X-Git-Url: http://sru.miketaylor.org.uk/?a=blobdiff_plain;f=src%2Ffilter_http_rewrite.cpp;h=fd1b8876a8801b3d14e7449dcc66b595db9035e0;hb=8bcd4c3e063e932b2f80f9491ec0af66d3da5c2e;hp=098c349bd95c6f3a51d234d19599a21762d13542;hpb=6bc88607a5726bcf482ceb23d0cf5e7dd9db8341;p=metaproxy-moved-to-github.git diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp index 098c349..fd1b887 100644 --- a/src/filter_http_rewrite.cpp +++ b/src/filter_http_rewrite.cpp @@ -28,6 +28,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include +#include #include @@ -61,6 +62,7 @@ namespace metaproxy_1 { std::string header; std::string attr; std::string tag; + bool reqline; RulePtr rule; }; @@ -76,15 +78,22 @@ namespace metaproxy_1 { std::map & vars) const; }; class HttpRewrite::Event : public HTMLParserEvent { - public: void openTagStart(const char *name); - void anyTagEnd(const char *name); + void anyTagEnd(const char *name, int close_it); void attribute(const char *tagName, const char *name, const char *value, int val_len); void closeTag(const char *name); void text(const char *value, int len); + const Phase *m_phase; + WRBUF m_w; + std::list::const_iterator enabled_within; + std::map &m_vars; + public: + Event(const Phase *p, std::map &vars); + ~Event(); + const char *result(); }; } } @@ -151,51 +160,54 @@ void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o, path += hreq->path; } - std::list::const_iterator it = within_list.begin(); - if (it != within_list.end()) - { - 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()) + std::list::const_iterator it = within_list.begin(); + for (; it != 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); + 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) + 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::list::const_iterator it = within_list.begin(); - if (it == within_list.end()) - continue; - RulePtr rule = it->rule; - - std::string out = rule->test_patterns(vars, sheader); - if (!out.empty()) + 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()); } } } @@ -207,34 +219,75 @@ void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o, { if (*content_buf) { + int i; + for (i = 0; i < *content_len; i++) + if ((*content_buf)[i] == 0) + return; // binary content. skip + HTMLParser parser; - Event ev; + Event ev(this, vars); std::string buf(*content_buf, *content_len); parser.parse(ev, buf.c_str()); - std::list::const_iterator it = within_list.begin(); - if (it != within_list.end()) - { - RulePtr rule = it->rule; - - 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(); - } - } + 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(); + enabled_within = m_phase->within_list.end(); +} + +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 *name) { + // check if there is + if (enabled_within == m_phase->within_list.end()) + { + std::list::const_iterator it = + m_phase->within_list.begin(); + for (; it != m_phase->within_list.end(); it++) + { + if (it->tag.length() > 0 && it->tag.compare(name) == 0) + { + enabled_within = it; + } + } + } + wrbuf_putc(m_w, '<'); + wrbuf_puts(m_w, name); } -void yf::HttpRewrite::Event::anyTagEnd(const char *name) +void yf::HttpRewrite::Event::anyTagEnd(const char *name, int close_it) { + if (close_it) + { + std::list::const_iterator it = enabled_within; + if (it != m_phase->within_list.end()) + { + if (it->tag.compare(name) == 0) + { + enabled_within = m_phase->within_list.end(); + } + } + } + if (close_it) + wrbuf_putc(m_w, '/'); + wrbuf_putc(m_w, '>'); } void yf::HttpRewrite::Event::attribute(const char *tagName, @@ -242,15 +295,90 @@ void yf::HttpRewrite::Event::attribute(const char *tagName, const char *value, int val_len) { -} + std::list::const_iterator it = m_phase->within_list.begin(); + bool subst = false; + for (; it != m_phase->within_list.end(); it++) + { + if (it->tag.length() == 0 || it->tag.compare(tagName) == 0) + { + 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(name) == 0) + subst = true; + } + } + if (subst) + break; + } + + wrbuf_putc(m_w, ' '); + wrbuf_puts(m_w, name); + wrbuf_puts(m_w, "=\""); + + 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, "\""); +} void yf::HttpRewrite::Event::closeTag(const char *name) { + std::list::const_iterator it = enabled_within; + if (it != m_phase->within_list.end()) + { + if (it->tag.compare(name) == 0) + { + enabled_within = m_phase->within_list.end(); + } + } + wrbuf_puts(m_w, "::const_iterator it = enabled_within; + bool subst = false; + + if (it != m_phase->within_list.end()) + { + 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::string output; + if (subst) + { + 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()); } @@ -473,14 +601,13 @@ 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")) { - static const char *names[5] = - { "header", "attr", "tag", "rule", 0 }; - std::string values[4]; + 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]; @@ -493,6 +620,7 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase) ("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