X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_http_rewrite.cpp;h=a615c2f3520c462490f15a217817dc8ee592a296;hb=dcf44e3b0a3feaa4bb651f3a7acaccdae35c2c46;hp=5259fc36c0bc3aeaf6ab612060e79853b7ad2df5;hpb=ec7e55c86b12403bf0b6e869339c24cbbd4fcdff;p=metaproxy-moved-to-github.git diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp index 5259fc3..a615c2f 100644 --- a/src/filter_http_rewrite.cpp +++ b/src/filter_http_rewrite.cpp @@ -21,64 +21,89 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include "filter_http_rewrite.hpp" +#include "html_parser.hpp" #include #include +#include #include #include +#include -#include #include -#if HAVE_SYS_TYPES_H -#include -#endif - namespace mp = metaproxy_1; namespace yf = mp::filter; namespace metaproxy_1 { namespace filter { - class HttpRewrite::RuleScope { - public: - std::vector tags; - std::vector attrs; - std::string content_type; - }; - class HttpRewrite::Rule { + class HttpRewrite::Replace { public: - enum Section { METHOD, HEADER, BODY }; - std::string regex; + boost::regex re; + boost::smatch what; std::string recipe; std::map group_index; - std::vector scopes; - Section section; const std::string search_replace( std::map & vars, - const std::string & txt) const; - std::string sub_vars ( + const std::string & txt); + std::string sub_vars( const std::map & vars) const; - void parse_groups(); + void parse_groups(std::string pattern); + }; + + class HttpRewrite::Rule { + public: + std::list replace_list; + const std::string test_patterns( + std::map & vars, + const std::string & txt); }; - class HttpRewrite::Rules { + class HttpRewrite::Within { public: - std::vector rules; - void rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq, + std::string header; + std::string attr; + std::string tag; + bool reqline; + RulePtr rule; + }; + + class HttpRewrite::Phase { + public: + Phase(); + std::list within_list; + int m_verbose; + void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq, std::map & vars) const; void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers, std::map & vars) const; - void rewrite_body (mp::odr & o, + void rewrite_body(mp::odr & o, char **content_buf, int *content_len, std::map & vars) const; - const std::string test_patterns( - std::map & vars, - const std::string & txt) const; + }; + class HttpRewrite::Event : public HTMLParserEvent { + void openTagStart(const char *tag, int tag_len); + void anyTagEnd(const char *tag, int tag_len, int close_it); + void attribute(const char *tag, int tag_len, + const char *attr, int attr_len, + const char *value, int val_len, + const char *sep); + void closeTag(const char *tag, int tag_len); + void text(const char *value, int len); + const Phase *m_phase; + WRBUF m_w; + std::stack::const_iterator> s_within; + std::map &m_vars; + public: + Event(const Phase *p, std::map &vars); + ~Event(); + const char *result(); }; } } -yf::HttpRewrite::HttpRewrite() : req_rules(new Rules), res_rules(new Rules) +yf::HttpRewrite::HttpRewrite() : + req_phase(new Phase), res_phase(new Phase) { } @@ -86,7 +111,7 @@ yf::HttpRewrite::~HttpRewrite() { } -void yf::HttpRewrite::process(mp::Package & package) const +void yf::HttpRewrite::process(mp::Package & package) const { yaz_log(YLOG_LOG, "HttpRewrite begins...."); Z_GDU *gdu = package.request().get(); @@ -97,12 +122,11 @@ void yf::HttpRewrite::process(mp::Package & package) const { Z_HTTP_Request *hreq = gdu->u.HTTP_Request; mp::odr o; - req_rules->rewrite_reqline(o, hreq, vars); + req_phase->rewrite_reqline(o, hreq, vars); yaz_log(YLOG_LOG, ">> Request headers"); - req_rules->rewrite_headers(o, hreq->headers, vars); - req_rules->rewrite_body(o, - &hreq->content_buf, &hreq->content_len, - vars); + req_phase->rewrite_headers(o, hreq->headers, vars); + req_phase->rewrite_body(o, + &hreq->content_buf, &hreq->content_len, vars); package.request() = gdu; } package.move(); @@ -113,22 +137,22 @@ void yf::HttpRewrite::process(mp::Package & package) const yaz_log(YLOG_LOG, "Response code %d", hres->code); mp::odr o; yaz_log(YLOG_LOG, "<< Respose headers"); - res_rules->rewrite_headers(o, hres->headers, vars); - res_rules->rewrite_body(o, &hres->content_buf, + res_phase->rewrite_headers(o, hres->headers, vars); + res_phase->rewrite_body(o, &hres->content_buf, &hres->content_len, vars); package.response() = gdu; } } -void yf::HttpRewrite::Rules::rewrite_reqline (mp::odr & o, +void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o, Z_HTTP_Request *hreq, - std::map & vars) const + std::map & vars) const { //rewrite the request line std::string path; if (strstr(hreq->path, "http://") == hreq->path) { - yaz_log(YLOG_LOG, "Path in the method line is absolute, " + yaz_log(YLOG_LOG, "Path in the method line is absolute, " "possibly a proxy request"); path += hreq->path; } @@ -137,87 +161,251 @@ void yf::HttpRewrite::Rules::rewrite_reqline (mp::odr & o, //TODO what about proto path += "http://"; path += z_HTTP_header_lookup(hreq->headers, "Host"); - path += hreq->path; - } - yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str()); - std::string npath = - test_patterns(vars, path); - if (!npath.empty()) - { - yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str()); - hreq->path = odr_strdup(o, npath.c_str()); + path += hreq->path; } + + + std::list::const_iterator it = within_list.begin(); + for (; it != within_list.end(); it++) + if (it->reqline) + { + RulePtr rule = it->rule; + yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str()); + std::string npath = rule->test_patterns(vars, path); + if (!npath.empty()) + { + yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str()); + hreq->path = odr_strdup(o, npath.c_str()); + } + } } -void yf::HttpRewrite::Rules::rewrite_headers(mp::odr & o, +void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o, Z_HTTP_Header *headers, - std::map & vars) const + std::map & vars) const { - for (Z_HTTP_Header *header = headers; - header != 0; - header = header->next) + for (Z_HTTP_Header *header = headers; header; header = header->next) { - std::string sheader(header->name); - sheader += ": "; - sheader += header->value; - yaz_log(YLOG_LOG, "%s: %s", header->name, header->value); - std::string out = test_patterns(vars, sheader); - if (!out.empty()) + std::list::const_iterator it = within_list.begin(); + for (; it != within_list.end(); it++) { - size_t pos = out.find(": "); - if (pos == std::string::npos) + if (it->header.length() > 0 && + yaz_strcasecmp(it->header.c_str(), header->name) == 0) { - yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring"); - continue; + std::string sheader(header->name); + sheader += ": "; + sheader += header->value; + + RulePtr rule = it->rule; + std::string out = rule->test_patterns(vars, sheader); + if (!out.empty()) + { + size_t pos = out.find(": "); + if (pos == std::string::npos) + { + yaz_log(YLOG_LOG, "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()); + } } - 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()); } } } -void yf::HttpRewrite::Rules::rewrite_body (mp::odr & o, - char **content_buf, +void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o, + char **content_buf, int *content_len, - std::map & vars) const + std::map & vars) const { if (*content_buf) { - std::string body(*content_buf); - std::string nbody = - test_patterns(vars, body); - if (!nbody.empty()) + int i; + for (i = 0; i < *content_len; i++) + if ((*content_buf)[i] == 0) + return; // binary content. skip + + HTMLParser parser; + Event ev(this, vars); + + parser.set_verbose(m_verbose); + + std::string buf(*content_buf, *content_len); + + parser.parse(ev, buf.c_str()); + const char *res = ev.result(); + *content_buf = odr_strdup(o, res); + *content_len = strlen(res); + } +} + +yf::HttpRewrite::Event::Event(const Phase *p, + std::map & vars + ) : m_phase(p), m_vars(vars) +{ + m_w = wrbuf_alloc(); +} + +yf::HttpRewrite::Event::~Event() +{ + wrbuf_destroy(m_w); +} + +const char *yf::HttpRewrite::Event::result() +{ + return wrbuf_cstr(m_w); +} + +void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len) +{ + wrbuf_putc(m_w, '<'); + wrbuf_write(m_w, tag, tag_len); + + std::string t(tag, tag_len); + std::list::const_iterator it = m_phase->within_list.begin(); + for (; it != m_phase->within_list.end(); it++) + { + if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(), + t.c_str()) == 0) { - *content_buf = odr_strdup(o, nbody.c_str()); - *content_len = nbody.size(); + std::vector attr; + boost::split(attr, it->attr, boost::is_any_of(",")); + size_t i; + for (i = 0; i < attr.size(); i++) + { + if (attr[i].compare("#text") == 0) + { + s_within.push(it); + return; + } + } } } } +void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len, + int close_it) +{ + if (close_it) + { + if (!s_within.empty()) + { + std::list::const_iterator it = s_within.top(); + std::string t(tag, tag_len); + if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0) + s_within.pop(); + } + } + if (close_it) + wrbuf_putc(m_w, '/'); + wrbuf_putc(m_w, '>'); +} + +void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len, + const char *attr, int attr_len, + const char *value, int val_len, + const char *sep) +{ + std::list::const_iterator it = m_phase->within_list.begin(); + bool subst = false; + + for (; it != m_phase->within_list.end(); it++) + { + std::string t(tag, tag_len); + if (it->tag.length() == 0 || + yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0) + { + std::string a(attr, attr_len); + std::vector attr; + boost::split(attr, it->attr, boost::is_any_of(",")); + size_t i; + for (i = 0; i < attr.size(); i++) + { + if (attr[i].compare("#text") && + yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0) + subst = true; + } + } + if (subst) + break; + } + + wrbuf_putc(m_w, ' '); + wrbuf_write(m_w, attr, attr_len); + if (value) + { + wrbuf_puts(m_w, "="); + wrbuf_puts(m_w, sep); + + std::string output; + if (subst) + { + std::string input(value, val_len); + output = it->rule->test_patterns(m_vars, input); + } + if (output.empty()) + wrbuf_write(m_w, value, val_len); + else + wrbuf_puts(m_w, output.c_str()); + wrbuf_puts(m_w, sep); + } +} + +void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len) +{ + if (!s_within.empty()) + { + std::list::const_iterator it = s_within.top(); + std::string t(tag, tag_len); + if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0) + s_within.pop(); + } + wrbuf_puts(m_w, "::const_iterator it = m_phase->within_list.end(); + if (!s_within.empty()) + it = s_within.top(); + std::string output; + if (it != m_phase->within_list.end()) + { + std::string input(value, len); + output = it->rule->test_patterns(m_vars, input); + } + if (output.empty()) + wrbuf_write(m_w, value, len); + else + wrbuf_puts(m_w, output.c_str()); +} + /** * Tests pattern from the vector in order and executes recipe on the first match. */ -const std::string yf::HttpRewrite::Rules::test_patterns( +const std::string yf::HttpRewrite::Rule::test_patterns( std::map & vars, - const std::string & txt) const + const std::string & txt) { - for (size_t i = 0; i < rules.size(); i++) + std::list::iterator it = replace_list.begin(); + + for (; it != replace_list.end(); it++) { - std::string out = rules[i].search_replace(vars, txt); + std::string out = it->search_replace(vars, txt); if (!out.empty()) return out; } return ""; } -const std::string yf::HttpRewrite::Rule::search_replace( +const std::string yf::HttpRewrite::Replace::search_replace( std::map & vars, - const std::string & txt) const + const std::string & txt) { - //exec regex against value - boost::regex re(regex); - boost::smatch what; std::string::const_iterator start, end; start = txt.begin(); end = txt.end(); @@ -230,16 +418,15 @@ const std::string yf::HttpRewrite::Rule::search_replace( //check if the group is named std::map::const_iterator it = group_index.find(i); - if (it != group_index.end()) + if (it != group_index.end()) { //it is - if (!what[i].str().empty()) - vars[it->second] = what[i]; + vars[it->second] = what[i]; } } //prepare replacement string std::string rvalue = sub_vars(vars); - yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'", + yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'", what.str(0).c_str(), rvalue.c_str()); out.append(start, what[0].first); out.append(rvalue); @@ -251,14 +438,16 @@ const std::string yf::HttpRewrite::Rule::search_replace( return out; } -void yf::HttpRewrite::Rule::parse_groups() +void yf::HttpRewrite::Replace::parse_groups(std::string pattern) { int gnum = 0; bool esc = false; - const std::string & str = regex; + const std::string &str = pattern; + std::string res; yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str()); for (size_t i = 0; i < str.size(); ++i) { + res += str[i]; if (!esc && str[i] == '\\') { esc = true; @@ -267,13 +456,15 @@ void yf::HttpRewrite::Rule::parse_groups() if (!esc && str[i] == '(') //group starts { gnum++; - if (i+1 < str.size() && str[i+1] == '?') //group with attrs + if (i+1 < str.size() && str[i+1] == '?') //group with attrs { - i++; + i++; if (i+1 < str.size() && str[i+1] == ':') //non-capturing { if (gnum > 0) gnum--; + res += str[i]; i++; + res += str[i]; continue; } if (i+1 < str.size() && str[i+1] == 'P') //optional, python @@ -286,18 +477,18 @@ void yf::HttpRewrite::Rule::parse_groups() while (++i < str.size()) { if (str[i] == '>') { term = true; break; } - if (!isalnum(str[i])) + if (!isalnum(str[i])) throw mp::filter::FilterException ("Only alphanumeric chars allowed, found " - " in '" - + str - + "' at " - + boost::lexical_cast(i)); + " in '" + + str + + "' at " + + boost::lexical_cast(i)); gname += str[i]; } if (!term) throw mp::filter::FilterException - ("Unterminated group name '" + gname + ("Unterminated group name '" + gname + " in '" + str +"'"); group_index[gnum] = gname; yaz_log(YLOG_LOG, "Found named group '%s' at $%d", @@ -307,10 +498,11 @@ void yf::HttpRewrite::Rule::parse_groups() } esc = false; } + re = res; } -std::string yf::HttpRewrite::Rule::sub_vars ( - const std::map & vars) const +std::string yf::HttpRewrite::Replace::sub_vars( + const std::map & vars) const { std::string out; bool esc = false; @@ -329,7 +521,7 @@ std::string yf::HttpRewrite::Rule::sub_vars ( ++i; std::string name; bool term = false; - while (++i < in.size()) + while (++i < in.size()) { if (in[i] == '}') { term = true; break; } name += in[i]; @@ -348,7 +540,7 @@ std::string yf::HttpRewrite::Rule::sub_vars ( { throw mp::filter::FilterException ("Malformed or trimmed var ref in '" - +in+"' at "+boost::lexical_cast(i)); + +in+"' at "+boost::lexical_cast(i)); } continue; } @@ -359,41 +551,95 @@ std::string yf::HttpRewrite::Rule::sub_vars ( return out; } -void yf::HttpRewrite::configure_rules(const xmlNode *ptr, - Rules & rules) +yf::HttpRewrite::Phase::Phase() : m_verbose(0) +{ +} + +void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase) { + static const char *names[2] = { "verbose", 0 }; + std::string values[1]; + values[0] = "0"; + mp::xml::parse_attr(ptr, names, values); + + phase.m_verbose = atoi(values[0].c_str()); + + std::map rules; for (ptr = ptr->children; ptr; ptr = ptr->next) { if (ptr->type != XML_ELEMENT_NODE) continue; - else if (!strcmp((const char *) ptr->name, "rewrite")) + else if (!strcmp((const char *) ptr->name, "rule")) { - Rule rule; - const struct _xmlAttr *attr; - for (attr = ptr->properties; attr; attr = attr->next) + static const char *names[2] = { "name", 0 }; + std::string values[1]; + values[0] = "default"; + mp::xml::parse_attr(ptr, names, values); + + RulePtr rule(new Rule); + for (xmlNode *p = ptr->children; p; p = p->next) { - if (!strcmp((const char *) attr->name, "from")) - rule.regex = mp::xml::get_text(attr->children); - else if (!strcmp((const char *) attr->name, "to")) - rule.recipe = mp::xml::get_text(attr->children); + if (p->type != XML_ELEMENT_NODE) + continue; + if (!strcmp((const char *) p->name, "rewrite")) + { + Replace replace; + std::string from; + const struct _xmlAttr *attr; + for (attr = p->properties; attr; attr = attr->next) + { + if (!strcmp((const char *) attr->name, "from")) + from = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "to")) + replace.recipe = mp::xml::get_text(attr->children); + else + throw mp::filter::FilterException + ("Bad attribute " + + std::string((const char *) attr->name) + + " in rewrite section of http_rewrite"); + } + yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'", + from.c_str(), replace.recipe.c_str()); + if (!from.empty()) + { + replace.parse_groups(from); + rule->replace_list.push_back(replace); + } + } else throw mp::filter::FilterException - ("Bad attribute " - + std::string((const char *) attr->name) - + " in rewrite section of http_rewrite"); + ("Bad element " + + std::string((const char *) p->name) + + " in http_rewrite filter"); } - yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'", - rule.regex.c_str(), rule.recipe.c_str()); - rule.parse_groups(); - if (!rule.regex.empty()) - rules.rules.push_back(rule); + rules[values[0]] = rule; + } + else if (!strcmp((const char *) ptr->name, "within")) + { + static const char *names[6] = + { "header", "attr", "tag", "rule", "reqline", 0 }; + std::string values[5]; + mp::xml::parse_attr(ptr, names, values); + Within w; + w.header = values[0]; + w.attr = values[1]; + w.tag = values[2]; + std::map::const_iterator it = + rules.find(values[3]); + if (it == rules.end()) + throw mp::filter::FilterException + ("Reference to non-existing rule '" + values[3] + + "' in http_rewrite filter"); + w.rule = it->second; + w.reqline = values[4] == "1"; + phase.within_list.push_back(w); } else { throw mp::filter::FilterException - ("Bad element o" + ("Bad element " + std::string((const char *) ptr->name) - + " in http_rewrite1 filter"); + + " in http_rewrite filter"); } } } @@ -407,11 +653,11 @@ void yf::HttpRewrite::configure(const xmlNode * ptr, bool test_only, continue; else if (!strcmp((const char *) ptr->name, "request")) { - configure_rules(ptr, *req_rules); + configure_phase(ptr, *req_phase); } else if (!strcmp((const char *) ptr->name, "response")) { - configure_rules(ptr, *res_rules); + configure_phase(ptr, *res_phase); } else {