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         std::cout << "Response " << 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 void yf::HttpRewrite::configure(const xmlNode* ptr, bool test_only, const char *path) {};
 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 (int 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         std::cout << "! Rewritten '"+what.str(0)+"' to '"+rvalue+"'\n";
 210         out += rhvalue;
 211         start = what[0].second; //move search forward
 212     }
 213     return out;
 214 }
 215
 216 void yf::HttpRewrite::parse_groups(
 217         const spair_vec & uri_pats,
 218         std::vector<std::map<int, std::string> > & groups_bynum_vec)
 219 {
 220     for (int h = 0; h < uri_pats.size(); h++)
 221     {
 222         int gnum = 0;
 223         bool esc = false;
 224         //regex is first, subpat is second
 225         std::string str = uri_pats[h].first;
 226         //for each pair we have an indexing map
 227         std::map<int, std::string> groups_bynum;
 228         for (int i = 0; i < str.size(); ++i)
 229         {
 230             if (!esc && str[i] == '\\')
 231             {
 232                 esc = true;
 233                 continue;
 234             }
 235             if (!esc && str[i] == '(') //group starts
 236             {
 237                 gnum++;
 238                 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 239                 {
 240                     i++;
 241                     if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 242                     {
 243                         if (gnum > 0) gnum--;
 244                         i++;
 245                         continue;
 246                     }
 247                     if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 248                         i++;
 249                     if (i+1 < str.size() && str[i+1] == '<') //named
 250                     {
 251                         i++;
 252                         std::string gname;
 253                         bool term = false;
 254                         while (++i < str.size())
 255                         {
 256                             if (str[i] == '>') { term = true; break; }
 257                             if (!isalnum(str[i]))
 258                                 throw mp::filter::FilterException
 259                                     ("Only alphanumeric chars allowed, found "
 260                                      " in '"
 261                                      + str
 262                                      + "' at "
 263                                      + boost::lexical_cast<std::string>(i));
 264                             gname += str[i];
 265                         }
 266                         if (!term)
 267                             throw mp::filter::FilterException
 268                                 ("Unterminated group name '" + gname
 269                                  + " in '" + str +"'");
 270                         groups_bynum[gnum] = gname;
 271                         std::cout << "Found named group '" << gname
 272                             << "' at $" << gnum << std::endl;
 273                     }
 274                 }
 275             }
 276             esc = false;
 277         }
 278         groups_bynum_vec.push_back(groups_bynum);
 279     }
 280 }
 281
 282 std::string yf::HttpRewrite::sub_vars (const std::string & in,
 283         const std::map<std::string, std::string> & vars)
 284 {
 285     std::string out;
 286     bool esc = false;
 287     for (int i = 0; i < in.size(); ++i)
 288     {
 289         if (!esc && in[i] == '\\')
 290         {
 291             esc = true;
 292             continue;
 293         }
 294         if (!esc && in[i] == '$') //var
 295         {
 296             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 297             {
 298                 ++i;
 299                 std::string name;
 300                 bool term = false;
 301                 while (++i < in.size())
 302                 {
 303                     if (in[i] == '}') { term = true; break; }
 304                     name += in[i];
 305                 }
 306                 if (!term) throw mp::filter::FilterException
 307                     ("Unterminated var ref in '"+in+"' at "
 308                      + boost::lexical_cast<std::string>(i));
 309                 std::map<std::string, std::string>::const_iterator it
 310                     = vars.find(name);
 311                 if (it != vars.end())
 312                 {
 313                     out += it->second;
 314                 }
 315             }
 316             else
 317             {
 318                 throw mp::filter::FilterException
 319                     ("Malformed or trimmed var ref in '"
 320                      +in+"' at "+boost::lexical_cast<std::string>(i));
 321             }
 322             continue;
 323         }
 324         //passthru
 325         out += in[i];
 326         esc = false;
 327     }
 328     return out;
 329 }
 330
 331 void yf::HttpRewrite::configure(
 332         const spair_vec req_uri_pats,
 333         const spair_vec res_uri_pats)
 334 {
 335     //TODO should we really copy them out?
 336     this->req_uri_pats = req_uri_pats;
 337     this->res_uri_pats = res_uri_pats;
 338     //pick up names
 339     parse_groups(req_uri_pats, req_groups_bynum);
 340     parse_groups(res_uri_pats, res_groups_bynum);
 341 }
 342
 343 static mp::filter::Base* filter_creator()
 344 {
 345     return new mp::filter::HttpRewrite;
 346 }
 347
 348 extern "C" {
 349     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 350         0,
 351         "http_rewrite",
 352         filter_creator
 353     };
 354 }
 355
 356
 357 /*
 358  * Local variables:
 359  * c-basic-offset: 4
 360  * c-file-style: "Stroustrup"
 361  * indent-tabs-mode: nil
 362  * End:
 363  * vim: shiftwidth=4 tabstop=8 expandtab
 364  */
 365