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     std::list<Content>::const_iterator cit = content_list.begin();
 279     for (; cit != content_list.end(); cit++)
 280     {
 281         yaz_log(YLOG_LOG, "rewrite_body: content_type=%s type=%s",
 282                 content_type, cit->type.c_str());
 283         if (cit->type != "headers"
 284             && regex_match(content_type, cit->content_re))
 285             break;
 286     }
 287     if (cit == content_list.end())
 288         return;
 289
 290     int i;
 291     for (i = 0; i < *content_len; i++)
 292         if ((*content_buf)[i] == 0)
 293             return;  // binary content. skip
 294
 295     std::string content(*content_buf, *content_len);
 296     cit->parse(m_verbose, content, vars);
 297     *content_buf = odr_strdup(o, content.c_str());
 298     *content_len = strlen(*content_buf);
 299 }
 300
 301 yf::HttpRewrite::Event::Event(const Content *p,
 302                               std::map<std::string, std::string> & vars
 303     ) : m_content(p), m_vars(vars)
 304 {
 305     m_w = wrbuf_alloc();
 306 }
 307
 308 yf::HttpRewrite::Event::~Event()
 309 {
 310     wrbuf_destroy(m_w);
 311 }
 312
 313 const char *yf::HttpRewrite::Event::result()
 314 {
 315     return wrbuf_cstr(m_w);
 316 }
 317
 318 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 319 {
 320     wrbuf_putc(m_w, '<');
 321     wrbuf_write(m_w, tag, tag_len);
 322
 323     std::string t(tag, tag_len);
 324     std::list<Within>::const_iterator it = m_content->within_list.begin();
 325     for (; it != m_content->within_list.end(); it++)
 326     {
 327         if (!it->tag.empty() && regex_match(t, it->tag))
 328         {
 329             if (!it->attr.empty() && regex_match("#text", it->attr))
 330             {
 331                 s_within.push(it);
 332                 return;
 333             }
 334         }
 335     }
 336 }
 337
 338 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 339                                        int close_it)
 340 {
 341     if (close_it)
 342     {
 343         if (!s_within.empty())
 344         {
 345             std::list<Within>::const_iterator it = s_within.top();
 346             std::string t(tag, tag_len);
 347             if (regex_match(t, it->tag))
 348                 s_within.pop();
 349         }
 350     }
 351     if (close_it)
 352         wrbuf_putc(m_w, '/');
 353     wrbuf_putc(m_w, '>');
 354 }
 355
 356 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 357                                        const char *attr, int attr_len,
 358                                        const char *value, int val_len,
 359                                        const char *sep)
 360 {
 361     std::list<Within>::const_iterator it = m_content->within_list.begin();
 362     bool subst = false;
 363
 364     for (; it != m_content->within_list.end(); it++)
 365     {
 366         std::string t(tag, tag_len);
 367         if (it->tag.empty() || regex_match(t, it->tag))
 368         {
 369             std::string a(attr, attr_len);
 370             if (!it->attr.empty() && regex_match(a, it->attr))
 371                 subst = true;
 372         }
 373         if (subst)
 374             break;
 375     }
 376
 377     wrbuf_putc(m_w, ' ');
 378     wrbuf_write(m_w, attr, attr_len);
 379     if (value)
 380     {
 381         wrbuf_puts(m_w, "=");
 382         wrbuf_puts(m_w, sep);
 383
 384         std::string output;
 385         if (subst)
 386         {
 387             std::string s(value, val_len);
 388             it->exec(m_vars, s, true);
 389             wrbuf_puts(m_w, s.c_str());
 390         }
 391         else
 392             wrbuf_write(m_w, value, val_len);
 393         wrbuf_puts(m_w, sep);
 394     }
 395 }
 396
 397 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 398 {
 399     if (!s_within.empty())
 400     {
 401         std::list<Within>::const_iterator it = s_within.top();
 402         std::string t(tag, tag_len);
 403         if (regex_match(t, it->tag))
 404             s_within.pop();
 405     }
 406     wrbuf_puts(m_w, "</");
 407     wrbuf_write(m_w, tag, tag_len);
 408 }
 409
 410 void yf::HttpRewrite::Event::text(const char *value, int len)
 411 {
 412     std::list<Within>::const_iterator it = m_content->within_list.end();
 413     if (!s_within.empty())
 414         it = s_within.top();
 415     if (it != m_content->within_list.end())
 416     {
 417         std::string s(value, len);
 418         it->exec(m_vars, s, false);
 419         wrbuf_puts(m_w, s.c_str());
 420     }
 421     else
 422         wrbuf_write(m_w, value, len);
 423 }
 424
 425 static bool embed_quoted_literal(
 426     std::string &content,
 427     std::map<std::string, std::string> &vars,
 428     mp::filter::HttpRewrite::RulePtr ruleptr,
 429     bool html_context)
 430 {
 431     bool replace = false;
 432     std::string res;
 433     const char *cp = content.c_str();
 434     const char *cp0 = cp;
 435     while (*cp)
 436     {
 437         if (html_context && !strncmp(cp, "&quot;", 6))
 438         {
 439             cp += 6;
 440             res.append(cp0, cp - cp0);
 441             cp0 = cp;
 442             while (*cp)
 443             {
 444                 if (!strncmp(cp, "&quot;", 6))
 445                     break;
 446                 if (*cp == '\n')
 447                     break;
 448                 cp++;
 449             }
 450             if (!*cp)
 451                 break;
 452             std::string s(cp0, cp - cp0);
 453             if (ruleptr->test_patterns(vars, s, true))
 454                 replace = true;
 455             cp0 = cp;
 456             res.append(s);
 457         }
 458         else if (*cp == '"' || *cp == '\'')
 459         {
 460             int m = *cp;
 461             cp++;
 462             res.append(cp0, cp - cp0);
 463             cp0 = cp;
 464             while (*cp)
 465             {
 466                 if (cp[-1] != '\\' && *cp == m)
 467                     break;
 468                 if (*cp == '\n')
 469                     break;
 470                 cp++;
 471             }
 472             if (!*cp)
 473                 break;
 474             std::string s(cp0, cp - cp0);
 475             if (ruleptr->test_patterns(vars, s, true))
 476                 replace = true;
 477             cp0 = cp;
 478             res.append(s);
 479         }
 480         else if (*cp == '/' && cp[1] == '/')
 481         {
 482             while (cp[1] && cp[1] != '\n')
 483                 cp++;
 484         }
 485         cp++;
 486     }
 487     res.append(cp0, cp - cp0);
 488     content = res;
 489     return replace;
 490 }
 491
 492 bool yf::HttpRewrite::Within::exec(
 493     std::map<std::string, std::string> & vars,
 494     std::string & txt, bool anchor) const
 495 {
 496     if (type == "quoted-literal")
 497     {
 498         return embed_quoted_literal(txt, vars, rule, true);
 499     }
 500     else
 501     {
 502         return rule->test_patterns(vars, txt, anchor);
 503     }
 504 }
 505
 506 bool yf::HttpRewrite::Rule::test_patterns(
 507     std::map<std::string, std::string> & vars,
 508     std::string & txt, bool anchor)
 509 {
 510     bool replaces = false;
 511     bool first = anchor;
 512     std::string out;
 513     std::string::const_iterator start, end;
 514     start = txt.begin();
 515     end = txt.end();
 516     while (1)
 517     {
 518         std::list<Replace>::iterator bit = replace_list.end();
 519         {
 520             std::string::const_iterator best_pos = txt.end();
 521             std::list<Replace>::iterator it = replace_list.begin();
 522             for (; it != replace_list.end(); it++)
 523             {
 524                 if (it->start_anchor && !first)
 525                     continue;
 526                 if (regex_search(start, end, it->what, it->re))
 527                 {
 528                     if (it->what[0].first < best_pos)
 529                     {
 530                         best_pos = it->what[0].first;
 531                         bit = it;
 532                     }
 533                 }
 534             }
 535             if (bit == replace_list.end())
 536                 break;
 537         }
 538         first = false;
 539         replaces = true;
 540         size_t i;
 541         for (i = 1; i < bit->what.size(); ++i)
 542         {
 543             //check if the group is named
 544             std::map<int, std::string>::const_iterator git
 545                 = bit->group_index.find(i);
 546             if (git != bit->group_index.end())
 547             {   //it is
 548                 vars[git->second] = bit->what[i];
 549             }
 550
 551         }
 552         //prepare replacement string
 553         std::string rvalue = bit->sub_vars(vars);
 554         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 555                 bit->what.str(0).c_str(), rvalue.c_str());
 556         out.append(start, bit->what[0].first);
 557         out.append(rvalue);
 558         start = bit->what[0].second; //move search forward
 559     }
 560     out.append(start, end);
 561     txt = out;
 562     return replaces;
 563 }
 564
 565 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 566 {
 567     int gnum = 0;
 568     bool esc = false;
 569     const std::string &str = pattern;
 570     std::string res;
 571     start_anchor = str[0] == '^';
 572     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 573     for (size_t i = 0; i < str.size(); ++i)
 574     {
 575         res += str[i];
 576         if (!esc && str[i] == '\\')
 577         {
 578             esc = true;
 579             continue;
 580         }
 581         if (!esc && str[i] == '(') //group starts
 582         {
 583             gnum++;
 584             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 585             {
 586                 i++;
 587                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 588                 {
 589                     if (gnum > 0) gnum--;
 590                     res += str[i];
 591                     i++;
 592                     res += str[i];
 593                     continue;
 594                 }
 595                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 596                     i++;
 597                 if (i+1 < str.size() && str[i+1] == '<') //named
 598                 {
 599                     i++;
 600                     std::string gname;
 601                     bool term = false;
 602                     while (++i < str.size())
 603                     {
 604                         if (str[i] == '>') { term = true; break; }
 605                         if (!isalnum(str[i]))
 606                             throw mp::filter::FilterException
 607                                 ("Only alphanumeric chars allowed, found "
 608                                  " in '"
 609                                  + str
 610                                  + "' at "
 611                                  + boost::lexical_cast<std::string>(i));
 612                         gname += str[i];
 613                     }
 614                     if (!term)
 615                         throw mp::filter::FilterException
 616                             ("Unterminated group name '" + gname
 617                              + " in '" + str +"'");
 618                     group_index[gnum] = gname;
 619                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 620                             gname.c_str(), gnum);
 621                 }
 622             }
 623         }
 624         esc = false;
 625     }
 626     re = res;
 627 }
 628
 629 std::string yf::HttpRewrite::Replace::sub_vars(
 630     const std::map<std::string, std::string> & vars) const
 631 {
 632     std::string out;
 633     bool esc = false;
 634     const std::string & in = recipe;
 635     for (size_t i = 0; i < in.size(); ++i)
 636     {
 637         if (!esc && in[i] == '\\')
 638         {
 639             esc = true;
 640             continue;
 641         }
 642         if (!esc && in[i] == '$') //var
 643         {
 644             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 645             {
 646                 ++i;
 647                 std::string name;
 648                 bool term = false;
 649                 while (++i < in.size())
 650                 {
 651                     if (in[i] == '}') { term = true; break; }
 652                     name += in[i];
 653                 }
 654                 if (!term) throw mp::filter::FilterException
 655                     ("Unterminated var ref in '"+in+"' at "
 656                      + boost::lexical_cast<std::string>(i));
 657                 std::map<std::string, std::string>::const_iterator it
 658                     = vars.find(name);
 659                 if (it != vars.end())
 660                 {
 661                     out += it->second;
 662                 }
 663             }
 664             else
 665             {
 666                 throw mp::filter::FilterException
 667                     ("Malformed or trimmed var ref in '"
 668                      +in+"' at "+boost::lexical_cast<std::string>(i));
 669             }
 670             continue;
 671         }
 672         //passthru
 673         out += in[i];
 674         esc = false;
 675     }
 676     return out;
 677 }
 678
 679 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 680 {
 681 }
 682
 683 void yf::HttpRewrite::Content::parse(
 684     int verbose,
 685     std::string &content,
 686     std::map<std::string, std::string> &vars) const
 687 {
 688     if (type == "html")
 689     {
 690         HTMLParser parser;
 691         Event ev(this, vars);
 692
 693         parser.set_verbose(verbose);
 694
 695         parser.parse(ev, content.c_str());
 696         content = ev.result();
 697     }
 698     if (type == "quoted-literal")
 699     {
 700         quoted_literal(content, vars);
 701     }
 702 }
 703
 704 void yf::HttpRewrite::Content::quoted_literal(
 705     std::string &content,
 706     std::map<std::string, std::string> &vars) const
 707 {
 708     std::list<Within>::const_iterator it = within_list.begin();
 709     if (it != within_list.end())
 710         embed_quoted_literal(content, vars, it->rule, false);
 711 }
 712
 713 void yf::HttpRewrite::Content::configure(
 714     const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
 715 {
 716     for (; ptr; ptr = ptr->next)
 717     {
 718         if (ptr->type != XML_ELEMENT_NODE)
 719             continue;
 720         if (!strcmp((const char *) ptr->name, "within"))
 721         {
 722             static const char *names[7] =
 723                 { "header", "attr", "tag", "rule", "reqline", "type", 0 };
 724             std::string values[6];
 725             mp::xml::parse_attr(ptr, names, values);
 726             Within w;
 727             if (values[0].length() > 0)
 728                 w.header.assign(values[0], boost::regex_constants::icase);
 729             if (values[1].length() > 0)
 730                 w.attr.assign(values[1], boost::regex_constants::icase);
 731             if (values[2].length() > 0)
 732                 w.tag.assign(values[2], boost::regex_constants::icase);
 733
 734             std::vector<std::string> rulenames;
 735             boost::split(rulenames, values[3], boost::is_any_of(","));
 736             if (rulenames.size() == 0)
 737             {
 738                 throw mp::filter::FilterException
 739                     ("Empty rule in '" + values[3] +
 740                      "' in http_rewrite filter");
 741             }
 742             else if (rulenames.size() == 1)
 743             {
 744                 std::map<std::string,RulePtr>::const_iterator it =
 745                     rules.find(rulenames[0]);
 746                 if (it == rules.end())
 747                     throw mp::filter::FilterException
 748                         ("Reference to non-existing rule '" + rulenames[0] +
 749                          "' in http_rewrite filter");
 750                 w.rule = it->second;
 751
 752             }
 753             else
 754             {
 755                 RulePtr rule(new Rule);
 756                 size_t i;
 757                 for (i = 0; i < rulenames.size(); i++)
 758                 {
 759                     std::map<std::string,RulePtr>::const_iterator it =
 760                         rules.find(rulenames[i]);
 761                     if (it == rules.end())
 762                         throw mp::filter::FilterException
 763                             ("Reference to non-existing rule '" + rulenames[i] +
 764                              "' in http_rewrite filter");
 765                     RulePtr subRule = it->second;
 766                     std::list<Replace>::iterator rit =
 767                         subRule->replace_list.begin();
 768                     for (; rit != subRule->replace_list.end(); rit++)
 769                         rule->replace_list.push_back(*rit);
 770                 }
 771                 w.rule = rule;
 772             }
 773             w.reqline = values[4] == "1";
 774             w.type = values[5];
 775             if (w.type.empty() || w.type == "quoted-literal")
 776                 ;
 777             else
 778                 throw mp::filter::FilterException
 779                     ("within type must be quoted-literal or none in "
 780                      " in http_rewrite filter");
 781             within_list.push_back(w);
 782         }
 783     }
 784 }
 785
 786 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 787 {
 788     static const char *names[2] = { "verbose", 0 };
 789     std::string values[1];
 790     values[0] = "0";
 791     mp::xml::parse_attr(ptr, names, values);
 792
 793     phase.m_verbose = atoi(values[0].c_str());
 794
 795     std::map<std::string, RulePtr > rules;
 796     for (ptr = ptr->children; ptr; ptr = ptr->next)
 797     {
 798         if (ptr->type != XML_ELEMENT_NODE)
 799             continue;
 800         else if (!strcmp((const char *) ptr->name, "rule"))
 801         {
 802             static const char *names[2] = { "name", 0 };
 803             std::string values[1];
 804             values[0] = "default";
 805             mp::xml::parse_attr(ptr, names, values);
 806
 807             RulePtr rule(new Rule);
 808             for (xmlNode *p = ptr->children; p; p = p->next)
 809             {
 810                 if (p->type != XML_ELEMENT_NODE)
 811                     continue;
 812                 if (!strcmp((const char *) p->name, "rewrite"))
 813                 {
 814                     Replace replace;
 815                     std::string from;
 816                     const struct _xmlAttr *attr;
 817                     for (attr = p->properties; attr; attr = attr->next)
 818                     {
 819                         if (!strcmp((const char *) attr->name,  "from"))
 820                             from = mp::xml::get_text(attr->children);
 821                         else if (!strcmp((const char *) attr->name,  "to"))
 822                             replace.recipe = mp::xml::get_text(attr->children);
 823                         else
 824                             throw mp::filter::FilterException
 825                                 ("Bad attribute "
 826                                  + std::string((const char *) attr->name)
 827                                  + " in rewrite section of http_rewrite");
 828                     }
 829                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 830                             from.c_str(), replace.recipe.c_str());
 831                     if (!from.empty())
 832                     {
 833                         replace.parse_groups(from);
 834                         rule->replace_list.push_back(replace);
 835                     }
 836                 }
 837                 else
 838                     throw mp::filter::FilterException
 839                         ("Bad element "
 840                          + std::string((const char *) p->name)
 841                          + " in http_rewrite filter");
 842             }
 843             rules[values[0]] = rule;
 844         }
 845         else if (!strcmp((const char *) ptr->name, "content"))
 846         {
 847             static const char *names[3] =
 848                 { "type", "mime", 0 };
 849             std::string values[2];
 850             mp::xml::parse_attr(ptr, names, values);
 851             if (values[0].empty())
 852             {
 853                     throw mp::filter::FilterException
 854                         ("Missing attribute, type for for element "
 855                          + std::string((const char *) ptr->name)
 856                          + " in http_rewrite filter");
 857             }
 858             Content c;
 859
 860             c.type = values[0];
 861             if (!values[1].empty())
 862                 c.content_re.assign(values[1], boost::regex::icase);
 863             c.configure(ptr->children, rules);
 864             phase.content_list.push_back(c);
 865         }
 866         else
 867         {
 868             throw mp::filter::FilterException
 869                 ("Bad element "
 870                  + std::string((const char *) ptr->name)
 871                  + " in http_rewrite filter");
 872         }
 873     }
 874 }
 875
 876 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 877         const char *path)
 878 {
 879     for (ptr = ptr->children; ptr; ptr = ptr->next)
 880     {
 881         if (ptr->type != XML_ELEMENT_NODE)
 882             continue;
 883         else if (!strcmp((const char *) ptr->name, "request"))
 884         {
 885             configure_phase(ptr, *req_phase);
 886         }
 887         else if (!strcmp((const char *) ptr->name, "response"))
 888         {
 889             configure_phase(ptr, *res_phase);
 890         }
 891         else
 892         {
 893             throw mp::filter::FilterException
 894                 ("Bad element "
 895                  + std::string((const char *) ptr->name)
 896                  + " in http_rewrite1 filter");
 897         }
 898     }
 899 }
 900
 901 static mp::filter::Base* filter_creator()
 902 {
 903     return new mp::filter::HttpRewrite;
 904 }
 905
 906 extern "C" {
 907     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 908         0,
 909         "http_rewrite",
 910         filter_creator
 911     };
 912 }
 913
 914
 915 /*
 916  * Local variables:
 917  * c-basic-offset: 4
 918  * c-file-style: "Stroustrup"
 919  * indent-tabs-mode: nil
 920  * End:
 921  * vim: shiftwidth=4 tabstop=8 expandtab
 922  */
 923