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 #include <boost/algorithm/string.hpp>
  32
  33 #include <map>
  34
  35 namespace mp = metaproxy_1;
  36 namespace yf = mp::filter;
  37
  38 namespace metaproxy_1 {
  39     namespace filter {
  40         class HttpRewrite::Replace {
  41         public:
  42             std::string regex;
  43             std::string recipe;
  44             std::map<int, std::string> group_index;
  45             const std::string search_replace(
  46                 std::map<std::string, std::string> & vars,
  47                 const std::string & txt) const;
  48             std::string sub_vars (
  49                 const std::map<std::string, std::string> & vars) const;
  50             void parse_groups();
  51         };
  52
  53         class HttpRewrite::Rule {
  54         public:
  55             std::list<Replace> replace_list;
  56             const std::string test_patterns(
  57                 std::map<std::string, std::string> & vars,
  58                 const std::string & txt) const;
  59         };
  60         class HttpRewrite::Within {
  61         public:
  62             std::string header;
  63             std::string attr;
  64             std::string tag;
  65             bool reqline;
  66             RulePtr rule;
  67         };
  68
  69         class HttpRewrite::Phase {
  70         public:
  71             std::list<Within> within_list;
  72             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  73                 std::map<std::string, std::string> & vars) const;
  74             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  75                 std::map<std::string, std::string> & vars) const;
  76             void rewrite_body(mp::odr & o,
  77                 char **content_buf, int *content_len,
  78                 std::map<std::string, std::string> & vars) const;
  79         };
  80         class HttpRewrite::Event : public HTMLParserEvent {
  81             void openTagStart(const char *name);
  82             void anyTagEnd(const char *name, int close_it);
  83             void attribute(const char *tagName,
  84                            const char *name,
  85                            const char *value,
  86                            int val_len);
  87             void closeTag(const char *name);
  88             void text(const char *value, int len);
  89             const Phase *m_phase;
  90             WRBUF m_w;
  91             std::list<Within>::const_iterator enabled_within;
  92             std::map<std::string, std::string> &m_vars;
  93         public:
  94             Event(const Phase *p, std::map<std::string, std::string> &vars);
  95             ~Event();
  96             const char *result();
  97         };
  98     }
  99 }
 100
 101 yf::HttpRewrite::HttpRewrite() :
 102     req_phase(new Phase), res_phase(new Phase)
 103 {
 104 }
 105
 106 yf::HttpRewrite::~HttpRewrite()
 107 {
 108 }
 109
 110 void yf::HttpRewrite::process(mp::Package & package) const
 111 {
 112     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 113     Z_GDU *gdu = package.request().get();
 114     //map of request/response vars
 115     std::map<std::string, std::string> vars;
 116     //we have an http req
 117     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 118     {
 119         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 120         mp::odr o;
 121         req_phase->rewrite_reqline(o, hreq, vars);
 122         yaz_log(YLOG_LOG, ">> Request headers");
 123         req_phase->rewrite_headers(o, hreq->headers, vars);
 124         req_phase->rewrite_body(o,
 125                 &hreq->content_buf, &hreq->content_len, vars);
 126         package.request() = gdu;
 127     }
 128     package.move();
 129     gdu = package.response().get();
 130     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 131     {
 132         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 133         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 134         mp::odr o;
 135         yaz_log(YLOG_LOG, "<< Respose headers");
 136         res_phase->rewrite_headers(o, hres->headers, vars);
 137         res_phase->rewrite_body(o, &hres->content_buf,
 138                 &hres->content_len, vars);
 139         package.response() = gdu;
 140     }
 141 }
 142
 143 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 144         Z_HTTP_Request *hreq,
 145         std::map<std::string, std::string> & vars) const
 146 {
 147     //rewrite the request line
 148     std::string path;
 149     if (strstr(hreq->path, "http://") == hreq->path)
 150     {
 151         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 152             "possibly a proxy request");
 153         path += hreq->path;
 154     }
 155     else
 156     {
 157         //TODO what about proto
 158         path += "http://";
 159         path += z_HTTP_header_lookup(hreq->headers, "Host");
 160         path += hreq->path;
 161     }
 162
 163
 164     std::list<Within>::const_iterator it = within_list.begin();
 165     for (; it != within_list.end(); it++)
 166         if (it->reqline)
 167         {
 168             RulePtr rule = it->rule;
 169             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 170             std::string npath = rule->test_patterns(vars, path);
 171             if (!npath.empty())
 172             {
 173                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 174                 hreq->path = odr_strdup(o, npath.c_str());
 175             }
 176         }
 177 }
 178
 179 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 180         Z_HTTP_Header *headers,
 181         std::map<std::string, std::string> & vars) const
 182 {
 183     for (Z_HTTP_Header *header = headers; header; header = header->next)
 184     {
 185         std::list<Within>::const_iterator it = within_list.begin();
 186         for (; it != within_list.end(); it++)
 187         {
 188             if (it->header.length() > 0 &&
 189                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 190             {
 191                 std::string sheader(header->name);
 192                 sheader += ": ";
 193                 sheader += header->value;
 194
 195                 RulePtr rule = it->rule;
 196                 std::string out = rule->test_patterns(vars, sheader);
 197                 if (!out.empty())
 198                 {
 199                     size_t pos = out.find(": ");
 200                     if (pos == std::string::npos)
 201                     {
 202                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 203                         continue;
 204                     }
 205                     header->name = odr_strdup(o, out.substr(0, pos).c_str());
 206                     header->value = odr_strdup(o,
 207                                                out.substr(pos + 2,
 208                                                           std::string::npos).c_str());
 209                 }
 210             }
 211         }
 212     }
 213 }
 214
 215 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 216         char **content_buf,
 217         int *content_len,
 218         std::map<std::string, std::string> & vars) const
 219 {
 220     if (*content_buf)
 221     {
 222         int i;
 223         for (i = 0; i < *content_len; i++)
 224             if ((*content_buf)[i] == 0)
 225                 return;  // binary content. skip
 226
 227         HTMLParser parser;
 228         Event ev(this, vars);
 229         std::string buf(*content_buf, *content_len);
 230
 231         parser.parse(ev, buf.c_str());
 232         const char *res = ev.result();
 233         *content_buf = odr_strdup(o, res);
 234         *content_len = strlen(res);
 235     }
 236 }
 237
 238 yf::HttpRewrite::Event::Event(const Phase *p,
 239                               std::map<std::string, std::string> & vars
 240     ) : m_phase(p), m_vars(vars)
 241 {
 242     m_w = wrbuf_alloc();
 243     enabled_within = m_phase->within_list.end();
 244 }
 245
 246 yf::HttpRewrite::Event::~Event()
 247 {
 248     wrbuf_destroy(m_w);
 249 }
 250
 251 const char *yf::HttpRewrite::Event::result()
 252 {
 253     return wrbuf_cstr(m_w);
 254 }
 255
 256 void yf::HttpRewrite::Event::openTagStart(const char *name)
 257 {
 258     // check if there is <within tag="x" .. />
 259     if (enabled_within == m_phase->within_list.end())
 260     {
 261         std::list<Within>::const_iterator it =
 262             m_phase->within_list.begin();
 263         for (; it != m_phase->within_list.end(); it++)
 264         {
 265             if (it->tag.length() > 0 && it->tag.compare(name) == 0)
 266             {
 267                 enabled_within = it;
 268             }
 269         }
 270     }
 271     wrbuf_putc(m_w, '<');
 272     wrbuf_puts(m_w, name);
 273 }
 274
 275 void yf::HttpRewrite::Event::anyTagEnd(const char *name, int close_it)
 276 {
 277     if (close_it)
 278     {
 279         std::list<Within>::const_iterator it = enabled_within;
 280         if (it != m_phase->within_list.end())
 281         {
 282             if (it->tag.compare(name) == 0)
 283             {
 284                 enabled_within = m_phase->within_list.end();
 285             }
 286         }
 287     }
 288     if (close_it)
 289         wrbuf_putc(m_w, '/');
 290     wrbuf_putc(m_w, '>');
 291 }
 292
 293 void yf::HttpRewrite::Event::attribute(const char *tagName,
 294                                          const char *name,
 295                                          const char *value,
 296                                          int val_len)
 297 {
 298     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 299     bool subst = false;
 300
 301     for (; it != m_phase->within_list.end(); it++)
 302     {
 303         if (it->tag.length() == 0 || it->tag.compare(tagName) == 0)
 304         {
 305             std::vector<std::string> attr;
 306             boost::split(attr, it->attr, boost::is_any_of(","));
 307             size_t i;
 308             for (i = 0; i < attr.size(); i++)
 309             {
 310                 if (attr[i].compare("#text") && attr[i].compare(name) == 0)
 311                     subst = true;
 312             }
 313         }
 314         if (subst)
 315             break;
 316     }
 317
 318     wrbuf_putc(m_w, ' ');
 319     wrbuf_puts(m_w, name);
 320     wrbuf_puts(m_w, "=\"");
 321
 322     std::string output;
 323     if (subst)
 324     {
 325         std::string input(value, val_len);
 326         output = it->rule->test_patterns(m_vars, input);
 327     }
 328     if (output.empty())
 329         wrbuf_write(m_w, value, val_len);
 330     else
 331         wrbuf_puts(m_w, output.c_str());
 332     wrbuf_puts(m_w, "\"");
 333 }
 334
 335 void yf::HttpRewrite::Event::closeTag(const char *name)
 336 {
 337     std::list<Within>::const_iterator it = enabled_within;
 338     if (it != m_phase->within_list.end())
 339     {
 340         if (it->tag.compare(name) == 0)
 341         {
 342             enabled_within = m_phase->within_list.end();
 343         }
 344     }
 345     wrbuf_puts(m_w, "</");
 346     wrbuf_puts(m_w, name);
 347 }
 348
 349 void yf::HttpRewrite::Event::text(const char *value, int len)
 350 {
 351     std::list<Within>::const_iterator it = enabled_within;
 352     bool subst = false;
 353
 354     if (it != m_phase->within_list.end())
 355     {
 356         subst = true;
 357         if (it->attr.length() > 0)
 358         {
 359             subst = false;
 360             std::vector<std::string> attr;
 361             boost::split(attr, it->attr, boost::is_any_of(","));
 362             size_t i;
 363             for (i = 0; i < attr.size(); i++)
 364             {
 365                 if (attr[i].compare("#text") == 0)
 366                 {
 367                     subst = true;
 368                 }
 369             }
 370         }
 371     }
 372     std::string output;
 373     if (subst)
 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
 385 /**
 386  * Tests pattern from the vector in order and executes recipe on
 387  the first match.
 388  */
 389 const std::string yf::HttpRewrite::Rule::test_patterns(
 390         std::map<std::string, std::string> & vars,
 391         const std::string & txt) const
 392 {
 393     std::list<Replace>::const_iterator it = replace_list.begin();
 394
 395     for (; it != replace_list.end(); it++)
 396     {
 397         std::string out = it->search_replace(vars, txt);
 398         if (!out.empty()) return out;
 399     }
 400     return "";
 401 }
 402
 403 const std::string yf::HttpRewrite::Replace::search_replace(
 404         std::map<std::string, std::string> & vars,
 405         const std::string & txt) const
 406 {
 407     //exec regex against value
 408     boost::regex re(regex);
 409     boost::smatch what;
 410     std::string::const_iterator start, end;
 411     start = txt.begin();
 412     end = txt.end();
 413     std::string out;
 414     while (regex_search(start, end, what, re)) //find next full match
 415     {
 416         size_t i;
 417         for (i = 1; i < what.size(); ++i)
 418         {
 419             //check if the group is named
 420             std::map<int, std::string>::const_iterator it
 421                 = group_index.find(i);
 422             if (it != group_index.end())
 423             {   //it is
 424                 if (!what[i].str().empty())
 425                     vars[it->second] = what[i];
 426             }
 427
 428         }
 429         //prepare replacement string
 430         std::string rvalue = sub_vars(vars);
 431         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 432                 what.str(0).c_str(), rvalue.c_str());
 433         out.append(start, what[0].first);
 434         out.append(rvalue);
 435         start = what[0].second; //move search forward
 436     }
 437     //if we had a match cat the last part
 438     if (start != txt.begin())
 439         out.append(start, end);
 440     return out;
 441 }
 442
 443 void yf::HttpRewrite::Replace::parse_groups()
 444 {
 445     int gnum = 0;
 446     bool esc = false;
 447     const std::string & str = regex;
 448     std::string res;
 449     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 450     for (size_t i = 0; i < str.size(); ++i)
 451     {
 452         res += str[i];
 453         if (!esc && str[i] == '\\')
 454         {
 455             esc = true;
 456             continue;
 457         }
 458         if (!esc && str[i] == '(') //group starts
 459         {
 460             gnum++;
 461             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 462             {
 463                 i++;
 464                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 465                 {
 466                     if (gnum > 0) gnum--;
 467                     res += str[i];
 468                     i++;
 469                     res += str[i];
 470                     continue;
 471                 }
 472                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 473                     i++;
 474                 if (i+1 < str.size() && str[i+1] == '<') //named
 475                 {
 476                     i++;
 477                     std::string gname;
 478                     bool term = false;
 479                     while (++i < str.size())
 480                     {
 481                         if (str[i] == '>') { term = true; break; }
 482                         if (!isalnum(str[i]))
 483                             throw mp::filter::FilterException
 484                                 ("Only alphanumeric chars allowed, found "
 485                                  " in '"
 486                                  + str
 487                                  + "' at "
 488                                  + boost::lexical_cast<std::string>(i));
 489                         gname += str[i];
 490                     }
 491                     if (!term)
 492                         throw mp::filter::FilterException
 493                             ("Unterminated group name '" + gname
 494                              + " in '" + str +"'");
 495                     group_index[gnum] = gname;
 496                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 497                             gname.c_str(), gnum);
 498                 }
 499             }
 500         }
 501         esc = false;
 502     }
 503     regex = res;
 504 }
 505
 506 std::string yf::HttpRewrite::Replace::sub_vars (
 507         const std::map<std::string, std::string> & vars) const
 508 {
 509     std::string out;
 510     bool esc = false;
 511     const std::string & in = recipe;
 512     for (size_t i = 0; i < in.size(); ++i)
 513     {
 514         if (!esc && in[i] == '\\')
 515         {
 516             esc = true;
 517             continue;
 518         }
 519         if (!esc && in[i] == '$') //var
 520         {
 521             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 522             {
 523                 ++i;
 524                 std::string name;
 525                 bool term = false;
 526                 while (++i < in.size())
 527                 {
 528                     if (in[i] == '}') { term = true; break; }
 529                     name += in[i];
 530                 }
 531                 if (!term) throw mp::filter::FilterException
 532                     ("Unterminated var ref in '"+in+"' at "
 533                      + boost::lexical_cast<std::string>(i));
 534                 std::map<std::string, std::string>::const_iterator it
 535                     = vars.find(name);
 536                 if (it != vars.end())
 537                 {
 538                     out += it->second;
 539                 }
 540             }
 541             else
 542             {
 543                 throw mp::filter::FilterException
 544                     ("Malformed or trimmed var ref in '"
 545                      +in+"' at "+boost::lexical_cast<std::string>(i));
 546             }
 547             continue;
 548         }
 549         //passthru
 550         out += in[i];
 551         esc = false;
 552     }
 553     return out;
 554 }
 555
 556
 557 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 558 {
 559     std::map<std::string, RulePtr > rules;
 560     for (ptr = ptr->children; ptr; ptr = ptr->next)
 561     {
 562         if (ptr->type != XML_ELEMENT_NODE)
 563             continue;
 564         else if (!strcmp((const char *) ptr->name, "rule"))
 565         {
 566             static const char *names[2] = { "name", 0 };
 567             std::string values[1];
 568             values[0] = "default";
 569             mp::xml::parse_attr(ptr, names, values);
 570
 571             RulePtr rule(new Rule);
 572             for (xmlNode *p = ptr->children; p; p = p->next)
 573             {
 574                 if (p->type != XML_ELEMENT_NODE)
 575                     continue;
 576                 if (!strcmp((const char *) p->name, "rewrite"))
 577                 {
 578                     Replace replace;
 579                     const struct _xmlAttr *attr;
 580                     for (attr = p->properties; attr; attr = attr->next)
 581                     {
 582                         if (!strcmp((const char *) attr->name,  "from"))
 583                             replace.regex = mp::xml::get_text(attr->children);
 584                         else if (!strcmp((const char *) attr->name,  "to"))
 585                             replace.recipe = mp::xml::get_text(attr->children);
 586                         else
 587                             throw mp::filter::FilterException
 588                                 ("Bad attribute "
 589                                  + std::string((const char *) attr->name)
 590                                  + " in rewrite section of http_rewrite");
 591                     }
 592                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 593                             replace.regex.c_str(), replace.recipe.c_str());
 594                     replace.parse_groups();
 595                     if (!replace.regex.empty())
 596                         rule->replace_list.push_back(replace);
 597                 }
 598                 else
 599                     throw mp::filter::FilterException
 600                         ("Bad element "
 601                          + std::string((const char *) p->name)
 602                          + " in http_rewrite filter");
 603             }
 604             rules[values[0]] = rule;
 605         }
 606         else if (!strcmp((const char *) ptr->name, "within"))
 607         {
 608             static const char *names[6] =
 609                 { "header", "attr", "tag", "rule", "reqline", 0 };
 610             std::string values[5];
 611             mp::xml::parse_attr(ptr, names, values);
 612             Within w;
 613             w.header = values[0];
 614             w.attr = values[1];
 615             w.tag = values[2];
 616             std::map<std::string,RulePtr>::const_iterator it =
 617                 rules.find(values[3]);
 618             if (it == rules.end())
 619                 throw mp::filter::FilterException
 620                     ("Reference to non-existing rule '" + values[3] +
 621                      "' in http_rewrite filter");
 622             w.rule = it->second;
 623             w.reqline = values[4] == "1";
 624             phase.within_list.push_back(w);
 625         }
 626         else
 627         {
 628             throw mp::filter::FilterException
 629                 ("Bad element "
 630                  + std::string((const char *) ptr->name)
 631                  + " in http_rewrite filter");
 632         }
 633     }
 634 }
 635
 636 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 637         const char *path)
 638 {
 639     for (ptr = ptr->children; ptr; ptr = ptr->next)
 640     {
 641         if (ptr->type != XML_ELEMENT_NODE)
 642             continue;
 643         else if (!strcmp((const char *) ptr->name, "request"))
 644         {
 645             configure_phase(ptr, *req_phase);
 646         }
 647         else if (!strcmp((const char *) ptr->name, "response"))
 648         {
 649             configure_phase(ptr, *res_phase);
 650         }
 651         else
 652         {
 653             throw mp::filter::FilterException
 654                 ("Bad element "
 655                  + std::string((const char *) ptr->name)
 656                  + " in http_rewrite1 filter");
 657         }
 658     }
 659 }
 660
 661 static mp::filter::Base* filter_creator()
 662 {
 663     return new mp::filter::HttpRewrite;
 664 }
 665
 666 extern "C" {
 667     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 668         0,
 669         "http_rewrite",
 670         filter_creator
 671     };
 672 }
 673
 674
 675 /*
 676  * Local variables:
 677  * c-basic-offset: 4
 678  * c-file-style: "Stroustrup"
 679  * indent-tabs-mode: nil
 680  * End:
 681  * vim: shiftwidth=4 tabstop=8 expandtab
 682  */
 683