/* This file is part of Metaproxy.
- Copyright (C) 2005-2013 Index Data
+ Copyright (C) Index Data
Metaproxy is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
#include <stdexcept>
#include "filter_http_client.hpp"
+#include "filter_http_rewrite.hpp"
#include <metaproxy/util.hpp>
-#include "router_chain.hpp"
+#include <metaproxy/router_chain.hpp>
#include <metaproxy/package.hpp>
#include <boost/regex.hpp>
#include <boost/lexical_cast.hpp>
+#include <yaz/log.h>
+
#define BOOST_AUTO_TEST_MAIN
#define BOOST_TEST_DYN_LINK
using namespace boost::unit_test;
namespace mp = metaproxy_1;
-class FilterHeaderRewrite: public mp::filter::Base {
-public:
- void process(mp::Package & package) const {
- Z_GDU *gdu = package.request().get();
- //map of request/response vars
- std::map<std::string, std::string> vars;
- //we have an http req
- if (gdu && gdu->which == Z_GDU_HTTP_Request)
- {
- Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
- mp::odr o;
- //rewrite the request line
- std::string path;
- if (strstr(hreq->path, "http://") == hreq->path)
- {
- std::cout << "Path in the method line is absolute, "
- "possibly a proxy request\n";
- path += hreq->path;
- }
- else
- {
- path += z_HTTP_header_lookup(hreq->headers, "Host");
- path += hreq->path;
- }
- std::cout << "Proxy request URL is " << path << std::endl;
- std::string npath =
- search_replace(vars, path, req_uri_rx, req_uri_pat);
- std::cout << "Resp request URL is " << npath << std::endl;
- if (!npath.empty())
- hreq->path = odr_strdup(o, npath.c_str());
- std::cout << ">> Request headers" << std::endl;
- //iterate headers
- for (Z_HTTP_Header *header = hreq->headers;
- header != 0;
- header = header->next)
- {
- std::cout << header->name << ": " << header->value << std::endl;
- std::string out = search_replace(vars,
- std::string(header->value),
- req_uri_rx, req_uri_pat);
- if (!out.empty())
- header->value = odr_strdup(o, out.c_str());
- }
- package.request() = gdu;
- }
- package.move();
- gdu = package.response().get();
- if (gdu && gdu->which == Z_GDU_HTTP_Response)
- {
- std::cout << "<< Respose headers" << std::endl;
- Z_HTTP_Response *hr = gdu->u.HTTP_Response;
- mp::odr o;
- //iterate headers
- for (Z_HTTP_Header *header = hr->headers;
- header != 0;
- header = header->next)
- {
- std::cout << header->name << ": " << header->value << std::endl;
- std::string out = search_replace(vars,
- std::string(header->value),
- resp_uri_rx, resp_uri_pat);
- if (!out.empty())
- header->value = odr_strdup(o, out.c_str());
- }
- package.response() = gdu;
- }
- };
+BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
+{
+ try
+ {
+ mp::RouterChain router;
+ mp::filter::HttpRewrite fhr;
- void configure(const xmlNode* ptr, bool test_only, const char *path) {};
+ std::string xmlconf =
+ "<?xml version='1.0'?>\n"
+ "<filter xmlns='http://indexdata.com/metaproxy'\n"
+ " id='rewrite1' type='http_rewrite'>\n"
+ " <request verbose=\"1\">\n"
+ " <rule name=\"null\"/>\n"
+ " <rule name=\"url\">\n"
+ " <rewrite from='"
+ "(?<proto>https?://)(?<pxhost>[^ /?#]+)/(?<pxpath>[^ /]+)"
+ "/(?<host>[^ /]+)(?<path>[^ ]*)'\n"
+ " to='${proto}${host}${path}' />\n"
+ " <rewrite from='(?:Host: )(.*)'\n"
+ " to='Host: ${host}' />\n"
+ " </rule>\n"
+ " <content type=\"headers\">\n"
+ " <within header=\"link\" rule=\"null\"/>\n"
+ " <within reqline=\"1\" rule=\"url\"/>\n"
+ " </content>\n"
+ " </request>\n"
+ " <response verbose=\"1\">\n"
+ " <rule name=\"null\"/>\n"
+ " <rule name=\"cx\">\n"
+ " <rewrite from='^cx' to='cy'/>\n"
+ " </rule>\n"
+ " <rule name=\"url\">\n"
+ " <rewrite from='foo' to='bar'/>\n"
+ " <rewrite from='"
+ "(?<proto>https?://)(?<host>[^/?# "'>]+)/(?<path>[^ "'>]+)'\n"
+ " to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
+ " </rule>\n"
+ " <content type=\"headers\">\n"
+ " <within header=\"link\" rule=\"url\"/>\n"
+ " </content>\n"
+ " <content type=\"html\" mime=\"text/xml|text/html\">\n"
+ " <within tag=\"body\" attr=\"background\" rule=\"null\"/>\n"
+ " <within tag=\"script\" attr=\"#text\" type=\"quoted-literal\" rule=\"url\"/>\n"
+ " <within tag=\"style\" attr=\"#text\" rule=\"url\"/>\n"
+ " <within attr=\"href|src\" rule=\"url,cx\"/>\n"
+ " <within attr=\"onclick\" type=\"quoted-literal\" rule=\"url\"/>\n"
+ " </content>\n"
+ " <content type=\"quoted-literal\" mime=\".*javascript\">\n"
+ " <within rule=\"url\"/>\n"
+ " </content>\n"
+ " </response>\n"
+ "</filter>\n"
+ ;
- const std::string search_replace(
- std::map<std::string, std::string> & vars,
- const std::string txt,
- const std::string & uri_re,
- const std::string & uri_pat) const
- {
- //exec regex against value
- boost::regex re(uri_re);
- boost::smatch what;
- std::string::const_iterator start, end;
- start = txt.begin();
- end = txt.end();
- std::string out;
- while (regex_search(start, end, what, re)) //find next full match
- {
- unsigned i;
- for (i = 1; i < what.size(); ++i)
- {
- //check if the group is named
- std::map<int, std::string>::const_iterator it
- = groups_by_num.find(i);
- if (it != groups_by_num.end())
- { //it is
- std::string name = it->second;
- vars[name] = what[i];
- }
+ std::cout << xmlconf;
- }
- //prepare replacement string
- std::string rvalue = sub_vars(uri_pat, vars);
- //rewrite value
- std::string rhvalue = what.prefix().str()
- + rvalue + what.suffix().str();
- std::cout << "! Rewritten '"+what.str(0)+"' to '"+rvalue+"'\n";
- out += rhvalue;
- start = what[0].second; //move search forward
- }
- return out;
- };
+ // reading and parsing XML conf
+ xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
+ BOOST_CHECK(doc);
+ xmlNode *root_element = xmlDocGetRootElement(doc);
+ fhr.configure(root_element, true, "");
+ xmlFreeDoc(doc);
- static void parse_groups(const std::string & str,
- std::map<int, std::string> & groups_bynum,
- std::map<std::string, int> & groups_byname)
- {
- int gnum = 0;
- bool esc = false;
- for (int i = 0; i < str.size(); ++i)
- {
- if (!esc && str[i] == '\\')
- {
- esc = true;
- continue;
- }
- if (!esc && str[i] == '(') //group starts
- {
- gnum++;
- if (i+1 < str.size() && str[i+1] == '?') //group with attrs
- {
- i++;
- if (i+1 < str.size() && str[i+1] == ':') //non-capturing
- {
- if (gnum > 0) gnum--;
- i++;
- continue;
- }
- if (i+1 < str.size() && str[i+1] == 'P') //optional, python
- i++;
- if (i+1 < str.size() && str[i+1] == '<') //named
- {
- i++;
- std::string gname;
- bool term = false;
- while (++i < str.size())
- {
- if (str[i] == '>') { term = true; break; }
- if (!isalnum(str[i]))
- throw mp::filter::FilterException
- ("Only alphanumeric chars allowed, found "
- " in '"
- + str
- + "' at "
- + boost::lexical_cast<std::string>(i));
- gname += str[i];
- }
- if (!term)
- throw mp::filter::FilterException
- ("Unterminated group name '" + gname
- + " in '" + str +"'");
- groups_bynum[gnum] = gname;
- groups_byname[gname] = gnum;
- std::cout << "Found named group '" << gname
- << "' at $" << gnum << std::endl;
- }
- }
- }
- esc = false;
- }
- }
+ router.append(fhr);
- static std::string sub_vars (const std::string & in,
- const std::map<std::string, std::string> & vars)
- {
- std::string out;
- bool esc = false;
- for (int i = 0; i < in.size(); ++i)
+ // create an http request
+ mp::Package pack;
+
+ mp::odr odr;
+ Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
+ "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
+
+ Z_HTTP_Request *hreq = gdu_req->u.HTTP_Request;
+ z_HTTP_header_set(odr, &hreq->headers,
+ "X-Metaproxy-SkipLink", ".* skiplink.com" );
+ pack.request() = gdu_req;
+
+ //create the http response
+
+ const char *resp_buf =
+ "HTTP/1.1 200 OK\r\n"
+ "Content-Type: text/html\r\n"
+ "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
+ "Link: </dir/file.xml>; rel=relative\r\n"
+ "\r\n"
+ "<html><head><title>Hello proxy!</title>"
+ "<style>"
+ "body {"
+ " background-image:url('http://targetsite/images/bg.png');"
+ "}"
+ "</style>"
+ "</head>"
+ "<script>var jslink=\"http://targetsite/webservice.xml\";"
+ "for (i = 0; i<foo; i++) ;\n"
+ "var some=\"foo\"; foo=1;"
+ "</script>"
+ "<body>"
+ "<p>Welcome to our website. It doesn't make it easy to get pro"
+ "xified"
+ "<a href=\"http://targetsite/page2.html\">"
+ " An absolute link</a>"
+ "<a target=_blank href=\"http://targetsite/page3.html\">"
+ " Another abs link</a>"
+ "<a href=\"/docs/page4.html\" />"
+ "<A href=\"cxcx\" />"
+ "<a HREF=\"cx \" onclick=\"foo("foo");\"/>\n"
+ "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
+ "</body></html>";
+
+ const char *resp_expected =
+ "HTTP/1.1 200 OK\r\n"
+ "Content-Length: 686\r\n"
+ "Content-Type: text/html\r\n"
+ "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
+ "Link: </dir/file.xml>; rel=relative\r\n"
+ "\r\n"
+ "<html><head><title>Hello proxy!</title>"
+ "<style>"
+ "body {"
+ " background-image:url('http://proxyhost/proxypath/targetsite/images/bg.png');"
+ "}"
+ "</style>"
+ "</head>"
+ "<script>var jslink=\"http://proxyhost/proxypath/targetsite/webservice.xml\";"
+ "for (i = 0; i<foo; i++) ;\n"
+ "var some=\"bar\"; foo=1;"
+ "</script>"
+ "<body>"
+ "<p>Welcome to our website. It doesn't make it easy to get pro"
+ "xified"
+ "<a href=\"http://proxyhost/proxypath/targetsite/page2.html\">"
+ " An absolute link</a>"
+ "<a target=_blank href=\"http://proxyhost/proxypath/targetsite/page3.html\">"
+ " Another abs link</a>"
+ "<a href=\"/docs/page4.html\"/>"
+ "<A href=\"cycx\"/>"
+ "<a HREF=\"cy \" onclick=\"foo("bar");\"/>\n"
+ "<a href=\"http://www.skiplink.com/page5.html\">skip</a>\n"
+ "</body></html>";
+
+ Z_GDU *gdu_res;
+ mp::odr dec(ODR_DECODE);
+ mp::odr enc(ODR_ENCODE);
+ odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
+ int r = z_GDU(dec, &gdu_res, 0, 0);
+
+ BOOST_CHECK(r);
+ if (r)
{
- if (!esc && in[i] == '\\')
- {
- esc = true;
- continue;
- }
- if (!esc && in[i] == '$') //var
+ BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
+
+ pack.response() = gdu_res;
+
+ //feed to the router
+ pack.router(router).move();
+
+ //analyze the response
+ Z_GDU *gdu_res_rew = pack.response().get();
+ BOOST_CHECK(gdu_res_rew);
+ BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
+
+ Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
+ BOOST_CHECK(hres);
+
+ z_GDU(enc, &gdu_res_rew, 0, 0);
+ char *resp_result;
+ int resp_result_len;
+ resp_result = odr_getbuf(enc, &resp_result_len, 0);
+
+ int equal = ((size_t) resp_result_len == strlen(resp_expected))
+ && !memcmp(resp_result, resp_expected, resp_result_len);
+ BOOST_CHECK(equal);
+
+ if (!equal)
{
- if (i+1 < in.size() && in[i+1] == '{') //ref prefix
- {
- ++i;
- std::string name;
- bool term = false;
- while (++i < in.size())
- {
- if (in[i] == '}') { term = true; break; }
- name += in[i];
- }
- if (!term) throw mp::filter::FilterException
- ("Unterminated var ref in '"+in+"' at "
- + boost::lexical_cast<std::string>(i));
- std::map<std::string, std::string>::const_iterator it
- = vars.find(name);
- if (it != vars.end())
- out += it->second;
- }
- else
- {
- throw mp::filter::FilterException
- ("Malformed or trimmed var ref in '"
- +in+"' at "+boost::lexical_cast<std::string>(i));
- }
- continue;
+ //compare buffers
+ std::cout << "Expected result:\n" << resp_expected << "\n";
+ std::cout << "Got result:\n";
+ fflush(stdout);
+ fwrite(resp_result, 1, resp_result_len, stdout);
+ fflush(stdout);
+ std::cout << "\nGot result buf len: " << resp_result_len
+ << "\n";
}
- //passthru
- out += in[i];
- esc = false;
}
- return out;
}
-
- void configure(
- const std::string & req_uri_rx,
- const std::string & req_uri_pat,
- const std::string & resp_uri_rx,
- const std::string & resp_uri_pat)
- {
- this->req_uri_rx = req_uri_rx;
- this->req_uri_pat = req_uri_pat;
- //pick up names
- parse_groups(req_uri_rx, groups_by_num, groups_by_name);
- this->resp_uri_rx = resp_uri_rx;
- this->resp_uri_pat = resp_uri_pat;
- };
-
-private:
- std::map<std::string, std::string> vars;
- std::string req_uri_rx;
- std::string resp_uri_rx;
- std::string req_uri_pat;
- std::string resp_uri_pat;
- std::map<int, std::string> groups_by_num;
- std::map<std::string, int> groups_by_name;
-
-};
-
-
-BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 )
-{
- try
- {
- FilterHeaderRewrite fhr;
- }
- catch ( ... ) {
+ catch (std::exception & e) {
+ std::cout << e.what();
+ std::cout << std::endl;
BOOST_CHECK (false);
}
}
try
{
mp::RouterChain router;
+ mp::filter::HttpRewrite fhr;
- FilterHeaderRewrite fhr;
- fhr.configure(
- "(?:http\\:\\/\\/s?)?(?<host>[A-Za-z.]+):(?<port>\\d+)",
- "http://${host}:${port}/somepath",
- //rewrite connection close
- "close",
- "open for ${host}");
+ std::string xmlconf =
+ "<?xml version='1.0'?>\n"
+ "<filter xmlns='http://indexdata.com/metaproxy'\n"
+ " id='rewrite1' type='http_rewrite'>\n"
+ " <request verbose=\"1\">\n"
+ " <rule name=\"null\"/>\n"
+ " <rule name=\"url\">\n"
+ " <rewrite from='"
+ "(?<proto>https?://)(?<pxhost>[^ /?#]+)/(?<pxpath>[^ /]+)"
+ "/(?<host>[^ /]+)(?<path>[^ ]*)'\n"
+ " to='${proto}${host}${path}' />\n"
+ " <rewrite from='(?:Host: )(.*)'\n"
+ " to='Host: ${host}' />\n"
+ " </rule>\n"
+ " <content type=\"headers\">\n"
+ " <within header=\"link\" rule=\"null\"/>\n"
+ " <within reqline=\"1\" rule=\"url\"/>\n"
+ " </content>\n"
+ " </request>\n"
+ " <response verbose=\"1\">\n"
+ " <rule name=\"null\"/>\n"
+ " <rule name=\"url\">\n"
+ " <rewrite from='foo' to='bar'/>\n"
+ " <rewrite from='^cx' to='cy'/>\n"
+ " <rewrite from='"
+ "(?<proto>https?://)(?<host>[^/?# "'>]+)/(?<path>[^ "'>]+)'\n"
+ " to='${proto}${pxhost}/${pxpath}/${host}/${path}' />\n"
+ " </rule>\n"
+ " <content type=\"headers\">\n"
+ " <within header=\"link\" rule=\"url\"/>\n"
+ " </content>\n"
+ " <content type=\"quoted-literal\" mime=\".*javascript\">\n"
+ " <within rule=\"url\"/>\n"
+ " </content>\n"
+ " </response>\n"
+ "</filter>\n"
+ ;
+
+ std::cout << xmlconf;
+
+ // reading and parsing XML conf
+ xmlDocPtr doc = xmlParseMemory(xmlconf.c_str(), xmlconf.size());
+ BOOST_CHECK(doc);
+ xmlNode *root_element = xmlDocGetRootElement(doc);
+ fhr.configure(root_element, true, "");
+ xmlFreeDoc(doc);
- mp::filter::HTTPClient hc;
-
router.append(fhr);
- router.append(hc);
// create an http request
mp::Package pack;
mp::odr odr;
- Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
- "http://localhost:80/~jakub/targetsite.php", 0, 1);
+ Z_GDU *gdu_req = z_get_HTTP_Request_uri(odr,
+ "http://proxyhost/proxypath/targetsite/page1.html", 0, 1);
pack.request() = gdu_req;
- //feed to the router
- pack.router(router).move();
+ //create the http response
- //analyze the response
- Z_GDU *gdu_res = pack.response().get();
- BOOST_CHECK(gdu_res);
- BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
-
- Z_HTTP_Response *hres = gdu_res->u.HTTP_Response;
- BOOST_CHECK(hres);
+ const char *resp_buf =
+ "HTTP/1.1 200 OK\r\n"
+ "Content-Type: application/javascript\r\n"
+ "Link: <http://targetsite/file.xml>; rel=absolute\r\n"
+ "Link: </dir/file.xml>; rel=relative\r\n"
+ "\r\n"
+ "// \"\n"
+ "my.location = 'http://targetsite/images/bg.png';\n"
+ "my.other = \"http://targetsite/images/fg.png\";\n"
+ "my.thrd = \"other\";\n"
+ "// \"http://targetsite/images/bg.png\n";
+
+ const char *resp_expected =
+ "HTTP/1.1 200 OK\r\n"
+ "Content-Length: 195\r\n"
+ "Content-Type: application/javascript\r\n"
+ "Link: <http://proxyhost/proxypath/targetsite/file.xml>; rel=absolute\r\n"
+ "Link: </dir/file.xml>; rel=relative\r\n"
+ "\r\n"
+ "// \"\n"
+ "my.location = 'http://proxyhost/proxypath/targetsite/images/bg.png';\n"
+ "my.other = \"http://proxyhost/proxypath/targetsite/images/fg.png\";\n"
+ "my.thrd = \"other\";\n"
+ "// \"http://targetsite/images/bg.png\n";
+
+ Z_GDU *gdu_res;
+ mp::odr dec(ODR_DECODE);
+ mp::odr enc(ODR_ENCODE);
+ odr_setbuf(dec, (char *) resp_buf, strlen(resp_buf), 0);
+ int r = z_GDU(dec, &gdu_res, 0, 0);
+
+ BOOST_CHECK(r);
+ if (r)
+ {
+ BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response);
+ pack.response() = gdu_res;
+
+ //feed to the router
+ pack.router(router).move();
+
+ //analyze the response
+ Z_GDU *gdu_res_rew = pack.response().get();
+ BOOST_CHECK(gdu_res_rew);
+ BOOST_CHECK_EQUAL(gdu_res_rew->which, Z_GDU_HTTP_Response);
+
+ Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response;
+ BOOST_CHECK(hres);
+
+ z_GDU(enc, &gdu_res_rew, 0, 0);
+ char *resp_result;
+ int resp_result_len;
+ resp_result = odr_getbuf(enc, &resp_result_len, 0);
+
+ int equal = ((size_t) resp_result_len == strlen(resp_expected))
+ && !memcmp(resp_result, resp_expected, resp_result_len);
+ BOOST_CHECK(equal);
+
+ if (!equal)
+ {
+ //compare buffers
+ std::cout << "Expected result:\n" << resp_expected << "\n";
+ std::cout << "Got result:\n";
+ fflush(stdout);
+ fwrite(resp_result, 1, resp_result_len, stdout);
+ fflush(stdout);
+ std::cout << "\nGot result buf len: " << resp_result_len
+ << "\n";
+ }
+ }
}
catch (std::exception & e) {
std::cout << e.what();
+ std::cout << std::endl;
BOOST_CHECK (false);
}
}