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     std::cout << "About to rewrite body: " << *content_buf;
 145     if (*content_buf)
 146     {
 147         std::string body(*content_buf);
 148         std::string nbody =
 149             test_patterns(vars, body, uri_pats, groups_bynum);
 150         if (!nbody.empty())
 151         {
 152             *content_buf = odr_strdup(o, nbody.c_str());
 153             *content_len = nbody.size();
 154         }
 155     }
 156 }
 157
 158 /**
 159  * Tests pattern from the vector in order and executes recipe on
 160  the first match.
 161  */
 162 const std::string yf::HttpRewrite::test_patterns(
 163         std::map<std::string, std::string> & vars,
 164         const std::string & txt,
 165         const spair_vec & uri_pats,
 166         const std::vector<std::map<int, std::string> > & groups_bynum_vec)
 167     const
 168 {
 169     for (unsigned i = 0; i < uri_pats.size(); i++)
 170     {
 171         std::string out = search_replace(vars, txt,
 172                 uri_pats[i].first, uri_pats[i].second,
 173                 groups_bynum_vec[i]);
 174         if (!out.empty()) return out;
 175     }
 176     return "";
 177 }
 178
 179
 180 const std::string yf::HttpRewrite::search_replace(
 181         std::map<std::string, std::string> & vars,
 182         const std::string & txt,
 183         const std::string & uri_re,
 184         const std::string & uri_pat,
 185         const std::map<int, std::string> & groups_bynum) const
 186 {
 187     //exec regex against value
 188     boost::regex re(uri_re);
 189     boost::smatch what;
 190     std::string::const_iterator start, end;
 191     start = txt.begin();
 192     end = txt.end();
 193     std::string out;
 194     while (regex_search(start, end, what, re)) //find next full match
 195     {
 196         unsigned i;
 197         for (i = 1; i < what.size(); ++i)
 198         {
 199             //check if the group is named
 200             std::map<int, std::string>::const_iterator it
 201                 = groups_bynum.find(i);
 202             if (it != groups_bynum.end())
 203             {   //it is
 204                 if (!what[i].str().empty())
 205                     vars[it->second] = what[i];
 206             }
 207
 208         }
 209         //prepare replacement string
 210         std::string rvalue = sub_vars(uri_pat, vars);
 211         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 212                 what.str(0).c_str(), rvalue.c_str());
 213         out.append(start, what[0].first);
 214         out.append(rvalue);
 215         start = what[0].second; //move search forward
 216     }
 217     //if we had a match cat the last part
 218     if (start != txt.begin())
 219         out.append(start, end);
 220     return out;
 221 }
 222
 223 void yf::HttpRewrite::parse_groups(
 224         const spair_vec & uri_pats,
 225         std::vector<std::map<int, std::string> > & groups_bynum_vec)
 226 {
 227     for (unsigned h = 0; h < uri_pats.size(); h++)
 228     {
 229         int gnum = 0;
 230         bool esc = false;
 231         //regex is first, subpat is second
 232         std::string str = uri_pats[h].first;
 233         //for each pair we have an indexing map
 234         std::map<int, std::string> groups_bynum;
 235         for (unsigned i = 0; i < str.size(); ++i)
 236         {
 237             if (!esc && str[i] == '\\')
 238             {
 239                 esc = true;
 240                 continue;
 241             }
 242             if (!esc && str[i] == '(') //group starts
 243             {
 244                 gnum++;
 245                 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 246                 {
 247                     i++;
 248                     if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 249                     {
 250                         if (gnum > 0) gnum--;
 251                         i++;
 252                         continue;
 253                     }
 254                     if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 255                         i++;
 256                     if (i+1 < str.size() && str[i+1] == '<') //named
 257                     {
 258                         i++;
 259                         std::string gname;
 260                         bool term = false;
 261                         while (++i < str.size())
 262                         {
 263                             if (str[i] == '>') { term = true; break; }
 264                             if (!isalnum(str[i]))
 265                                 throw mp::filter::FilterException
 266                                     ("Only alphanumeric chars allowed, found "
 267                                      " in '"
 268                                      + str
 269                                      + "' at "
 270                                      + boost::lexical_cast<std::string>(i));
 271                             gname += str[i];
 272                         }
 273                         if (!term)
 274                             throw mp::filter::FilterException
 275                                 ("Unterminated group name '" + gname
 276                                  + " in '" + str +"'");
 277                         groups_bynum[gnum] = gname;
 278                         yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 279                                 gname.c_str(), gnum);
 280                     }
 281                 }
 282             }
 283             esc = false;
 284         }
 285         groups_bynum_vec.push_back(groups_bynum);
 286     }
 287 }
 288
 289 std::string yf::HttpRewrite::sub_vars (const std::string & in,
 290         const std::map<std::string, std::string> & vars)
 291 {
 292     std::string out;
 293     bool esc = false;
 294     for (unsigned i = 0; i < in.size(); ++i)
 295     {
 296         if (!esc && in[i] == '\\')
 297         {
 298             esc = true;
 299             continue;
 300         }
 301         if (!esc && in[i] == '$') //var
 302         {
 303             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 304             {
 305                 ++i;
 306                 std::string name;
 307                 bool term = false;
 308                 while (++i < in.size())
 309                 {
 310                     if (in[i] == '}') { term = true; break; }
 311                     name += in[i];
 312                 }
 313                 if (!term) throw mp::filter::FilterException
 314                     ("Unterminated var ref in '"+in+"' at "
 315                      + boost::lexical_cast<std::string>(i));
 316                 std::map<std::string, std::string>::const_iterator it
 317                     = vars.find(name);
 318                 if (it != vars.end())
 319                 {
 320                     out += it->second;
 321                 }
 322             }
 323             else
 324             {
 325                 throw mp::filter::FilterException
 326                     ("Malformed or trimmed var ref in '"
 327                      +in+"' at "+boost::lexical_cast<std::string>(i));
 328             }
 329             continue;
 330         }
 331         //passthru
 332         out += in[i];
 333         esc = false;
 334     }
 335     return out;
 336 }
 337
 338 void yf::HttpRewrite::configure(
 339         const spair_vec req_uri_pats,
 340         const spair_vec res_uri_pats)
 341 {
 342     //TODO should we really copy them out?
 343     this->req_uri_pats = req_uri_pats;
 344     this->res_uri_pats = res_uri_pats;
 345     //pick up names
 346     parse_groups(req_uri_pats, req_groups_bynum);
 347     parse_groups(res_uri_pats, res_groups_bynum);
 348 }
 349
 350
 351 static void configure_rules(const xmlNode *ptr, yf::HttpRewrite::spair_vec & dest)
 352 {
 353     for (ptr = ptr->children; ptr; ptr = ptr->next)
 354     {
 355         if (ptr->type != XML_ELEMENT_NODE)
 356             continue;
 357         else if (!strcmp((const char *) ptr->name, "rewrite"))
 358         {
 359             std::string from, to;
 360             const struct _xmlAttr *attr;
 361             for (attr = ptr->properties; attr; attr = attr->next)
 362             {
 363                 if (!strcmp((const char *) attr->name,  "from"))
 364                     from = mp::xml::get_text(attr->children);
 365                 else if (!strcmp((const char *) attr->name,  "to"))
 366                     to = mp::xml::get_text(attr->children);
 367                 else
 368                     throw mp::filter::FilterException
 369                         ("Bad attribute "
 370                          + std::string((const char *) attr->name)
 371                          + " in rewrite section of http_rewrite");
 372             }
 373             yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 374                     from.c_str(), to.c_str());
 375             if (!from.empty())
 376                 dest.push_back(std::make_pair(from, to));
 377         }
 378         else
 379         {
 380             throw mp::filter::FilterException
 381                 ("Bad element o"
 382                  + std::string((const char *) ptr->name)
 383                  + " in http_rewrite1 filter");
 384         }
 385     }
 386 }
 387
 388 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 389         const char *path)
 390 {
 391     spair_vec req_uri_pats;
 392     spair_vec res_uri_pats;
 393     for (ptr = ptr->children; ptr; ptr = ptr->next)
 394     {
 395         if (ptr->type != XML_ELEMENT_NODE)
 396             continue;
 397         else if (!strcmp((const char *) ptr->name, "request"))
 398         {
 399             configure_rules(ptr, req_uri_pats);
 400         }
 401         else if (!strcmp((const char *) ptr->name, "response"))
 402         {
 403             configure_rules(ptr, res_uri_pats);
 404         }
 405         else
 406         {
 407             throw mp::filter::FilterException
 408                 ("Bad element "
 409                  + std::string((const char *) ptr->name)
 410                  + " in http_rewrite1 filter");
 411         }
 412     }
 413     configure(req_uri_pats, res_uri_pats);
 414 }
 415
 416 static mp::filter::Base* filter_creator()
 417 {
 418     return new mp::filter::HttpRewrite;
 419 }
 420
 421 extern "C" {
 422     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 423         0,
 424         "http_rewrite",
 425         filter_creator
 426     };
 427 }
 428
 429
 430 /*
 431  * Local variables:
 432  * c-basic-offset: 4
 433  * c-file-style: "Stroustrup"
 434  * indent-tabs-mode: nil
 435  * End:
 436  * vim: shiftwidth=4 tabstop=8 expandtab
 437  */
 438