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 <map>
  32
  33 #if HAVE_SYS_TYPES_H
  34 #include <sys/types.h>
  35 #endif
  36
  37 namespace mp = metaproxy_1;
  38 namespace yf = mp::filter;
  39
  40 namespace metaproxy_1 {
  41     namespace filter {
  42         class HttpRewrite::Replace {
  43         public:
  44             std::string regex;
  45             std::string recipe;
  46             std::map<int, std::string> group_index;
  47             const std::string search_replace(
  48                 std::map<std::string, std::string> & vars,
  49                 const std::string & txt) const;
  50             std::string sub_vars (
  51                 const std::map<std::string, std::string> & vars) const;
  52             void parse_groups();
  53         };
  54
  55         class HttpRewrite::Rule {
  56         public:
  57             std::list<Replace> replace_list;
  58             const std::string test_patterns(
  59                 std::map<std::string, std::string> & vars,
  60                 const std::string & txt) const;
  61         };
  62         class HttpRewrite::Within {
  63         public:
  64             std::string header;
  65             std::string attr;
  66             std::string tag;
  67             RulePtr rule;
  68         };
  69
  70         class HttpRewrite::Phase {
  71         public:
  72             std::list<Within> within_list;
  73             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  74                 std::map<std::string, std::string> & vars) const;
  75             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  76                 std::map<std::string, std::string> & vars) const;
  77             void rewrite_body(mp::odr & o,
  78                 char **content_buf, int *content_len,
  79                 std::map<std::string, std::string> & vars) const;
  80         };
  81     }
  82 }
  83
  84 yf::HttpRewrite::HttpRewrite() :
  85     req_phase(new Phase), res_phase(new Phase)
  86 {
  87 }
  88
  89 yf::HttpRewrite::~HttpRewrite()
  90 {
  91 }
  92
  93 void yf::HttpRewrite::process(mp::Package & package) const
  94 {
  95     yaz_log(YLOG_LOG, "HttpRewrite begins....");
  96     Z_GDU *gdu = package.request().get();
  97     //map of request/response vars
  98     std::map<std::string, std::string> vars;
  99     //we have an http req
 100     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 101     {
 102         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 103         mp::odr o;
 104         req_phase->rewrite_reqline(o, hreq, vars);
 105         yaz_log(YLOG_LOG, ">> Request headers");
 106         req_phase->rewrite_headers(o, hreq->headers, vars);
 107         req_phase->rewrite_body(o,
 108                 &hreq->content_buf, &hreq->content_len, vars);
 109         package.request() = gdu;
 110     }
 111     package.move();
 112     gdu = package.response().get();
 113     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 114     {
 115         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 116         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 117         mp::odr o;
 118         yaz_log(YLOG_LOG, "<< Respose headers");
 119         res_phase->rewrite_headers(o, hres->headers, vars);
 120         res_phase->rewrite_body(o, &hres->content_buf,
 121                 &hres->content_len, vars);
 122         package.response() = gdu;
 123     }
 124 }
 125
 126 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 127         Z_HTTP_Request *hreq,
 128         std::map<std::string, std::string> & vars) const
 129 {
 130     //rewrite the request line
 131     std::string path;
 132     if (strstr(hreq->path, "http://") == hreq->path)
 133     {
 134         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 135             "possibly a proxy request");
 136         path += hreq->path;
 137     }
 138     else
 139     {
 140         //TODO what about proto
 141         path += "http://";
 142         path += z_HTTP_header_lookup(hreq->headers, "Host");
 143         path += hreq->path;
 144     }
 145
 146     std::list<Within>::const_iterator it = within_list.begin();
 147     if (it != within_list.end())
 148     {
 149         RulePtr rule = it->rule;
 150
 151         yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 152         std::string npath = rule->test_patterns(vars, path);
 153         if (!npath.empty())
 154         {
 155             yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 156             hreq->path = odr_strdup(o, npath.c_str());
 157         }
 158     }
 159 }
 160
 161 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 162         Z_HTTP_Header *headers,
 163         std::map<std::string, std::string> & vars) const
 164 {
 165     for (Z_HTTP_Header *header = headers;
 166             header != 0;
 167             header = header->next)
 168     {
 169         std::string sheader(header->name);
 170         sheader += ": ";
 171         sheader += header->value;
 172         yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
 173
 174         std::list<Within>::const_iterator it = within_list.begin();
 175         if (it == within_list.end())
 176             continue;
 177         RulePtr rule = it->rule;
 178
 179         std::string out = rule->test_patterns(vars, sheader);
 180         if (!out.empty())
 181         {
 182             size_t pos = out.find(": ");
 183             if (pos == std::string::npos)
 184             {
 185                 yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 186                 continue;
 187             }
 188             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 189             header->value = odr_strdup(o, out.substr(pos+2,
 190                                                      std::string::npos).c_str());
 191         }
 192     }
 193 }
 194
 195 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 196         char **content_buf,
 197         int *content_len,
 198         std::map<std::string, std::string> & vars) const
 199 {
 200     if (*content_buf)
 201     {
 202
 203         std::list<Within>::const_iterator it = within_list.begin();
 204         if (it != within_list.end())
 205         {
 206             RulePtr rule = it->rule;
 207
 208             std::string body(*content_buf);
 209             std::string nbody = rule->test_patterns(vars, body);
 210             if (!nbody.empty())
 211             {
 212                 *content_buf = odr_strdup(o, nbody.c_str());
 213                 *content_len = nbody.size();
 214             }
 215         }
 216     }
 217 }
 218
 219 /**
 220  * Tests pattern from the vector in order and executes recipe on
 221  the first match.
 222  */
 223 const std::string yf::HttpRewrite::Rule::test_patterns(
 224         std::map<std::string, std::string> & vars,
 225         const std::string & txt) const
 226 {
 227     std::list<Replace>::const_iterator it = replace_list.begin();
 228
 229     for (; it != replace_list.end(); it++)
 230     {
 231         std::string out = it->search_replace(vars, txt);
 232         if (!out.empty()) return out;
 233     }
 234     return "";
 235 }
 236
 237 const std::string yf::HttpRewrite::Replace::search_replace(
 238         std::map<std::string, std::string> & vars,
 239         const std::string & txt) const
 240 {
 241     //exec regex against value
 242     boost::regex re(regex);
 243     boost::smatch what;
 244     std::string::const_iterator start, end;
 245     start = txt.begin();
 246     end = txt.end();
 247     std::string out;
 248     while (regex_search(start, end, what, re)) //find next full match
 249     {
 250         size_t i;
 251         for (i = 1; i < what.size(); ++i)
 252         {
 253             //check if the group is named
 254             std::map<int, std::string>::const_iterator it
 255                 = group_index.find(i);
 256             if (it != group_index.end())
 257             {   //it is
 258                 if (!what[i].str().empty())
 259                     vars[it->second] = what[i];
 260             }
 261
 262         }
 263         //prepare replacement string
 264         std::string rvalue = sub_vars(vars);
 265         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 266                 what.str(0).c_str(), rvalue.c_str());
 267         out.append(start, what[0].first);
 268         out.append(rvalue);
 269         start = what[0].second; //move search forward
 270     }
 271     //if we had a match cat the last part
 272     if (start != txt.begin())
 273         out.append(start, end);
 274     return out;
 275 }
 276
 277 void yf::HttpRewrite::Replace::parse_groups()
 278 {
 279     int gnum = 0;
 280     bool esc = false;
 281     const std::string & str = regex;
 282     std::string res;
 283     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 284     for (size_t i = 0; i < str.size(); ++i)
 285     {
 286         res += str[i];
 287         if (!esc && str[i] == '\\')
 288         {
 289             esc = true;
 290             continue;
 291         }
 292         if (!esc && str[i] == '(') //group starts
 293         {
 294             gnum++;
 295             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 296             {
 297                 i++;
 298                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 299                 {
 300                     if (gnum > 0) gnum--;
 301                     res += str[i];
 302                     i++;
 303                     res += str[i];
 304                     continue;
 305                 }
 306                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 307                     i++;
 308                 if (i+1 < str.size() && str[i+1] == '<') //named
 309                 {
 310                     i++;
 311                     std::string gname;
 312                     bool term = false;
 313                     while (++i < str.size())
 314                     {
 315                         if (str[i] == '>') { term = true; break; }
 316                         if (!isalnum(str[i]))
 317                             throw mp::filter::FilterException
 318                                 ("Only alphanumeric chars allowed, found "
 319                                  " in '"
 320                                  + str
 321                                  + "' at "
 322                                  + boost::lexical_cast<std::string>(i));
 323                         gname += str[i];
 324                     }
 325                     if (!term)
 326                         throw mp::filter::FilterException
 327                             ("Unterminated group name '" + gname
 328                              + " in '" + str +"'");
 329                     group_index[gnum] = gname;
 330                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 331                             gname.c_str(), gnum);
 332                 }
 333             }
 334         }
 335         esc = false;
 336     }
 337     regex = res;
 338 }
 339
 340 std::string yf::HttpRewrite::Replace::sub_vars (
 341         const std::map<std::string, std::string> & vars) const
 342 {
 343     std::string out;
 344     bool esc = false;
 345     const std::string & in = recipe;
 346     for (size_t i = 0; i < in.size(); ++i)
 347     {
 348         if (!esc && in[i] == '\\')
 349         {
 350             esc = true;
 351             continue;
 352         }
 353         if (!esc && in[i] == '$') //var
 354         {
 355             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 356             {
 357                 ++i;
 358                 std::string name;
 359                 bool term = false;
 360                 while (++i < in.size())
 361                 {
 362                     if (in[i] == '}') { term = true; break; }
 363                     name += in[i];
 364                 }
 365                 if (!term) throw mp::filter::FilterException
 366                     ("Unterminated var ref in '"+in+"' at "
 367                      + boost::lexical_cast<std::string>(i));
 368                 std::map<std::string, std::string>::const_iterator it
 369                     = vars.find(name);
 370                 if (it != vars.end())
 371                 {
 372                     out += it->second;
 373                 }
 374             }
 375             else
 376             {
 377                 throw mp::filter::FilterException
 378                     ("Malformed or trimmed var ref in '"
 379                      +in+"' at "+boost::lexical_cast<std::string>(i));
 380             }
 381             continue;
 382         }
 383         //passthru
 384         out += in[i];
 385         esc = false;
 386     }
 387     return out;
 388 }
 389
 390
 391 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 392 {
 393     std::map<std::string, RulePtr > rules;
 394     for (ptr = ptr->children; ptr; ptr = ptr->next)
 395     {
 396         if (ptr->type != XML_ELEMENT_NODE)
 397             continue;
 398         else if (!strcmp((const char *) ptr->name, "rule"))
 399         {
 400             static const char *names[2] = { "name", 0 };
 401             std::string values[1];
 402             values[0] = "default";
 403             mp::xml::parse_attr(ptr, names, values);
 404
 405             RulePtr rule(new Rule);
 406             for (xmlNode *p = ptr->children; p; p = p->next)
 407             {
 408                 if (p->type != XML_ELEMENT_NODE)
 409                     continue;
 410                 if (!strcmp((const char *) p->name, "rewrite"))
 411                 {
 412                     Replace replace;
 413                     const struct _xmlAttr *attr;
 414                     for (attr = p->properties; attr; attr = attr->next)
 415                     {
 416                         if (!strcmp((const char *) attr->name,  "from"))
 417                             replace.regex = mp::xml::get_text(attr->children);
 418                         else if (!strcmp((const char *) attr->name,  "to"))
 419                             replace.recipe = mp::xml::get_text(attr->children);
 420                         else
 421                             throw mp::filter::FilterException
 422                                 ("Bad attribute "
 423                                  + std::string((const char *) attr->name)
 424                                  + " in rewrite section of http_rewrite");
 425                     }
 426                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 427                             replace.regex.c_str(), replace.recipe.c_str());
 428                     replace.parse_groups();
 429                     if (!replace.regex.empty())
 430                         rule->replace_list.push_back(replace);
 431                 }
 432                 else
 433                     throw mp::filter::FilterException
 434                         ("Bad element "
 435                          + std::string((const char *) p->name)
 436                          + " in http_rewrite filter");
 437             }
 438             if (!rule->replace_list.empty())
 439                 rules[values[0]] = rule;
 440         }
 441         else if (!strcmp((const char *) ptr->name, "within"))
 442         {
 443             static const char *names[5] =
 444                 { "header", "attr", "tag", "rule", 0 };
 445             std::string values[4];
 446             mp::xml::parse_attr(ptr, names, values);
 447             Within w;
 448             w.header = values[0];
 449             w.attr = values[1];
 450             w.tag = values[2];
 451             std::map<std::string,RulePtr>::const_iterator it =
 452                 rules.find(values[3]);
 453             if (it == rules.end())
 454                 throw mp::filter::FilterException
 455                     ("Reference to non-existing rule '" + values[3] +
 456                      "' in http_rewrite filter");
 457             w.rule = it->second;
 458             phase.within_list.push_back(w);
 459         }
 460         else
 461         {
 462             throw mp::filter::FilterException
 463                 ("Bad element "
 464                  + std::string((const char *) ptr->name)
 465                  + " in http_rewrite filter");
 466         }
 467     }
 468 }
 469
 470 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 471         const char *path)
 472 {
 473     for (ptr = ptr->children; ptr; ptr = ptr->next)
 474     {
 475         if (ptr->type != XML_ELEMENT_NODE)
 476             continue;
 477         else if (!strcmp((const char *) ptr->name, "request"))
 478         {
 479             configure_phase(ptr, *req_phase);
 480         }
 481         else if (!strcmp((const char *) ptr->name, "response"))
 482         {
 483             configure_phase(ptr, *res_phase);
 484         }
 485         else
 486         {
 487             throw mp::filter::FilterException
 488                 ("Bad element "
 489                  + std::string((const char *) ptr->name)
 490                  + " in http_rewrite1 filter");
 491         }
 492     }
 493 }
 494
 495 static mp::filter::Base* filter_creator()
 496 {
 497     return new mp::filter::HttpRewrite;
 498 }
 499
 500 extern "C" {
 501     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 502         0,
 503         "http_rewrite",
 504         filter_creator
 505     };
 506 }
 507
 508
 509 /*
 510  * Local variables:
 511  * c-basic-offset: 4
 512  * c-file-style: "Stroustrup"
 513  * indent-tabs-mode: nil
 514  * End:
 515  * vim: shiftwidth=4 tabstop=8 expandtab
 516  */
 517