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, req_uri_pats, req_groups_bynum);
  63         rewrite_body(o, &hreq->content_buf, &hreq->content_len, vars,
  64                 req_uri_pats, req_groups_bynum);
  65         package.request() = gdu;
  66     }
  67     package.move();
  68     gdu = package.response().get();
  69     if (gdu && gdu->which == Z_GDU_HTTP_Response)
  70     {
  71         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
  72         yaz_log(YLOG_LOG, "Response code %d", hres->code);
  73         mp::odr o;
  74         yaz_log(YLOG_LOG, "<< Respose headers");
  75         rewrite_headers(o, hres->headers, vars, res_uri_pats, res_groups_bynum);
  76         rewrite_body(o, &hres->content_buf, &hres->content_len, vars,
  77                 res_uri_pats, res_groups_bynum);
  78         package.response() = gdu;
  79     }
  80 }
  81
  82 void yf::HttpRewrite::rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq,
  83         std::map<std::string, std::string> & vars) const
  84 {
  85     //rewrite the request line
  86     std::string path;
  87     if (strstr(hreq->path, "http://") == hreq->path)
  88     {
  89         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
  90             "possibly a proxy request");
  91         path += hreq->path;
  92     }
  93     else
  94     {
  95         //TODO what about proto
  96         path += "http://";
  97         path += z_HTTP_header_lookup(hreq->headers, "Host");
  98         path += hreq->path;
  99     }
 100     yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 101     std::string npath =
 102         test_patterns(vars, path, req_uri_pats, req_groups_bynum);
 103     if (!npath.empty())
 104     {
 105         yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 106         hreq->path = odr_strdup(o, npath.c_str());
 107     }
 108 }
 109
 110 void yf::HttpRewrite::rewrite_headers (mp::odr & o, Z_HTTP_Header *headers,
 111         std::map<std::string, std::string> & vars,
 112         const spair_vec & uri_pats,
 113         const std::vector<std::map<int, std::string> > & groups_bynum) const
 114 {
 115     for (Z_HTTP_Header *header = headers;
 116             header != 0;
 117             header = header->next)
 118     {
 119         std::string sheader(header->name);
 120         sheader += ": ";
 121         sheader += header->value;
 122         yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
 123         std::string out = test_patterns(vars, sheader, uri_pats, groups_bynum);
 124         if (!out.empty())
 125         {
 126             size_t pos = out.find(": ");
 127             if (pos == std::string::npos)
 128             {
 129                 yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 130                 continue;
 131             }
 132             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 133             header->value = odr_strdup(o, out.substr(pos+2,
 134                         std::string::npos).c_str());
 135         }
 136     }
 137 }
 138
 139 void yf::HttpRewrite::rewrite_body (mp::odr & o, char **content_buf, int *content_len,
 140         std::map<std::string, std::string> & vars,
 141         const spair_vec & uri_pats,
 142         const std::vector<std::map<int, std::string> > & groups_bynum) const
 143 {
 144     if (*content_buf)
 145     {
 146         std::string body(*content_buf);
 147         std::string nbody =
 148             test_patterns(vars, body, uri_pats, groups_bynum);
 149         if (!nbody.empty())
 150         {
 151             *content_buf = odr_strdup(o, nbody.c_str());
 152             *content_len = nbody.size();
 153         }
 154     }
 155 }
 156
 157 /**
 158  * Tests pattern from the vector in order and executes recipe on
 159  the first match.
 160  */
 161 const std::string yf::HttpRewrite::test_patterns(
 162         std::map<std::string, std::string> & vars,
 163         const std::string & txt,
 164         const spair_vec & uri_pats,
 165         const std::vector<std::map<int, std::string> > & groups_bynum_vec)
 166     const
 167 {
 168     for (unsigned i = 0; i < uri_pats.size(); i++)
 169     {
 170         std::string out = search_replace(vars, txt,
 171                 uri_pats[i].first, uri_pats[i].second,
 172                 groups_bynum_vec[i]);
 173         if (!out.empty()) return out;
 174     }
 175     return "";
 176 }
 177
 178
 179 const std::string yf::HttpRewrite::search_replace(
 180         std::map<std::string, std::string> & vars,
 181         const std::string & txt,
 182         const std::string & uri_re,
 183         const std::string & uri_pat,
 184         const std::map<int, std::string> & groups_bynum) const
 185 {
 186     //exec regex against value
 187     boost::regex re(uri_re);
 188     boost::smatch what;
 189     std::string::const_iterator start, end;
 190     start = txt.begin();
 191     end = txt.end();
 192     std::string out;
 193     while (regex_search(start, end, what, re)) //find next full match
 194     {
 195         unsigned i;
 196         for (i = 1; i < what.size(); ++i)
 197         {
 198             //check if the group is named
 199             std::map<int, std::string>::const_iterator it
 200                 = groups_bynum.find(i);
 201             if (it != groups_bynum.end())
 202             {   //it is
 203                 if (!what[i].str().empty())
 204                     vars[it->second] = what[i];
 205             }
 206
 207         }
 208         //prepare replacement string
 209         std::string rvalue = sub_vars(uri_pat, vars);
 210         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 211                 what.str(0).c_str(), rvalue.c_str());
 212         out.append(start, what[0].first);
 213         out.append(rvalue);
 214         start = what[0].second; //move search forward
 215     }
 216     //if we had a match cat the last part
 217     if (start != txt.begin())
 218         out.append(start, end);
 219     return out;
 220 }
 221
 222 void yf::HttpRewrite::parse_groups(
 223         const spair_vec & uri_pats,
 224         std::vector<std::map<int, std::string> > & groups_bynum_vec)
 225 {
 226     for (unsigned h = 0; h < uri_pats.size(); h++)
 227     {
 228         int gnum = 0;
 229         bool esc = false;
 230         //regex is first, subpat is second
 231         std::string str = uri_pats[h].first;
 232         //for each pair we have an indexing map
 233         std::map<int, std::string> groups_bynum;
 234         for (unsigned i = 0; i < str.size(); ++i)
 235         {
 236             if (!esc && str[i] == '\\')
 237             {
 238                 esc = true;
 239                 continue;
 240             }
 241             if (!esc && str[i] == '(') //group starts
 242             {
 243                 gnum++;
 244                 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 245                 {
 246                     i++;
 247                     if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 248                     {
 249                         if (gnum > 0) gnum--;
 250                         i++;
 251                         continue;
 252                     }
 253                     if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 254                         i++;
 255                     if (i+1 < str.size() && str[i+1] == '<') //named
 256                     {
 257                         i++;
 258                         std::string gname;
 259                         bool term = false;
 260                         while (++i < str.size())
 261                         {
 262                             if (str[i] == '>') { term = true; break; }
 263                             if (!isalnum(str[i]))
 264                                 throw mp::filter::FilterException
 265                                     ("Only alphanumeric chars allowed, found "
 266                                      " in '"
 267                                      + str
 268                                      + "' at "
 269                                      + boost::lexical_cast<std::string>(i));
 270                             gname += str[i];
 271                         }
 272                         if (!term)
 273                             throw mp::filter::FilterException
 274                                 ("Unterminated group name '" + gname
 275                                  + " in '" + str +"'");
 276                         groups_bynum[gnum] = gname;
 277                         yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 278                                 gname.c_str(), gnum);
 279                     }
 280                 }
 281             }
 282             esc = false;
 283         }
 284         groups_bynum_vec.push_back(groups_bynum);
 285     }
 286 }
 287
 288 std::string yf::HttpRewrite::sub_vars (const std::string & in,
 289         const std::map<std::string, std::string> & vars)
 290 {
 291     std::string out;
 292     bool esc = false;
 293     for (unsigned i = 0; i < in.size(); ++i)
 294     {
 295         if (!esc && in[i] == '\\')
 296         {
 297             esc = true;
 298             continue;
 299         }
 300         if (!esc && in[i] == '$') //var
 301         {
 302             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 303             {
 304                 ++i;
 305                 std::string name;
 306                 bool term = false;
 307                 while (++i < in.size())
 308                 {
 309                     if (in[i] == '}') { term = true; break; }
 310                     name += in[i];
 311                 }
 312                 if (!term) throw mp::filter::FilterException
 313                     ("Unterminated var ref in '"+in+"' at "
 314                      + boost::lexical_cast<std::string>(i));
 315                 std::map<std::string, std::string>::const_iterator it
 316                     = vars.find(name);
 317                 if (it != vars.end())
 318                 {
 319                     out += it->second;
 320                 }
 321             }
 322             else
 323             {
 324                 throw mp::filter::FilterException
 325                     ("Malformed or trimmed var ref in '"
 326                      +in+"' at "+boost::lexical_cast<std::string>(i));
 327             }
 328             continue;
 329         }
 330         //passthru
 331         out += in[i];
 332         esc = false;
 333     }
 334     return out;
 335 }
 336
 337 void yf::HttpRewrite::configure(
 338         const spair_vec req_uri_pats,
 339         const spair_vec res_uri_pats)
 340 {
 341     //TODO should we really copy them out?
 342     this->req_uri_pats = req_uri_pats;
 343     this->res_uri_pats = res_uri_pats;
 344     //pick up names
 345     parse_groups(req_uri_pats, req_groups_bynum);
 346     parse_groups(res_uri_pats, res_groups_bynum);
 347 }
 348
 349
 350 static void configure_rules(const xmlNode *ptr, yf::HttpRewrite::spair_vec & dest)
 351 {
 352     for (ptr = ptr->children; ptr; ptr = ptr->next)
 353     {
 354         if (ptr->type != XML_ELEMENT_NODE)
 355             continue;
 356         else if (!strcmp((const char *) ptr->name, "rewrite"))
 357         {
 358             std::string from, to;
 359             const struct _xmlAttr *attr;
 360             for (attr = ptr->properties; attr; attr = attr->next)
 361             {
 362                 if (!strcmp((const char *) attr->name,  "from"))
 363                     from = mp::xml::get_text(attr->children);
 364                 else if (!strcmp((const char *) attr->name,  "to"))
 365                     to = mp::xml::get_text(attr->children);
 366                 else
 367                     throw mp::filter::FilterException
 368                         ("Bad attribute "
 369                          + std::string((const char *) attr->name)
 370                          + " in rewrite section of http_rewrite");
 371             }
 372             yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 373                     from.c_str(), to.c_str());
 374             if (!from.empty())
 375                 dest.push_back(std::make_pair(from, to));
 376         }
 377         else
 378         {
 379             throw mp::filter::FilterException
 380                 ("Bad element o"
 381                  + std::string((const char *) ptr->name)
 382                  + " in http_rewrite1 filter");
 383         }
 384     }
 385 }
 386
 387 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 388         const char *path)
 389 {
 390     spair_vec req_uri_pats;
 391     spair_vec res_uri_pats;
 392     for (ptr = ptr->children; ptr; ptr = ptr->next)
 393     {
 394         if (ptr->type != XML_ELEMENT_NODE)
 395             continue;
 396         else if (!strcmp((const char *) ptr->name, "request"))
 397         {
 398             configure_rules(ptr, req_uri_pats);
 399         }
 400         else if (!strcmp((const char *) ptr->name, "response"))
 401         {
 402             configure_rules(ptr, res_uri_pats);
 403         }
 404         else
 405         {
 406             throw mp::filter::FilterException
 407                 ("Bad element "
 408                  + std::string((const char *) ptr->name)
 409                  + " in http_rewrite1 filter");
 410         }
 411     }
 412     configure(req_uri_pats, res_uri_pats);
 413 }
 414
 415 static mp::filter::Base* filter_creator()
 416 {
 417     return new mp::filter::HttpRewrite;
 418 }
 419
 420 extern "C" {
 421     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 422         0,
 423         "http_rewrite",
 424         filter_creator
 425     };
 426 }
 427
 428
 429 /*
 430  * Local variables:
 431  * c-basic-offset: 4
 432  * c-file-style: "Stroustrup"
 433  * indent-tabs-mode: nil
 434  * End:
 435  * vim: shiftwidth=4 tabstop=8 expandtab
 436  */
 437