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 <stack>
  30 #include <boost/regex.hpp>
  31 #include <boost/lexical_cast.hpp>
  32 #include <boost/algorithm/string.hpp>
  33
  34 #include <map>
  35
  36 namespace mp = metaproxy_1;
  37 namespace yf = mp::filter;
  38
  39 namespace metaproxy_1 {
  40     namespace filter {
  41         class HttpRewrite::Replace {
  42         public:
  43             boost::regex re;
  44             boost::smatch what;
  45             std::string recipe;
  46             std::map<int, std::string> group_index;
  47             std::string sub_vars(
  48                 const std::map<std::string, std::string> & vars) const;
  49             void parse_groups(std::string pattern);
  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);
  58         };
  59         class HttpRewrite::Within {
  60         public:
  61             std::string header;
  62             std::string attr;
  63             std::string tag;
  64             bool reqline;
  65             RulePtr rule;
  66         };
  67
  68         class HttpRewrite::Phase {
  69         public:
  70             Phase();
  71             std::list<Within> within_list;
  72             int m_verbose;
  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         class HttpRewrite::Event : public HTMLParserEvent {
  82             void openTagStart(const char *tag, int tag_len);
  83             void anyTagEnd(const char *tag, int tag_len, int close_it);
  84             void attribute(const char *tag, int tag_len,
  85                            const char *attr, int attr_len,
  86                            const char *value, int val_len,
  87                            const char *sep);
  88             void closeTag(const char *tag, int tag_len);
  89             void text(const char *value, int len);
  90             const Phase *m_phase;
  91             WRBUF m_w;
  92             std::stack<std::list<Within>::const_iterator> s_within;
  93             std::map<std::string, std::string> &m_vars;
  94         public:
  95             Event(const Phase *p, std::map<std::string, std::string> &vars);
  96             ~Event();
  97             const char *result();
  98         };
  99     }
 100 }
 101
 102 yf::HttpRewrite::HttpRewrite() :
 103     req_phase(new Phase), res_phase(new Phase)
 104 {
 105 }
 106
 107 yf::HttpRewrite::~HttpRewrite()
 108 {
 109 }
 110
 111 void yf::HttpRewrite::process(mp::Package & package) const
 112 {
 113     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 114     Z_GDU *gdu = package.request().get();
 115     //map of request/response vars
 116     std::map<std::string, std::string> vars;
 117     //we have an http req
 118     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 119     {
 120         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 121         mp::odr o;
 122         req_phase->rewrite_reqline(o, hreq, vars);
 123         yaz_log(YLOG_LOG, ">> Request headers");
 124         req_phase->rewrite_headers(o, hreq->headers, vars);
 125         req_phase->rewrite_body(o,
 126                 &hreq->content_buf, &hreq->content_len, vars);
 127         package.request() = gdu;
 128     }
 129     package.move();
 130     gdu = package.response().get();
 131     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 132     {
 133         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 134         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 135         mp::odr o;
 136         yaz_log(YLOG_LOG, "<< Respose headers");
 137         res_phase->rewrite_headers(o, hres->headers, vars);
 138         res_phase->rewrite_body(o, &hres->content_buf,
 139                 &hres->content_len, vars);
 140         package.response() = gdu;
 141     }
 142 }
 143
 144 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 145         Z_HTTP_Request *hreq,
 146         std::map<std::string, std::string> & vars) const
 147 {
 148     //rewrite the request line
 149     std::string path;
 150     if (strstr(hreq->path, "http://") == hreq->path)
 151     {
 152         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 153             "possibly a proxy request");
 154         path += hreq->path;
 155     }
 156     else
 157     {
 158         //TODO what about proto
 159         path += "http://";
 160         path += z_HTTP_header_lookup(hreq->headers, "Host");
 161         path += hreq->path;
 162     }
 163
 164
 165     std::list<Within>::const_iterator it = within_list.begin();
 166     for (; it != within_list.end(); it++)
 167         if (it->reqline)
 168         {
 169             RulePtr rule = it->rule;
 170             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 171             std::string npath = rule->test_patterns(vars, path);
 172             if (!npath.empty())
 173             {
 174                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 175                 hreq->path = odr_strdup(o, npath.c_str());
 176             }
 177         }
 178 }
 179
 180 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 181         Z_HTTP_Header *headers,
 182         std::map<std::string, std::string> & vars) const
 183 {
 184     for (Z_HTTP_Header *header = headers; header; header = header->next)
 185     {
 186         std::list<Within>::const_iterator it = within_list.begin();
 187         for (; it != within_list.end(); it++)
 188         {
 189             if (it->header.length() > 0 &&
 190                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 191             {
 192                 std::string sheader(header->name);
 193                 sheader += ": ";
 194                 sheader += header->value;
 195
 196                 RulePtr rule = it->rule;
 197                 std::string out = rule->test_patterns(vars, sheader);
 198                 if (!out.empty())
 199                 {
 200                     size_t pos = out.find(": ");
 201                     if (pos == std::string::npos)
 202                     {
 203                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 204                         continue;
 205                     }
 206                     header->name = odr_strdup(o, out.substr(0, pos).c_str());
 207                     header->value = odr_strdup(o,
 208                                                out.substr(pos + 2,
 209                                                           std::string::npos).c_str());
 210                 }
 211             }
 212         }
 213     }
 214 }
 215
 216 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 217         char **content_buf,
 218         int *content_len,
 219         std::map<std::string, std::string> & vars) const
 220 {
 221     if (*content_buf)
 222     {
 223         int i;
 224         for (i = 0; i < *content_len; i++)
 225             if ((*content_buf)[i] == 0)
 226                 return;  // binary content. skip
 227
 228         HTMLParser parser;
 229         Event ev(this, vars);
 230
 231         parser.set_verbose(m_verbose);
 232
 233         std::string buf(*content_buf, *content_len);
 234
 235         parser.parse(ev, buf.c_str());
 236         const char *res = ev.result();
 237         *content_buf = odr_strdup(o, res);
 238         *content_len = strlen(res);
 239     }
 240 }
 241
 242 yf::HttpRewrite::Event::Event(const Phase *p,
 243                               std::map<std::string, std::string> & vars
 244     ) : m_phase(p), m_vars(vars)
 245 {
 246     m_w = wrbuf_alloc();
 247 }
 248
 249 yf::HttpRewrite::Event::~Event()
 250 {
 251     wrbuf_destroy(m_w);
 252 }
 253
 254 const char *yf::HttpRewrite::Event::result()
 255 {
 256     return wrbuf_cstr(m_w);
 257 }
 258
 259 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 260 {
 261     wrbuf_putc(m_w, '<');
 262     wrbuf_write(m_w, tag, tag_len);
 263
 264     std::string t(tag, tag_len);
 265     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 266     for (; it != m_phase->within_list.end(); it++)
 267     {
 268         if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
 269                                                    t.c_str()) == 0)
 270         {
 271             std::vector<std::string> attr;
 272             boost::split(attr, it->attr, boost::is_any_of(","));
 273             size_t i;
 274             for (i = 0; i < attr.size(); i++)
 275             {
 276                 if (attr[i].compare("#text") == 0)
 277                 {
 278                     s_within.push(it);
 279                     return;
 280                 }
 281             }
 282         }
 283     }
 284 }
 285
 286 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 287                                        int close_it)
 288 {
 289     if (close_it)
 290     {
 291         if (!s_within.empty())
 292         {
 293             std::list<Within>::const_iterator it = s_within.top();
 294             std::string t(tag, tag_len);
 295             if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 296                 s_within.pop();
 297         }
 298     }
 299     if (close_it)
 300         wrbuf_putc(m_w, '/');
 301     wrbuf_putc(m_w, '>');
 302 }
 303
 304 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 305                                        const char *attr, int attr_len,
 306                                        const char *value, int val_len,
 307                                        const char *sep)
 308 {
 309     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 310     bool subst = false;
 311
 312     for (; it != m_phase->within_list.end(); it++)
 313     {
 314         std::string t(tag, tag_len);
 315         if (it->tag.length() == 0 ||
 316             yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 317         {
 318             std::string a(attr, attr_len);
 319             std::vector<std::string> attr;
 320             boost::split(attr, it->attr, boost::is_any_of(","));
 321             size_t i;
 322             for (i = 0; i < attr.size(); i++)
 323             {
 324                 if (attr[i].compare("#text") &&
 325                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 326                     subst = true;
 327             }
 328         }
 329         if (subst)
 330             break;
 331     }
 332
 333     wrbuf_putc(m_w, ' ');
 334     wrbuf_write(m_w, attr, attr_len);
 335     if (value)
 336     {
 337         wrbuf_puts(m_w, "=");
 338         wrbuf_puts(m_w, sep);
 339
 340         std::string output;
 341         if (subst)
 342         {
 343             std::string input(value, val_len);
 344             output = it->rule->test_patterns(m_vars, input);
 345         }
 346         if (output.empty())
 347             wrbuf_write(m_w, value, val_len);
 348         else
 349             wrbuf_puts(m_w, output.c_str());
 350         wrbuf_puts(m_w, sep);
 351     }
 352 }
 353
 354 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 355 {
 356     if (!s_within.empty())
 357     {
 358         std::list<Within>::const_iterator it = s_within.top();
 359         std::string t(tag, tag_len);
 360         if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 361             s_within.pop();
 362     }
 363     wrbuf_puts(m_w, "</");
 364     wrbuf_write(m_w, tag, tag_len);
 365 }
 366
 367 void yf::HttpRewrite::Event::text(const char *value, int len)
 368 {
 369     std::list<Within>::const_iterator it = m_phase->within_list.end();
 370     if (!s_within.empty())
 371         it = s_within.top();
 372     std::string output;
 373     if (it != m_phase->within_list.end())
 374     {
 375         std::string input(value, len);
 376         output = it->rule->test_patterns(m_vars, input);
 377     }
 378     if (output.empty())
 379         wrbuf_write(m_w, value, len);
 380     else
 381         wrbuf_puts(m_w, output.c_str());
 382 }
 383
 384 const std::string yf::HttpRewrite::Rule::test_patterns(
 385         std::map<std::string, std::string> & vars,
 386         const std::string & txt)
 387 {
 388     std::string out;
 389     std::string::const_iterator start, end;
 390     start = txt.begin();
 391     end = txt.end();
 392     while (1)
 393     {
 394         std::list<Replace>::iterator bit = replace_list.end();
 395         {
 396             std::string::const_iterator best_pos = txt.end();
 397             std::list<Replace>::iterator it = replace_list.begin();
 398             for (; it != replace_list.end(); it++)
 399             {
 400                 if (regex_search(start, end, it->what, it->re))
 401                 {
 402                     if (it->what[0].first < best_pos)
 403                     {
 404                         best_pos = it->what[0].first;
 405                         bit = it;
 406                     }
 407                 }
 408             }
 409             if (bit == replace_list.end())
 410                 break;
 411         }
 412
 413         size_t i;
 414         for (i = 1; i < bit->what.size(); ++i)
 415         {
 416             //check if the group is named
 417             std::map<int, std::string>::const_iterator git
 418                 = bit->group_index.find(i);
 419             if (git != bit->group_index.end())
 420             {   //it is
 421                 vars[git->second] = bit->what[i];
 422             }
 423
 424         }
 425         //prepare replacement string
 426         std::string rvalue = bit->sub_vars(vars);
 427         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 428                 bit->what.str(0).c_str(), rvalue.c_str());
 429         out.append(start, bit->what[0].first);
 430         out.append(rvalue);
 431         start = bit->what[0].second; //move search forward
 432     }
 433     if (start != txt.begin())
 434         out.append(start, end);
 435     return out;
 436 }
 437
 438 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 439 {
 440     int gnum = 0;
 441     bool esc = false;
 442     const std::string &str = pattern;
 443     std::string res;
 444     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 445     for (size_t i = 0; i < str.size(); ++i)
 446     {
 447         res += str[i];
 448         if (!esc && str[i] == '\\')
 449         {
 450             esc = true;
 451             continue;
 452         }
 453         if (!esc && str[i] == '(') //group starts
 454         {
 455             gnum++;
 456             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 457             {
 458                 i++;
 459                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 460                 {
 461                     if (gnum > 0) gnum--;
 462                     res += str[i];
 463                     i++;
 464                     res += str[i];
 465                     continue;
 466                 }
 467                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 468                     i++;
 469                 if (i+1 < str.size() && str[i+1] == '<') //named
 470                 {
 471                     i++;
 472                     std::string gname;
 473                     bool term = false;
 474                     while (++i < str.size())
 475                     {
 476                         if (str[i] == '>') { term = true; break; }
 477                         if (!isalnum(str[i]))
 478                             throw mp::filter::FilterException
 479                                 ("Only alphanumeric chars allowed, found "
 480                                  " in '"
 481                                  + str
 482                                  + "' at "
 483                                  + boost::lexical_cast<std::string>(i));
 484                         gname += str[i];
 485                     }
 486                     if (!term)
 487                         throw mp::filter::FilterException
 488                             ("Unterminated group name '" + gname
 489                              + " in '" + str +"'");
 490                     group_index[gnum] = gname;
 491                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 492                             gname.c_str(), gnum);
 493                 }
 494             }
 495         }
 496         esc = false;
 497     }
 498     re = res;
 499 }
 500
 501 std::string yf::HttpRewrite::Replace::sub_vars(
 502     const std::map<std::string, std::string> & vars) const
 503 {
 504     std::string out;
 505     bool esc = false;
 506     const std::string & in = recipe;
 507     for (size_t i = 0; i < in.size(); ++i)
 508     {
 509         if (!esc && in[i] == '\\')
 510         {
 511             esc = true;
 512             continue;
 513         }
 514         if (!esc && in[i] == '$') //var
 515         {
 516             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 517             {
 518                 ++i;
 519                 std::string name;
 520                 bool term = false;
 521                 while (++i < in.size())
 522                 {
 523                     if (in[i] == '}') { term = true; break; }
 524                     name += in[i];
 525                 }
 526                 if (!term) throw mp::filter::FilterException
 527                     ("Unterminated var ref in '"+in+"' at "
 528                      + boost::lexical_cast<std::string>(i));
 529                 std::map<std::string, std::string>::const_iterator it
 530                     = vars.find(name);
 531                 if (it != vars.end())
 532                 {
 533                     out += it->second;
 534                 }
 535             }
 536             else
 537             {
 538                 throw mp::filter::FilterException
 539                     ("Malformed or trimmed var ref in '"
 540                      +in+"' at "+boost::lexical_cast<std::string>(i));
 541             }
 542             continue;
 543         }
 544         //passthru
 545         out += in[i];
 546         esc = false;
 547     }
 548     return out;
 549 }
 550
 551 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 552 {
 553 }
 554
 555 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 556 {
 557     static const char *names[2] = { "verbose", 0 };
 558     std::string values[1];
 559     values[0] = "0";
 560     mp::xml::parse_attr(ptr, names, values);
 561
 562     phase.m_verbose = atoi(values[0].c_str());
 563
 564     std::map<std::string, RulePtr > rules;
 565     for (ptr = ptr->children; ptr; ptr = ptr->next)
 566     {
 567         if (ptr->type != XML_ELEMENT_NODE)
 568             continue;
 569         else if (!strcmp((const char *) ptr->name, "rule"))
 570         {
 571             static const char *names[2] = { "name", 0 };
 572             std::string values[1];
 573             values[0] = "default";
 574             mp::xml::parse_attr(ptr, names, values);
 575
 576             RulePtr rule(new Rule);
 577             for (xmlNode *p = ptr->children; p; p = p->next)
 578             {
 579                 if (p->type != XML_ELEMENT_NODE)
 580                     continue;
 581                 if (!strcmp((const char *) p->name, "rewrite"))
 582                 {
 583                     Replace replace;
 584                     std::string from;
 585                     const struct _xmlAttr *attr;
 586                     for (attr = p->properties; attr; attr = attr->next)
 587                     {
 588                         if (!strcmp((const char *) attr->name,  "from"))
 589                             from = mp::xml::get_text(attr->children);
 590                         else if (!strcmp((const char *) attr->name,  "to"))
 591                             replace.recipe = mp::xml::get_text(attr->children);
 592                         else
 593                             throw mp::filter::FilterException
 594                                 ("Bad attribute "
 595                                  + std::string((const char *) attr->name)
 596                                  + " in rewrite section of http_rewrite");
 597                     }
 598                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 599                             from.c_str(), replace.recipe.c_str());
 600                     if (!from.empty())
 601                     {
 602                         replace.parse_groups(from);
 603                         rule->replace_list.push_back(replace);
 604                     }
 605                 }
 606                 else
 607                     throw mp::filter::FilterException
 608                         ("Bad element "
 609                          + std::string((const char *) p->name)
 610                          + " in http_rewrite filter");
 611             }
 612             rules[values[0]] = rule;
 613         }
 614         else if (!strcmp((const char *) ptr->name, "within"))
 615         {
 616             static const char *names[6] =
 617                 { "header", "attr", "tag", "rule", "reqline", 0 };
 618             std::string values[5];
 619             mp::xml::parse_attr(ptr, names, values);
 620             Within w;
 621             w.header = values[0];
 622             w.attr = values[1];
 623             w.tag = values[2];
 624             std::map<std::string,RulePtr>::const_iterator it =
 625                 rules.find(values[3]);
 626             if (it == rules.end())
 627                 throw mp::filter::FilterException
 628                     ("Reference to non-existing rule '" + values[3] +
 629                      "' in http_rewrite filter");
 630             w.rule = it->second;
 631             w.reqline = values[4] == "1";
 632             phase.within_list.push_back(w);
 633         }
 634         else
 635         {
 636             throw mp::filter::FilterException
 637                 ("Bad element "
 638                  + std::string((const char *) ptr->name)
 639                  + " in http_rewrite filter");
 640         }
 641     }
 642 }
 643
 644 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 645         const char *path)
 646 {
 647     for (ptr = ptr->children; ptr; ptr = ptr->next)
 648     {
 649         if (ptr->type != XML_ELEMENT_NODE)
 650             continue;
 651         else if (!strcmp((const char *) ptr->name, "request"))
 652         {
 653             configure_phase(ptr, *req_phase);
 654         }
 655         else if (!strcmp((const char *) ptr->name, "response"))
 656         {
 657             configure_phase(ptr, *res_phase);
 658         }
 659         else
 660         {
 661             throw mp::filter::FilterException
 662                 ("Bad element "
 663                  + std::string((const char *) ptr->name)
 664                  + " in http_rewrite1 filter");
 665         }
 666     }
 667 }
 668
 669 static mp::filter::Base* filter_creator()
 670 {
 671     return new mp::filter::HttpRewrite;
 672 }
 673
 674 extern "C" {
 675     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 676         0,
 677         "http_rewrite",
 678         filter_creator
 679     };
 680 }
 681
 682
 683 /*
 684  * Local variables:
 685  * c-basic-offset: 4
 686  * c-file-style: "Stroustrup"
 687  * indent-tabs-mode: nil
 688  * End:
 689  * vim: shiftwidth=4 tabstop=8 expandtab
 690  */
 691