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
  25 #include <yaz/zgdu.h>
  26 #include <yaz/log.h>
  27
  28 #include <boost/regex.hpp>
  29 #include <boost/lexical_cast.hpp>
  30
  31 #include <vector>
  32 #include <map>
  33
  34 #if HAVE_SYS_TYPES_H
  35 #include <sys/types.h>
  36 #endif
  37
  38 namespace mp = metaproxy_1;
  39 namespace yf = mp::filter;
  40
  41 namespace metaproxy_1 {
  42     namespace filter {
  43         class HttpRewrite::RuleScope {
  44         public:
  45             std::vector<std::string> tags;
  46             std::vector<std::string> attrs;
  47             std::string content_type;
  48         };
  49         class HttpRewrite::Rule {
  50         public:
  51             enum Section { METHOD, HEADER, BODY };
  52             std::string regex;
  53             std::string recipe;
  54             std::map<int, std::string> group_index;
  55             std::vector<RuleScope> scopes;
  56             Section section;
  57             const std::string search_replace(
  58                 std::map<std::string, std::string> & vars,
  59                 const std::string & txt) const;
  60             std::string sub_vars (
  61                 const std::map<std::string, std::string> & vars) const;
  62             void parse_groups();
  63         };
  64         class HttpRewrite::Rules {
  65         public:
  66             std::vector<Rule> rules;
  67             void rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq,
  68                 std::map<std::string, std::string> & vars) const;
  69             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
  70                 std::map<std::string, std::string> & vars) const;
  71             void rewrite_body (mp::odr & o,
  72                 char **content_buf, int *content_len,
  73                 std::map<std::string, std::string> & vars) const;
  74             const std::string test_patterns(
  75                 std::map<std::string, std::string> & vars,
  76                 const std::string & txt) const;
  77         };
  78     }
  79 }
  80
  81 yf::HttpRewrite::HttpRewrite() : req_rules(new Rules), res_rules(new Rules)
  82 {
  83 }
  84
  85 yf::HttpRewrite::~HttpRewrite()
  86 {
  87 }
  88
  89 void yf::HttpRewrite::process(mp::Package & package) const
  90 {
  91     yaz_log(YLOG_LOG, "HttpRewrite begins....");
  92     Z_GDU *gdu = package.request().get();
  93     //map of request/response vars
  94     std::map<std::string, std::string> vars;
  95     //we have an http req
  96     if (gdu && gdu->which == Z_GDU_HTTP_Request)
  97     {
  98         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
  99         mp::odr o;
 100         req_rules->rewrite_reqline(o, hreq, vars);
 101         yaz_log(YLOG_LOG, ">> Request headers");
 102         req_rules->rewrite_headers(o, hreq->headers, vars);
 103         req_rules->rewrite_body(o,
 104                 &hreq->content_buf, &hreq->content_len,
 105                 vars);
 106         package.request() = gdu;
 107     }
 108     package.move();
 109     gdu = package.response().get();
 110     if (gdu && gdu->which == Z_GDU_HTTP_Response)
 111     {
 112         Z_HTTP_Response *hres = gdu->u.HTTP_Response;
 113         yaz_log(YLOG_LOG, "Response code %d", hres->code);
 114         mp::odr o;
 115         yaz_log(YLOG_LOG, "<< Respose headers");
 116         res_rules->rewrite_headers(o, hres->headers, vars);
 117         res_rules->rewrite_body(o, &hres->content_buf,
 118                 &hres->content_len, vars);
 119         package.response() = gdu;
 120     }
 121 }
 122
 123 void yf::HttpRewrite::Rules::rewrite_reqline (mp::odr & o,
 124         Z_HTTP_Request *hreq,
 125         std::map<std::string, std::string> & vars) const
 126 {
 127     //rewrite the request line
 128     std::string path;
 129     if (strstr(hreq->path, "http://") == hreq->path)
 130     {
 131         yaz_log(YLOG_LOG, "Path in the method line is absolute, "
 132             "possibly a proxy request");
 133         path += hreq->path;
 134     }
 135     else
 136     {
 137         //TODO what about proto
 138         path += "http://";
 139         path += z_HTTP_header_lookup(hreq->headers, "Host");
 140         path += hreq->path;
 141     }
 142     yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
 143     std::string npath =
 144         test_patterns(vars, path);
 145     if (!npath.empty())
 146     {
 147         yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
 148         hreq->path = odr_strdup(o, npath.c_str());
 149     }
 150 }
 151
 152 void yf::HttpRewrite::Rules::rewrite_headers(mp::odr & o,
 153         Z_HTTP_Header *headers,
 154         std::map<std::string, std::string> & vars) const
 155 {
 156     for (Z_HTTP_Header *header = headers;
 157             header != 0;
 158             header = header->next)
 159     {
 160         std::string sheader(header->name);
 161         sheader += ": ";
 162         sheader += header->value;
 163         yaz_log(YLOG_LOG, "%s: %s", header->name, header->value);
 164         std::string out = test_patterns(vars, sheader);
 165         if (!out.empty())
 166         {
 167             size_t pos = out.find(": ");
 168             if (pos == std::string::npos)
 169             {
 170                 yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
 171                 continue;
 172             }
 173             header->name = odr_strdup(o, out.substr(0, pos).c_str());
 174             header->value = odr_strdup(o, out.substr(pos+2,
 175                         std::string::npos).c_str());
 176         }
 177     }
 178 }
 179
 180 void yf::HttpRewrite::Rules::rewrite_body (mp::odr & o,
 181         char **content_buf,
 182         int *content_len,
 183         std::map<std::string, std::string> & vars) const
 184 {
 185     if (*content_buf)
 186     {
 187         std::string body(*content_buf);
 188         std::string nbody =
 189             test_patterns(vars, body);
 190         if (!nbody.empty())
 191         {
 192             *content_buf = odr_strdup(o, nbody.c_str());
 193             *content_len = nbody.size();
 194         }
 195     }
 196 }
 197
 198 /**
 199  * Tests pattern from the vector in order and executes recipe on
 200  the first match.
 201  */
 202 const std::string yf::HttpRewrite::Rules::test_patterns(
 203         std::map<std::string, std::string> & vars,
 204         const std::string & txt) const
 205 {
 206     for (size_t i = 0; i < rules.size(); i++)
 207     {
 208         std::string out = rules[i].search_replace(vars, txt);
 209         if (!out.empty()) return out;
 210     }
 211     return "";
 212 }
 213
 214 const std::string yf::HttpRewrite::Rule::search_replace(
 215         std::map<std::string, std::string> & vars,
 216         const std::string & txt) const
 217 {
 218     //exec regex against value
 219     boost::regex re(regex);
 220     boost::smatch what;
 221     std::string::const_iterator start, end;
 222     start = txt.begin();
 223     end = txt.end();
 224     std::string out;
 225     while (regex_search(start, end, what, re)) //find next full match
 226     {
 227         size_t i;
 228         for (i = 1; i < what.size(); ++i)
 229         {
 230             //check if the group is named
 231             std::map<int, std::string>::const_iterator it
 232                 = group_index.find(i);
 233             if (it != group_index.end())
 234             {   //it is
 235                 if (!what[i].str().empty())
 236                     vars[it->second] = what[i];
 237             }
 238
 239         }
 240         //prepare replacement string
 241         std::string rvalue = sub_vars(vars);
 242         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
 243                 what.str(0).c_str(), rvalue.c_str());
 244         out.append(start, what[0].first);
 245         out.append(rvalue);
 246         start = what[0].second; //move search forward
 247     }
 248     //if we had a match cat the last part
 249     if (start != txt.begin())
 250         out.append(start, end);
 251     return out;
 252 }
 253
 254 void yf::HttpRewrite::Rule::parse_groups()
 255 {
 256     int gnum = 0;
 257     bool esc = false;
 258     const std::string & str = regex;
 259     std::string res;
 260     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
 261     for (size_t i = 0; i < str.size(); ++i)
 262     {
 263         res += str[i];
 264         if (!esc && str[i] == '\\')
 265         {
 266             esc = true;
 267             continue;
 268         }
 269         if (!esc && str[i] == '(') //group starts
 270         {
 271             gnum++;
 272             if (i+1 < str.size() && str[i+1] == '?') //group with attrs
 273             {
 274                 i++;
 275                 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
 276                 {
 277                     if (gnum > 0) gnum--;
 278                     res += str[i];
 279                     i++;
 280                     res += str[i];
 281                     continue;
 282                 }
 283                 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
 284                     i++;
 285                 if (i+1 < str.size() && str[i+1] == '<') //named
 286                 {
 287                     i++;
 288                     std::string gname;
 289                     bool term = false;
 290                     while (++i < str.size())
 291                     {
 292                         if (str[i] == '>') { term = true; break; }
 293                         if (!isalnum(str[i]))
 294                             throw mp::filter::FilterException
 295                                 ("Only alphanumeric chars allowed, found "
 296                                  " in '"
 297                                  + str
 298                                  + "' at "
 299                                  + boost::lexical_cast<std::string>(i));
 300                         gname += str[i];
 301                     }
 302                     if (!term)
 303                         throw mp::filter::FilterException
 304                             ("Unterminated group name '" + gname
 305                              + " in '" + str +"'");
 306                     group_index[gnum] = gname;
 307                     yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
 308                             gname.c_str(), gnum);
 309                 }
 310             }
 311         }
 312         esc = false;
 313     }
 314     regex = res;
 315 }
 316
 317 std::string yf::HttpRewrite::Rule::sub_vars (
 318         const std::map<std::string, std::string> & vars) const
 319 {
 320     std::string out;
 321     bool esc = false;
 322     const std::string & in = recipe;
 323     for (size_t i = 0; i < in.size(); ++i)
 324     {
 325         if (!esc && in[i] == '\\')
 326         {
 327             esc = true;
 328             continue;
 329         }
 330         if (!esc && in[i] == '$') //var
 331         {
 332             if (i+1 < in.size() && in[i+1] == '{') //ref prefix
 333             {
 334                 ++i;
 335                 std::string name;
 336                 bool term = false;
 337                 while (++i < in.size())
 338                 {
 339                     if (in[i] == '}') { term = true; break; }
 340                     name += in[i];
 341                 }
 342                 if (!term) throw mp::filter::FilterException
 343                     ("Unterminated var ref in '"+in+"' at "
 344                      + boost::lexical_cast<std::string>(i));
 345                 std::map<std::string, std::string>::const_iterator it
 346                     = vars.find(name);
 347                 if (it != vars.end())
 348                 {
 349                     out += it->second;
 350                 }
 351             }
 352             else
 353             {
 354                 throw mp::filter::FilterException
 355                     ("Malformed or trimmed var ref in '"
 356                      +in+"' at "+boost::lexical_cast<std::string>(i));
 357             }
 358             continue;
 359         }
 360         //passthru
 361         out += in[i];
 362         esc = false;
 363     }
 364     return out;
 365 }
 366
 367 void yf::HttpRewrite::configure_rules(const xmlNode *ptr,
 368         Rules & rules)
 369 {
 370     for (ptr = ptr->children; ptr; ptr = ptr->next)
 371     {
 372         if (ptr->type != XML_ELEMENT_NODE)
 373             continue;
 374         else if (!strcmp((const char *) ptr->name, "rewrite"))
 375         {
 376             Rule rule;
 377             const struct _xmlAttr *attr;
 378             for (attr = ptr->properties; attr; attr = attr->next)
 379             {
 380                 if (!strcmp((const char *) attr->name,  "from"))
 381                     rule.regex = mp::xml::get_text(attr->children);
 382                 else if (!strcmp((const char *) attr->name,  "to"))
 383                     rule.recipe = mp::xml::get_text(attr->children);
 384                 else
 385                     throw mp::filter::FilterException
 386                         ("Bad attribute "
 387                          + std::string((const char *) attr->name)
 388                          + " in rewrite section of http_rewrite");
 389             }
 390             yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
 391                     rule.regex.c_str(), rule.recipe.c_str());
 392             rule.parse_groups();
 393             if (!rule.regex.empty())
 394                 rules.rules.push_back(rule);
 395         }
 396         else
 397         {
 398             throw mp::filter::FilterException
 399                 ("Bad element o"
 400                  + std::string((const char *) ptr->name)
 401                  + " in http_rewrite1 filter");
 402         }
 403     }
 404 }
 405
 406 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
 407         const char *path)
 408 {
 409     for (ptr = ptr->children; ptr; ptr = ptr->next)
 410     {
 411         if (ptr->type != XML_ELEMENT_NODE)
 412             continue;
 413         else if (!strcmp((const char *) ptr->name, "request"))
 414         {
 415             configure_rules(ptr, *req_rules);
 416         }
 417         else if (!strcmp((const char *) ptr->name, "response"))
 418         {
 419             configure_rules(ptr, *res_rules);
 420         }
 421         else
 422         {
 423             throw mp::filter::FilterException
 424                 ("Bad element "
 425                  + std::string((const char *) ptr->name)
 426                  + " in http_rewrite1 filter");
 427         }
 428     }
 429 }
 430
 431 static mp::filter::Base* filter_creator()
 432 {
 433     return new mp::filter::HttpRewrite;
 434 }
 435
 436 extern "C" {
 437     struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
 438         0,
 439         "http_rewrite",
 440         filter_creator
 441     };
 442 }
 443
 444
 445 /*
 446  * Local variables:
 447  * c-basic-offset: 4
 448  * c-file-style: "Stroustrup"
 449  * indent-tabs-mode: nil
 450  * End:
 451  * vim: shiftwidth=4 tabstop=8 expandtab
 452  */
 453