src/filter_http_rewrite.cpp

   1 /* This file is part of Metaproxy.
   2    Copyright (C) 2005-2013 Index Data
   3
   4 Metaproxy is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Metaproxy is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17 */
  18
  19 #include "config.hpp"
  20 #include <metaproxy/filter.hpp>
  21 #include <metaproxy/package.hpp>
  22 #include <metaproxy/util.hpp>
  23 #include "filter_http_rewrite.hpp"
  24 #include "html_parser.hpp"
  25
  26 #include <yaz/zgdu.h>
  27 #include <yaz/log.h>
  28
  29 #include <stack>
  30 #include <boost/regex.hpp>
  31 #include <boost/lexical_cast.hpp>
  32 #include <boost/algorithm/string.hpp>
  33
  34 #include <map>
  35
  36 namespace mp = metaproxy_1;
  37 namespace yf = mp::filter;
  38
  39 namespace metaproxy_1 {
  40     namespace filter {
  41         class HttpRewrite::Replace {
  42         public:
  43             bool start_anchor;
  44             boost::regex re;
  45             boost::smatch what;
  46             std::string recipe;
  47             std::map<int, std::string> group_index;
  48             std::string sub_vars(
  49                 const std::map<std::string, std::string> & vars) const;
  50             void parse_groups(std::string pattern);
  51         };
  52
  53         class HttpRewrite::Rule {
  54         public:
  55             std::list<Replace> replace_list;
  56             bool test_patterns(
  57                 std::map<std::string, std::string> &vars,
  58                 std::string &txt, bool anchor);
  59         };
  60         class HttpRewrite::Within {
  61         public:
  62             boost::regex header;
  63             boost::regex attr;
  64             boost::regex tag;
  65             std::string type;
  66             bool reqline;
  67             RulePtr rule;
  68             bool exec(std::map<std::string, std::string> &vars,
  69                       std::string &txt, bool anchor) const;
  70         };
  71
  72         class HttpRewrite::Content {
  73         public:
  74             std::string type;
  75             boost::regex content_re;
  76             std::list<Within> within_list;
  77             void configure(const xmlNode *ptr,
  78                            std::map<std::string, RulePtr > &rules);
  79             void quoted_literal(std::string &content,
  80                                 std::map<std::string, std::string> &vars) const;
  81             void parse(int verbose, std::string &content,
  82                        std::map<std::string, std::string> & vars) const;
  83         };
  84         class HttpRewrite::Phase {
  85         public:
  86             Phase();
  87             int m_verbose;
  88             std::list<Content> content_list;
  89             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
  90                 std::map<std::string, std::string> & vars) const;
  91             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  92                 std::map<std::string, std::string> & vars) const;
  93             void rewrite_body(mp::odr & o,
  94                               const char *content_type,
  95                               char **content_buf, int *content_len,
  96                               std::map<std::string, std::string> & vars) const;
  97         };
  98         class HttpRewrite::Event : public HTMLParserEvent {
  99             void openTagStart(const char *tag, int tag_len);
 100             void anyTagEnd(const char *tag, int tag_len, int close_it);
 101             void attribute(const char *tag, int tag_len,
 102                            const char *attr, int attr_len,
 103                            const char *value, int val_len,
 104                            const char *sep);
 105             void closeTag(const char *tag, int tag_len);
 106             void text(const char *value, int len);
 107             const Content *m_content;
 108             WRBUF m_w;
 109             std::stack<std::list<Within>::const_iterator> s_within;
 110             std::map<std::string, std::string> &m_vars;
 111         public:
 112             Event(const Content *p, std::map<std::string, std::string> &vars);
 113             ~Event();
 114             const char *result();
 115         };
 116     }
 117 }
 118
 119 yf::HttpRewrite::HttpRewrite() :
 120     req_phase(new Phase), res_phase(new Phase)
 121 {
 122 }
 123
 124 yf::HttpRewrite::~HttpRewrite()
 125 {
 126 }
 127
 128 void yf::HttpRewrite::process(mp::Package & package) const
 129 {
 130     yaz_log(YLOG_LOG, "HttpRewrite begins....");
 131     Z_GDU *gdu = package.request().get();
 132     //map of request/response vars
 133     std::map<std::string, std::string> vars;
 134     //we have an http req
 135     if (gdu && gdu->which == Z_GDU_HTTP_Request)
 136     {
 137         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
 138         mp::odr o;
 139         req_phase->rewrite_reqline(o, hreq, vars);
 140         yaz_log(YLOG_LOG, ">> Request headers");
 141         req_phase->rewrite_headers(o, hreq->headers, vars);
 142         req_phase->rewrite_body(o,
 143                                 z_HTTP_header_lookup(hreq->headers,
 144                                                      "Content-Type"),
 145                                 &hreq->content_buf, &hreq->content_len,
 146                                 vars);
 147         package.request() = gdu;
 148     }
 149     package.move();
 150     gdu = package.response().get();
 151     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 152     {
 153         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 154         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 155         mp::odr o;
 156         yaz_log(YLOG_LOG, "<< Respose headers");
 157         res_phase->rewrite_headers(o, hres->headers, vars);
 158         res_phase->rewrite_body(o,
 159                                 z_HTTP_header_lookup(hres->headers,
 160                                                      "Content-Type"),
 161                                 &hres->content_buf, &hres->content_len,
 162                                 vars);
 163         package.response() = gdu;
 164     }
 165 }
 166
 167 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
 168         Z_HTTP_Request *hreq,
 169         std::map<std::string, std::string> & vars) const
 170 {
 171     //rewrite the request line
 172     std::string path;
 173     if (strstr(hreq->path, "http://") == hreq->path)
 174     {
 175         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 176             "possibly a proxy request");
 177         path += hreq->path;
 178     }
 179     else
 180     {
 181         //TODO what about proto
 182         const char *host = z_HTTP_header_lookup(hreq->headers, "Host");
 183         if (!host)
 184             return;
 185
 186         path += "http://";
 187         path += host;
 188         path += hreq->path;
 189     }
 190
 191     std::list<Content>::const_iterator cit = content_list.begin();
 192     for (; cit != content_list.end(); cit++)
 193         if (cit->type == "headers")
 194             break;
 195
 196     if (cit == content_list.end())
 197         return;
 198
 199     std::list<Within>::const_iterator it = cit->within_list.begin();
 200     for (; it != cit->within_list.end(); it++)
 201         if (it->reqline)
 202         {
 203             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 204             if (it->exec(vars, path, true))
 205             {
 206                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", path.c_str());
 207                 hreq->path = odr_strdup(o, path.c_str());
 208             }
 209         }
 210 }
 211
 212 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
 213         Z_HTTP_Header *headers,
 214         std::map<std::string, std::string> & vars) const
 215 {
 216     std::list<Content>::const_iterator cit = content_list.begin();
 217     for (; cit != content_list.end(); cit++)
 218         if (cit->type == "headers")
 219             break;
 220
 221     if (cit == content_list.end())
 222         return;
 223
 224     for (Z_HTTP_Header *header = headers; header; header = header->next)
 225     {
 226         std::list<Within>::const_iterator it = cit->within_list.begin();
 227         for (; it != cit->within_list.end(); it++)
 228         {
 229             if (!it->header.empty() &&
 230                 regex_match(header->name, it->header))
 231             {
 232 #ifdef OLDHEADERMATCH
 233                 // Matches and replaces the whole header line.
 234                 // This is good if you want to play with the header name too,
 235                 // but useless for patterns that want to anchor to the beginning
 236                 // or end of the header value, as we want to do with host-relative
 237                 // links. This code should probably be removed.
 238                 std::string sheader(header->name);
 239                 sheader += ": ";
 240                 sheader += header->value;
 241
 242                 if (it->exec(vars, sheader, true))
 243                 {
 244                     size_t pos = sheader.find(": ");
 245                     if (pos == std::string::npos)
 246                     {
 247                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 248                         continue;
 249                     }
 250                     header->name = odr_strdup(
 251                         o, sheader.substr(0, pos).c_str());
 252                     header->value = odr_strdup(
 253                         o, sheader.substr(pos + 2, std::string::npos).c_str());
 254                 }
 255 #else
 256                 // Match and replace only the header value
 257                 std::string hval(header->value);
 258                 if (it->exec(vars, hval, true))
 259                 {
 260                     header->value = odr_strdup(o, hval.c_str());
 261                 }
 262
 263 #endif
 264             }
 265         }
 266     }
 267 }
 268
 269 void yf::HttpRewrite::Phase::rewrite_body(
 270     mp::odr &o,
 271     const char *content_type,
 272     char **content_buf,
 273     int *content_len,
 274     std::map<std::string, std::string> & vars) const
 275 {
 276     if (*content_len == 0)
 277         return;
 278     if (!content_type) {
 279         yaz_log(YLOG_LOG, "rewrite_body: null content_type, can not rewrite");
 280         return;
 281     }
 282     std::list<Content>::const_iterator cit = content_list.begin();
 283     for (; cit != content_list.end(); cit++)
 284     {
 285         yaz_log(YLOG_LOG, "rewrite_body: content_type=%s type=%s",
 286                 content_type, cit->type.c_str());
 287         if (cit->type != "headers"
 288             && regex_match(content_type, cit->content_re))
 289             break;
 290     }
 291     if (cit == content_list.end()) {
 292         yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
 293                 content_type );
 294         return;
 295     }
 296
 297     int i;
 298     for (i = 0; i < *content_len; i++)
 299         if ((*content_buf)[i] == 0) {
 300             yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
 301             return;  // binary content. skip
 302         }
 303
 304     std::string content(*content_buf, *content_len);
 305     cit->parse(m_verbose, content, vars);
 306     *content_buf = odr_strdup(o, content.c_str());
 307     *content_len = strlen(*content_buf);
 308 }
 309
 310 yf::HttpRewrite::Event::Event(const Content *p,
 311                               std::map<std::string, std::string> & vars
 312     ) : m_content(p), m_vars(vars)
 313 {
 314     m_w = wrbuf_alloc();
 315 }
 316
 317 yf::HttpRewrite::Event::~Event()
 318 {
 319     wrbuf_destroy(m_w);
 320 }
 321
 322 const char *yf::HttpRewrite::Event::result()
 323 {
 324     return wrbuf_cstr(m_w);
 325 }
 326
 327 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 328 {
 329     wrbuf_putc(m_w, '<');
 330     wrbuf_write(m_w, tag, tag_len);
 331
 332     std::string t(tag, tag_len);
 333     std::list<Within>::const_iterator it = m_content->within_list.begin();
 334     for (; it != m_content->within_list.end(); it++)
 335     {
 336         if (!it->tag.empty() && regex_match(t, it->tag))
 337         {
 338             if (!it->attr.empty() && regex_match("#text", it->attr))
 339             {
 340                 s_within.push(it);
 341                 return;
 342             }
 343         }
 344     }
 345 }
 346
 347 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 348                                        int close_it)
 349 {
 350     if (close_it)
 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 (regex_match(t, it->tag))
 357                 s_within.pop();
 358         }
 359     }
 360     if (close_it)
 361         wrbuf_putc(m_w, '/');
 362     wrbuf_putc(m_w, '>');
 363 }
 364
 365 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 366                                        const char *attr, int attr_len,
 367                                        const char *value, int val_len,
 368                                        const char *sep)
 369 {
 370     std::list<Within>::const_iterator it = m_content->within_list.begin();
 371     bool subst = false;
 372
 373     for (; it != m_content->within_list.end(); it++)
 374     {
 375         std::string t(tag, tag_len);
 376         if (it->tag.empty() || regex_match(t, it->tag))
 377         {
 378             std::string a(attr, attr_len);
 379             if (!it->attr.empty() && regex_match(a, it->attr))
 380                 subst = true;
 381         }
 382         if (subst)
 383             break;
 384     }
 385
 386     wrbuf_putc(m_w, ' ');
 387     wrbuf_write(m_w, attr, attr_len);
 388     if (value)
 389     {
 390         wrbuf_puts(m_w, "=");
 391         wrbuf_puts(m_w, sep);
 392
 393         std::string output;
 394         if (subst)
 395         {
 396             std::string s(value, val_len);
 397             it->exec(m_vars, s, true);
 398             wrbuf_puts(m_w, s.c_str());
 399         }
 400         else
 401             wrbuf_write(m_w, value, val_len);
 402         wrbuf_puts(m_w, sep);
 403     }
 404 }
 405
 406 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 407 {
 408     if (!s_within.empty())
 409     {
 410         std::list<Within>::const_iterator it = s_within.top();
 411         std::string t(tag, tag_len);
 412         if (regex_match(t, it->tag))
 413             s_within.pop();
 414     }
 415     wrbuf_puts(m_w, "</");
 416     wrbuf_write(m_w, tag, tag_len);
 417 }
 418
 419 void yf::HttpRewrite::Event::text(const char *value, int len)
 420 {
 421     std::list<Within>::const_iterator it = m_content->within_list.end();
 422     if (!s_within.empty())
 423         it = s_within.top();
 424     if (it != m_content->within_list.end())
 425     {
 426         std::string s(value, len);
 427         it->exec(m_vars, s, false);
 428         wrbuf_puts(m_w, s.c_str());
 429     }
 430     else
 431         wrbuf_write(m_w, value, len);
 432 }
 433
 434 static bool embed_quoted_literal(
 435     std::string &content,
 436     std::map<std::string, std::string> &vars,
 437     mp::filter::HttpRewrite::RulePtr ruleptr,
 438     bool html_context)
 439 {
 440     bool replace = false;
 441     std::string res;
 442     const char *cp = content.c_str();
 443     const char *cp0 = cp;
 444     while (*cp)
 445     {
 446         if (html_context && !strncmp(cp, "&quot;", 6))
 447         {
 448             cp += 6;
 449             res.append(cp0, cp - cp0);
 450             cp0 = cp;
 451             while (*cp)
 452             {
 453                 if (!strncmp(cp, "&quot;", 6))
 454                     break;
 455                 if (*cp == '\n')
 456                     break;
 457                 cp++;
 458             }
 459             if (!*cp)
 460                 break;
 461             std::string s(cp0, cp - cp0);
 462             if (ruleptr->test_patterns(vars, s, true))
 463                 replace = true;
 464             cp0 = cp;
 465             res.append(s);
 466         }
 467         else if (*cp == '"' || *cp == '\'')
 468         {
 469             int m = *cp;
 470             cp++;
 471             res.append(cp0, cp - cp0);
 472             cp0 = cp;
 473             while (*cp)
 474             {
 475                 if (cp[-1] != '\\' && *cp == m)
 476                     break;
 477                 if (*cp == '\n')
 478                     break;
 479                 cp++;
 480             }
 481             if (!*cp)
 482                 break;
 483             std::string s(cp0, cp - cp0);
 484             if (ruleptr->test_patterns(vars, s, true))
 485                 replace = true;
 486             cp0 = cp;
 487             res.append(s);
 488         }
 489         else if (*cp == '/' && cp[1] == '/')
 490         {
 491             while (cp[1] && cp[1] != '\n')
 492                 cp++;
 493         }
 494         cp++;
 495     }
 496     res.append(cp0, cp - cp0);
 497     content = res;
 498     return replace;
 499 }
 500
 501 bool yf::HttpRewrite::Within::exec(
 502     std::map<std::string, std::string> & vars,
 503     std::string & txt, bool anchor) const
 504 {
 505     if (type == "quoted-literal")
 506     {
 507         return embed_quoted_literal(txt, vars, rule, true);
 508     }
 509     else
 510     {
 511         return rule->test_patterns(vars, txt, anchor);
 512     }
 513 }
 514
 515 bool yf::HttpRewrite::Rule::test_patterns(
 516     std::map<std::string, std::string> & vars,
 517     std::string & txt, bool anchor)
 518 {
 519     bool replaces = false;
 520     bool first = anchor;
 521     std::string out;
 522     std::string::const_iterator start, end;
 523     start = txt.begin();
 524     end = txt.end();
 525     while (1)
 526     {
 527         std::list<Replace>::iterator bit = replace_list.end();
 528         {
 529             std::string::const_iterator best_pos = txt.end();
 530             std::list<Replace>::iterator it = replace_list.begin();
 531             for (; it != replace_list.end(); it++)
 532             {
 533                 if (it->start_anchor && !first)
 534                     continue;
 535                 if (regex_search(start, end, it->what, it->re))
 536                 {
 537                     if (it->what[0].first < best_pos)
 538                     {
 539                         best_pos = it->what[0].first;
 540                         bit = it;
 541                     }
 542                 }
 543             }
 544             if (bit == replace_list.end())
 545                 break;
 546         }
 547         first = false;
 548         replaces = true;
 549         size_t i;
 550         for (i = 1; i < bit->what.size(); ++i)
 551         {
 552             //check if the group is named
 553             std::map<int, std::string>::const_iterator git
 554                 = bit->group_index.find(i);
 555             if (git != bit->group_index.end())
 556             {   //it is
 557                 vars[git->second] = bit->what[i];
 558             }
 559
 560         }
 561         //prepare replacement string
 562         std::string rvalue = bit->sub_vars(vars);
 563         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 564                 bit->what.str(0).c_str(), rvalue.c_str());
 565         out.append(start, bit->what[0].first);
 566         out.append(rvalue);
 567         start = bit->what[0].second; //move search forward
 568     }
 569     out.append(start, end);
 570     txt = out;
 571     return replaces;
 572 }
 573
 574 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 575 {
 576     int gnum = 0;
 577     bool esc = false;
 578     const std::string &str = pattern;
 579     std::string res;
 580     start_anchor = str[0] == '^';
 581     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 582     for (size_t i = 0; i < str.size(); ++i)
 583     {
 584         res += str[i];
 585         if (!esc && str[i] == '\\')
 586         {
 587             esc = true;
 588             continue;
 589         }
 590         if (!esc && str[i] == '(') //group starts
 591         {
 592             gnum++;
 593             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 594             {
 595                 i++;
 596                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 597                 {
 598                     if (gnum > 0) gnum--;
 599                     res += str[i];
 600                     i++;
 601                     res += str[i];
 602                     continue;
 603                 }
 604                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 605                     i++;
 606                 if (i+1 < str.size() && str[i+1] == '<') //named
 607                 {
 608                     i++;
 609                     std::string gname;
 610                     bool term = false;
 611                     while (++i < str.size())
 612                     {
 613                         if (str[i] == '>') { term = true; break; }
 614                         if (!isalnum(str[i]))
 615                             throw mp::filter::FilterException
 616                                 ("Only alphanumeric chars allowed, found "
 617                                  " in '"
 618                                  + str
 619                                  + "' at "
 620                                  + boost::lexical_cast<std::string>(i));
 621                         gname += str[i];
 622                     }
 623                     if (!term)
 624                         throw mp::filter::FilterException
 625                             ("Unterminated group name '" + gname
 626                              + " in '" + str +"'");
 627                     group_index[gnum] = gname;
 628                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 629                             gname.c_str(), gnum);
 630                 }
 631             }
 632         }
 633         esc = false;
 634     }
 635     re = res;
 636 }
 637
 638 std::string yf::HttpRewrite::Replace::sub_vars(
 639     const std::map<std::string, std::string> & vars) const
 640 {
 641     std::string out;
 642     bool esc = false;
 643     const std::string & in = recipe;
 644     for (size_t i = 0; i < in.size(); ++i)
 645     {
 646         if (!esc && in[i] == '\\')
 647         {
 648             esc = true;
 649             continue;
 650         }
 651         if (!esc && in[i] == '$') //var
 652         {
 653             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 654             {
 655                 ++i;
 656                 std::string name;
 657                 bool term = false;
 658                 while (++i < in.size())
 659                 {
 660                     if (in[i] == '}') { term = true; break; }
 661                     name += in[i];
 662                 }
 663                 if (!term) throw mp::filter::FilterException
 664                     ("Unterminated var ref in '"+in+"' at "
 665                      + boost::lexical_cast<std::string>(i));
 666                 std::map<std::string, std::string>::const_iterator it
 667                     = vars.find(name);
 668                 if (it != vars.end())
 669                 {
 670                     out += it->second;
 671                 }
 672             }
 673             else
 674             {
 675                 throw mp::filter::FilterException
 676                     ("Malformed or trimmed var ref in '"
 677                      +in+"' at "+boost::lexical_cast<std::string>(i));
 678             }
 679             continue;
 680         }
 681         //passthru
 682         out += in[i];
 683         esc = false;
 684     }
 685     return out;
 686 }
 687
 688 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 689 {
 690 }
 691
 692 void yf::HttpRewrite::Content::parse(
 693     int verbose,
 694     std::string &content,
 695     std::map<std::string, std::string> &vars) const
 696 {
 697     if (type == "html")
 698     {
 699         HTMLParser parser;
 700         Event ev(this, vars);
 701
 702         parser.set_verbose(verbose);
 703
 704         parser.parse(ev, content.c_str());
 705         content = ev.result();
 706     }
 707     if (type == "quoted-literal")
 708     {
 709         quoted_literal(content, vars);
 710     }
 711 }
 712
 713 void yf::HttpRewrite::Content::quoted_literal(
 714     std::string &content,
 715     std::map<std::string, std::string> &vars) const
 716 {
 717     std::list<Within>::const_iterator it = within_list.begin();
 718     if (it != within_list.end())
 719         embed_quoted_literal(content, vars, it->rule, false);
 720 }
 721
 722 void yf::HttpRewrite::Content::configure(
 723     const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
 724 {
 725     for (; ptr; ptr = ptr->next)
 726     {
 727         if (ptr->type != XML_ELEMENT_NODE)
 728             continue;
 729         if (!strcmp((const char *) ptr->name, "within"))
 730         {
 731             static const char *names[7] =
 732                 { "header", "attr", "tag", "rule", "reqline", "type", 0 };
 733             std::string values[6];
 734             mp::xml::parse_attr(ptr, names, values);
 735             Within w;
 736             if (values[0].length() > 0)
 737                 w.header.assign(values[0], boost::regex_constants::icase);
 738             if (values[1].length() > 0)
 739                 w.attr.assign(values[1], boost::regex_constants::icase);
 740             if (values[2].length() > 0)
 741                 w.tag.assign(values[2], boost::regex_constants::icase);
 742
 743             std::vector<std::string> rulenames;
 744             boost::split(rulenames, values[3], boost::is_any_of(","));
 745             if (rulenames.size() == 0)
 746             {
 747                 throw mp::filter::FilterException
 748                     ("Empty rule in '" + values[3] +
 749                      "' in http_rewrite filter");
 750             }
 751             else if (rulenames.size() == 1)
 752             {
 753                 std::map<std::string,RulePtr>::const_iterator it =
 754                     rules.find(rulenames[0]);
 755                 if (it == rules.end())
 756                     throw mp::filter::FilterException
 757                         ("Reference to non-existing rule '" + rulenames[0] +
 758                          "' in http_rewrite filter");
 759                 w.rule = it->second;
 760
 761             }
 762             else
 763             {
 764                 RulePtr rule(new Rule);
 765                 size_t i;
 766                 for (i = 0; i < rulenames.size(); i++)
 767                 {
 768                     std::map<std::string,RulePtr>::const_iterator it =
 769                         rules.find(rulenames[i]);
 770                     if (it == rules.end())
 771                         throw mp::filter::FilterException
 772                             ("Reference to non-existing rule '" + rulenames[i] +
 773                              "' in http_rewrite filter");
 774                     RulePtr subRule = it->second;
 775                     std::list<Replace>::iterator rit =
 776                         subRule->replace_list.begin();
 777                     for (; rit != subRule->replace_list.end(); rit++)
 778                         rule->replace_list.push_back(*rit);
 779                 }
 780                 w.rule = rule;
 781             }
 782             w.reqline = values[4] == "1";
 783             w.type = values[5];
 784             if (w.type.empty() || w.type == "quoted-literal")
 785                 ;
 786             else
 787                 throw mp::filter::FilterException
 788                     ("within type must be quoted-literal or none in "
 789                      " in http_rewrite filter");
 790             within_list.push_back(w);
 791         }
 792     }
 793 }
 794
 795 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 796 {
 797     static const char *names[2] = { "verbose", 0 };
 798     std::string values[1];
 799     values[0] = "0";
 800     mp::xml::parse_attr(ptr, names, values);
 801
 802     phase.m_verbose = atoi(values[0].c_str());
 803
 804     std::map<std::string, RulePtr > rules;
 805     for (ptr = ptr->children; ptr; ptr = ptr->next)
 806     {
 807         if (ptr->type != XML_ELEMENT_NODE)
 808             continue;
 809         else if (!strcmp((const char *) ptr->name, "rule"))
 810         {
 811             static const char *names[2] = { "name", 0 };
 812             std::string values[1];
 813             values[0] = "default";
 814             mp::xml::parse_attr(ptr, names, values);
 815
 816             RulePtr rule(new Rule);
 817             for (xmlNode *p = ptr->children; p; p = p->next)
 818             {
 819                 if (p->type != XML_ELEMENT_NODE)
 820                     continue;
 821                 if (!strcmp((const char *) p->name, "rewrite"))
 822                 {
 823                     Replace replace;
 824                     std::string from;
 825                     const struct _xmlAttr *attr;
 826                     for (attr = p->properties; attr; attr = attr->next)
 827                     {
 828                         if (!strcmp((const char *) attr->name,  "from"))
 829                             from = mp::xml::get_text(attr->children);
 830                         else if (!strcmp((const char *) attr->name,  "to"))
 831                             replace.recipe = mp::xml::get_text(attr->children);
 832                         else
 833                             throw mp::filter::FilterException
 834                                 ("Bad attribute "
 835                                  + std::string((const char *) attr->name)
 836                                  + " in rewrite section of http_rewrite");
 837                     }
 838                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 839                             from.c_str(), replace.recipe.c_str());
 840                     if (!from.empty())
 841                     {
 842                         replace.parse_groups(from);
 843                         rule->replace_list.push_back(replace);
 844                     }
 845                 }
 846                 else
 847                     throw mp::filter::FilterException
 848                         ("Bad element "
 849                          + std::string((const char *) p->name)
 850                          + " in http_rewrite filter");
 851             }
 852             rules[values[0]] = rule;
 853         }
 854         else if (!strcmp((const char *) ptr->name, "content"))
 855         {
 856             static const char *names[3] =
 857                 { "type", "mime", 0 };
 858             std::string values[2];
 859             mp::xml::parse_attr(ptr, names, values);
 860             if (values[0].empty())
 861             {
 862                     throw mp::filter::FilterException
 863                         ("Missing attribute, type for for element "
 864                          + std::string((const char *) ptr->name)
 865                          + " in http_rewrite filter");
 866             }
 867             Content c;
 868
 869             c.type = values[0];
 870             if (!values[1].empty())
 871                 c.content_re.assign(values[1], boost::regex::icase);
 872             c.configure(ptr->children, rules);
 873             phase.content_list.push_back(c);
 874         }
 875         else
 876         {
 877             throw mp::filter::FilterException
 878                 ("Bad element "
 879                  + std::string((const char *) ptr->name)
 880                  + " in http_rewrite filter");
 881         }
 882     }
 883 }
 884
 885 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 886         const char *path)
 887 {
 888     for (ptr = ptr->children; ptr; ptr = ptr->next)
 889     {
 890         if (ptr->type != XML_ELEMENT_NODE)
 891             continue;
 892         else if (!strcmp((const char *) ptr->name, "request"))
 893         {
 894             configure_phase(ptr, *req_phase);
 895         }
 896         else if (!strcmp((const char *) ptr->name, "response"))
 897         {
 898             configure_phase(ptr, *res_phase);
 899         }
 900         else
 901         {
 902             throw mp::filter::FilterException
 903                 ("Bad element "
 904                  + std::string((const char *) ptr->name)
 905                  + " in http_rewrite1 filter");
 906         }
 907     }
 908 }
 909
 910 static mp::filter::Base* filter_creator()
 911 {
 912     return new mp::filter::HttpRewrite;
 913 }
 914
 915 extern "C" {
 916     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 917         0,
 918         "http_rewrite",
 919         filter_creator
 920     };
 921 }
 922
 923
 924 /*
 925  * Local variables:
 926  * c-basic-offset: 4
 927  * c-file-style: "Stroustrup"
 928  * indent-tabs-mode: nil
 929  * End:
 930  * vim: shiftwidth=4 tabstop=8 expandtab
 931  */
 932