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             std::string regex;
  44             std::string recipe;
  45             std::map<int, std::string> group_index;
  46             const std::string search_replace(
  47                 std::map<std::string, std::string> & vars,
  48                 const std::string & txt) const;
  49             std::string sub_vars(
  50                 const std::map<std::string, std::string> & vars) const;
  51             void parse_groups();
  52         };
  53
  54         class HttpRewrite::Rule {
  55         public:
  56             std::list<Replace> replace_list;
  57             const std::string test_patterns(
  58                 std::map<std::string, std::string> & vars,
  59                 const std::string & txt) const;
  60         };
  61         class HttpRewrite::Within {
  62         public:
  63             std::string header;
  64             std::string attr;
  65             std::string tag;
  66             bool reqline;
  67             RulePtr rule;
  68         };
  69
  70         class HttpRewrite::Phase {
  71         public:
  72             Phase();
  73             std::list<Within> within_list;
  74             int m_verbose;
  75             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  76                 std::map<std::string, std::string> & vars) const;
  77             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  78                 std::map<std::string, std::string> & vars) const;
  79             void rewrite_body(mp::odr & o,
  80                 char **content_buf, int *content_len,
  81                 std::map<std::string, std::string> & vars) const;
  82         };
  83         class HttpRewrite::Event : public HTMLParserEvent {
  84             void openTagStart(const char *tag, int tag_len);
  85             void anyTagEnd(const char *tag, int tag_len, int close_it);
  86             void attribute(const char *tag, int tag_len,
  87                            const char *attr, int attr_len,
  88                            const char *value, int val_len,
  89                            const char *sep);
  90             void closeTag(const char *tag, int tag_len);
  91             void text(const char *value, int len);
  92             const Phase *m_phase;
  93             WRBUF m_w;
  94             std::stack<std::list<Within>::const_iterator> s_within;
  95             std::map<std::string, std::string> &m_vars;
  96         public:
  97             Event(const Phase *p, std::map<std::string, std::string> &vars);
  98             ~Event();
  99             const char *result();
 100         };
 101     }
 102 }
 103
 104 yf::HttpRewrite::HttpRewrite() :
 105     req_phase(new Phase), res_phase(new Phase)
 106 {
 107 }
 108
 109 yf::HttpRewrite::~HttpRewrite()
 110 {
 111 }
 112
 113 void yf::HttpRewrite::process(mp::Package & package) const
 114 {
 115     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 116     Z_GDU *gdu = package.request().get();
 117     //map of request/response vars
 118     std::map<std::string, std::string> vars;
 119     //we have an http req
 120     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 121     {
 122         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 123         mp::odr o;
 124         req_phase->rewrite_reqline(o, hreq, vars);
 125         yaz_log(YLOG_LOG, ">> Request headers");
 126         req_phase->rewrite_headers(o, hreq->headers, vars);
 127         req_phase->rewrite_body(o,
 128                 &hreq->content_buf, &hreq->content_len, vars);
 129         package.request() = gdu;
 130     }
 131     package.move();
 132     gdu = package.response().get();
 133     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 134     {
 135         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 136         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 137         mp::odr o;
 138         yaz_log(YLOG_LOG, "<< Respose headers");
 139         res_phase->rewrite_headers(o, hres->headers, vars);
 140         res_phase->rewrite_body(o, &hres->content_buf,
 141                 &hres->content_len, vars);
 142         package.response() = gdu;
 143     }
 144 }
 145
 146 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 147         Z_HTTP_Request *hreq,
 148         std::map<std::string, std::string> & vars) const
 149 {
 150     //rewrite the request line
 151     std::string path;
 152     if (strstr(hreq->path, "http://") == hreq->path)
 153     {
 154         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 155             "possibly a proxy request");
 156         path += hreq->path;
 157     }
 158     else
 159     {
 160         //TODO what about proto
 161         path += "http://";
 162         path += z_HTTP_header_lookup(hreq->headers, "Host");
 163         path += hreq->path;
 164     }
 165
 166
 167     std::list<Within>::const_iterator it = within_list.begin();
 168     for (; it != within_list.end(); it++)
 169         if (it->reqline)
 170         {
 171             RulePtr rule = it->rule;
 172             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 173             std::string npath = rule->test_patterns(vars, path);
 174             if (!npath.empty())
 175             {
 176                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 177                 hreq->path = odr_strdup(o, npath.c_str());
 178             }
 179         }
 180 }
 181
 182 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 183         Z_HTTP_Header *headers,
 184         std::map<std::string, std::string> & vars) const
 185 {
 186     for (Z_HTTP_Header *header = headers; header; header = header->next)
 187     {
 188         std::list<Within>::const_iterator it = within_list.begin();
 189         for (; it != within_list.end(); it++)
 190         {
 191             if (it->header.length() > 0 &&
 192                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 193             {
 194                 std::string sheader(header->name);
 195                 sheader += ": ";
 196                 sheader += header->value;
 197
 198                 RulePtr rule = it->rule;
 199                 std::string out = rule->test_patterns(vars, sheader);
 200                 if (!out.empty())
 201                 {
 202                     size_t pos = out.find(": ");
 203                     if (pos == std::string::npos)
 204                     {
 205                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 206                         continue;
 207                     }
 208                     header->name = odr_strdup(o, out.substr(0, pos).c_str());
 209                     header->value = odr_strdup(o,
 210                                                out.substr(pos + 2,
 211                                                           std::string::npos).c_str());
 212                 }
 213             }
 214         }
 215     }
 216 }
 217
 218 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 219         char **content_buf,
 220         int *content_len,
 221         std::map<std::string, std::string> & vars) const
 222 {
 223     if (*content_buf)
 224     {
 225         int i;
 226         for (i = 0; i < *content_len; i++)
 227             if ((*content_buf)[i] == 0)
 228                 return;  // binary content. skip
 229
 230         HTMLParser parser;
 231         Event ev(this, vars);
 232
 233         parser.set_verbose(m_verbose);
 234
 235         std::string buf(*content_buf, *content_len);
 236
 237         parser.parse(ev, buf.c_str());
 238         const char *res = ev.result();
 239         *content_buf = odr_strdup(o, res);
 240         *content_len = strlen(res);
 241     }
 242 }
 243
 244 yf::HttpRewrite::Event::Event(const Phase *p,
 245                               std::map<std::string, std::string> & vars
 246     ) : m_phase(p), m_vars(vars)
 247 {
 248     m_w = wrbuf_alloc();
 249 }
 250
 251 yf::HttpRewrite::Event::~Event()
 252 {
 253     wrbuf_destroy(m_w);
 254 }
 255
 256 const char *yf::HttpRewrite::Event::result()
 257 {
 258     return wrbuf_cstr(m_w);
 259 }
 260
 261 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 262 {
 263     wrbuf_putc(m_w, '<');
 264     wrbuf_write(m_w, tag, tag_len);
 265
 266     std::string t(tag, tag_len);
 267     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 268     for (; it != m_phase->within_list.end(); it++)
 269     {
 270         if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
 271                                                    t.c_str()) == 0)
 272         {
 273             std::vector<std::string> attr;
 274             boost::split(attr, it->attr, boost::is_any_of(","));
 275             size_t i;
 276             for (i = 0; i < attr.size(); i++)
 277             {
 278                 if (attr[i].compare("#text") == 0)
 279                 {
 280                     s_within.push(it);
 281                     return;
 282                 }
 283             }
 284         }
 285     }
 286 }
 287
 288 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 289                                        int close_it)
 290 {
 291     if (close_it)
 292     {
 293         if (!s_within.empty())
 294         {
 295             std::list<Within>::const_iterator it = s_within.top();
 296             std::string t(tag, tag_len);
 297             if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 298                 s_within.pop();
 299         }
 300     }
 301     if (close_it)
 302         wrbuf_putc(m_w, '/');
 303     wrbuf_putc(m_w, '>');
 304 }
 305
 306 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 307                                        const char *attr, int attr_len,
 308                                        const char *value, int val_len,
 309                                        const char *sep)
 310 {
 311     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 312     bool subst = false;
 313
 314     for (; it != m_phase->within_list.end(); it++)
 315     {
 316         std::string t(tag, tag_len);
 317         if (it->tag.length() == 0 ||
 318             yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 319         {
 320             std::string a(attr, attr_len);
 321             std::vector<std::string> attr;
 322             boost::split(attr, it->attr, boost::is_any_of(","));
 323             size_t i;
 324             for (i = 0; i < attr.size(); i++)
 325             {
 326                 if (attr[i].compare("#text") &&
 327                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 328                     subst = true;
 329             }
 330         }
 331         if (subst)
 332             break;
 333     }
 334
 335     wrbuf_putc(m_w, ' ');
 336     wrbuf_write(m_w, attr, attr_len);
 337     if (value)
 338     {
 339         wrbuf_puts(m_w, "=");
 340         wrbuf_puts(m_w, sep);
 341
 342         std::string output;
 343         if (subst)
 344         {
 345             std::string input(value, val_len);
 346             output = it->rule->test_patterns(m_vars, input);
 347         }
 348         if (output.empty())
 349             wrbuf_write(m_w, value, val_len);
 350         else
 351             wrbuf_puts(m_w, output.c_str());
 352         wrbuf_puts(m_w, sep);
 353     }
 354 }
 355
 356 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 357 {
 358     if (!s_within.empty())
 359     {
 360         std::list<Within>::const_iterator it = s_within.top();
 361         std::string t(tag, tag_len);
 362         if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 363             s_within.pop();
 364     }
 365     wrbuf_puts(m_w, "</");
 366     wrbuf_write(m_w, tag, tag_len);
 367 }
 368
 369 void yf::HttpRewrite::Event::text(const char *value, int len)
 370 {
 371     std::list<Within>::const_iterator it = m_phase->within_list.end();
 372     if (!s_within.empty())
 373         it = s_within.top();
 374     std::string output;
 375     if (it != m_phase->within_list.end())
 376     {
 377         std::string input(value, len);
 378         output = it->rule->test_patterns(m_vars, input);
 379     }
 380     if (output.empty())
 381         wrbuf_write(m_w, value, len);
 382     else
 383         wrbuf_puts(m_w, output.c_str());
 384 }
 385
 386
 387 /**
 388  * Tests pattern from the vector in order and executes recipe on
 389  the first match.
 390  */
 391 const std::string yf::HttpRewrite::Rule::test_patterns(
 392         std::map<std::string, std::string> & vars,
 393         const std::string & txt) const
 394 {
 395     std::list<Replace>::const_iterator it = replace_list.begin();
 396
 397     for (; it != replace_list.end(); it++)
 398     {
 399         std::string out = it->search_replace(vars, txt);
 400         if (!out.empty()) return out;
 401     }
 402     return "";
 403 }
 404
 405 const std::string yf::HttpRewrite::Replace::search_replace(
 406         std::map<std::string, std::string> & vars,
 407         const std::string & txt) const
 408 {
 409     //exec regex against value
 410     boost::regex re(regex);
 411     boost::smatch what;
 412     std::string::const_iterator start, end;
 413     start = txt.begin();
 414     end = txt.end();
 415     std::string out;
 416     while (regex_search(start, end, what, re)) //find next full match
 417     {
 418         size_t i;
 419         for (i = 1; i < what.size(); ++i)
 420         {
 421             //check if the group is named
 422             std::map<int, std::string>::const_iterator it
 423                 = group_index.find(i);
 424             if (it != group_index.end())
 425             {   //it is
 426                 vars[it->second] = what[i];
 427             }
 428
 429         }
 430         //prepare replacement string
 431         std::string rvalue = sub_vars(vars);
 432         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 433                 what.str(0).c_str(), rvalue.c_str());
 434         out.append(start, what[0].first);
 435         out.append(rvalue);
 436         start = what[0].second; //move search forward
 437     }
 438     //if we had a match cat the last part
 439     if (start != txt.begin())
 440         out.append(start, end);
 441     return out;
 442 }
 443
 444 void yf::HttpRewrite::Replace::parse_groups()
 445 {
 446     int gnum = 0;
 447     bool esc = false;
 448     const std::string & str = regex;
 449     std::string res;
 450     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 451     for (size_t i = 0; i < str.size(); ++i)
 452     {
 453         res += str[i];
 454         if (!esc && str[i] == '\\')
 455         {
 456             esc = true;
 457             continue;
 458         }
 459         if (!esc && str[i] == '(') //group starts
 460         {
 461             gnum++;
 462             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 463             {
 464                 i++;
 465                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 466                 {
 467                     if (gnum > 0) gnum--;
 468                     res += str[i];
 469                     i++;
 470                     res += str[i];
 471                     continue;
 472                 }
 473                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 474                     i++;
 475                 if (i+1 < str.size() && str[i+1] == '<') //named
 476                 {
 477                     i++;
 478                     std::string gname;
 479                     bool term = false;
 480                     while (++i < str.size())
 481                     {
 482                         if (str[i] == '>') { term = true; break; }
 483                         if (!isalnum(str[i]))
 484                             throw mp::filter::FilterException
 485                                 ("Only alphanumeric chars allowed, found "
 486                                  " in '"
 487                                  + str
 488                                  + "' at "
 489                                  + boost::lexical_cast<std::string>(i));
 490                         gname += str[i];
 491                     }
 492                     if (!term)
 493                         throw mp::filter::FilterException
 494                             ("Unterminated group name '" + gname
 495                              + " in '" + str +"'");
 496                     group_index[gnum] = gname;
 497                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 498                             gname.c_str(), gnum);
 499                 }
 500             }
 501         }
 502         esc = false;
 503     }
 504     regex = res;
 505 }
 506
 507 std::string yf::HttpRewrite::Replace::sub_vars(
 508     const std::map<std::string, std::string> & vars) const
 509 {
 510     std::string out;
 511     bool esc = false;
 512     const std::string & in = recipe;
 513     for (size_t i = 0; i < in.size(); ++i)
 514     {
 515         if (!esc && in[i] == '\\')
 516         {
 517             esc = true;
 518             continue;
 519         }
 520         if (!esc && in[i] == '$') //var
 521         {
 522             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 523             {
 524                 ++i;
 525                 std::string name;
 526                 bool term = false;
 527                 while (++i < in.size())
 528                 {
 529                     if (in[i] == '}') { term = true; break; }
 530                     name += in[i];
 531                 }
 532                 if (!term) throw mp::filter::FilterException
 533                     ("Unterminated var ref in '"+in+"' at "
 534                      + boost::lexical_cast<std::string>(i));
 535                 std::map<std::string, std::string>::const_iterator it
 536                     = vars.find(name);
 537                 if (it != vars.end())
 538                 {
 539                     out += it->second;
 540                 }
 541             }
 542             else
 543             {
 544                 throw mp::filter::FilterException
 545                     ("Malformed or trimmed var ref in '"
 546                      +in+"' at "+boost::lexical_cast<std::string>(i));
 547             }
 548             continue;
 549         }
 550         //passthru
 551         out += in[i];
 552         esc = false;
 553     }
 554     return out;
 555 }
 556
 557 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 558 {
 559 }
 560
 561 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 562 {
 563     static const char *names[2] = { "verbose", 0 };
 564     std::string values[1];
 565     values[0] = "0";
 566     mp::xml::parse_attr(ptr, names, values);
 567
 568     phase.m_verbose = atoi(values[0].c_str());
 569
 570     std::map<std::string, RulePtr > rules;
 571     for (ptr = ptr->children; ptr; ptr = ptr->next)
 572     {
 573         if (ptr->type != XML_ELEMENT_NODE)
 574             continue;
 575         else if (!strcmp((const char *) ptr->name, "rule"))
 576         {
 577             static const char *names[2] = { "name", 0 };
 578             std::string values[1];
 579             values[0] = "default";
 580             mp::xml::parse_attr(ptr, names, values);
 581
 582             RulePtr rule(new Rule);
 583             for (xmlNode *p = ptr->children; p; p = p->next)
 584             {
 585                 if (p->type != XML_ELEMENT_NODE)
 586                     continue;
 587                 if (!strcmp((const char *) p->name, "rewrite"))
 588                 {
 589                     Replace replace;
 590                     const struct _xmlAttr *attr;
 591                     for (attr = p->properties; attr; attr = attr->next)
 592                     {
 593                         if (!strcmp((const char *) attr->name,  "from"))
 594                             replace.regex = mp::xml::get_text(attr->children);
 595                         else if (!strcmp((const char *) attr->name,  "to"))
 596                             replace.recipe = mp::xml::get_text(attr->children);
 597                         else
 598                             throw mp::filter::FilterException
 599                                 ("Bad attribute "
 600                                  + std::string((const char *) attr->name)
 601                                  + " in rewrite section of http_rewrite");
 602                     }
 603                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 604                             replace.regex.c_str(), replace.recipe.c_str());
 605                     replace.parse_groups();
 606                     if (!replace.regex.empty())
 607                         rule->replace_list.push_back(replace);
 608                 }
 609                 else
 610                     throw mp::filter::FilterException
 611                         ("Bad element "
 612                          + std::string((const char *) p->name)
 613                          + " in http_rewrite filter");
 614             }
 615             rules[values[0]] = rule;
 616         }
 617         else if (!strcmp((const char *) ptr->name, "within"))
 618         {
 619             static const char *names[6] =
 620                 { "header", "attr", "tag", "rule", "reqline", 0 };
 621             std::string values[5];
 622             mp::xml::parse_attr(ptr, names, values);
 623             Within w;
 624             w.header = values[0];
 625             w.attr = values[1];
 626             w.tag = values[2];
 627             std::map<std::string,RulePtr>::const_iterator it =
 628                 rules.find(values[3]);
 629             if (it == rules.end())
 630                 throw mp::filter::FilterException
 631                     ("Reference to non-existing rule '" + values[3] +
 632                      "' in http_rewrite filter");
 633             w.rule = it->second;
 634             w.reqline = values[4] == "1";
 635             phase.within_list.push_back(w);
 636         }
 637         else
 638         {
 639             throw mp::filter::FilterException
 640                 ("Bad element "
 641                  + std::string((const char *) ptr->name)
 642                  + " in http_rewrite filter");
 643         }
 644     }
 645 }
 646
 647 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 648         const char *path)
 649 {
 650     for (ptr = ptr->children; ptr; ptr = ptr->next)
 651     {
 652         if (ptr->type != XML_ELEMENT_NODE)
 653             continue;
 654         else if (!strcmp((const char *) ptr->name, "request"))
 655         {
 656             configure_phase(ptr, *req_phase);
 657         }
 658         else if (!strcmp((const char *) ptr->name, "response"))
 659         {
 660             configure_phase(ptr, *res_phase);
 661         }
 662         else
 663         {
 664             throw mp::filter::FilterException
 665                 ("Bad element "
 666                  + std::string((const char *) ptr->name)
 667                  + " in http_rewrite1 filter");
 668         }
 669     }
 670 }
 671
 672 static mp::filter::Base* filter_creator()
 673 {
 674     return new mp::filter::HttpRewrite;
 675 }
 676
 677 extern "C" {
 678     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 679         0,
 680         "http_rewrite",
 681         filter_creator
 682     };
 683 }
 684
 685
 686 /*
 687  * Local variables:
 688  * c-basic-offset: 4
 689  * c-file-style: "Stroustrup"
 690  * indent-tabs-mode: nil
 691  * End:
 692  * vim: shiftwidth=4 tabstop=8 expandtab
 693  */
 694