X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_http_rewrite.cpp;h=ddcac86f8791dba523271d361c2e6d5a63b4519a;hb=aed0227951a728103325801ebae842b96f57745f;hp=9d672fb8da54a733552036eeccd152b5238bcfa0;hpb=213d7bae1233185595186f881a150ee238dac64f;p=metaproxy-moved-to-github.git diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp index 9d672fb..ddcac86 100644 --- a/src/filter_http_rewrite.cpp +++ b/src/filter_http_rewrite.cpp @@ -26,6 +26,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include +#include #include #include #include @@ -39,15 +40,14 @@ namespace metaproxy_1 { namespace filter { class HttpRewrite::Replace { public: - std::string regex; + bool start_anchor; + boost::regex re; + boost::smatch what; std::string recipe; std::map group_index; - const std::string search_replace( - std::map & vars, - const std::string & txt) const; - std::string sub_vars ( + std::string sub_vars( const std::map & vars) const; - void parse_groups(); + void parse_groups(std::string pattern); }; class HttpRewrite::Rule { @@ -55,41 +55,54 @@ namespace metaproxy_1 { std::list replace_list; const std::string test_patterns( std::map & vars, - const std::string & txt) const; + const std::string & txt, bool anchor); }; class HttpRewrite::Within { public: std::string header; std::string attr; std::string tag; + bool reqline; RulePtr rule; }; - class HttpRewrite::Phase { + class HttpRewrite::Content { public: + std::string type; + boost::regex content_re; std::list within_list; + void configure(const xmlNode *ptr, + std::map &rules); + }; + class HttpRewrite::Phase { + public: + Phase(); + int m_verbose; + std::list content_list; 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, - char **content_buf, int *content_len, - std::map & vars) const; + const char *content_type, + char **content_buf, int *content_len, + std::map & vars) const; }; class HttpRewrite::Event : public HTMLParserEvent { - void openTagStart(const char *name); - void anyTagEnd(const char *name); - void attribute(const char *tagName, - const char *name, - const char *value, - int val_len); - void closeTag(const char *name); + 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; + const Content *m_content; WRBUF m_w; - std::list::const_iterator enabled_within; + std::stack::const_iterator> s_within; + std::map &m_vars; public: - Event(const Phase *p); + Event(const Content *p, std::map &vars); ~Event(); const char *result(); }; @@ -120,7 +133,10 @@ void yf::HttpRewrite::process(mp::Package & package) const yaz_log(YLOG_LOG, ">> Request headers"); req_phase->rewrite_headers(o, hreq->headers, vars); req_phase->rewrite_body(o, - &hreq->content_buf, &hreq->content_len, vars); + z_HTTP_header_lookup(hreq->headers, + "Content-Type"), + &hreq->content_buf, &hreq->content_len, + vars); package.request() = gdu; } package.move(); @@ -132,8 +148,11 @@ void yf::HttpRewrite::process(mp::Package & package) const mp::odr o; yaz_log(YLOG_LOG, "<< Respose headers"); res_phase->rewrite_headers(o, hres->headers, vars); - res_phase->rewrite_body(o, &hres->content_buf, - &hres->content_len, vars); + res_phase->rewrite_body(o, + z_HTTP_header_lookup(hres->headers, + "Content-Type"), + &hres->content_buf, &hres->content_len, + vars); package.response() = gdu; } } @@ -153,96 +172,130 @@ void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o, else { //TODO what about proto + const char *host = z_HTTP_header_lookup(hreq->headers, "Host"); + if (!host) + return; + path += "http://"; - path += z_HTTP_header_lookup(hreq->headers, "Host"); + path += host; path += hreq->path; } - std::list::const_iterator it = within_list.begin(); - if (it != within_list.end()) - { - RulePtr rule = it->rule; + std::list::const_iterator cit = content_list.begin(); + for (; cit != content_list.end(); cit++) + if (cit->type == "headers") + break; + + if (cit == content_list.end()) + return; - yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str()); - std::string npath = rule->test_patterns(vars, path); - if (!npath.empty()) + std::list::const_iterator it = cit->within_list.begin(); + for (; it != cit->within_list.end(); it++) + if (it->reqline) { - yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str()); - hreq->path = odr_strdup(o, npath.c_str()); + RulePtr rule = it->rule; + yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str()); + std::string npath = rule->test_patterns(vars, path, true); + 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::Phase::rewrite_headers(mp::odr & o, Z_HTTP_Header *headers, std::map & vars) const { - for (Z_HTTP_Header *header = headers; - header != 0; - header = header->next) - { - std::string sheader(header->name); - sheader += ": "; - sheader += header->value; - yaz_log(YLOG_LOG, "%s: %s", header->name, header->value); + std::list::const_iterator cit = content_list.begin(); + for (; cit != content_list.end(); cit++) + if (cit->type == "headers") + break; - std::list::const_iterator it = within_list.begin(); - if (it == within_list.end()) - continue; - RulePtr rule = it->rule; + if (cit == content_list.end()) + return; - std::string out = rule->test_patterns(vars, sheader); - if (!out.empty()) + for (Z_HTTP_Header *header = headers; header; header = header->next) + { + std::list::const_iterator it = cit->within_list.begin(); + for (; it != cit->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, true); + 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::Phase::rewrite_body(mp::odr & o, - char **content_buf, - int *content_len, - std::map & vars) const +void yf::HttpRewrite::Phase::rewrite_body( + mp::odr &o, + const char *content_type, + char **content_buf, + int *content_len, + std::map & vars) const { + std::list::const_iterator cit = content_list.begin(); + for (; cit != content_list.end(); cit++) + { + yaz_log(YLOG_LOG, "rewrite_body: content_type=%s type=%s", + content_type, cit->type.c_str()); + if (cit->type != "headers" + && regex_match(content_type, cit->content_re)) + break; + } + if (cit == content_list.end()) + return; + if (*content_buf) { - HTMLParser parser; - Event ev(this); - std::string buf(*content_buf, *content_len); + int i; + for (i = 0; i < *content_len; i++) + if ((*content_buf)[i] == 0) + return; // binary content. skip - parser.parse(ev, buf.c_str()); - std::cout << "RES\n" << ev.result() << std::endl; - std::cout << "-----" << std::endl; + if (cit->type == "html") + { + HTMLParser parser; + Event ev(&*cit, vars); + parser.set_verbose(m_verbose); - std::list::const_iterator it = within_list.begin(); - if (it != within_list.end()) - { - RulePtr rule = it->rule; + std::string buf(*content_buf, *content_len); - std::string body(*content_buf); - std::string nbody = rule->test_patterns(vars, body); - if (!nbody.empty()) - { - *content_buf = odr_strdup(o, nbody.c_str()); - *content_len = nbody.size(); - } + 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) : m_phase(p) +yf::HttpRewrite::Event::Event(const Content *p, + std::map & vars + ) : m_content(p), m_vars(vars) { m_w = wrbuf_alloc(); - enabled_within = m_phase->within_list.end(); } yf::HttpRewrite::Event::~Event() @@ -255,186 +308,195 @@ const char *yf::HttpRewrite::Event::result() return wrbuf_cstr(m_w); } -void yf::HttpRewrite::Event::openTagStart(const char *name) +void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len) { - // check if there is - if (enabled_within == m_phase->within_list.end()) + wrbuf_putc(m_w, '<'); + wrbuf_write(m_w, tag, tag_len); + + std::string t(tag, tag_len); + std::list::const_iterator it = m_content->within_list.begin(); + for (; it != m_content->within_list.end(); it++) { - 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) { - if (it->tag.length() > 0 && it->tag.compare(name) == 0) + std::vector attr; + boost::split(attr, it->attr, boost::is_any_of(",")); + size_t i; + for (i = 0; i < attr.size(); i++) { - enabled_within = it; + if (attr[i].compare("#text") == 0) + { + s_within.push(it); + return; + } } } } - wrbuf_putc(m_w, '<'); - wrbuf_puts(m_w, name); } -void yf::HttpRewrite::Event::anyTagEnd(const char *name) +void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len, + int close_it) { - std::list::const_iterator it = enabled_within; - if (it != m_phase->within_list.end()) + if (close_it) { - if (it->tag.compare(name) == 0) + if (!s_within.empty()) { - enabled_within = m_phase->within_list.end(); + 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 *tagName, - const char *name, - const char *value, - int val_len) +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 = enabled_within; + std::list::const_iterator it = m_content->within_list.begin(); bool subst = false; - if (it == m_phase->within_list.end()) + for (; it != m_content->within_list.end(); it++) { - // no active within tag.. see if a attr rule without tag applies - it = m_phase->within_list.begin(); - 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) { - if (it->attr.length() > 0 && it->tag.length() == 0) - break; - } - } - if (it != m_phase->within_list.end()) - { - 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") && attr[i].compare(tagName) == 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++) { - subst = true; + 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_puts(m_w, name); - wrbuf_puts(m_w, "\""); - wrbuf_write(m_w, value, val_len); - if (subst) - wrbuf_puts(m_w, " SUBST"); - wrbuf_puts(m_w, "\""); -} - -void yf::HttpRewrite::Event::closeTag(const char *name) -{ - std::list::const_iterator it = enabled_within; - if (it != m_phase->within_list.end()) + wrbuf_write(m_w, attr, attr_len); + if (value) { - if (it->tag.compare(name) == 0) + wrbuf_puts(m_w, "="); + wrbuf_puts(m_w, sep); + + std::string output; + if (subst) { - enabled_within = m_phase->within_list.end(); + std::string input(value, val_len); + output = it->rule->test_patterns(m_vars, input, true); } + if (output.empty()) + wrbuf_write(m_w, value, val_len); + else + wrbuf_puts(m_w, output.c_str()); + wrbuf_puts(m_w, sep); } - wrbuf_puts(m_w, "::const_iterator it = enabled_within; - bool subst = false; - - if (it != m_phase->within_list.end()) + if (!s_within.empty()) { - subst = true; - if (it->attr.length() > 0) - { - subst = false; - 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) - { - subst = true; - } - } - } + 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_write(m_w, value, len); - if (subst) - wrbuf_puts(m_w, ""); + wrbuf_puts(m_w, " & vars, - const std::string & txt) const +void yf::HttpRewrite::Event::text(const char *value, int len) { - std::list::const_iterator it = replace_list.begin(); - - for (; it != replace_list.end(); it++) + std::list::const_iterator it = m_content->within_list.end(); + if (!s_within.empty()) + it = s_within.top(); + std::string output; + if (it != m_content->within_list.end()) { - std::string out = it->search_replace(vars, txt); - if (!out.empty()) return out; + std::string input(value, len); + output = it->rule->test_patterns(m_vars, input, false); } - return ""; + if (output.empty()) + wrbuf_write(m_w, value, len); + else + wrbuf_puts(m_w, output.c_str()); } -const std::string yf::HttpRewrite::Replace::search_replace( +const std::string yf::HttpRewrite::Rule::test_patterns( std::map & vars, - const std::string & txt) const + const std::string & txt, bool anchor) { - //exec regex against value - boost::regex re(regex); - boost::smatch what; + bool first = anchor; + std::string out; 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 + while (1) { + std::list::iterator bit = replace_list.end(); + { + std::string::const_iterator best_pos = txt.end(); + std::list::iterator it = replace_list.begin(); + for (; it != replace_list.end(); it++) + { + if (it->start_anchor && !first) + continue; + if (regex_search(start, end, it->what, it->re)) + { + if (it->what[0].first < best_pos) + { + best_pos = it->what[0].first; + bit = it; + } + } + } + if (bit == replace_list.end()) + break; + } + first = false; size_t i; - for (i = 1; i < what.size(); ++i) + for (i = 1; i < bit->what.size(); ++i) { //check if the group is named - std::map::const_iterator it - = group_index.find(i); - if (it != group_index.end()) + std::map::const_iterator git + = bit->group_index.find(i); + if (git != bit->group_index.end()) { //it is - if (!what[i].str().empty()) - vars[it->second] = what[i]; + vars[git->second] = bit->what[i]; } } //prepare replacement string - std::string rvalue = sub_vars(vars); + std::string rvalue = bit->sub_vars(vars); yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'", - what.str(0).c_str(), rvalue.c_str()); - out.append(start, what[0].first); + bit->what.str(0).c_str(), rvalue.c_str()); + out.append(start, bit->what[0].first); out.append(rvalue); - start = what[0].second; //move search forward + start = bit->what[0].second; //move search forward } - //if we had a match cat the last part if (start != txt.begin()) out.append(start, end); return out; } -void yf::HttpRewrite::Replace::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; + start_anchor = str[0] == '^'; yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str()); for (size_t i = 0; i < str.size(); ++i) { @@ -489,11 +551,11 @@ void yf::HttpRewrite::Replace::parse_groups() } esc = false; } - regex = res; + re = res; } -std::string yf::HttpRewrite::Replace::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; @@ -542,9 +604,49 @@ std::string yf::HttpRewrite::Replace::sub_vars ( return out; } +yf::HttpRewrite::Phase::Phase() : m_verbose(0) +{ +} + +void yf::HttpRewrite::Content::configure( + const xmlNode *ptr, std::map &rules) +{ + for (; ptr; ptr = ptr->next) + { + if (ptr->type != XML_ELEMENT_NODE) + continue; + 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"; + within_list.push_back(w); + } + } +} 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) { @@ -565,11 +667,12 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase) 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")) - replace.regex = mp::xml::get_text(attr->children); + from = mp::xml::get_text(attr->children); else if (!strcmp((const char *) attr->name, "to")) replace.recipe = mp::xml::get_text(attr->children); else @@ -579,10 +682,12 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase) + " in rewrite section of http_rewrite"); } yaz_log(YLOG_LOG, "Found rewrite rule from '%s' to '%s'", - replace.regex.c_str(), replace.recipe.c_str()); - replace.parse_groups(); - if (!replace.regex.empty()) + 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 @@ -590,27 +695,28 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase) + std::string((const char *) p->name) + " in http_rewrite filter"); } - if (!rule->replace_list.empty()) - rules[values[0]] = rule; + rules[values[0]] = rule; } - else if (!strcmp((const char *) ptr->name, "within")) + else if (!strcmp((const char *) ptr->name, "content")) { - static const char *names[5] = - { "header", "attr", "tag", "rule", 0 }; - std::string values[4]; + static const char *names[3] = + { "type", "mime", 0 }; + std::string values[2]; 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; - phase.within_list.push_back(w); + if (values[0].empty()) + { + throw mp::filter::FilterException + ("Missing attribute, type for for element " + + std::string((const char *) ptr->name) + + " in http_rewrite filter"); + } + Content c; + + c.type = values[0]; + // if (!values[1].empty()) + c.content_re = values[1]; + c.configure(ptr->children, rules); + phase.content_list.push_back(c); } else {