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 #include "html_parser.hpp"
  25
  26 #include <yaz/zgdu.h>
  27 #include <yaz/log.h>
  28
  29 #include <boost/regex.hpp>
  30 #include <boost/lexical_cast.hpp>
  31
  32 #include <map>
  33
  34 namespace mp = metaproxy_1;
  35 namespace yf = mp::filter;
  36
  37 namespace metaproxy_1 {
  38     namespace filter {
  39         class HttpRewrite::Replace {
  40         public:
  41             std::string regex;
  42             std::string recipe;
  43             std::map<int, std::string> group_index;
  44             const std::string search_replace(
  45                 std::map<std::string, std::string> & vars,
  46                 const std::string & txt) const;
  47             std::string sub_vars (
  48                 const std::map<std::string, std::string> & vars) const;
  49             void parse_groups();
  50         };
  51
  52         class HttpRewrite::Rule {
  53         public:
  54             std::list<Replace> replace_list;
  55             const std::string test_patterns(
  56                 std::map<std::string, std::string> & vars,
  57                 const std::string & txt) const;
  58         };
  59         class HttpRewrite::Within {
  60         public:
  61             std::string header;
  62             std::string attr;
  63             std::string tag;
  64             RulePtr rule;
  65         };
  66
  67         class HttpRewrite::Phase {
  68         public:
  69             std::list<Within> within_list;
  70             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  71                 std::map<std::string, std::string> & vars) const;
  72             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  73                 std::map<std::string, std::string> & vars) const;
  74             void rewrite_body(mp::odr & o,
  75                 char **content_buf, int *content_len,
  76                 std::map<std::string, std::string> & vars) const;
  77         };
  78         class HttpRewrite::Event : public HTMLParserEvent {
  79         public:
  80             void openTagStart(const char *name);
  81             void anyTagEnd(const char *name);
  82             void attribute(const char *tagName,
  83                            const char *name,
  84                            const char *value,
  85                            int val_len);
  86             void closeTag(const char *name);
  87             void text(const char *value, int len);
  88         };
  89     }
  90 }
  91
  92 yf::HttpRewrite::HttpRewrite() :
  93     req_phase(new Phase), res_phase(new Phase)
  94 {
  95 }
  96
  97 yf::HttpRewrite::~HttpRewrite()
  98 {
  99 }
 100
 101 void yf::HttpRewrite::process(mp::Package & package) const
 102 {
 103     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 104     Z_GDU *gdu = package.request().get();
 105     //map of request/response vars
 106     std::map<std::string, std::string> vars;
 107     //we have an http req
 108     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 109     {
 110         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 111         mp::odr o;
 112         req_phase->rewrite_reqline(o, hreq, vars);
 113         yaz_log(YLOG_LOG, ">> Request headers");
 114         req_phase->rewrite_headers(o, hreq->headers, vars);
 115         req_phase->rewrite_body(o,
 116                 &hreq->content_buf, &hreq->content_len, vars);
 117         package.request() = gdu;
 118     }
 119     package.move();
 120     gdu = package.response().get();
 121     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 122     {
 123         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 124         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 125         mp::odr o;
 126         yaz_log(YLOG_LOG, "<< Respose headers");
 127         res_phase->rewrite_headers(o, hres->headers, vars);
 128         res_phase->rewrite_body(o, &hres->content_buf,
 129                 &hres->content_len, vars);
 130         package.response() = gdu;
 131     }
 132 }
 133
 134 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 135         Z_HTTP_Request *hreq,
 136         std::map<std::string, std::string> & vars) const
 137 {
 138     //rewrite the request line
 139     std::string path;
 140     if (strstr(hreq->path, "http://") == hreq->path)
 141     {
 142         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 143             "possibly a proxy request");
 144         path += hreq->path;
 145     }
 146     else
 147     {
 148         //TODO what about proto
 149         path += "http://";
 150         path += z_HTTP_header_lookup(hreq->headers, "Host");
 151         path += hreq->path;
 152     }
 153
 154     std::list<Within>::const_iterator it = within_list.begin();
 155     if (it != within_list.end())
 156     {
 157         RulePtr rule = it->rule;
 158
 159         yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 160         std::string npath = rule->test_patterns(vars, path);
 161         if (!npath.empty())
 162         {
 163             yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 164             hreq->path = odr_strdup(o, npath.c_str());
 165         }
 166     }
 167 }
 168
 169 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 170         Z_HTTP_Header *headers,
 171         std::map<std::string, std::string> & vars) const
 172 {
 173     for (Z_HTTP_Header *header = headers;
 174             header != 0;
 175             header = header->next)
 176     {
 177         std::string sheader(header->name);
 178         sheader += ": ";
 179         sheader += header->value;
 180         yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
 181
 182         std::list<Within>::const_iterator it = within_list.begin();
 183         if (it == within_list.end())
 184             continue;
 185         RulePtr rule = it->rule;
 186
 187         std::string out = rule->test_patterns(vars, sheader);
 188         if (!out.empty())
 189         {
 190             size_t pos = out.find(": ");
 191             if (pos == std::string::npos)
 192             {
 193                 yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 194                 continue;
 195             }
 196             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 197             header->value = odr_strdup(o, out.substr(pos+2,
 198                                                      std::string::npos).c_str());
 199         }
 200     }
 201 }
 202
 203 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 204         char **content_buf,
 205         int *content_len,
 206         std::map<std::string, std::string> & vars) const
 207 {
 208     if (*content_buf)
 209     {
 210         HTMLParser parser;
 211         Event ev;
 212         std::string buf(*content_buf, *content_len);
 213
 214         parser.parse(ev, buf.c_str());
 215         std::list<Within>::const_iterator it = within_list.begin();
 216         if (it != within_list.end())
 217         {
 218             RulePtr rule = it->rule;
 219
 220             std::string body(*content_buf);
 221             std::string nbody = rule->test_patterns(vars, body);
 222             if (!nbody.empty())
 223             {
 224                 *content_buf = odr_strdup(o, nbody.c_str());
 225                 *content_len = nbody.size();
 226             }
 227         }
 228     }
 229 }
 230
 231
 232 void yf::HttpRewrite::Event::openTagStart(const char *name)
 233 {
 234 }
 235
 236 void yf::HttpRewrite::Event::anyTagEnd(const char *name)
 237 {
 238 }
 239
 240 void yf::HttpRewrite::Event::attribute(const char *tagName,
 241                                          const char *name,
 242                                          const char *value,
 243                                          int val_len)
 244 {
 245 }
 246
 247
 248 void yf::HttpRewrite::Event::closeTag(const char *name)
 249 {
 250 }
 251
 252 void yf::HttpRewrite::Event::text(const char *value, int len)
 253 {
 254 }
 255
 256
 257 /**
 258  * Tests pattern from the vector in order and executes recipe on
 259  the first match.
 260  */
 261 const std::string yf::HttpRewrite::Rule::test_patterns(
 262         std::map<std::string, std::string> & vars,
 263         const std::string & txt) const
 264 {
 265     std::list<Replace>::const_iterator it = replace_list.begin();
 266
 267     for (; it != replace_list.end(); it++)
 268     {
 269         std::string out = it->search_replace(vars, txt);
 270         if (!out.empty()) return out;
 271     }
 272     return "";
 273 }
 274
 275 const std::string yf::HttpRewrite::Replace::search_replace(
 276         std::map<std::string, std::string> & vars,
 277         const std::string & txt) const
 278 {
 279     //exec regex against value
 280     boost::regex re(regex);
 281     boost::smatch what;
 282     std::string::const_iterator start, end;
 283     start = txt.begin();
 284     end = txt.end();
 285     std::string out;
 286     while (regex_search(start, end, what, re)) //find next full match
 287     {
 288         size_t i;
 289         for (i = 1; i < what.size(); ++i)
 290         {
 291             //check if the group is named
 292             std::map<int, std::string>::const_iterator it
 293                 = group_index.find(i);
 294             if (it != group_index.end())
 295             {   //it is
 296                 if (!what[i].str().empty())
 297                     vars[it->second] = what[i];
 298             }
 299
 300         }
 301         //prepare replacement string
 302         std::string rvalue = sub_vars(vars);
 303         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 304                 what.str(0).c_str(), rvalue.c_str());
 305         out.append(start, what[0].first);
 306         out.append(rvalue);
 307         start = what[0].second; //move search forward
 308     }
 309     //if we had a match cat the last part
 310     if (start != txt.begin())
 311         out.append(start, end);
 312     return out;
 313 }
 314
 315 void yf::HttpRewrite::Replace::parse_groups()
 316 {
 317     int gnum = 0;
 318     bool esc = false;
 319     const std::string & str = regex;
 320     std::string res;
 321     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 322     for (size_t i = 0; i < str.size(); ++i)
 323     {
 324         res += str[i];
 325         if (!esc && str[i] == '\\')
 326         {
 327             esc = true;
 328             continue;
 329         }
 330         if (!esc && str[i] == '(') //group starts
 331         {
 332             gnum++;
 333             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 334             {
 335                 i++;
 336                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 337                 {
 338                     if (gnum > 0) gnum--;
 339                     res += str[i];
 340                     i++;
 341                     res += str[i];
 342                     continue;
 343                 }
 344                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 345                     i++;
 346                 if (i+1 < str.size() && str[i+1] == '<') //named
 347                 {
 348                     i++;
 349                     std::string gname;
 350                     bool term = false;
 351                     while (++i < str.size())
 352                     {
 353                         if (str[i] == '>') { term = true; break; }
 354                         if (!isalnum(str[i]))
 355                             throw mp::filter::FilterException
 356                                 ("Only alphanumeric chars allowed, found "
 357                                  " in '"
 358                                  + str
 359                                  + "' at "
 360                                  + boost::lexical_cast<std::string>(i));
 361                         gname += str[i];
 362                     }
 363                     if (!term)
 364                         throw mp::filter::FilterException
 365                             ("Unterminated group name '" + gname
 366                              + " in '" + str +"'");
 367                     group_index[gnum] = gname;
 368                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 369                             gname.c_str(), gnum);
 370                 }
 371             }
 372         }
 373         esc = false;
 374     }
 375     regex = res;
 376 }
 377
 378 std::string yf::HttpRewrite::Replace::sub_vars (
 379         const std::map<std::string, std::string> & vars) const
 380 {
 381     std::string out;
 382     bool esc = false;
 383     const std::string & in = recipe;
 384     for (size_t i = 0; i < in.size(); ++i)
 385     {
 386         if (!esc && in[i] == '\\')
 387         {
 388             esc = true;
 389             continue;
 390         }
 391         if (!esc && in[i] == '$') //var
 392         {
 393             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 394             {
 395                 ++i;
 396                 std::string name;
 397                 bool term = false;
 398                 while (++i < in.size())
 399                 {
 400                     if (in[i] == '}') { term = true; break; }
 401                     name += in[i];
 402                 }
 403                 if (!term) throw mp::filter::FilterException
 404                     ("Unterminated var ref in '"+in+"' at "
 405                      + boost::lexical_cast<std::string>(i));
 406                 std::map<std::string, std::string>::const_iterator it
 407                     = vars.find(name);
 408                 if (it != vars.end())
 409                 {
 410                     out += it->second;
 411                 }
 412             }
 413             else
 414             {
 415                 throw mp::filter::FilterException
 416                     ("Malformed or trimmed var ref in '"
 417                      +in+"' at "+boost::lexical_cast<std::string>(i));
 418             }
 419             continue;
 420         }
 421         //passthru
 422         out += in[i];
 423         esc = false;
 424     }
 425     return out;
 426 }
 427
 428
 429 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 430 {
 431     std::map<std::string, RulePtr > rules;
 432     for (ptr = ptr->children; ptr; ptr = ptr->next)
 433     {
 434         if (ptr->type != XML_ELEMENT_NODE)
 435             continue;
 436         else if (!strcmp((const char *) ptr->name, "rule"))
 437         {
 438             static const char *names[2] = { "name", 0 };
 439             std::string values[1];
 440             values[0] = "default";
 441             mp::xml::parse_attr(ptr, names, values);
 442
 443             RulePtr rule(new Rule);
 444             for (xmlNode *p = ptr->children; p; p = p->next)
 445             {
 446                 if (p->type != XML_ELEMENT_NODE)
 447                     continue;
 448                 if (!strcmp((const char *) p->name, "rewrite"))
 449                 {
 450                     Replace replace;
 451                     const struct _xmlAttr *attr;
 452                     for (attr = p->properties; attr; attr = attr->next)
 453                     {
 454                         if (!strcmp((const char *) attr->name,  "from"))
 455                             replace.regex = mp::xml::get_text(attr->children);
 456                         else if (!strcmp((const char *) attr->name,  "to"))
 457                             replace.recipe = mp::xml::get_text(attr->children);
 458                         else
 459                             throw mp::filter::FilterException
 460                                 ("Bad attribute "
 461                                  + std::string((const char *) attr->name)
 462                                  + " in rewrite section of http_rewrite");
 463                     }
 464                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 465                             replace.regex.c_str(), replace.recipe.c_str());
 466                     replace.parse_groups();
 467                     if (!replace.regex.empty())
 468                         rule->replace_list.push_back(replace);
 469                 }
 470                 else
 471                     throw mp::filter::FilterException
 472                         ("Bad element "
 473                          + std::string((const char *) p->name)
 474                          + " in http_rewrite filter");
 475             }
 476             if (!rule->replace_list.empty())
 477                 rules[values[0]] = rule;
 478         }
 479         else if (!strcmp((const char *) ptr->name, "within"))
 480         {
 481             static const char *names[5] =
 482                 { "header", "attr", "tag", "rule", 0 };
 483             std::string values[4];
 484             mp::xml::parse_attr(ptr, names, values);
 485             Within w;
 486             w.header = values[0];
 487             w.attr = values[1];
 488             w.tag = values[2];
 489             std::map<std::string,RulePtr>::const_iterator it =
 490                 rules.find(values[3]);
 491             if (it == rules.end())
 492                 throw mp::filter::FilterException
 493                     ("Reference to non-existing rule '" + values[3] +
 494                      "' in http_rewrite filter");
 495             w.rule = it->second;
 496             phase.within_list.push_back(w);
 497         }
 498         else
 499         {
 500             throw mp::filter::FilterException
 501                 ("Bad element "
 502                  + std::string((const char *) ptr->name)
 503                  + " in http_rewrite filter");
 504         }
 505     }
 506 }
 507
 508 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 509         const char *path)
 510 {
 511     for (ptr = ptr->children; ptr; ptr = ptr->next)
 512     {
 513         if (ptr->type != XML_ELEMENT_NODE)
 514             continue;
 515         else if (!strcmp((const char *) ptr->name, "request"))
 516         {
 517             configure_phase(ptr, *req_phase);
 518         }
 519         else if (!strcmp((const char *) ptr->name, "response"))
 520         {
 521             configure_phase(ptr, *res_phase);
 522         }
 523         else
 524         {
 525             throw mp::filter::FilterException
 526                 ("Bad element "
 527                  + std::string((const char *) ptr->name)
 528                  + " in http_rewrite1 filter");
 529         }
 530     }
 531 }
 532
 533 static mp::filter::Base* filter_creator()
 534 {
 535     return new mp::filter::HttpRewrite;
 536 }
 537
 538 extern "C" {
 539     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 540         0,
 541         "http_rewrite",
 542         filter_creator
 543     };
 544 }
 545
 546
 547 /*
 548  * Local variables:
 549  * c-basic-offset: 4
 550  * c-file-style: "Stroustrup"
 551  * indent-tabs-mode: nil
 552  * End:
 553  * vim: shiftwidth=4 tabstop=8 expandtab
 554  */
 555