1 /* This file is part of Metaproxy.
2 Copyright (C) 2005-2013 Index Data
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
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
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
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"
30 #include <boost/regex.hpp>
31 #include <boost/lexical_cast.hpp>
32 #include <boost/algorithm/string.hpp>
36 namespace mp = metaproxy_1;
37 namespace yf = mp::filter;
39 namespace metaproxy_1 {
41 class HttpRewrite::Replace {
47 std::map<int, std::string> group_index;
49 const std::map<std::string, std::string> & vars) const;
50 void parse_groups(std::string pattern);
53 class HttpRewrite::Rule {
55 std::list<Replace> replace_list;
57 std::map<std::string, std::string> &vars,
58 std::string &txt, bool anchor);
60 class HttpRewrite::Within {
68 bool exec(std::map<std::string, std::string> &vars,
69 std::string &txt, bool anchor) const;
72 class HttpRewrite::Content {
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;
84 class HttpRewrite::Phase {
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;
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,
105 void closeTag(const char *tag, int tag_len);
106 void text(const char *value, int len);
107 const Content *m_content;
109 std::stack<std::list<Within>::const_iterator> s_within;
110 std::map<std::string, std::string> &m_vars;
112 Event(const Content *p, std::map<std::string, std::string> &vars);
114 const char *result();
119 yf::HttpRewrite::HttpRewrite() :
120 req_phase(new Phase), res_phase(new Phase)
124 yf::HttpRewrite::~HttpRewrite()
128 void yf::HttpRewrite::process(mp::Package & package) const
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)
137 Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
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,
145 &hreq->content_buf, &hreq->content_len,
147 package.request() = gdu;
150 gdu = package.response().get();
151 if (gdu && gdu->which == Z_GDU_HTTP_Response)
153 Z_HTTP_Response *hres = gdu->u.HTTP_Response;
154 yaz_log(YLOG_LOG, "Response code %d", hres->code);
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,
161 &hres->content_buf, &hres->content_len,
163 package.response() = gdu;
167 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
168 Z_HTTP_Request *hreq,
169 std::map<std::string, std::string> & vars) const
171 //rewrite the request line
173 if (strstr(hreq->path, "http://") == hreq->path)
175 yaz_log(YLOG_LOG, "Path in the method line is absolute, "
176 "possibly a proxy request");
181 //TODO what about proto
182 const char *host = z_HTTP_header_lookup(hreq->headers, "Host");
191 std::list<Content>::const_iterator cit = content_list.begin();
192 for (; cit != content_list.end(); cit++)
193 if (cit->type == "headers")
196 if (cit == content_list.end())
199 std::list<Within>::const_iterator it = cit->within_list.begin();
200 for (; it != cit->within_list.end(); it++)
203 yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
204 if (it->exec(vars, path, true))
206 yaz_log(YLOG_LOG, "Rewritten request URL is %s", path.c_str());
207 hreq->path = odr_strdup(o, path.c_str());
212 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
213 Z_HTTP_Header *headers,
214 std::map<std::string, std::string> & vars) const
216 std::list<Content>::const_iterator cit = content_list.begin();
217 for (; cit != content_list.end(); cit++)
218 if (cit->type == "headers")
221 if (cit == content_list.end())
224 for (Z_HTTP_Header *header = headers; header; header = header->next)
226 std::list<Within>::const_iterator it = cit->within_list.begin();
227 for (; it != cit->within_list.end(); it++)
229 if (!it->header.empty() &&
230 regex_match(header->name, it->header))
232 // Match and replace only the header value
233 std::string hval(header->value);
234 if (it->exec(vars, hval, true))
236 header->value = odr_strdup(o, hval.c_str());
243 void yf::HttpRewrite::Phase::rewrite_body(
245 const char *content_type,
248 std::map<std::string, std::string> & vars) const
250 if (*content_len == 0)
253 yaz_log(YLOG_LOG, "rewrite_body: null content_type, can not rewrite");
256 std::list<Content>::const_iterator cit = content_list.begin();
257 for (; cit != content_list.end(); cit++)
259 yaz_log(YLOG_LOG, "rewrite_body: content_type=%s type=%s",
260 content_type, cit->type.c_str());
261 if (cit->type != "headers"
262 && regex_match(content_type, cit->content_re))
265 if (cit == content_list.end()) {
266 yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
272 for (i = 0; i < *content_len; i++)
273 if ((*content_buf)[i] == 0) {
274 yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
275 return; // binary content. skip
278 std::string content(*content_buf, *content_len);
279 cit->parse(m_verbose, content, vars);
280 *content_buf = odr_strdup(o, content.c_str());
281 *content_len = strlen(*content_buf);
284 yf::HttpRewrite::Event::Event(const Content *p,
285 std::map<std::string, std::string> & vars
286 ) : m_content(p), m_vars(vars)
291 yf::HttpRewrite::Event::~Event()
296 const char *yf::HttpRewrite::Event::result()
298 return wrbuf_cstr(m_w);
301 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
303 wrbuf_putc(m_w, '<');
304 wrbuf_write(m_w, tag, tag_len);
306 std::string t(tag, tag_len);
307 std::list<Within>::const_iterator it = m_content->within_list.begin();
308 for (; it != m_content->within_list.end(); it++)
310 if (!it->tag.empty() && regex_match(t, it->tag))
312 if (!it->attr.empty() && regex_match("#text", it->attr))
321 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
326 if (!s_within.empty())
328 std::list<Within>::const_iterator it = s_within.top();
329 std::string t(tag, tag_len);
330 if (regex_match(t, it->tag))
335 wrbuf_putc(m_w, '/');
336 wrbuf_putc(m_w, '>');
339 void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
340 const char *attr, int attr_len,
341 const char *value, int val_len,
344 std::list<Within>::const_iterator it = m_content->within_list.begin();
347 for (; it != m_content->within_list.end(); it++)
349 std::string t(tag, tag_len);
350 if (it->tag.empty() || regex_match(t, it->tag))
352 std::string a(attr, attr_len);
353 if (!it->attr.empty() && regex_match(a, it->attr))
360 wrbuf_putc(m_w, ' ');
361 wrbuf_write(m_w, attr, attr_len);
364 wrbuf_puts(m_w, "=");
365 wrbuf_puts(m_w, sep);
370 std::string s(value, val_len);
371 it->exec(m_vars, s, true);
372 wrbuf_puts(m_w, s.c_str());
375 wrbuf_write(m_w, value, val_len);
376 wrbuf_puts(m_w, sep);
380 void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
382 if (!s_within.empty())
384 std::list<Within>::const_iterator it = s_within.top();
385 std::string t(tag, tag_len);
386 if (regex_match(t, it->tag))
389 wrbuf_puts(m_w, "</");
390 wrbuf_write(m_w, tag, tag_len);
393 void yf::HttpRewrite::Event::text(const char *value, int len)
395 std::list<Within>::const_iterator it = m_content->within_list.end();
396 if (!s_within.empty())
398 if (it != m_content->within_list.end())
400 std::string s(value, len);
401 it->exec(m_vars, s, false);
402 wrbuf_puts(m_w, s.c_str());
405 wrbuf_write(m_w, value, len);
408 static bool embed_quoted_literal(
409 std::string &content,
410 std::map<std::string, std::string> &vars,
411 mp::filter::HttpRewrite::RulePtr ruleptr,
414 bool replace = false;
416 const char *cp = content.c_str();
417 const char *cp0 = cp;
420 if (html_context && !strncmp(cp, """, 6))
423 res.append(cp0, cp - cp0);
427 if (!strncmp(cp, """, 6))
435 std::string s(cp0, cp - cp0);
436 if (ruleptr->test_patterns(vars, s, true))
441 else if (*cp == '"' || *cp == '\'')
445 res.append(cp0, cp - cp0);
449 if (cp[-1] != '\\' && *cp == m)
457 std::string s(cp0, cp - cp0);
458 if (ruleptr->test_patterns(vars, s, true))
463 else if (*cp == '/' && cp[1] == '/')
465 while (cp[1] && cp[1] != '\n')
470 res.append(cp0, cp - cp0);
475 bool yf::HttpRewrite::Within::exec(
476 std::map<std::string, std::string> & vars,
477 std::string & txt, bool anchor) const
479 if (type == "quoted-literal")
481 return embed_quoted_literal(txt, vars, rule, true);
485 return rule->test_patterns(vars, txt, anchor);
489 bool yf::HttpRewrite::Rule::test_patterns(
490 std::map<std::string, std::string> & vars,
491 std::string & txt, bool anchor)
493 bool replaces = false;
496 std::string::const_iterator start, end;
501 std::list<Replace>::iterator bit = replace_list.end();
503 std::string::const_iterator best_pos = txt.end();
504 std::list<Replace>::iterator it = replace_list.begin();
505 for (; it != replace_list.end(); it++)
507 if (it->start_anchor && !first)
509 if (regex_search(start, end, it->what, it->re))
511 if (it->what[0].first < best_pos)
513 best_pos = it->what[0].first;
518 if (bit == replace_list.end())
524 for (i = 1; i < bit->what.size(); ++i)
526 //check if the group is named
527 std::map<int, std::string>::const_iterator git
528 = bit->group_index.find(i);
529 if (git != bit->group_index.end())
531 vars[git->second] = bit->what[i];
535 //prepare replacement string
536 std::string rvalue = bit->sub_vars(vars);
537 yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
538 bit->what.str(0).c_str(), rvalue.c_str());
539 out.append(start, bit->what[0].first);
541 start = bit->what[0].second; //move search forward
543 out.append(start, end);
548 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
552 const std::string &str = pattern;
554 start_anchor = str[0] == '^';
555 yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
556 for (size_t i = 0; i < str.size(); ++i)
559 if (!esc && str[i] == '\\')
564 if (!esc && str[i] == '(') //group starts
567 if (i+1 < str.size() && str[i+1] == '?') //group with attrs
570 if (i+1 < str.size() && str[i+1] == ':') //non-capturing
572 if (gnum > 0) gnum--;
578 if (i+1 < str.size() && str[i+1] == 'P') //optional, python
580 if (i+1 < str.size() && str[i+1] == '<') //named
585 while (++i < str.size())
587 if (str[i] == '>') { term = true; break; }
588 if (!isalnum(str[i]))
589 throw mp::filter::FilterException
590 ("Only alphanumeric chars allowed, found "
594 + boost::lexical_cast<std::string>(i));
598 throw mp::filter::FilterException
599 ("Unterminated group name '" + gname
600 + " in '" + str +"'");
601 group_index[gnum] = gname;
602 yaz_log(YLOG_LOG, "Found named group '%s' at $%d",
603 gname.c_str(), gnum);
612 std::string yf::HttpRewrite::Replace::sub_vars(
613 const std::map<std::string, std::string> & vars) const
617 const std::string & in = recipe;
618 for (size_t i = 0; i < in.size(); ++i)
620 if (!esc && in[i] == '\\')
625 if (!esc && in[i] == '$') //var
627 if (i+1 < in.size() && in[i+1] == '{') //ref prefix
632 while (++i < in.size())
634 if (in[i] == '}') { term = true; break; }
637 if (!term) throw mp::filter::FilterException
638 ("Unterminated var ref in '"+in+"' at "
639 + boost::lexical_cast<std::string>(i));
640 std::map<std::string, std::string>::const_iterator it
642 if (it != vars.end())
649 throw mp::filter::FilterException
650 ("Malformed or trimmed var ref in '"
651 +in+"' at "+boost::lexical_cast<std::string>(i));
662 yf::HttpRewrite::Phase::Phase() : m_verbose(0)
666 void yf::HttpRewrite::Content::parse(
668 std::string &content,
669 std::map<std::string, std::string> &vars) const
674 Event ev(this, vars);
676 parser.set_verbose(verbose);
678 parser.parse(ev, content.c_str());
679 content = ev.result();
681 if (type == "quoted-literal")
683 quoted_literal(content, vars);
687 void yf::HttpRewrite::Content::quoted_literal(
688 std::string &content,
689 std::map<std::string, std::string> &vars) const
691 std::list<Within>::const_iterator it = within_list.begin();
692 if (it != within_list.end())
693 embed_quoted_literal(content, vars, it->rule, false);
696 void yf::HttpRewrite::Content::configure(
697 const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
699 for (; ptr; ptr = ptr->next)
701 if (ptr->type != XML_ELEMENT_NODE)
703 if (!strcmp((const char *) ptr->name, "within"))
705 static const char *names[7] =
706 { "header", "attr", "tag", "rule", "reqline", "type", 0 };
707 std::string values[6];
708 mp::xml::parse_attr(ptr, names, values);
710 if (values[0].length() > 0)
711 w.header.assign(values[0], boost::regex_constants::icase);
712 if (values[1].length() > 0)
713 w.attr.assign(values[1], boost::regex_constants::icase);
714 if (values[2].length() > 0)
715 w.tag.assign(values[2], boost::regex_constants::icase);
717 std::vector<std::string> rulenames;
718 boost::split(rulenames, values[3], boost::is_any_of(","));
719 if (rulenames.size() == 0)
721 throw mp::filter::FilterException
722 ("Empty rule in '" + values[3] +
723 "' in http_rewrite filter");
725 else if (rulenames.size() == 1)
727 std::map<std::string,RulePtr>::const_iterator it =
728 rules.find(rulenames[0]);
729 if (it == rules.end())
730 throw mp::filter::FilterException
731 ("Reference to non-existing rule '" + rulenames[0] +
732 "' in http_rewrite filter");
738 RulePtr rule(new Rule);
740 for (i = 0; i < rulenames.size(); i++)
742 std::map<std::string,RulePtr>::const_iterator it =
743 rules.find(rulenames[i]);
744 if (it == rules.end())
745 throw mp::filter::FilterException
746 ("Reference to non-existing rule '" + rulenames[i] +
747 "' in http_rewrite filter");
748 RulePtr subRule = it->second;
749 std::list<Replace>::iterator rit =
750 subRule->replace_list.begin();
751 for (; rit != subRule->replace_list.end(); rit++)
752 rule->replace_list.push_back(*rit);
756 w.reqline = values[4] == "1";
758 if (w.type.empty() || w.type == "quoted-literal")
761 throw mp::filter::FilterException
762 ("within type must be quoted-literal or none in "
763 " in http_rewrite filter");
764 within_list.push_back(w);
769 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
771 static const char *names[2] = { "verbose", 0 };
772 std::string values[1];
774 mp::xml::parse_attr(ptr, names, values);
776 phase.m_verbose = atoi(values[0].c_str());
778 std::map<std::string, RulePtr > rules;
779 for (ptr = ptr->children; ptr; ptr = ptr->next)
781 if (ptr->type != XML_ELEMENT_NODE)
783 else if (!strcmp((const char *) ptr->name, "rule"))
785 static const char *names[2] = { "name", 0 };
786 std::string values[1];
787 values[0] = "default";
788 mp::xml::parse_attr(ptr, names, values);
790 RulePtr rule(new Rule);
791 for (xmlNode *p = ptr->children; p; p = p->next)
793 if (p->type != XML_ELEMENT_NODE)
795 if (!strcmp((const char *) p->name, "rewrite"))
799 const struct _xmlAttr *attr;
800 for (attr = p->properties; attr; attr = attr->next)
802 if (!strcmp((const char *) attr->name, "from"))
803 from = mp::xml::get_text(attr->children);
804 else if (!strcmp((const char *) attr->name, "to"))
805 replace.recipe = mp::xml::get_text(attr->children);
807 throw mp::filter::FilterException
809 + std::string((const char *) attr->name)
810 + " in rewrite section of http_rewrite");
812 yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'",
813 from.c_str(), replace.recipe.c_str());
816 replace.parse_groups(from);
817 rule->replace_list.push_back(replace);
821 throw mp::filter::FilterException
823 + std::string((const char *) p->name)
824 + " in http_rewrite filter");
826 rules[values[0]] = rule;
828 else if (!strcmp((const char *) ptr->name, "content"))
830 static const char *names[3] =
831 { "type", "mime", 0 };
832 std::string values[2];
833 mp::xml::parse_attr(ptr, names, values);
834 if (values[0].empty())
836 throw mp::filter::FilterException
837 ("Missing attribute, type for for element "
838 + std::string((const char *) ptr->name)
839 + " in http_rewrite filter");
844 if (!values[1].empty())
845 c.content_re.assign(values[1], boost::regex::icase);
846 c.configure(ptr->children, rules);
847 phase.content_list.push_back(c);
851 throw mp::filter::FilterException
853 + std::string((const char *) ptr->name)
854 + " in http_rewrite filter");
859 void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only,
862 for (ptr = ptr->children; ptr; ptr = ptr->next)
864 if (ptr->type != XML_ELEMENT_NODE)
866 else if (!strcmp((const char *) ptr->name, "request"))
868 configure_phase(ptr, *req_phase);
870 else if (!strcmp((const char *) ptr->name, "response"))
872 configure_phase(ptr, *res_phase);
876 throw mp::filter::FilterException
878 + std::string((const char *) ptr->name)
879 + " in http_rewrite1 filter");
884 static mp::filter::Base* filter_creator()
886 return new mp::filter::HttpRewrite;
890 struct metaproxy_1_filter_struct metaproxy_1_filter_http_rewrite = {
901 * c-file-style: "Stroustrup"
902 * indent-tabs-mode: nil
904 * vim: shiftwidth=4 tabstop=8 expandtab