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         yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
 289                 content_type );
 290         return;
 291     }
 292
 293     int i;
 294     for (i = 0; i < *content_len; i++)
 295         if ((*content_buf)[i] == 0) {
 296             yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
 297             return;  // binary content. skip
 298         }
 299
 300     std::string content(*content_buf, *content_len);
 301     cit->parse(m_verbose, content, vars);
 302     *content_buf = odr_strdup(o, content.c_str());
 303     *content_len = strlen(*content_buf);
 304 }
 305
 306 yf::HttpRewrite::Event::Event(const Content *p,
 307                               std::map<std::string, std::string> & vars
 308     ) : m_content(p), m_vars(vars)
 309 {
 310     m_w = wrbuf_alloc();
 311 }
 312
 313 yf::HttpRewrite::Event::~Event()
 314 {
 315     wrbuf_destroy(m_w);
 316 }
 317
 318 const char *yf::HttpRewrite::Event::result()
 319 {
 320     return wrbuf_cstr(m_w);
 321 }
 322
 323 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 324 {
 325     wrbuf_putc(m_w, '<');
 326     wrbuf_write(m_w, tag, tag_len);
 327
 328     std::string t(tag, tag_len);
 329     std::list<Within>::const_iterator it = m_content->within_list.begin();
 330     for (; it != m_content->within_list.end(); it++)
 331     {
 332         if (!it->tag.empty() && regex_match(t, it->tag))
 333         {
 334             if (!it->attr.empty() && regex_match("#text", it->attr))
 335             {
 336                 s_within.push(it);
 337                 return;
 338             }
 339         }
 340     }
 341 }
 342
 343 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 344                                        int close_it)
 345 {
 346     if (close_it)
 347     {
 348         if (!s_within.empty())
 349         {
 350             std::list<Within>::const_iterator it = s_within.top();
 351             std::string t(tag, tag_len);
 352             if (regex_match(t, it->tag))
 353                 s_within.pop();
 354         }
 355     }
 356     if (close_it)
 357         wrbuf_putc(m_w, '/');
 358     wrbuf_putc(m_w, '>');
 359 }
 360
 361 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 362                                        const char *attr, int attr_len,
 363                                        const char *value, int val_len,
 364                                        const char *sep)
 365 {
 366     std::list<Within>::const_iterator it = m_content->within_list.begin();
 367     bool subst = false;
 368
 369     for (; it != m_content->within_list.end(); it++)
 370     {
 371         std::string t(tag, tag_len);
 372         if (it->tag.empty() || regex_match(t, it->tag))
 373         {
 374             std::string a(attr, attr_len);
 375             if (!it->attr.empty() && regex_match(a, it->attr))
 376                 subst = true;
 377         }
 378         if (subst)
 379             break;
 380     }
 381
 382     wrbuf_putc(m_w, ' ');
 383     wrbuf_write(m_w, attr, attr_len);
 384     if (value)
 385     {
 386         wrbuf_puts(m_w, "=");
 387         wrbuf_puts(m_w, sep);
 388
 389         std::string output;
 390         if (subst)
 391         {
 392             std::string s(value, val_len);
 393             it->exec(m_vars, s, true);
 394             wrbuf_puts(m_w, s.c_str());
 395         }
 396         else
 397             wrbuf_write(m_w, value, val_len);
 398         wrbuf_puts(m_w, sep);
 399     }
 400 }
 401
 402 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 403 {
 404     if (!s_within.empty())
 405     {
 406         std::list<Within>::const_iterator it = s_within.top();
 407         std::string t(tag, tag_len);
 408         if (regex_match(t, it->tag))
 409             s_within.pop();
 410     }
 411     wrbuf_puts(m_w, "</");
 412     wrbuf_write(m_w, tag, tag_len);
 413 }
 414
 415 void yf::HttpRewrite::Event::text(const char *value, int len)
 416 {
 417     std::list<Within>::const_iterator it = m_content->within_list.end();
 418     if (!s_within.empty())
 419         it = s_within.top();
 420     if (it != m_content->within_list.end())
 421     {
 422         std::string s(value, len);
 423         it->exec(m_vars, s, false);
 424         wrbuf_puts(m_w, s.c_str());
 425     }
 426     else
 427         wrbuf_write(m_w, value, len);
 428 }
 429
 430 static bool embed_quoted_literal(
 431     std::string &content,
 432     std::map<std::string, std::string> &vars,
 433     mp::filter::HttpRewrite::RulePtr ruleptr,
 434     bool html_context)
 435 {
 436     bool replace = false;
 437     std::string res;
 438     const char *cp = content.c_str();
 439     const char *cp0 = cp;
 440     while (*cp)
 441     {
 442         if (html_context && !strncmp(cp, "&quot;", 6))
 443         {
 444             cp += 6;
 445             res.append(cp0, cp - cp0);
 446             cp0 = cp;
 447             while (*cp)
 448             {
 449                 if (!strncmp(cp, "&quot;", 6))
 450                     break;
 451                 if (*cp == '\n')
 452                     break;
 453                 cp++;
 454             }
 455             if (!*cp)
 456                 break;
 457             std::string s(cp0, cp - cp0);
 458             if (ruleptr->test_patterns(vars, s, true))
 459                 replace = true;
 460             cp0 = cp;
 461             res.append(s);
 462         }
 463         else if (*cp == '"' || *cp == '\'')
 464         {
 465             int m = *cp;
 466             cp++;
 467             res.append(cp0, cp - cp0);
 468             cp0 = cp;
 469             while (*cp)
 470             {
 471                 if (cp[-1] != '\\' && *cp == m)
 472                     break;
 473                 if (*cp == '\n')
 474                     break;
 475                 cp++;
 476             }
 477             if (!*cp)
 478                 break;
 479             std::string s(cp0, cp - cp0);
 480             if (ruleptr->test_patterns(vars, s, true))
 481                 replace = true;
 482             cp0 = cp;
 483             res.append(s);
 484         }
 485         else if (*cp == '/' && cp[1] == '/')
 486         {
 487             while (cp[1] && cp[1] != '\n')
 488                 cp++;
 489         }
 490         cp++;
 491     }
 492     res.append(cp0, cp - cp0);
 493     content = res;
 494     return replace;
 495 }
 496
 497 bool yf::HttpRewrite::Within::exec(
 498     std::map<std::string, std::string> & vars,
 499     std::string & txt, bool anchor) const
 500 {
 501     if (type == "quoted-literal")
 502     {
 503         return embed_quoted_literal(txt, vars, rule, true);
 504     }
 505     else
 506     {
 507         return rule->test_patterns(vars, txt, anchor);
 508     }
 509 }
 510
 511 bool yf::HttpRewrite::Rule::test_patterns(
 512     std::map<std::string, std::string> & vars,
 513     std::string & txt, bool anchor)
 514 {
 515     bool replaces = false;
 516     bool first = anchor;
 517     std::string out;
 518     std::string::const_iterator start, end;
 519     start = txt.begin();
 520     end = txt.end();
 521     while (1)
 522     {
 523         std::list<Replace>::iterator bit = replace_list.end();
 524         {
 525             std::string::const_iterator best_pos = txt.end();
 526             std::list<Replace>::iterator it = replace_list.begin();
 527             for (; it != replace_list.end(); it++)
 528             {
 529                 if (it->start_anchor && !first)
 530                     continue;
 531                 if (regex_search(start, end, it->what, it->re))
 532                 {
 533                     if (it->what[0].first < best_pos)
 534                     {
 535                         best_pos = it->what[0].first;
 536                         bit = it;
 537                     }
 538                 }
 539             }
 540             if (bit == replace_list.end())
 541                 break;
 542         }
 543         first = false;
 544         replaces = true;
 545         size_t i;
 546         for (i = 1; i < bit->what.size(); ++i)
 547         {
 548             //check if the group is named
 549             std::map<int, std::string>::const_iterator git
 550                 = bit->group_index.find(i);
 551             if (git != bit->group_index.end())
 552             {   //it is
 553                 vars[git->second] = bit->what[i];
 554             }
 555
 556         }
 557         //prepare replacement string
 558         std::string rvalue = bit->sub_vars(vars);
 559         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 560                 bit->what.str(0).c_str(), rvalue.c_str());
 561         out.append(start, bit->what[0].first);
 562         out.append(rvalue);
 563         start = bit->what[0].second; //move search forward
 564     }
 565     out.append(start, end);
 566     txt = out;
 567     return replaces;
 568 }
 569
 570 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 571 {
 572     int gnum = 0;
 573     bool esc = false;
 574     const std::string &str = pattern;
 575     std::string res;
 576     start_anchor = str[0] == '^';
 577     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 578     for (size_t i = 0; i < str.size(); ++i)
 579     {
 580         res += str[i];
 581         if (!esc && str[i] == '\\')
 582         {
 583             esc = true;
 584             continue;
 585         }
 586         if (!esc && str[i] == '(') //group starts
 587         {
 588             gnum++;
 589             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 590             {
 591                 i++;
 592                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 593                 {
 594                     if (gnum > 0) gnum--;
 595                     res += str[i];
 596                     i++;
 597                     res += str[i];
 598                     continue;
 599                 }
 600                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 601                     i++;
 602                 if (i+1 < str.size() && str[i+1] == '<') //named
 603                 {
 604                     i++;
 605                     std::string gname;
 606                     bool term = false;
 607                     while (++i < str.size())
 608                     {
 609                         if (str[i] == '>') { term = true; break; }
 610                         if (!isalnum(str[i]))
 611                             throw mp::filter::FilterException
 612                                 ("Only alphanumeric chars allowed, found "
 613                                  " in '"
 614                                  + str
 615                                  + "' at "
 616                                  + boost::lexical_cast<std::string>(i));
 617                         gname += str[i];
 618                     }
 619                     if (!term)
 620                         throw mp::filter::FilterException
 621                             ("Unterminated group name '" + gname
 622                              + " in '" + str +"'");
 623                     group_index[gnum] = gname;
 624                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 625                             gname.c_str(), gnum);
 626                 }
 627             }
 628         }
 629         esc = false;
 630     }
 631     re = res;
 632 }
 633
 634 std::string yf::HttpRewrite::Replace::sub_vars(
 635     const std::map<std::string, std::string> & vars) const
 636 {
 637     std::string out;
 638     bool esc = false;
 639     const std::string & in = recipe;
 640     for (size_t i = 0; i < in.size(); ++i)
 641     {
 642         if (!esc && in[i] == '\\')
 643         {
 644             esc = true;
 645             continue;
 646         }
 647         if (!esc && in[i] == '$') //var
 648         {
 649             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 650             {
 651                 ++i;
 652                 std::string name;
 653                 bool term = false;
 654                 while (++i < in.size())
 655                 {
 656                     if (in[i] == '}') { term = true; break; }
 657                     name += in[i];
 658                 }
 659                 if (!term) throw mp::filter::FilterException
 660                     ("Unterminated var ref in '"+in+"' at "
 661                      + boost::lexical_cast<std::string>(i));
 662                 std::map<std::string, std::string>::const_iterator it
 663                     = vars.find(name);
 664                 if (it != vars.end())
 665                 {
 666                     out += it->second;
 667                 }
 668             }
 669             else
 670             {
 671                 throw mp::filter::FilterException
 672                     ("Malformed or trimmed var ref in '"
 673                      +in+"' at "+boost::lexical_cast<std::string>(i));
 674             }
 675             continue;
 676         }
 677         //passthru
 678         out += in[i];
 679         esc = false;
 680     }
 681     return out;
 682 }
 683
 684 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 685 {
 686 }
 687
 688 void yf::HttpRewrite::Content::parse(
 689     int verbose,
 690     std::string &content,
 691     std::map<std::string, std::string> &vars) const
 692 {
 693     if (type == "html")
 694     {
 695         HTMLParser parser;
 696         Event ev(this, vars);
 697
 698         parser.set_verbose(verbose);
 699
 700         parser.parse(ev, content.c_str());
 701         content = ev.result();
 702     }
 703     if (type == "quoted-literal")
 704     {
 705         quoted_literal(content, vars);
 706     }
 707 }
 708
 709 void yf::HttpRewrite::Content::quoted_literal(
 710     std::string &content,
 711     std::map<std::string, std::string> &vars) const
 712 {
 713     std::list<Within>::const_iterator it = within_list.begin();
 714     if (it != within_list.end())
 715         embed_quoted_literal(content, vars, it->rule, false);
 716 }
 717
 718 void yf::HttpRewrite::Content::configure(
 719     const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
 720 {
 721     for (; ptr; ptr = ptr->next)
 722     {
 723         if (ptr->type != XML_ELEMENT_NODE)
 724             continue;
 725         if (!strcmp((const char *) ptr->name, "within"))
 726         {
 727             static const char *names[7] =
 728                 { "header", "attr", "tag", "rule", "reqline", "type", 0 };
 729             std::string values[6];
 730             mp::xml::parse_attr(ptr, names, values);
 731             Within w;
 732             if (values[0].length() > 0)
 733                 w.header.assign(values[0], boost::regex_constants::icase);
 734             if (values[1].length() > 0)
 735                 w.attr.assign(values[1], boost::regex_constants::icase);
 736             if (values[2].length() > 0)
 737                 w.tag.assign(values[2], boost::regex_constants::icase);
 738
 739             std::vector<std::string> rulenames;
 740             boost::split(rulenames, values[3], boost::is_any_of(","));
 741             if (rulenames.size() == 0)
 742             {
 743                 throw mp::filter::FilterException
 744                     ("Empty rule in '" + values[3] +
 745                      "' in http_rewrite filter");
 746             }
 747             else if (rulenames.size() == 1)
 748             {
 749                 std::map<std::string,RulePtr>::const_iterator it =
 750                     rules.find(rulenames[0]);
 751                 if (it == rules.end())
 752                     throw mp::filter::FilterException
 753                         ("Reference to non-existing rule '" + rulenames[0] +
 754                          "' in http_rewrite filter");
 755                 w.rule = it->second;
 756
 757             }
 758             else
 759             {
 760                 RulePtr rule(new Rule);
 761                 size_t i;
 762                 for (i = 0; i < rulenames.size(); i++)
 763                 {
 764                     std::map<std::string,RulePtr>::const_iterator it =
 765                         rules.find(rulenames[i]);
 766                     if (it == rules.end())
 767                         throw mp::filter::FilterException
 768                             ("Reference to non-existing rule '" + rulenames[i] +
 769                              "' in http_rewrite filter");
 770                     RulePtr subRule = it->second;
 771                     std::list<Replace>::iterator rit =
 772                         subRule->replace_list.begin();
 773                     for (; rit != subRule->replace_list.end(); rit++)
 774                         rule->replace_list.push_back(*rit);
 775                 }
 776                 w.rule = rule;
 777             }
 778             w.reqline = values[4] == "1";
 779             w.type = values[5];
 780             if (w.type.empty() || w.type == "quoted-literal")
 781                 ;
 782             else
 783                 throw mp::filter::FilterException
 784                     ("within type must be quoted-literal or none in "
 785                      " in http_rewrite filter");
 786             within_list.push_back(w);
 787         }
 788     }
 789 }
 790
 791 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 792 {
 793     static const char *names[2] = { "verbose", 0 };
 794     std::string values[1];
 795     values[0] = "0";
 796     mp::xml::parse_attr(ptr, names, values);
 797
 798     phase.m_verbose = atoi(values[0].c_str());
 799
 800     std::map<std::string, RulePtr > rules;
 801     for (ptr = ptr->children; ptr; ptr = ptr->next)
 802     {
 803         if (ptr->type != XML_ELEMENT_NODE)
 804             continue;
 805         else if (!strcmp((const char *) ptr->name, "rule"))
 806         {
 807             static const char *names[2] = { "name", 0 };
 808             std::string values[1];
 809             values[0] = "default";
 810             mp::xml::parse_attr(ptr, names, values);
 811
 812             RulePtr rule(new Rule);
 813             for (xmlNode *p = ptr->children; p; p = p->next)
 814             {
 815                 if (p->type != XML_ELEMENT_NODE)
 816                     continue;
 817                 if (!strcmp((const char *) p->name, "rewrite"))
 818                 {
 819                     Replace replace;
 820                     std::string from;
 821                     const struct _xmlAttr *attr;
 822                     for (attr = p->properties; attr; attr = attr->next)
 823                     {
 824                         if (!strcmp((const char *) attr->name,  "from"))
 825                             from = mp::xml::get_text(attr->children);
 826                         else if (!strcmp((const char *) attr->name,  "to"))
 827                             replace.recipe = mp::xml::get_text(attr->children);
 828                         else
 829                             throw mp::filter::FilterException
 830                                 ("Bad attribute "
 831                                  + std::string((const char *) attr->name)
 832                                  + " in rewrite section of http_rewrite");
 833                     }
 834                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 835                             from.c_str(), replace.recipe.c_str());
 836                     if (!from.empty())
 837                     {
 838                         replace.parse_groups(from);
 839                         rule->replace_list.push_back(replace);
 840                     }
 841                 }
 842                 else
 843                     throw mp::filter::FilterException
 844                         ("Bad element "
 845                          + std::string((const char *) p->name)
 846                          + " in http_rewrite filter");
 847             }
 848             rules[values[0]] = rule;
 849         }
 850         else if (!strcmp((const char *) ptr->name, "content"))
 851         {
 852             static const char *names[3] =
 853                 { "type", "mime", 0 };
 854             std::string values[2];
 855             mp::xml::parse_attr(ptr, names, values);
 856             if (values[0].empty())
 857             {
 858                     throw mp::filter::FilterException
 859                         ("Missing attribute, type for for element "
 860                          + std::string((const char *) ptr->name)
 861                          + " in http_rewrite filter");
 862             }
 863             Content c;
 864
 865             c.type = values[0];
 866             if (!values[1].empty())
 867                 c.content_re.assign(values[1], boost::regex::icase);
 868             c.configure(ptr->children, rules);
 869             phase.content_list.push_back(c);
 870         }
 871         else
 872         {
 873             throw mp::filter::FilterException
 874                 ("Bad element "
 875                  + std::string((const char *) ptr->name)
 876                  + " in http_rewrite filter");
 877         }
 878     }
 879 }
 880
 881 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 882         const char *path)
 883 {
 884     for (ptr = ptr->children; ptr; ptr = ptr->next)
 885     {
 886         if (ptr->type != XML_ELEMENT_NODE)
 887             continue;
 888         else if (!strcmp((const char *) ptr->name, "request"))
 889         {
 890             configure_phase(ptr, *req_phase);
 891         }
 892         else if (!strcmp((const char *) ptr->name, "response"))
 893         {
 894             configure_phase(ptr, *res_phase);
 895         }
 896         else
 897         {
 898             throw mp::filter::FilterException
 899                 ("Bad element "
 900                  + std::string((const char *) ptr->name)
 901                  + " in http_rewrite1 filter");
 902         }
 903     }
 904 }
 905
 906 static mp::filter::Base* filter_creator()
 907 {
 908     return new mp::filter::HttpRewrite;
 909 }
 910
 911 extern "C" {
 912     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 913         0,
 914         "http_rewrite",
 915         filter_creator
 916     };
 917 }
 918
 919
 920 /*
 921  * Local variables:
 922  * c-basic-offset: 4
 923  * c-file-style: "Stroustrup"
 924  * indent-tabs-mode: nil
 925  * End:
 926  * vim: shiftwidth=4 tabstop=8 expandtab
 927  */
 928