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             bool start_anchor;
  44             boost::regex re;
  45             boost::smatch what;
  46             std::string recipe;
  47             std::map<int, std::string> group_index;
  48             std::string sub_vars(
  49                 const std::map<std::string, std::string> & vars) const;
  50             void parse_groups(std::string pattern);
  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, bool anchor);
  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             Phase();
  72             std::list<Within> within_list;
  73             int m_verbose;
  74             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  75                 std::map<std::string, std::string> & vars) const;
  76             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  77                 std::map<std::string, std::string> & vars) const;
  78             void rewrite_body(mp::odr & o,
  79                 char **content_buf, int *content_len,
  80                 std::map<std::string, std::string> & vars) const;
  81         };
  82         class HttpRewrite::Event : public HTMLParserEvent {
  83             void openTagStart(const char *tag, int tag_len);
  84             void anyTagEnd(const char *tag, int tag_len, int close_it);
  85             void attribute(const char *tag, int tag_len,
  86                            const char *attr, int attr_len,
  87                            const char *value, int val_len,
  88                            const char *sep);
  89             void closeTag(const char *tag, int tag_len);
  90             void text(const char *value, int len);
  91             const Phase *m_phase;
  92             WRBUF m_w;
  93             std::stack<std::list<Within>::const_iterator> s_within;
  94             std::map<std::string, std::string> &m_vars;
  95         public:
  96             Event(const Phase *p, std::map<std::string, std::string> &vars);
  97             ~Event();
  98             const char *result();
  99         };
 100     }
 101 }
 102
 103 yf::HttpRewrite::HttpRewrite() :
 104     req_phase(new Phase), res_phase(new Phase)
 105 {
 106 }
 107
 108 yf::HttpRewrite::~HttpRewrite()
 109 {
 110 }
 111
 112 void yf::HttpRewrite::process(mp::Package & package) const
 113 {
 114     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 115     Z_GDU *gdu = package.request().get();
 116     //map of request/response vars
 117     std::map<std::string, std::string> vars;
 118     //we have an http req
 119     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 120     {
 121         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 122         mp::odr o;
 123         req_phase->rewrite_reqline(o, hreq, vars);
 124         yaz_log(YLOG_LOG, ">> Request headers");
 125         req_phase->rewrite_headers(o, hreq->headers, vars);
 126         req_phase->rewrite_body(o,
 127                 &hreq->content_buf, &hreq->content_len, vars);
 128         package.request() = gdu;
 129     }
 130     package.move();
 131     gdu = package.response().get();
 132     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 133     {
 134         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 135         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 136         mp::odr o;
 137         yaz_log(YLOG_LOG, "<< Respose headers");
 138         res_phase->rewrite_headers(o, hres->headers, vars);
 139         res_phase->rewrite_body(o, &hres->content_buf,
 140                 &hres->content_len, vars);
 141         package.response() = gdu;
 142     }
 143 }
 144
 145 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 146         Z_HTTP_Request *hreq,
 147         std::map<std::string, std::string> & vars) const
 148 {
 149     //rewrite the request line
 150     std::string path;
 151     if (strstr(hreq->path, "http://") == hreq->path)
 152     {
 153         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 154             "possibly a proxy request");
 155         path += hreq->path;
 156     }
 157     else
 158     {
 159         //TODO what about proto
 160         const char *host = z_HTTP_header_lookup(hreq->headers, "Host");
 161         if (!host)
 162             return;
 163
 164         path += "http://";
 165         path += host;
 166         path += hreq->path;
 167     }
 168
 169
 170     std::list<Within>::const_iterator it = within_list.begin();
 171     for (; it != within_list.end(); it++)
 172         if (it->reqline)
 173         {
 174             RulePtr rule = it->rule;
 175             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 176             std::string npath = rule->test_patterns(vars, path, true);
 177             if (!npath.empty())
 178             {
 179                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 180                 hreq->path = odr_strdup(o, npath.c_str());
 181             }
 182         }
 183 }
 184
 185 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 186         Z_HTTP_Header *headers,
 187         std::map<std::string, std::string> & vars) const
 188 {
 189     for (Z_HTTP_Header *header = headers; header; header = header->next)
 190     {
 191         std::list<Within>::const_iterator it = within_list.begin();
 192         for (; it != within_list.end(); it++)
 193         {
 194             if (it->header.length() > 0 &&
 195                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 196             {
 197                 std::string sheader(header->name);
 198                 sheader += ": ";
 199                 sheader += header->value;
 200
 201                 RulePtr rule = it->rule;
 202                 std::string out = rule->test_patterns(vars, sheader, true);
 203                 if (!out.empty())
 204                 {
 205                     size_t pos = out.find(": ");
 206                     if (pos == std::string::npos)
 207                     {
 208                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 209                         continue;
 210                     }
 211                     header->name = odr_strdup(o, out.substr(0, pos).c_str());
 212                     header->value = odr_strdup(o,
 213                                                out.substr(pos + 2,
 214                                                           std::string::npos).c_str());
 215                 }
 216             }
 217         }
 218     }
 219 }
 220
 221 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 222         char **content_buf,
 223         int *content_len,
 224         std::map<std::string, std::string> & vars) const
 225 {
 226     if (*content_buf)
 227     {
 228         int i;
 229         for (i = 0; i < *content_len; i++)
 230             if ((*content_buf)[i] == 0)
 231                 return;  // binary content. skip
 232
 233         HTMLParser parser;
 234         Event ev(this, vars);
 235
 236         parser.set_verbose(m_verbose);
 237
 238         std::string buf(*content_buf, *content_len);
 239
 240         parser.parse(ev, buf.c_str());
 241         const char *res = ev.result();
 242         *content_buf = odr_strdup(o, res);
 243         *content_len = strlen(res);
 244     }
 245 }
 246
 247 yf::HttpRewrite::Event::Event(const Phase *p,
 248                               std::map<std::string, std::string> & vars
 249     ) : m_phase(p), m_vars(vars)
 250 {
 251     m_w = wrbuf_alloc();
 252 }
 253
 254 yf::HttpRewrite::Event::~Event()
 255 {
 256     wrbuf_destroy(m_w);
 257 }
 258
 259 const char *yf::HttpRewrite::Event::result()
 260 {
 261     return wrbuf_cstr(m_w);
 262 }
 263
 264 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 265 {
 266     wrbuf_putc(m_w, '<');
 267     wrbuf_write(m_w, tag, tag_len);
 268
 269     std::string t(tag, tag_len);
 270     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 271     for (; it != m_phase->within_list.end(); it++)
 272     {
 273         if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
 274                                                    t.c_str()) == 0)
 275         {
 276             std::vector<std::string> attr;
 277             boost::split(attr, it->attr, boost::is_any_of(","));
 278             size_t i;
 279             for (i = 0; i < attr.size(); i++)
 280             {
 281                 if (attr[i].compare("#text") == 0)
 282                 {
 283                     s_within.push(it);
 284                     return;
 285                 }
 286             }
 287         }
 288     }
 289 }
 290
 291 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 292                                        int close_it)
 293 {
 294     if (close_it)
 295     {
 296         if (!s_within.empty())
 297         {
 298             std::list<Within>::const_iterator it = s_within.top();
 299             std::string t(tag, tag_len);
 300             if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 301                 s_within.pop();
 302         }
 303     }
 304     if (close_it)
 305         wrbuf_putc(m_w, '/');
 306     wrbuf_putc(m_w, '>');
 307 }
 308
 309 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 310                                        const char *attr, int attr_len,
 311                                        const char *value, int val_len,
 312                                        const char *sep)
 313 {
 314     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 315     bool subst = false;
 316
 317     for (; it != m_phase->within_list.end(); it++)
 318     {
 319         std::string t(tag, tag_len);
 320         if (it->tag.length() == 0 ||
 321             yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 322         {
 323             std::string a(attr, attr_len);
 324             std::vector<std::string> attr;
 325             boost::split(attr, it->attr, boost::is_any_of(","));
 326             size_t i;
 327             for (i = 0; i < attr.size(); i++)
 328             {
 329                 if (attr[i].compare("#text") &&
 330                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 331                     subst = true;
 332             }
 333         }
 334         if (subst)
 335             break;
 336     }
 337
 338     wrbuf_putc(m_w, ' ');
 339     wrbuf_write(m_w, attr, attr_len);
 340     if (value)
 341     {
 342         wrbuf_puts(m_w, "=");
 343         wrbuf_puts(m_w, sep);
 344
 345         std::string output;
 346         if (subst)
 347         {
 348             std::string input(value, val_len);
 349             output = it->rule->test_patterns(m_vars, input, true);
 350         }
 351         if (output.empty())
 352             wrbuf_write(m_w, value, val_len);
 353         else
 354             wrbuf_puts(m_w, output.c_str());
 355         wrbuf_puts(m_w, sep);
 356     }
 357 }
 358
 359 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 360 {
 361     if (!s_within.empty())
 362     {
 363         std::list<Within>::const_iterator it = s_within.top();
 364         std::string t(tag, tag_len);
 365         if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 366             s_within.pop();
 367     }
 368     wrbuf_puts(m_w, "</");
 369     wrbuf_write(m_w, tag, tag_len);
 370 }
 371
 372 void yf::HttpRewrite::Event::text(const char *value, int len)
 373 {
 374     std::list<Within>::const_iterator it = m_phase->within_list.end();
 375     if (!s_within.empty())
 376         it = s_within.top();
 377     std::string output;
 378     if (it != m_phase->within_list.end())
 379     {
 380         std::string input(value, len);
 381         output = it->rule->test_patterns(m_vars, input, false);
 382     }
 383     if (output.empty())
 384         wrbuf_write(m_w, value, len);
 385     else
 386         wrbuf_puts(m_w, output.c_str());
 387 }
 388
 389 const std::string yf::HttpRewrite::Rule::test_patterns(
 390         std::map<std::string, std::string> & vars,
 391         const std::string & txt, bool anchor)
 392 {
 393     bool first = anchor;
 394     std::string out;
 395     std::string::const_iterator start, end;
 396     start = txt.begin();
 397     end = txt.end();
 398     while (1)
 399     {
 400         std::list<Replace>::iterator bit = replace_list.end();
 401         {
 402             std::string::const_iterator best_pos = txt.end();
 403             std::list<Replace>::iterator it = replace_list.begin();
 404             for (; it != replace_list.end(); it++)
 405             {
 406                 if (it->start_anchor && !first)
 407                     continue;
 408                 if (regex_search(start, end, it->what, it->re))
 409                 {
 410                     if (it->what[0].first < best_pos)
 411                     {
 412                         best_pos = it->what[0].first;
 413                         bit = it;
 414                     }
 415                 }
 416             }
 417             if (bit == replace_list.end())
 418                 break;
 419         }
 420         first = false;
 421         size_t i;
 422         for (i = 1; i < bit->what.size(); ++i)
 423         {
 424             //check if the group is named
 425             std::map<int, std::string>::const_iterator git
 426                 = bit->group_index.find(i);
 427             if (git != bit->group_index.end())
 428             {   //it is
 429                 vars[git->second] = bit->what[i];
 430             }
 431
 432         }
 433         //prepare replacement string
 434         std::string rvalue = bit->sub_vars(vars);
 435         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 436                 bit->what.str(0).c_str(), rvalue.c_str());
 437         out.append(start, bit->what[0].first);
 438         out.append(rvalue);
 439         start = bit->what[0].second; //move search forward
 440     }
 441     if (start != txt.begin())
 442         out.append(start, end);
 443     return out;
 444 }
 445
 446 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 447 {
 448     int gnum = 0;
 449     bool esc = false;
 450     const std::string &str = pattern;
 451     std::string res;
 452     start_anchor = str[0] == '^';
 453     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 454     for (size_t i = 0; i < str.size(); ++i)
 455     {
 456         res += str[i];
 457         if (!esc && str[i] == '\\')
 458         {
 459             esc = true;
 460             continue;
 461         }
 462         if (!esc && str[i] == '(') //group starts
 463         {
 464             gnum++;
 465             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 466             {
 467                 i++;
 468                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 469                 {
 470                     if (gnum > 0) gnum--;
 471                     res += str[i];
 472                     i++;
 473                     res += str[i];
 474                     continue;
 475                 }
 476                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 477                     i++;
 478                 if (i+1 < str.size() && str[i+1] == '<') //named
 479                 {
 480                     i++;
 481                     std::string gname;
 482                     bool term = false;
 483                     while (++i < str.size())
 484                     {
 485                         if (str[i] == '>') { term = true; break; }
 486                         if (!isalnum(str[i]))
 487                             throw mp::filter::FilterException
 488                                 ("Only alphanumeric chars allowed, found "
 489                                  " in '"
 490                                  + str
 491                                  + "' at "
 492                                  + boost::lexical_cast<std::string>(i));
 493                         gname += str[i];
 494                     }
 495                     if (!term)
 496                         throw mp::filter::FilterException
 497                             ("Unterminated group name '" + gname
 498                              + " in '" + str +"'");
 499                     group_index[gnum] = gname;
 500                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 501                             gname.c_str(), gnum);
 502                 }
 503             }
 504         }
 505         esc = false;
 506     }
 507     re = res;
 508 }
 509
 510 std::string yf::HttpRewrite::Replace::sub_vars(
 511     const std::map<std::string, std::string> & vars) const
 512 {
 513     std::string out;
 514     bool esc = false;
 515     const std::string & in = recipe;
 516     for (size_t i = 0; i < in.size(); ++i)
 517     {
 518         if (!esc && in[i] == '\\')
 519         {
 520             esc = true;
 521             continue;
 522         }
 523         if (!esc && in[i] == '$') //var
 524         {
 525             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 526             {
 527                 ++i;
 528                 std::string name;
 529                 bool term = false;
 530                 while (++i < in.size())
 531                 {
 532                     if (in[i] == '}') { term = true; break; }
 533                     name += in[i];
 534                 }
 535                 if (!term) throw mp::filter::FilterException
 536                     ("Unterminated var ref in '"+in+"' at "
 537                      + boost::lexical_cast<std::string>(i));
 538                 std::map<std::string, std::string>::const_iterator it
 539                     = vars.find(name);
 540                 if (it != vars.end())
 541                 {
 542                     out += it->second;
 543                 }
 544             }
 545             else
 546             {
 547                 throw mp::filter::FilterException
 548                     ("Malformed or trimmed var ref in '"
 549                      +in+"' at "+boost::lexical_cast<std::string>(i));
 550             }
 551             continue;
 552         }
 553         //passthru
 554         out += in[i];
 555         esc = false;
 556     }
 557     return out;
 558 }
 559
 560 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 561 {
 562 }
 563
 564 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 565 {
 566     static const char *names[2] = { "verbose", 0 };
 567     std::string values[1];
 568     values[0] = "0";
 569     mp::xml::parse_attr(ptr, names, values);
 570
 571     phase.m_verbose = atoi(values[0].c_str());
 572
 573     std::map<std::string, RulePtr > rules;
 574     for (ptr = ptr->children; ptr; ptr = ptr->next)
 575     {
 576         if (ptr->type != XML_ELEMENT_NODE)
 577             continue;
 578         else if (!strcmp((const char *) ptr->name, "rule"))
 579         {
 580             static const char *names[2] = { "name", 0 };
 581             std::string values[1];
 582             values[0] = "default";
 583             mp::xml::parse_attr(ptr, names, values);
 584
 585             RulePtr rule(new Rule);
 586             for (xmlNode *p = ptr->children; p; p = p->next)
 587             {
 588                 if (p->type != XML_ELEMENT_NODE)
 589                     continue;
 590                 if (!strcmp((const char *) p->name, "rewrite"))
 591                 {
 592                     Replace replace;
 593                     std::string from;
 594                     const struct _xmlAttr *attr;
 595                     for (attr = p->properties; attr; attr = attr->next)
 596                     {
 597                         if (!strcmp((const char *) attr->name,  "from"))
 598                             from = mp::xml::get_text(attr->children);
 599                         else if (!strcmp((const char *) attr->name,  "to"))
 600                             replace.recipe = mp::xml::get_text(attr->children);
 601                         else
 602                             throw mp::filter::FilterException
 603                                 ("Bad attribute "
 604                                  + std::string((const char *) attr->name)
 605                                  + " in rewrite section of http_rewrite");
 606                     }
 607                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 608                             from.c_str(), replace.recipe.c_str());
 609                     if (!from.empty())
 610                     {
 611                         replace.parse_groups(from);
 612                         rule->replace_list.push_back(replace);
 613                     }
 614                 }
 615                 else
 616                     throw mp::filter::FilterException
 617                         ("Bad element "
 618                          + std::string((const char *) p->name)
 619                          + " in http_rewrite filter");
 620             }
 621             rules[values[0]] = rule;
 622         }
 623         else if (!strcmp((const char *) ptr->name, "within"))
 624         {
 625             static const char *names[6] =
 626                 { "header", "attr", "tag", "rule", "reqline", 0 };
 627             std::string values[5];
 628             mp::xml::parse_attr(ptr, names, values);
 629             Within w;
 630             w.header = values[0];
 631             w.attr = values[1];
 632             w.tag = values[2];
 633             std::map<std::string,RulePtr>::const_iterator it =
 634                 rules.find(values[3]);
 635             if (it == rules.end())
 636                 throw mp::filter::FilterException
 637                     ("Reference to non-existing rule '" + values[3] +
 638                      "' in http_rewrite filter");
 639             w.rule = it->second;
 640             w.reqline = values[4] == "1";
 641             phase.within_list.push_back(w);
 642         }
 643         else
 644         {
 645             throw mp::filter::FilterException
 646                 ("Bad element "
 647                  + std::string((const char *) ptr->name)
 648                  + " in http_rewrite filter");
 649         }
 650     }
 651 }
 652
 653 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 654         const char *path)
 655 {
 656     for (ptr = ptr->children; ptr; ptr = ptr->next)
 657     {
 658         if (ptr->type != XML_ELEMENT_NODE)
 659             continue;
 660         else if (!strcmp((const char *) ptr->name, "request"))
 661         {
 662             configure_phase(ptr, *req_phase);
 663         }
 664         else if (!strcmp((const char *) ptr->name, "response"))
 665         {
 666             configure_phase(ptr, *res_phase);
 667         }
 668         else
 669         {
 670             throw mp::filter::FilterException
 671                 ("Bad element "
 672                  + std::string((const char *) ptr->name)
 673                  + " in http_rewrite1 filter");
 674         }
 675     }
 676 }
 677
 678 static mp::filter::Base* filter_creator()
 679 {
 680     return new mp::filter::HttpRewrite;
 681 }
 682
 683 extern "C" {
 684     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 685         0,
 686         "http_rewrite",
 687         filter_creator
 688     };
 689 }
 690
 691
 692 /*
 693  * Local variables:
 694  * c-basic-offset: 4
 695  * c-file-style: "Stroustrup"
 696  * indent-tabs-mode: nil
 697  * End:
 698  * vim: shiftwidth=4 tabstop=8 expandtab
 699  */
 700