src/filter_http_rewrite.cpp

   1 /* This file is part of Metaproxy.
   2    Copyright (C) 2005-2013 Index Data
   3
   4 Metaproxy is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Metaproxy is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17 */
  18
  19 #include "config.hpp"
  20 #include <metaproxy/filter.hpp>
  21 #include <metaproxy/package.hpp>
  22 #include <metaproxy/util.hpp>
  23 #include "filter_http_rewrite.hpp"
  24 #include "html_parser.hpp"
  25
  26 #include <yaz/zgdu.h>
  27 #include <yaz/log.h>
  28
  29 #include <stack>
  30 #include <boost/regex.hpp>
  31 #include <boost/lexical_cast.hpp>
  32 #include <boost/algorithm/string.hpp>
  33
  34 #include <map>
  35
  36 namespace mp = metaproxy_1;
  37 namespace yf = mp::filter;
  38
  39 namespace metaproxy_1 {
  40     namespace filter {
  41         class HttpRewrite::Replace {
  42         public:
  43             boost::regex re;
  44             boost::smatch what;
  45             std::string recipe;
  46             std::map<int, std::string> group_index;
  47             const std::string search_replace(
  48                 std::map<std::string, std::string> & vars,
  49                 const std::string & txt);
  50             std::string sub_vars(
  51                 const std::map<std::string, std::string> & vars) const;
  52             void parse_groups(std::string pattern);
  53         };
  54
  55         class HttpRewrite::Rule {
  56         public:
  57             std::list<Replace> replace_list;
  58             const std::string test_patterns(
  59                 std::map<std::string, std::string> & vars,
  60                 const std::string & txt);
  61         };
  62         class HttpRewrite::Within {
  63         public:
  64             std::string header;
  65             std::string attr;
  66             std::string tag;
  67             bool reqline;
  68             RulePtr rule;
  69         };
  70
  71         class HttpRewrite::Phase {
  72         public:
  73             Phase();
  74             std::list<Within> within_list;
  75             int m_verbose;
  76             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  77                 std::map<std::string, std::string> & vars) const;
  78             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  79                 std::map<std::string, std::string> & vars) const;
  80             void rewrite_body(mp::odr & o,
  81                 char **content_buf, int *content_len,
  82                 std::map<std::string, std::string> & vars) const;
  83         };
  84         class HttpRewrite::Event : public HTMLParserEvent {
  85             void openTagStart(const char *tag, int tag_len);
  86             void anyTagEnd(const char *tag, int tag_len, int close_it);
  87             void attribute(const char *tag, int tag_len,
  88                            const char *attr, int attr_len,
  89                            const char *value, int val_len,
  90                            const char *sep);
  91             void closeTag(const char *tag, int tag_len);
  92             void text(const char *value, int len);
  93             const Phase *m_phase;
  94             WRBUF m_w;
  95             std::stack<std::list<Within>::const_iterator> s_within;
  96             std::map<std::string, std::string> &m_vars;
  97         public:
  98             Event(const Phase *p, std::map<std::string, std::string> &vars);
  99             ~Event();
 100             const char *result();
 101         };
 102     }
 103 }
 104
 105 yf::HttpRewrite::HttpRewrite() :
 106     req_phase(new Phase), res_phase(new Phase)
 107 {
 108 }
 109
 110 yf::HttpRewrite::~HttpRewrite()
 111 {
 112 }
 113
 114 void yf::HttpRewrite::process(mp::Package & package) const
 115 {
 116     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 117     Z_GDU *gdu = package.request().get();
 118     //map of request/response vars
 119     std::map<std::string, std::string> vars;
 120     //we have an http req
 121     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 122     {
 123         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 124         mp::odr o;
 125         req_phase->rewrite_reqline(o, hreq, vars);
 126         yaz_log(YLOG_LOG, ">> Request headers");
 127         req_phase->rewrite_headers(o, hreq->headers, vars);
 128         req_phase->rewrite_body(o,
 129                 &hreq->content_buf, &hreq->content_len, vars);
 130         package.request() = gdu;
 131     }
 132     package.move();
 133     gdu = package.response().get();
 134     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 135     {
 136         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 137         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 138         mp::odr o;
 139         yaz_log(YLOG_LOG, "<< Respose headers");
 140         res_phase->rewrite_headers(o, hres->headers, vars);
 141         res_phase->rewrite_body(o, &hres->content_buf,
 142                 &hres->content_len, vars);
 143         package.response() = gdu;
 144     }
 145 }
 146
 147 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 148         Z_HTTP_Request *hreq,
 149         std::map<std::string, std::string> & vars) const
 150 {
 151     //rewrite the request line
 152     std::string path;
 153     if (strstr(hreq->path, "http://") == hreq->path)
 154     {
 155         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 156             "possibly a proxy request");
 157         path += hreq->path;
 158     }
 159     else
 160     {
 161         //TODO what about proto
 162         path += "http://";
 163         path += z_HTTP_header_lookup(hreq->headers, "Host");
 164         path += hreq->path;
 165     }
 166
 167
 168     std::list<Within>::const_iterator it = within_list.begin();
 169     for (; it != within_list.end(); it++)
 170         if (it->reqline)
 171         {
 172             RulePtr rule = it->rule;
 173             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 174             std::string npath = rule->test_patterns(vars, path);
 175             if (!npath.empty())
 176             {
 177                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 178                 hreq->path = odr_strdup(o, npath.c_str());
 179             }
 180         }
 181 }
 182
 183 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 184         Z_HTTP_Header *headers,
 185         std::map<std::string, std::string> & vars) const
 186 {
 187     for (Z_HTTP_Header *header = headers; header; header = header->next)
 188     {
 189         std::list<Within>::const_iterator it = within_list.begin();
 190         for (; it != within_list.end(); it++)
 191         {
 192             if (it->header.length() > 0 &&
 193                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 194             {
 195                 std::string sheader(header->name);
 196                 sheader += ": ";
 197                 sheader += header->value;
 198
 199                 RulePtr rule = it->rule;
 200                 std::string out = rule->test_patterns(vars, sheader);
 201                 if (!out.empty())
 202                 {
 203                     size_t pos = out.find(": ");
 204                     if (pos == std::string::npos)
 205                     {
 206                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 207                         continue;
 208                     }
 209                     header->name = odr_strdup(o, out.substr(0, pos).c_str());
 210                     header->value = odr_strdup(o,
 211                                                out.substr(pos + 2,
 212                                                           std::string::npos).c_str());
 213                 }
 214             }
 215         }
 216     }
 217 }
 218
 219 void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 220         char **content_buf,
 221         int *content_len,
 222         std::map<std::string, std::string> & vars) const
 223 {
 224     if (*content_buf)
 225     {
 226         int i;
 227         for (i = 0; i < *content_len; i++)
 228             if ((*content_buf)[i] == 0)
 229                 return;  // binary content. skip
 230
 231         HTMLParser parser;
 232         Event ev(this, vars);
 233
 234         parser.set_verbose(m_verbose);
 235
 236         std::string buf(*content_buf, *content_len);
 237
 238         parser.parse(ev, buf.c_str());
 239         const char *res = ev.result();
 240         *content_buf = odr_strdup(o, res);
 241         *content_len = strlen(res);
 242     }
 243 }
 244
 245 yf::HttpRewrite::Event::Event(const Phase *p,
 246                               std::map<std::string, std::string> & vars
 247     ) : m_phase(p), m_vars(vars)
 248 {
 249     m_w = wrbuf_alloc();
 250 }
 251
 252 yf::HttpRewrite::Event::~Event()
 253 {
 254     wrbuf_destroy(m_w);
 255 }
 256
 257 const char *yf::HttpRewrite::Event::result()
 258 {
 259     return wrbuf_cstr(m_w);
 260 }
 261
 262 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 263 {
 264     wrbuf_putc(m_w, '<');
 265     wrbuf_write(m_w, tag, tag_len);
 266
 267     std::string t(tag, tag_len);
 268     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 269     for (; it != m_phase->within_list.end(); it++)
 270     {
 271         if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
 272                                                    t.c_str()) == 0)
 273         {
 274             std::vector<std::string> attr;
 275             boost::split(attr, it->attr, boost::is_any_of(","));
 276             size_t i;
 277             for (i = 0; i < attr.size(); i++)
 278             {
 279                 if (attr[i].compare("#text") == 0)
 280                 {
 281                     s_within.push(it);
 282                     return;
 283                 }
 284             }
 285         }
 286     }
 287 }
 288
 289 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 290                                        int close_it)
 291 {
 292     if (close_it)
 293     {
 294         if (!s_within.empty())
 295         {
 296             std::list<Within>::const_iterator it = s_within.top();
 297             std::string t(tag, tag_len);
 298             if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 299                 s_within.pop();
 300         }
 301     }
 302     if (close_it)
 303         wrbuf_putc(m_w, '/');
 304     wrbuf_putc(m_w, '>');
 305 }
 306
 307 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 308                                        const char *attr, int attr_len,
 309                                        const char *value, int val_len,
 310                                        const char *sep)
 311 {
 312     std::list<Within>::const_iterator it = m_phase->within_list.begin();
 313     bool subst = false;
 314
 315     for (; it != m_phase->within_list.end(); it++)
 316     {
 317         std::string t(tag, tag_len);
 318         if (it->tag.length() == 0 ||
 319             yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 320         {
 321             std::string a(attr, attr_len);
 322             std::vector<std::string> attr;
 323             boost::split(attr, it->attr, boost::is_any_of(","));
 324             size_t i;
 325             for (i = 0; i < attr.size(); i++)
 326             {
 327                 if (attr[i].compare("#text") &&
 328                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 329                     subst = true;
 330             }
 331         }
 332         if (subst)
 333             break;
 334     }
 335
 336     wrbuf_putc(m_w, ' ');
 337     wrbuf_write(m_w, attr, attr_len);
 338     if (value)
 339     {
 340         wrbuf_puts(m_w, "=");
 341         wrbuf_puts(m_w, sep);
 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, sep);
 354     }
 355 }
 356
 357 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 358 {
 359     if (!s_within.empty())
 360     {
 361         std::list<Within>::const_iterator it = s_within.top();
 362         std::string t(tag, tag_len);
 363         if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
 364             s_within.pop();
 365     }
 366     wrbuf_puts(m_w, "</");
 367     wrbuf_write(m_w, tag, tag_len);
 368 }
 369
 370 void yf::HttpRewrite::Event::text(const char *value, int len)
 371 {
 372     std::list<Within>::const_iterator it = m_phase->within_list.end();
 373     if (!s_within.empty())
 374         it = s_within.top();
 375     std::string output;
 376     if (it != m_phase->within_list.end())
 377     {
 378         std::string input(value, len);
 379         output = it->rule->test_patterns(m_vars, input);
 380     }
 381     if (output.empty())
 382         wrbuf_write(m_w, value, len);
 383     else
 384         wrbuf_puts(m_w, output.c_str());
 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)
 394 {
 395     std::list<Replace>::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)
 408 {
 409     std::string::const_iterator start, end;
 410     start = txt.begin();
 411     end = txt.end();
 412     std::string out;
 413     while (regex_search(start, end, what, re)) //find next full match
 414     {
 415         size_t i;
 416         for (i = 1; i < what.size(); ++i)
 417         {
 418             //check if the group is named
 419             std::map<int, std::string>::const_iterator it
 420                 = group_index.find(i);
 421             if (it != group_index.end())
 422             {   //it is
 423                 vars[it->second] = what[i];
 424             }
 425
 426         }
 427         //prepare replacement string
 428         std::string rvalue = sub_vars(vars);
 429         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 430                 what.str(0).c_str(), rvalue.c_str());
 431         out.append(start, what[0].first);
 432         out.append(rvalue);
 433         start = what[0].second; //move search forward
 434     }
 435     //if we had a match cat the last part
 436     if (start != txt.begin())
 437         out.append(start, end);
 438     return out;
 439 }
 440
 441 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 442 {
 443     int gnum = 0;
 444     bool esc = false;
 445     const std::string &str = pattern;
 446     std::string res;
 447     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 448     for (size_t i = 0; i < str.size(); ++i)
 449     {
 450         res += str[i];
 451         if (!esc && str[i] == '\\')
 452         {
 453             esc = true;
 454             continue;
 455         }
 456         if (!esc && str[i] == '(') //group starts
 457         {
 458             gnum++;
 459             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 460             {
 461                 i++;
 462                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 463                 {
 464                     if (gnum > 0) gnum--;
 465                     res += str[i];
 466                     i++;
 467                     res += str[i];
 468                     continue;
 469                 }
 470                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 471                     i++;
 472                 if (i+1 < str.size() && str[i+1] == '<') //named
 473                 {
 474                     i++;
 475                     std::string gname;
 476                     bool term = false;
 477                     while (++i < str.size())
 478                     {
 479                         if (str[i] == '>') { term = true; break; }
 480                         if (!isalnum(str[i]))
 481                             throw mp::filter::FilterException
 482                                 ("Only alphanumeric chars allowed, found "
 483                                  " in '"
 484                                  + str
 485                                  + "' at "
 486                                  + boost::lexical_cast<std::string>(i));
 487                         gname += str[i];
 488                     }
 489                     if (!term)
 490                         throw mp::filter::FilterException
 491                             ("Unterminated group name '" + gname
 492                              + " in '" + str +"'");
 493                     group_index[gnum] = gname;
 494                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 495                             gname.c_str(), gnum);
 496                 }
 497             }
 498         }
 499         esc = false;
 500     }
 501     re = res;
 502 }
 503
 504 std::string yf::HttpRewrite::Replace::sub_vars(
 505     const std::map<std::string, std::string> & vars) const
 506 {
 507     std::string out;
 508     bool esc = false;
 509     const std::string & in = recipe;
 510     for (size_t i = 0; i < in.size(); ++i)
 511     {
 512         if (!esc && in[i] == '\\')
 513         {
 514             esc = true;
 515             continue;
 516         }
 517         if (!esc && in[i] == '$') //var
 518         {
 519             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 520             {
 521                 ++i;
 522                 std::string name;
 523                 bool term = false;
 524                 while (++i < in.size())
 525                 {
 526                     if (in[i] == '}') { term = true; break; }
 527                     name += in[i];
 528                 }
 529                 if (!term) throw mp::filter::FilterException
 530                     ("Unterminated var ref in '"+in+"' at "
 531                      + boost::lexical_cast<std::string>(i));
 532                 std::map<std::string, std::string>::const_iterator it
 533                     = vars.find(name);
 534                 if (it != vars.end())
 535                 {
 536                     out += it->second;
 537                 }
 538             }
 539             else
 540             {
 541                 throw mp::filter::FilterException
 542                     ("Malformed or trimmed var ref in '"
 543                      +in+"' at "+boost::lexical_cast<std::string>(i));
 544             }
 545             continue;
 546         }
 547         //passthru
 548         out += in[i];
 549         esc = false;
 550     }
 551     return out;
 552 }
 553
 554 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 555 {
 556 }
 557
 558 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 559 {
 560     static const char *names[2] = { "verbose", 0 };
 561     std::string values[1];
 562     values[0] = "0";
 563     mp::xml::parse_attr(ptr, names, values);
 564
 565     phase.m_verbose = atoi(values[0].c_str());
 566
 567     std::map<std::string, RulePtr > rules;
 568     for (ptr = ptr->children; ptr; ptr = ptr->next)
 569     {
 570         if (ptr->type != XML_ELEMENT_NODE)
 571             continue;
 572         else if (!strcmp((const char *) ptr->name, "rule"))
 573         {
 574             static const char *names[2] = { "name", 0 };
 575             std::string values[1];
 576             values[0] = "default";
 577             mp::xml::parse_attr(ptr, names, values);
 578
 579             RulePtr rule(new Rule);
 580             for (xmlNode *p = ptr->children; p; p = p->next)
 581             {
 582                 if (p->type != XML_ELEMENT_NODE)
 583                     continue;
 584                 if (!strcmp((const char *) p->name, "rewrite"))
 585                 {
 586                     Replace replace;
 587                     std::string from;
 588                     const struct _xmlAttr *attr;
 589                     for (attr = p->properties; attr; attr = attr->next)
 590                     {
 591                         if (!strcmp((const char *) attr->name,  "from"))
 592                             from = mp::xml::get_text(attr->children);
 593                         else if (!strcmp((const char *) attr->name,  "to"))
 594                             replace.recipe = mp::xml::get_text(attr->children);
 595                         else
 596                             throw mp::filter::FilterException
 597                                 ("Bad attribute "
 598                                  + std::string((const char *) attr->name)
 599                                  + " in rewrite section of http_rewrite");
 600                     }
 601                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 602                             from.c_str(), replace.recipe.c_str());
 603                     if (!from.empty())
 604                     {
 605                         replace.parse_groups(from);
 606                         rule->replace_list.push_back(replace);
 607                     }
 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