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             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             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::list<Within>::const_iterator enabled_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     enabled_within = m_phase->within_list.end();
 248 }
 249
 250 yf::HttpRewrite::Event::~Event()
 251 {
 252     wrbuf_destroy(m_w);
 253 }
 254
 255 const char *yf::HttpRewrite::Event::result()
 256 {
 257     return wrbuf_cstr(m_w);
 258 }
 259
 260 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 261 {
 262     // check if there is <within tag="x" .. />
 263     if (enabled_within == m_phase->within_list.end())
 264     {
 265         std::string t(tag, tag_len);
 266         std::list<Within>::const_iterator it =
 267             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                         enabled_within = it;
 281                         break;
 282                     }
 283                 }
 284             }
 285         }
 286     }
 287     wrbuf_putc(m_w, '<');
 288     wrbuf_write(m_w, tag, tag_len);
 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         std::list<Within>::const_iterator it = enabled_within;
 297         if (it != m_phase->within_list.end())
 298         {
 299             std::string t(tag, tag_len);
 300             if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 301             {
 302                 enabled_within = m_phase->within_list.end();
 303             }
 304         }
 305     }
 306     if (close_it)
 307         wrbuf_putc(m_w, '/');
 308     wrbuf_putc(m_w, '>');
 309 }
 310
 311 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 312                                        const char *attr, int attr_len,
 313                                        const char *value, int val_len)
 314 {
 315     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 316     bool subst = false;
 317
 318     for (; it != m_phase->within_list.end(); it++)
 319     {
 320         std::string t(tag, tag_len);
 321         if (it->tag.length() == 0 ||
 322             yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 323         {
 324             std::string a(attr, attr_len);
 325             std::vector<std::string> attr;
 326             boost::split(attr, it->attr, boost::is_any_of(","));
 327             size_t i;
 328             for (i = 0; i < attr.size(); i++)
 329             {
 330                 if (attr[i].compare("#text") &&
 331                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 332                     subst = true;
 333             }
 334         }
 335         if (subst)
 336             break;
 337     }
 338
 339     wrbuf_putc(m_w, ' ');
 340     wrbuf_write(m_w, attr, attr_len);
 341     wrbuf_puts(m_w, "=\"");
 342
 343     std::string output;
 344     if (subst)
 345     {
 346         std::string input(value, val_len);
 347         output = it->rule->test_patterns(m_vars, input);
 348     }
 349     if (output.empty())
 350         wrbuf_write(m_w, value, val_len);
 351     else
 352         wrbuf_puts(m_w, output.c_str());
 353     wrbuf_puts(m_w, "\"");
 354 }
 355
 356 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 357 {
 358     std::list<Within>::const_iterator it = enabled_within;
 359     if (it != m_phase->within_list.end())
 360     {
 361         std::string t(tag, tag_len);
 362         if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 363         {
 364             enabled_within = m_phase->within_list.end();
 365         }
 366     }
 367     wrbuf_puts(m_w, "</");
 368     wrbuf_write(m_w, tag, tag_len);
 369 }
 370
 371 void yf::HttpRewrite::Event::text(const char *value, int len)
 372 {
 373     std::list<Within>::const_iterator it = enabled_within;
 374     bool subst = false;
 375
 376     if (it != m_phase->within_list.end())
 377     {
 378         subst = true;
 379         if (it->attr.length() > 0)
 380         {
 381             subst = false;
 382             std::vector<std::string> attr;
 383             boost::split(attr, it->attr, boost::is_any_of(","));
 384             size_t i;
 385             for (i = 0; i < attr.size(); i++)
 386             {
 387                 if (attr[i].compare("#text") == 0)
 388                 {
 389                     subst = true;
 390                 }
 391             }
 392         }
 393     }
 394     std::string output;
 395     if (subst)
 396     {
 397         std::string input(value, len);
 398         output = it->rule->test_patterns(m_vars, input);
 399     }
 400     if (output.empty())
 401         wrbuf_write(m_w, value, len);
 402     else
 403         wrbuf_puts(m_w, output.c_str());
 404 }
 405
 406
 407 /**
 408  * Tests pattern from the vector in order and executes recipe on
 409  the first match.
 410  */
 411 const std::string yf::HttpRewrite::Rule::test_patterns(
 412         std::map<std::string, std::string> & vars,
 413         const std::string & txt) const
 414 {
 415     std::list<Replace>::const_iterator it = replace_list.begin();
 416
 417     for (; it != replace_list.end(); it++)
 418     {
 419         std::string out = it->search_replace(vars, txt);
 420         if (!out.empty()) return out;
 421     }
 422     return "";
 423 }
 424
 425 const std::string yf::HttpRewrite::Replace::search_replace(
 426         std::map<std::string, std::string> & vars,
 427         const std::string & txt) const
 428 {
 429     //exec regex against value
 430     boost::regex re(regex);
 431     boost::smatch what;
 432     std::string::const_iterator start, end;
 433     start = txt.begin();
 434     end = txt.end();
 435     std::string out;
 436     while (regex_search(start, end, what, re)) //find next full match
 437     {
 438         size_t i;
 439         for (i = 1; i < what.size(); ++i)
 440         {
 441             //check if the group is named
 442             std::map<int, std::string>::const_iterator it
 443                 = group_index.find(i);
 444             if (it != group_index.end())
 445             {   //it is
 446                 if (!what[i].str().empty())
 447                     vars[it->second] = what[i];
 448             }
 449
 450         }
 451         //prepare replacement string
 452         std::string rvalue = sub_vars(vars);
 453         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 454                 what.str(0).c_str(), rvalue.c_str());
 455         out.append(start, what[0].first);
 456         out.append(rvalue);
 457         start = what[0].second; //move search forward
 458     }
 459     //if we had a match cat the last part
 460     if (start != txt.begin())
 461         out.append(start, end);
 462     return out;
 463 }
 464
 465 void yf::HttpRewrite::Replace::parse_groups()
 466 {
 467     int gnum = 0;
 468     bool esc = false;
 469     const std::string & str = regex;
 470     std::string res;
 471     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 472     for (size_t i = 0; i < str.size(); ++i)
 473     {
 474         res += str[i];
 475         if (!esc && str[i] == '\\')
 476         {
 477             esc = true;
 478             continue;
 479         }
 480         if (!esc && str[i] == '(') //group starts
 481         {
 482             gnum++;
 483             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 484             {
 485                 i++;
 486                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 487                 {
 488                     if (gnum > 0) gnum--;
 489                     res += str[i];
 490                     i++;
 491                     res += str[i];
 492                     continue;
 493                 }
 494                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 495                     i++;
 496                 if (i+1 < str.size() && str[i+1] == '<') //named
 497                 {
 498                     i++;
 499                     std::string gname;
 500                     bool term = false;
 501                     while (++i < str.size())
 502                     {
 503                         if (str[i] == '>') { term = true; break; }
 504                         if (!isalnum(str[i]))
 505                             throw mp::filter::FilterException
 506                                 ("Only alphanumeric chars allowed, found "
 507                                  " in '"
 508                                  + str
 509                                  + "' at "
 510                                  + boost::lexical_cast<std::string>(i));
 511                         gname += str[i];
 512                     }
 513                     if (!term)
 514                         throw mp::filter::FilterException
 515                             ("Unterminated group name '" + gname
 516                              + " in '" + str +"'");
 517                     group_index[gnum] = gname;
 518                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 519                             gname.c_str(), gnum);
 520                 }
 521             }
 522         }
 523         esc = false;
 524     }
 525     regex = res;
 526 }
 527
 528 std::string yf::HttpRewrite::Replace::sub_vars (
 529         const std::map<std::string, std::string> & vars) const
 530 {
 531     std::string out;
 532     bool esc = false;
 533     const std::string & in = recipe;
 534     for (size_t i = 0; i < in.size(); ++i)
 535     {
 536         if (!esc && in[i] == '\\')
 537         {
 538             esc = true;
 539             continue;
 540         }
 541         if (!esc && in[i] == '$') //var
 542         {
 543             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 544             {
 545                 ++i;
 546                 std::string name;
 547                 bool term = false;
 548                 while (++i < in.size())
 549                 {
 550                     if (in[i] == '}') { term = true; break; }
 551                     name += in[i];
 552                 }
 553                 if (!term) throw mp::filter::FilterException
 554                     ("Unterminated var ref in '"+in+"' at "
 555                      + boost::lexical_cast<std::string>(i));
 556                 std::map<std::string, std::string>::const_iterator it
 557                     = vars.find(name);
 558                 if (it != vars.end())
 559                 {
 560                     out += it->second;
 561                 }
 562             }
 563             else
 564             {
 565                 throw mp::filter::FilterException
 566                     ("Malformed or trimmed var ref in '"
 567                      +in+"' at "+boost::lexical_cast<std::string>(i));
 568             }
 569             continue;
 570         }
 571         //passthru
 572         out += in[i];
 573         esc = false;
 574     }
 575     return out;
 576 }
 577
 578 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 579 {
 580 }
 581
 582 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 583 {
 584     static const char *names[2] = { "verbose", 0 };
 585     std::string values[1];
 586     values[0] = "0";
 587     mp::xml::parse_attr(ptr, names, values);
 588
 589     phase.m_verbose = atoi(values[0].c_str());
 590
 591     std::map<std::string, RulePtr > rules;
 592     for (ptr = ptr->children; ptr; ptr = ptr->next)
 593     {
 594         if (ptr->type != XML_ELEMENT_NODE)
 595             continue;
 596         else if (!strcmp((const char *) ptr->name, "rule"))
 597         {
 598             static const char *names[2] = { "name", 0 };
 599             std::string values[1];
 600             values[0] = "default";
 601             mp::xml::parse_attr(ptr, names, values);
 602
 603             RulePtr rule(new Rule);
 604             for (xmlNode *p = ptr->children; p; p = p->next)
 605             {
 606                 if (p->type != XML_ELEMENT_NODE)
 607                     continue;
 608                 if (!strcmp((const char *) p->name, "rewrite"))
 609                 {
 610                     Replace replace;
 611                     const struct _xmlAttr *attr;
 612                     for (attr = p->properties; attr; attr = attr->next)
 613                     {
 614                         if (!strcmp((const char *) attr->name,  "from"))
 615                             replace.regex = mp::xml::get_text(attr->children);
 616                         else if (!strcmp((const char *) attr->name,  "to"))
 617                             replace.recipe = mp::xml::get_text(attr->children);
 618                         else
 619                             throw mp::filter::FilterException
 620                                 ("Bad attribute "
 621                                  + std::string((const char *) attr->name)
 622                                  + " in rewrite section of http_rewrite");
 623                     }
 624                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 625                             replace.regex.c_str(), replace.recipe.c_str());
 626                     replace.parse_groups();
 627                     if (!replace.regex.empty())
 628                         rule->replace_list.push_back(replace);
 629                 }
 630                 else
 631                     throw mp::filter::FilterException
 632                         ("Bad element "
 633                          + std::string((const char *) p->name)
 634                          + " in http_rewrite filter");
 635             }
 636             rules[values[0]] = rule;
 637         }
 638         else if (!strcmp((const char *) ptr->name, "within"))
 639         {
 640             static const char *names[6] =
 641                 { "header", "attr", "tag", "rule", "reqline", 0 };
 642             std::string values[5];
 643             mp::xml::parse_attr(ptr, names, values);
 644             Within w;
 645             w.header = values[0];
 646             w.attr = values[1];
 647             w.tag = values[2];
 648             std::map<std::string,RulePtr>::const_iterator it =
 649                 rules.find(values[3]);
 650             if (it == rules.end())
 651                 throw mp::filter::FilterException
 652                     ("Reference to non-existing rule '" + values[3] +
 653                      "' in http_rewrite filter");
 654             w.rule = it->second;
 655             w.reqline = values[4] == "1";
 656             phase.within_list.push_back(w);
 657         }
 658         else
 659         {
 660             throw mp::filter::FilterException
 661                 ("Bad element "
 662                  + std::string((const char *) ptr->name)
 663                  + " in http_rewrite filter");
 664         }
 665     }
 666 }
 667
 668 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 669         const char *path)
 670 {
 671     for (ptr = ptr->children; ptr; ptr = ptr->next)
 672     {
 673         if (ptr->type != XML_ELEMENT_NODE)
 674             continue;
 675         else if (!strcmp((const char *) ptr->name, "request"))
 676         {
 677             configure_phase(ptr, *req_phase);
 678         }
 679         else if (!strcmp((const char *) ptr->name, "response"))
 680         {
 681             configure_phase(ptr, *res_phase);
 682         }
 683         else
 684         {
 685             throw mp::filter::FilterException
 686                 ("Bad element "
 687                  + std::string((const char *) ptr->name)
 688                  + " in http_rewrite1 filter");
 689         }
 690     }
 691 }
 692
 693 static mp::filter::Base* filter_creator()
 694 {
 695     return new mp::filter::HttpRewrite;
 696 }
 697
 698 extern "C" {
 699     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 700         0,
 701         "http_rewrite",
 702         filter_creator
 703     };
 704 }
 705
 706
 707 /*
 708  * Local variables:
 709  * c-basic-offset: 4
 710  * c-file-style: "Stroustrup"
 711  * indent-tabs-mode: nil
 712  * End:
 713  * vim: shiftwidth=4 tabstop=8 expandtab
 714  */
 715