X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ftest_filter_rewrite.cpp;h=110d6144037ec4d8d29d7e92f4fb2990c22fd67f;hb=586d78659d671683f33ec55f4a7d32b28e345ccd;hp=eee048d79b9a2928d4a84ac2427cfbd804b2d6f8;hpb=c106731ba24ab8570ca249809c7aab46364c303a;p=metaproxy-moved-to-github.git diff --git a/src/test_filter_rewrite.cpp b/src/test_filter_rewrite.cpp index eee048d..110d614 100644 --- a/src/test_filter_rewrite.cpp +++ b/src/test_filter_rewrite.cpp @@ -1,5 +1,5 @@ /* 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 @@ -21,13 +21,16 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include "filter_http_client.hpp" +#include "filter_http_rewrite.hpp" #include -#include "router_chain.hpp" +#include #include #include #include +#include + #define BOOST_AUTO_TEST_MAIN #define BOOST_TEST_DYN_LINK @@ -36,322 +39,195 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA using namespace boost::unit_test; namespace mp = metaproxy_1; -typedef std::pair string_pair; -typedef std::vector spair_vec; -typedef spair_vec::iterator spv_iter; - -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 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; - std::cout << ">> Request headers" << std::endl; - rewrite_reqline(o, hreq, vars); - rewrite_headers(o, hreq->headers, vars); - rewrite_body(o, &hreq->content_buf, &hreq->content_len, vars); - package.request() = gdu; - } - package.move(); - gdu = package.response().get(); - if (gdu && gdu->which == Z_GDU_HTTP_Response) - { - Z_HTTP_Response *hres = gdu->u.HTTP_Response; - std::cout << "Response " << hres->code; - std::cout << "<< Respose headers" << std::endl; - mp::odr o; - rewrite_headers(o, hres->headers, vars); - rewrite_body(o, &hres->content_buf, &hres->content_len, vars); - package.response() = gdu; - } - } - - void rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq, - std::map & vars) const - { - //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 - { - //TODO what about proto - path += z_HTTP_header_lookup(hreq->headers, "Host"); - path += hreq->path; - } - std::cout << "Proxy request URL is " << path << std::endl; - std::string npath = - test_patterns(vars, path, req_uri_pats, req_groups_bynum); - std::cout << "Resp request URL is " << npath << std::endl; - if (!npath.empty()) - hreq->path = odr_strdup(o, npath.c_str()); - } - - void rewrite_headers (mp::odr & o, Z_HTTP_Header *headers, - std::map & vars) const +BOOST_AUTO_TEST_CASE( test_filter_rewrite_1 ) +{ + try { - for (Z_HTTP_Header *header = headers; - header != 0; - header = header->next) - { - std::string sheader(header->name); - sheader += ": "; - sheader += header->value; - std::cout << header->name << ": " << header->value << std::endl; - std::string out = test_patterns(vars, - sheader, - req_uri_pats, req_groups_bynum); - if (!out.empty()) - { - size_t pos = out.find(": "); - if (pos == std::string::npos) - { - std::cout << "Header malformed during rewrite, ignoring"; - continue; - } - header->name = odr_strdup(o, out.substr(0, pos).c_str()); - header->value = odr_strdup(o, out.substr(pos+2, - std::string::npos).c_str()); - } - } - } + mp::RouterChain router; + mp::filter::HttpRewrite fhr; + + std::string xmlconf = + "\n" + "\n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + "\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); - void rewrite_body (mp::odr & o, char **content_buf, int *content_len, - std::map & vars) const - { - if (*content_buf) - { - std::string body(*content_buf); - std::string nbody = - test_patterns(vars, body, req_uri_pats, req_groups_bynum); - if (!nbody.empty()) - { - *content_buf = odr_strdup(o, nbody.c_str()); - *content_len = nbody.size(); - } - } - } + router.append(fhr); + // create an http request + mp::Package pack; - void configure(const xmlNode* ptr, bool test_only, const char *path) {}; + 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; - /** - * Tests pattern from the vector in order and executes recipe on - the first match. - */ - const std::string test_patterns( - std::map & vars, - const std::string & txt, - const spair_vec & uri_pats, - const std::vector > & groups_bynum_vec) - const - { - for (int i = 0; i < uri_pats.size(); i++) + //create the http response + + const char *resp_buf = + "HTTP/1.1 200 OK\r\n" + "Content-Type: text/html\r\n" + "Link: ; rel=absolute\r\n" + "Link: ; rel=relative\r\n" + "\r\n" + "Hello proxy!" + "" + "" + "" + "" + "

Welcome to our website. It doesn't make it easy to get pro" + "xified" + "" + " An absolute link" + "" + " Another abs link" + "" + "" + "\n" + "skip\n" + ""; + + const char *resp_expected = + "HTTP/1.1 200 OK\r\n" + "Content-Length: 686\r\n" + "Content-Type: text/html\r\n" + "Link: ; rel=absolute\r\n" + "Link: ; rel=relative\r\n" + "\r\n" + "Hello proxy!" + "" + "" + "" + "" + "

Welcome to our website. It doesn't make it easy to get pro" + "xified" + "" + " An absolute link" + "" + " Another abs link" + "" + "" + "\n" + "skip\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) { - std::string out = search_replace(vars, txt, - uri_pats[i].first, uri_pats[i].second, - groups_bynum_vec[i]); - if (!out.empty()) return out; - } - return ""; - } + BOOST_CHECK_EQUAL(gdu_res->which, Z_GDU_HTTP_Response); + pack.response() = gdu_res; - const std::string search_replace( - std::map & vars, - const std::string & txt, - const std::string & uri_re, - const std::string & uri_pat, - const std::map & groups_bynum) 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::const_iterator it - = groups_bynum.find(i); - if (it != groups_bynum.end()) - { //it is - std::string name = it->second; - if (!what[i].str().empty()) - vars[name] = what[i]; - } + //feed to the router + pack.router(router).move(); - } - //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; - } + //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); - static void parse_groups( - const spair_vec & uri_pats, - std::vector > & groups_bynum_vec) - { - for (int h = 0; h < uri_pats.size(); h++) - { - int gnum = 0; - bool esc = false; - //regex is first, subpat is second - std::string str = uri_pats[h].first; - //for each pair we have an indexing map - std::map groups_bynum; - 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(i)); - gname += str[i]; - } - if (!term) - throw mp::filter::FilterException - ("Unterminated group name '" + gname - + " in '" + str +"'"); - groups_bynum[gnum] = gname; - std::cout << "Found named group '" << gname - << "' at $" << gnum << std::endl; - } - } - } - esc = false; - } - groups_bynum_vec.push_back(groups_bynum); - } - } + Z_HTTP_Response *hres = gdu_res_rew->u.HTTP_Response; + BOOST_CHECK(hres); - static std::string sub_vars (const std::string & in, - const std::map & vars) - { - std::string out; - bool esc = false; - for (int i = 0; i < in.size(); ++i) - { - if (!esc && in[i] == '\\') - { - esc = true; - continue; - } - if (!esc && in[i] == '$') //var + 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(i)); - std::map::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(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 spair_vec req_uri_pats, - const spair_vec res_uri_pats) - { - //TODO should we really copy them out? - this->req_uri_pats = req_uri_pats; - this->res_uri_pats = res_uri_pats; - //pick up names - parse_groups(req_uri_pats, req_groups_bynum); - parse_groups(res_uri_pats, res_groups_bynum); - }; - -private: - spair_vec req_uri_pats; - spair_vec res_uri_pats; - std::vector > req_groups_bynum; - std::vector > res_groups_bynum; - -}; - - -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); } } @@ -361,56 +237,142 @@ BOOST_AUTO_TEST_CASE( test_filter_rewrite_2 ) try { mp::RouterChain router; + mp::filter::HttpRewrite fhr; + + std::string xmlconf = + "\n" + "\n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + " \n" + "\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); - FilterHeaderRewrite fhr; - - spair_vec vec_req; - vec_req.push_back(std::make_pair( - "(?http\\:\\/\\/s?)(?[^\\/?#]+)\\/(?[^\\/]+)" - "\\/(?[^\\/]+)(?.*)", - "${proto}${host}${path}" - )); - vec_req.push_back(std::make_pair( - "(?:Host\\: )(.*)", - "Host: localhost" - )); - - spair_vec vec_res; - vec_res.push_back(std::make_pair( - "(?http\\:\\/\\/s?)(?[^\\/?#]+)\\/(?[^ >]+)", - "http://${pxhost}/${pxpath}/${host}/${path}" - )); - - fhr.configure(vec_req, vec_res); - - 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://proxyhost/proxypath/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 + + const char *resp_buf = + "HTTP/1.1 200 OK\r\n" + "Content-Type: application/javascript\r\n" + "Link: ; rel=absolute\r\n" + "Link: ; 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: ; rel=absolute\r\n" + "Link: ; 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); - //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); + 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); } }