
 327         out += in[i];
 328         esc = false;
 329     }
 330     return out;
 331 }
 332
 333 void yf::HttpRewrite::configure(
 334         const spair_vec req_uri_pats,
 335         const spair_vec res_uri_pats)
 336 {
 337     //TODO should we really copy them out?
 338     this->req_uri_pats = req_uri_pats;
 339     this->res_uri_pats = res_uri_pats;
 340     //pick up names
 341     parse_groups(req_uri_pats, req_groups_bynum);
 342     parse_groups(res_uri_pats, res_groups_bynum);
 343 }
 344
 345
 346 static void configure_rules(const xmlNode *ptr, yf::HttpRewrite::spair_vec & dest)
 347 {
 348     for (ptr = ptr->children; ptr; ptr = ptr->next)
 349     {
 350         if (ptr->type != XML_ELEMENT_NODE)
 351             continue;
 352         else if (!strcmp((const char *) ptr->name, "rewrite"))
 353         {
 354             std::string from, to;
 355             const struct _xmlAttr *attr;
 356             for (attr = ptr->properties; attr; attr = attr->next)
 357             {
 358                 if (!strcmp((const char *) attr->name,  "from"))
 359                     from = mp::xml::get_text(attr->children);
 360                 else if (!strcmp((const char *) attr->name,  "to"))
 361                     to = mp::xml::get_text(attr->children);
 362                 else
 363                     throw mp::filter::FilterException
 364                         ("Bad attribute "
 365                          + std::string((const char *) attr->name)
 366                          + " in rewrite section of http_rewrite");
 367             }
 368             yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 369                     from.c_str(), to.c_str());
 370             if (!from.empty())
 371                 dest.push_back(std::make_pair(from, to));
 372         }
 373         else
 374         {
 375             throw mp::filter::FilterException
 376                 ("Bad element o"
 377                  + std::string((const char *) ptr->name)
 378                  + " in http_rewrite1 filter");
 379         }
 380     }
 381 }
 382
 383 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 384         const char *path)
 385 {
 386     spair_vec req_uri_pats;
 387     spair_vec res_uri_pats;
 388     for (ptr = ptr->children; ptr; ptr = ptr->next)
 389     {
 390         if (ptr->type != XML_ELEMENT_NODE)
 391             continue;
 392         else if (!strcmp((const char *) ptr->name, "request"))
 393         {
 394             configure_rules(ptr, req_uri_pats);
 395         }
 396         else if (!strcmp((const char *) ptr->name, "response"))
 397         {
 398             configure_rules(ptr, res_uri_pats);
 399         }
 400         else
 401         {
 402             throw mp::filter::FilterException
 403                 ("Bad element "
 404                  + std::string((const char *) ptr->name)
 405                  + " in http_rewrite1 filter");
 406         }
 407     }
 408     configure(req_uri_pats, res_uri_pats);
 409 }
 410
 411 static mp::filter::Base* filter_creator()
 412 {
 413     return new mp::filter::HttpRewrite;
 414 }
 415
 416 extern "C" {
 417     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 418         0,
 419         "http_rewrite",
 420         filter_creator
 421     };
 422 }
 423
 424
 425 /*
 426  * Local variables:
 427  * c-basic-offset: 4
 428  * c-file-style: "Stroustrup"
 429  * indent-tabs-mode: nil
 430  * End:
 431  * vim: shiftwidth=4 tabstop=8 expandtab
 432  */
 433