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