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             std::string header;
  63             std::string 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.length() > 0 &&
 230                 yaz_strcasecmp(it->header.c_str(), header->name) == 0)
 231             {
 232                 std::string sheader(header->name);
 233                 sheader += ": ";
 234                 sheader += header->value;
 235
 236                 if (it->exec(vars, sheader, true))
 237                 {
 238                     size_t pos = sheader.find(": ");
 239                     if (pos == std::string::npos)
 240                     {
 241                         yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 242                         continue;
 243                     }
 244                     header->name = odr_strdup(
 245                         o, sheader.substr(0, pos).c_str());
 246                     header->value = odr_strdup(
 247                         o, sheader.substr(pos + 2, std::string::npos).c_str());
 248                 }
 249             }
 250         }
 251     }
 252 }
 253
 254 void yf::HttpRewrite::Phase::rewrite_body(
 255     mp::odr &o,
 256     const char *content_type,
 257     char **content_buf,
 258     int *content_len,
 259     std::map<std::string, std::string> & vars) const
 260 {
 261     if (*content_len == 0)
 262         return;
 263     std::list<Content>::const_iterator cit = content_list.begin();
 264     for (; cit != content_list.end(); cit++)
 265     {
 266         yaz_log(YLOG_LOG, "rewrite_body: content_type=%s type=%s",
 267                 content_type, cit->type.c_str());
 268         if (cit->type != "headers"
 269             && regex_match(content_type, cit->content_re))
 270             break;
 271     }
 272     if (cit == content_list.end())
 273         return;
 274
 275     int i;
 276     for (i = 0; i < *content_len; i++)
 277         if ((*content_buf)[i] == 0)
 278             return;  // binary content. skip
 279
 280     std::string content(*content_buf, *content_len);
 281     cit->parse(m_verbose, content, vars);
 282     *content_buf = odr_strdup(o, content.c_str());
 283     *content_len = strlen(*content_buf);
 284 }
 285
 286 yf::HttpRewrite::Event::Event(const Content *p,
 287                               std::map<std::string, std::string> & vars
 288     ) : m_content(p), m_vars(vars)
 289 {
 290     m_w = wrbuf_alloc();
 291 }
 292
 293 yf::HttpRewrite::Event::~Event()
 294 {
 295     wrbuf_destroy(m_w);
 296 }
 297
 298 const char *yf::HttpRewrite::Event::result()
 299 {
 300     return wrbuf_cstr(m_w);
 301 }
 302
 303 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 304 {
 305     wrbuf_putc(m_w, '<');
 306     wrbuf_write(m_w, tag, tag_len);
 307
 308     std::string t(tag, tag_len);
 309     std::list<Within>::const_iterator it = m_content->within_list.begin();
 310     for (; it != m_content->within_list.end(); it++)
 311     {
 312         if (!it->tag.empty() && regex_match(t, it->tag))
 313         {
 314             std::vector<std::string> attr;
 315             boost::split(attr, it->attr, boost::is_any_of(","));
 316             size_t i;
 317             for (i = 0; i < attr.size(); i++)
 318             {
 319                 if (attr[i].compare("#text") == 0)
 320                 {
 321                     s_within.push(it);
 322                     return;
 323                 }
 324             }
 325         }
 326     }
 327 }
 328
 329 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 330                                        int close_it)
 331 {
 332     if (close_it)
 333     {
 334         if (!s_within.empty())
 335         {
 336             std::list<Within>::const_iterator it = s_within.top();
 337             std::string t(tag, tag_len);
 338             if (regex_match(t, it->tag))
 339                 s_within.pop();
 340         }
 341     }
 342     if (close_it)
 343         wrbuf_putc(m_w, '/');
 344     wrbuf_putc(m_w, '>');
 345 }
 346
 347 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 348                                        const char *attr, int attr_len,
 349                                        const char *value, int val_len,
 350                                        const char *sep)
 351 {
 352     std::list<Within>::const_iterator it = m_content->within_list.begin();
 353     bool subst = false;
 354
 355     for (; it != m_content->within_list.end(); it++)
 356     {
 357         std::string t(tag, tag_len);
 358         if (it->tag.empty() || regex_match(t, it->tag))
 359         {
 360             std::string a(attr, attr_len);
 361             std::vector<std::string> attr;
 362             boost::split(attr, it->attr, boost::is_any_of(","));
 363             size_t i;
 364             for (i = 0; i < attr.size(); i++)
 365             {
 366                 if (attr[i].compare("#text") &&
 367                     yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
 368                     subst = true;
 369             }
 370         }
 371         if (subst)
 372             break;
 373     }
 374
 375     wrbuf_putc(m_w, ' ');
 376     wrbuf_write(m_w, attr, attr_len);
 377     if (value)
 378     {
 379         wrbuf_puts(m_w, "=");
 380         wrbuf_puts(m_w, sep);
 381
 382         std::string output;
 383         if (subst)
 384         {
 385             std::string s(value, val_len);
 386             it->exec(m_vars, s, true);
 387             wrbuf_puts(m_w, s.c_str());
 388         }
 389         else
 390             wrbuf_write(m_w, value, val_len);
 391         wrbuf_puts(m_w, sep);
 392     }
 393 }
 394
 395 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 396 {
 397     if (!s_within.empty())
 398     {
 399         std::list<Within>::const_iterator it = s_within.top();
 400         std::string t(tag, tag_len);
 401         if (regex_match(t, it->tag))
 402             s_within.pop();
 403     }
 404     wrbuf_puts(m_w, "</");
 405     wrbuf_write(m_w, tag, tag_len);
 406 }
 407
 408 void yf::HttpRewrite::Event::text(const char *value, int len)
 409 {
 410     std::list<Within>::const_iterator it = m_content->within_list.end();
 411     if (!s_within.empty())
 412         it = s_within.top();
 413     if (it != m_content->within_list.end())
 414     {
 415         std::string s(value, len);
 416         it->exec(m_vars, s, false);
 417         wrbuf_puts(m_w, s.c_str());
 418     }
 419     else
 420         wrbuf_write(m_w, value, len);
 421 }
 422
 423 static bool embed_quoted_literal(
 424     std::string &content,
 425     std::map<std::string, std::string> &vars,
 426     mp::filter::HttpRewrite::RulePtr ruleptr,
 427     bool html_context)
 428 {
 429     bool replace = false;
 430     std::string res;
 431     const char *cp = content.c_str();
 432     const char *cp0 = cp;
 433     while (*cp)
 434     {
 435         if (html_context && !strncmp(cp, "&quot;", 6))
 436         {
 437             cp += 6;
 438             res.append(cp0, cp - cp0);
 439             cp0 = cp;
 440             while (*cp)
 441             {
 442                 if (!strncmp(cp, "&quot;", 6))
 443                     break;
 444                 if (*cp == '\n')
 445                     break;
 446                 cp++;
 447             }
 448             if (!*cp)
 449                 break;
 450             std::string s(cp0, cp - cp0);
 451             if (ruleptr->test_patterns(vars, s, true))
 452                 replace = true;
 453             cp0 = cp;
 454             res.append(s);
 455         }
 456         else if (*cp == '"' || *cp == '\'')
 457         {
 458             int m = *cp;
 459             cp++;
 460             res.append(cp0, cp - cp0);
 461             cp0 = cp;
 462             while (*cp)
 463             {
 464                 if (cp[-1] != '\\' && *cp == m)
 465                     break;
 466                 if (*cp == '\n')
 467                     break;
 468                 cp++;
 469             }
 470             if (!*cp)
 471                 break;
 472             std::string s(cp0, cp - cp0);
 473             if (ruleptr->test_patterns(vars, s, true))
 474                 replace = true;
 475             cp0 = cp;
 476             res.append(s);
 477         }
 478         else if (*cp == '/' && cp[1] == '/')
 479         {
 480             while (cp[1] && cp[1] != '\n')
 481                 cp++;
 482         }
 483         cp++;
 484     }
 485     res.append(cp0, cp - cp0);
 486     content = res;
 487     return replace;
 488 }
 489
 490 bool yf::HttpRewrite::Within::exec(
 491     std::map<std::string, std::string> & vars,
 492     std::string & txt, bool anchor) const
 493 {
 494     if (type == "quoted-literal")
 495     {
 496         return embed_quoted_literal(txt, vars, rule, true);
 497     }
 498     else
 499     {
 500         return rule->test_patterns(vars, txt, anchor);
 501     }
 502 }
 503
 504 bool yf::HttpRewrite::Rule::test_patterns(
 505     std::map<std::string, std::string> & vars,
 506     std::string & txt, bool anchor)
 507 {
 508     bool replaces = false;
 509     bool first = anchor;
 510     std::string out;
 511     std::string::const_iterator start, end;
 512     start = txt.begin();
 513     end = txt.end();
 514     while (1)
 515     {
 516         std::list<Replace>::iterator bit = replace_list.end();
 517         {
 518             std::string::const_iterator best_pos = txt.end();
 519             std::list<Replace>::iterator it = replace_list.begin();
 520             for (; it != replace_list.end(); it++)
 521             {
 522                 if (it->start_anchor && !first)
 523                     continue;
 524                 if (regex_search(start, end, it->what, it->re))
 525                 {
 526                     if (it->what[0].first < best_pos)
 527                     {
 528                         best_pos = it->what[0].first;
 529                         bit = it;
 530                     }
 531                 }
 532             }
 533             if (bit == replace_list.end())
 534                 break;
 535         }
 536         first = false;
 537         replaces = true;
 538         size_t i;
 539         for (i = 1; i < bit->what.size(); ++i)
 540         {
 541             //check if the group is named
 542             std::map<int, std::string>::const_iterator git
 543                 = bit->group_index.find(i);
 544             if (git != bit->group_index.end())
 545             {   //it is
 546                 vars[git->second] = bit->what[i];
 547             }
 548
 549         }
 550         //prepare replacement string
 551         std::string rvalue = bit->sub_vars(vars);
 552         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 553                 bit->what.str(0).c_str(), rvalue.c_str());
 554         out.append(start, bit->what[0].first);
 555         out.append(rvalue);
 556         start = bit->what[0].second; //move search forward
 557     }
 558     out.append(start, end);
 559     txt = out;
 560     return replaces;
 561 }
 562
 563 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
 564 {
 565     int gnum = 0;
 566     bool esc = false;
 567     const std::string &str = pattern;
 568     std::string res;
 569     start_anchor = str[0] == '^';
 570     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 571     for (size_t i = 0; i < str.size(); ++i)
 572     {
 573         res += str[i];
 574         if (!esc && str[i] == '\\')
 575         {
 576             esc = true;
 577             continue;
 578         }
 579         if (!esc && str[i] == '(') //group starts
 580         {
 581             gnum++;
 582             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 583             {
 584                 i++;
 585                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 586                 {
 587                     if (gnum > 0) gnum--;
 588                     res += str[i];
 589                     i++;
 590                     res += str[i];
 591                     continue;
 592                 }
 593                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 594                     i++;
 595                 if (i+1 < str.size() && str[i+1] == '<') //named
 596                 {
 597                     i++;
 598                     std::string gname;
 599                     bool term = false;
 600                     while (++i < str.size())
 601                     {
 602                         if (str[i] == '>') { term = true; break; }
 603                         if (!isalnum(str[i]))
 604                             throw mp::filter::FilterException
 605                                 ("Only alphanumeric chars allowed, found "
 606                                  " in '"
 607                                  + str
 608                                  + "' at "
 609                                  + boost::lexical_cast<std::string>(i));
 610                         gname += str[i];
 611                     }
 612                     if (!term)
 613                         throw mp::filter::FilterException
 614                             ("Unterminated group name '" + gname
 615                              + " in '" + str +"'");
 616                     group_index[gnum] = gname;
 617                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 618                             gname.c_str(), gnum);
 619                 }
 620             }
 621         }
 622         esc = false;
 623     }
 624     re = res;
 625 }
 626
 627 std::string yf::HttpRewrite::Replace::sub_vars(
 628     const std::map<std::string, std::string> & vars) const
 629 {
 630     std::string out;
 631     bool esc = false;
 632     const std::string & in = recipe;
 633     for (size_t i = 0; i < in.size(); ++i)
 634     {
 635         if (!esc && in[i] == '\\')
 636         {
 637             esc = true;
 638             continue;
 639         }
 640         if (!esc && in[i] == '$') //var
 641         {
 642             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 643             {
 644                 ++i;
 645                 std::string name;
 646                 bool term = false;
 647                 while (++i < in.size())
 648                 {
 649                     if (in[i] == '}') { term = true; break; }
 650                     name += in[i];
 651                 }
 652                 if (!term) throw mp::filter::FilterException
 653                     ("Unterminated var ref in '"+in+"' at "
 654                      + boost::lexical_cast<std::string>(i));
 655                 std::map<std::string, std::string>::const_iterator it
 656                     = vars.find(name);
 657                 if (it != vars.end())
 658                 {
 659                     out += it->second;
 660                 }
 661             }
 662             else
 663             {
 664                 throw mp::filter::FilterException
 665                     ("Malformed or trimmed var ref in '"
 666                      +in+"' at "+boost::lexical_cast<std::string>(i));
 667             }
 668             continue;
 669         }
 670         //passthru
 671         out += in[i];
 672         esc = false;
 673     }
 674     return out;
 675 }
 676
 677 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 678 {
 679 }
 680
 681 void yf::HttpRewrite::Content::parse(
 682     int verbose,
 683     std::string &content,
 684     std::map<std::string, std::string> &vars) const
 685 {
 686     if (type == "html")
 687     {
 688         HTMLParser parser;
 689         Event ev(this, vars);
 690
 691         parser.set_verbose(verbose);
 692
 693         parser.parse(ev, content.c_str());
 694         content = ev.result();
 695     }
 696     if (type == "quoted-literal")
 697     {
 698         quoted_literal(content, vars);
 699     }
 700 }
 701
 702 void yf::HttpRewrite::Content::quoted_literal(
 703     std::string &content,
 704     std::map<std::string, std::string> &vars) const
 705 {
 706     std::list<Within>::const_iterator it = within_list.begin();
 707     if (it != within_list.end())
 708         embed_quoted_literal(content, vars, it->rule, false);
 709 }
 710
 711 void yf::HttpRewrite::Content::configure(
 712     const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
 713 {
 714     for (; ptr; ptr = ptr->next)
 715     {
 716         if (ptr->type != XML_ELEMENT_NODE)
 717             continue;
 718         if (!strcmp((const char *) ptr->name, "within"))
 719         {
 720             static const char *names[7] =
 721                 { "header", "attr", "tag", "rule", "reqline", "type", 0 };
 722             std::string values[6];
 723             mp::xml::parse_attr(ptr, names, values);
 724             Within w;
 725             w.header = values[0];
 726             w.attr = values[1];
 727             if (values[2].length() > 0)
 728                 w.tag = values[2];
 729             std::map<std::string,RulePtr>::const_iterator it =
 730                 rules.find(values[3]);
 731             if (it == rules.end())
 732                 throw mp::filter::FilterException
 733                     ("Reference to non-existing rule '" + values[3] +
 734                      "' in http_rewrite filter");
 735             w.rule = it->second;
 736             w.reqline = values[4] == "1";
 737             w.type = values[5];
 738             if (w.type.empty() || w.type == "quoted-literal")
 739                 ;
 740             else
 741                 throw mp::filter::FilterException
 742                     ("within type must be quoted-literal or none in "
 743                      " in http_rewrite filter");
 744             within_list.push_back(w);
 745         }
 746     }
 747 }
 748
 749 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 750 {
 751     static const char *names[2] = { "verbose", 0 };
 752     std::string values[1];
 753     values[0] = "0";
 754     mp::xml::parse_attr(ptr, names, values);
 755
 756     phase.m_verbose = atoi(values[0].c_str());
 757
 758     std::map<std::string, RulePtr > rules;
 759     for (ptr = ptr->children; ptr; ptr = ptr->next)
 760     {
 761         if (ptr->type != XML_ELEMENT_NODE)
 762             continue;
 763         else if (!strcmp((const char *) ptr->name, "rule"))
 764         {
 765             static const char *names[2] = { "name", 0 };
 766             std::string values[1];
 767             values[0] = "default";
 768             mp::xml::parse_attr(ptr, names, values);
 769
 770             RulePtr rule(new Rule);
 771             for (xmlNode *p = ptr->children; p; p = p->next)
 772             {
 773                 if (p->type != XML_ELEMENT_NODE)
 774                     continue;
 775                 if (!strcmp((const char *) p->name, "rewrite"))
 776                 {
 777                     Replace replace;
 778                     std::string from;
 779                     const struct _xmlAttr *attr;
 780                     for (attr = p->properties; attr; attr = attr->next)
 781                     {
 782                         if (!strcmp((const char *) attr->name,  "from"))
 783                             from = mp::xml::get_text(attr->children);
 784                         else if (!strcmp((const char *) attr->name,  "to"))
 785                             replace.recipe = mp::xml::get_text(attr->children);
 786                         else
 787                             throw mp::filter::FilterException
 788                                 ("Bad attribute "
 789                                  + std::string((const char *) attr->name)
 790                                  + " in rewrite section of http_rewrite");
 791                     }
 792                     yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 793                             from.c_str(), replace.recipe.c_str());
 794                     if (!from.empty())
 795                     {
 796                         replace.parse_groups(from);
 797                         rule->replace_list.push_back(replace);
 798                     }
 799                 }
 800                 else
 801                     throw mp::filter::FilterException
 802                         ("Bad element "
 803                          + std::string((const char *) p->name)
 804                          + " in http_rewrite filter");
 805             }
 806             rules[values[0]] = rule;
 807         }
 808         else if (!strcmp((const char *) ptr->name, "content"))
 809         {
 810             static const char *names[3] =
 811                 { "type", "mime", 0 };
 812             std::string values[2];
 813             mp::xml::parse_attr(ptr, names, values);
 814             if (values[0].empty())
 815             {
 816                     throw mp::filter::FilterException
 817                         ("Missing attribute, type for for element "
 818                          + std::string((const char *) ptr->name)
 819                          + " in http_rewrite filter");
 820             }
 821             Content c;
 822
 823             c.type = values[0];
 824             // if (!values[1].empty())
 825                 c.content_re = values[1];
 826             c.configure(ptr->children, rules);
 827             phase.content_list.push_back(c);
 828         }
 829         else
 830         {
 831             throw mp::filter::FilterException
 832                 ("Bad element "
 833                  + std::string((const char *) ptr->name)
 834                  + " in http_rewrite filter");
 835         }
 836     }
 837 }
 838
 839 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 840         const char *path)
 841 {
 842     for (ptr = ptr->children; ptr; ptr = ptr->next)
 843     {
 844         if (ptr->type != XML_ELEMENT_NODE)
 845             continue;
 846         else if (!strcmp((const char *) ptr->name, "request"))
 847         {
 848             configure_phase(ptr, *req_phase);
 849         }
 850         else if (!strcmp((const char *) ptr->name, "response"))
 851         {
 852             configure_phase(ptr, *res_phase);
 853         }
 854         else
 855         {
 856             throw mp::filter::FilterException
 857                 ("Bad element "
 858                  + std::string((const char *) ptr->name)
 859                  + " in http_rewrite1 filter");
 860         }
 861     }
 862 }
 863
 864 static mp::filter::Base* filter_creator()
 865 {
 866     return new mp::filter::HttpRewrite;
 867 }
 868
 869 extern "C" {
 870     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 871         0,
 872         "http_rewrite",
 873         filter_creator
 874     };
 875 }
 876
 877
 878 /*
 879  * Local variables:
 880  * c-basic-offset: 4
 881  * c-file-style: "Stroustrup"
 882  * indent-tabs-mode: nil
 883  * End:
 884  * vim: shiftwidth=4 tabstop=8 expandtab
 885  */
 886