X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_http_rewrite.cpp;h=1badb44cee656f8a0744288b30307b207cb3a38e;hb=586d78659d671683f33ec55f4a7d32b28e345ccd;hp=90e3152b0362195a0b221c37a60bd10ae16c8650;hpb=41ddc790e888e2b0cfe5da66a34e26e737e20b54;p=metaproxy-moved-to-github.git

diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp
index 90e3152..1badb44 100644
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_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
@@ -26,6 +26,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <yaz/zgdu.h>
 #include <yaz/log.h>
 
+#include <stack>
 #include <boost/regex.hpp>
 #include <boost/lexical_cast.hpp>
 #include <boost/algorithm/string.hpp>
@@ -39,60 +40,85 @@ namespace metaproxy_1 {
     namespace filter {
         class HttpRewrite::Replace {
         public:
-            std::string regex;
+            bool start_anchor;
+            boost::regex re;
             std::string recipe;
             std::map<int, std::string> group_index;
-            const std::string search_replace(
-                std::map<std::string, std::string> & vars,
-                const std::string & txt) const;
-            std::string sub_vars (
+            std::string sub_vars(
                 const std::map<std::string, std::string> & vars) const;
-            void parse_groups();
+            void parse_groups(std::string pattern);
         };
 
         class HttpRewrite::Rule {
         public:
             std::list<Replace> replace_list;
-            const std::string test_patterns(
-                std::map<std::string, std::string> & vars,
-                const std::string & txt) const;
+            bool test_patterns(
+                std::map<std::string, std::string> &vars,
+                std::string &txt, bool anchor,
+                std::list<boost::regex> &skip_list);
         };
         class HttpRewrite::Within {
         public:
-            std::string header;
-            std::string attr;
-            std::string tag;
+            boost::regex header;
+            boost::regex attr;
+            boost::regex tag;
+            std::string type;
             bool reqline;
             RulePtr rule;
+            bool exec(std::map<std::string, std::string> &vars,
+                      std::string &txt, bool anchor,
+                      std::list<boost::regex> &skip_list) const;
         };
 
+        class HttpRewrite::Content {
+        public:
+            std::string type;
+            boost::regex content_re;
+            std::list<Within> within_list;
+            void configure(const xmlNode *ptr,
+                           std::map<std::string, RulePtr > &rules);
+            void quoted_literal(std::string &content,
+                                std::map<std::string, std::string> &vars,
+                                std::list<boost::regex> & skip_list) const;
+            void parse(int verbose, std::string &content,
+                       std::map<std::string, std::string> & vars,
+                       std::list<boost::regex> & skip_list ) const;
+        };
         class HttpRewrite::Phase {
         public:
             Phase();
-            std::list<Within> within_list;
             int m_verbose;
+            std::list<Content> content_list;
+            void read_skip_headers(Z_HTTP_Request *hreq,
+                                   std::list<boost::regex> &skip_list);
             void rewrite_reqline(mp::odr & o, Z_HTTP_Request *hreq,
                 std::map<std::string, std::string> & vars) const;
             void rewrite_headers(mp::odr & o, Z_HTTP_Header *headers,
                 std::map<std::string, std::string> & vars) const;
             void rewrite_body(mp::odr & o,
-                char **content_buf, int *content_len,
-                std::map<std::string, std::string> & vars) const;
+                              const char *content_type,
+                              char **content_buf, int *content_len,
+                              std::map<std::string, std::string> & vars,
+                              std::list<boost::regex> & skip_list ) 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 *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<Within>::const_iterator enabled_within;
+            std::stack<std::list<Within>::const_iterator> s_within;
             std::map<std::string, std::string> &m_vars;
+            std::list<boost::regex> & m_skips;
         public:
-            Event(const Phase *p, std::map<std::string, std::string> &vars);
+            Event(const Content *p,
+                  std::map<std::string, std::string> &vars,
+                  std::list<boost::regex> & skip_list );
             ~Event();
             const char *result();
         };
@@ -115,15 +141,22 @@ void yf::HttpRewrite::process(mp::Package & package) const
     //map of request/response vars
     std::map<std::string, std::string> vars;
     //we have an http req
+
+    std::list<boost::regex> skip_list;
+    
     if (gdu && gdu->which == Z_GDU_HTTP_Request)
     {
         Z_HTTP_Request *hreq = gdu->u.HTTP_Request;
         mp::odr o;
         req_phase->rewrite_reqline(o, hreq, vars);
+        res_phase->read_skip_headers(hreq, skip_list);  
         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, skip_list);
         package.request() = gdu;
     }
     package.move();
@@ -135,12 +168,53 @@ 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, skip_list);
         package.response() = gdu;
     }
 }
 
+// Read (and remove) the X-Metaproxy-SkipLink headers
+void yf::HttpRewrite::Phase::read_skip_headers(Z_HTTP_Request *hreq,
+                                 std::list<boost::regex> &skip_list )
+{
+    std::string url(hreq->path);
+    if ( url.substr(0,7) != "http://" )
+    { // path was relative, as it usually is
+        const char *host =  z_HTTP_header_lookup(hreq->headers, "Host");
+        if (host)
+          url = "http://" + std::string(host) + hreq->path ;
+    }
+
+    while ( const char *hv = z_HTTP_header_remove( &(hreq->headers),
+        "X-Metaproxy-SkipLink") )
+    {
+        yaz_log(YLOG_LOG,"Found SkipLink '%s'", hv );
+        const char *p = strchr(hv,' ');
+        if (!p)
+            continue; // should not happen
+        std::string page(hv,p);
+        std::string link(p+1);
+        boost::regex pagere(page);
+        if ( boost::regex_search(url, pagere) )
+        {
+            yaz_log(YLOG_LOG,"SkipLink '%s' matches URL %s",
+                    page.c_str(), url.c_str() );
+            boost::regex linkre(link);
+            skip_list.push_back(linkre);
+        }
+        else
+        {
+            yaz_log(YLOG_LOG,"SkipLink ignored, '%s' does not match '%s'",
+                    url.c_str(), page.c_str() );
+        }
+    }
+}
+
+
 void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
         Z_HTTP_Request *hreq,
         std::map<std::string, std::string> & vars) const
@@ -156,95 +230,117 @@ 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<Content>::const_iterator cit = content_list.begin();
+    for (; cit != content_list.end(); cit++)
+        if (cit->type == "headers")
+            break;
 
-    std::list<Within>::const_iterator it = within_list.begin();
-    for (; it != within_list.end(); it++)
+    if (cit == content_list.end())
+        return;
+
+    std::list<Within>::const_iterator it = cit->within_list.begin();
+    for (; it != cit->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())
+            std::list<boost::regex> dummy_skip_list; // no skips here!
+            if (it->exec(vars, path, true, dummy_skip_list))
             {
-                yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
-                hreq->path = odr_strdup(o, npath.c_str());
+                yaz_log(YLOG_LOG, "Rewritten request URL is %s", path.c_str());
+                hreq->path = odr_strdup(o, path.c_str());
             }
         }
 }
 
 void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
         Z_HTTP_Header *headers,
-        std::map<std::string, std::string> & vars) const
+        std::map<std::string, std::string> & vars ) const
 {
+    std::list<Content>::const_iterator cit = content_list.begin();
+    for (; cit != content_list.end(); cit++)
+        if (cit->type == "headers")
+            break;
+
+    if (cit == content_list.end())
+        return;
+
     for (Z_HTTP_Header *header = headers; header; header = header->next)
     {
-        std::list<Within>::const_iterator it = within_list.begin();
-        for (; it != within_list.end(); it++)
+        std::list<Within>::const_iterator it = cit->within_list.begin();
+        for (; it != cit->within_list.end(); it++)
         {
-            if (it->header.length() > 0 &&
-                yaz_strcasecmp(it->header.c_str(), header->name) == 0)
+            if (!it->header.empty() &&
+                regex_match(header->name, it->header))
             {
-                std::string sheader(header->name);
-                sheader += ": ";
-                sheader += header->value;
-
-                RulePtr rule = it->rule;
-                std::string out = rule->test_patterns(vars, sheader);
-                if (!out.empty())
+                // Match and replace only the header value
+                std::string hval(header->value);
+                std::list<boost::regex> dummy_skip_list; // no skips here!
+                if (it->exec(vars, hval, true, dummy_skip_list))
                 {
-                    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->value = odr_strdup(o, hval.c_str());
                 }
             }
         }
     }
 }
 
-void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
-        char **content_buf,
-        int *content_len,
-        std::map<std::string, std::string> & vars) const
+void yf::HttpRewrite::Phase::rewrite_body(
+    mp::odr &o,
+    const char *content_type,
+    char **content_buf,
+    int *content_len,
+    std::map<std::string, std::string> & vars,
+    std::list<boost::regex> & skip_list ) const
 {
-    if (*content_buf)
+    if (*content_len == 0)
+        return;
+    if (!content_type) {
+        yaz_log(YLOG_LOG, "rewrite_body: null content_type, can not rewrite");
+        return;
+    }
+    std::list<Content>::const_iterator cit = content_list.begin();
+    for (; cit != content_list.end(); cit++)
     {
-        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);
+        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()) {
+        yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
+                content_type );  
+        return;
+    }
 
-        std::string buf(*content_buf, *content_len);
+    int i;
+    for (i = 0; i < *content_len; i++)
+        if ((*content_buf)[i] == 0) {
+            yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
+            return;  // binary content. skip
+        }
 
-        parser.parse(ev, buf.c_str());
-        const char *res = ev.result();
-        *content_buf = odr_strdup(o, res);
-        *content_len = strlen(res);
-    }
+    std::string content(*content_buf, *content_len);
+    cit->parse(m_verbose, content, vars, skip_list);
+    *content_buf = odr_strdup(o, content.c_str());
+    *content_len = strlen(*content_buf);
 }
 
-yf::HttpRewrite::Event::Event(const Phase *p,
-                              std::map<std::string, std::string> & vars
-    ) : m_phase(p), m_vars(vars)
+yf::HttpRewrite::Event::Event(const Content *p,
+                              std::map<std::string, std::string> & vars,
+                              std::list<boost::regex> & skip_list 
+    ) : m_content(p), m_vars(vars), m_skips(skip_list)
 {
     m_w = wrbuf_alloc();
-    enabled_within = m_phase->within_list.end();
 }
 
 yf::HttpRewrite::Event::~Event()
@@ -259,33 +355,22 @@ const char *yf::HttpRewrite::Event::result()
 
 void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
 {
-    // check if there is <within tag="x" .. />
-    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<Within>::const_iterator it = m_content->within_list.begin();
+    for (; it != m_content->within_list.end(); it++)
     {
-        std::string t(tag, tag_len);
-        std::list<Within>::const_iterator it =
-            m_phase->within_list.begin();
-        for (; it != m_phase->within_list.end(); it++)
+        if (!it->tag.empty() && regex_match(t, it->tag))
         {
-            if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
-                                                       t.c_str()) == 0)
+            if (!it->attr.empty() && regex_match("#text", it->attr))
             {
-                std::vector<std::string> 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)
-                    {
-                        enabled_within = it;
-                        break;
-                    }
-                }
+                s_within.push(it);
+                return;
             }
         }
     }
-    wrbuf_putc(m_w, '<');
-    wrbuf_write(m_w, tag, tag_len);
 }
 
 void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
@@ -293,14 +378,12 @@ void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
 {
     if (close_it)
     {
-        std::list<Within>::const_iterator it = enabled_within;
-        if (it != m_phase->within_list.end())
+        if (!s_within.empty())
         {
+            std::list<Within>::const_iterator it = s_within.top();
             std::string t(tag, tag_len);
-            if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
-            {
-                enabled_within = m_phase->within_list.end();
-            }
+            if (regex_match(t, it->tag))
+                s_within.pop();
         }
     }
     if (close_it)
@@ -310,27 +393,20 @@ void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_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 *value, int val_len,
+                                       const char *sep)
 {
-    std::list<Within>::const_iterator it = m_phase->within_list.begin();
+    std::list<Within>::const_iterator it = m_content->within_list.begin();
     bool subst = false;
 
-    for (; it != m_phase->within_list.end(); it++)
+    for (; it != m_content->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->tag.empty() || regex_match(t, it->tag))
         {
             std::string a(attr, attr_len);
-            std::vector<std::string> 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 (!it->attr.empty() && regex_match(a, it->attr))
+                subst = true;
         }
         if (subst)
             break;
@@ -338,31 +414,32 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
 
     wrbuf_putc(m_w, ' ');
     wrbuf_write(m_w, attr, attr_len);
-    wrbuf_puts(m_w, "=\"");
-
-    std::string output;
-    if (subst)
+    if (value)
     {
-        std::string input(value, val_len);
-        output = it->rule->test_patterns(m_vars, input);
+        wrbuf_puts(m_w, "=");
+        wrbuf_puts(m_w, sep);
+
+        std::string output;
+        if (subst)
+        {
+            std::string s(value, val_len);
+            it->exec(m_vars, s, true, m_skips);
+            wrbuf_puts(m_w, s.c_str());
+        }
+        else
+            wrbuf_write(m_w, value, val_len);
+        wrbuf_puts(m_w, sep);
     }
-    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 *tag, int tag_len)
 {
-    std::list<Within>::const_iterator it = enabled_within;
-    if (it != m_phase->within_list.end())
+    if (!s_within.empty())
     {
+        std::list<Within>::const_iterator it = s_within.top();
         std::string t(tag, tag_len);
-        if (yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
-        {
-            enabled_within = m_phase->within_list.end();
-        }
+        if (regex_match(t, it->tag))
+            s_within.pop();
     }
     wrbuf_puts(m_w, "</");
     wrbuf_write(m_w, tag, tag_len);
@@ -370,104 +447,191 @@ void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
 
 void yf::HttpRewrite::Event::text(const char *value, int len)
 {
-    std::list<Within>::const_iterator it = enabled_within;
-    bool subst = false;
+    std::list<Within>::const_iterator it = m_content->within_list.end();
+    if (!s_within.empty())
+        it = s_within.top();
+    if (it != m_content->within_list.end())
+    {
+        std::string s(value, len);
+        it->exec(m_vars, s, false, m_skips);
+        wrbuf_puts(m_w, s.c_str());
+    }
+    else
+        wrbuf_write(m_w, value, len);
+}
 
-    if (it != m_phase->within_list.end())
+static bool embed_quoted_literal(
+    std::string &content,
+    std::map<std::string, std::string> &vars,
+    mp::filter::HttpRewrite::RulePtr ruleptr,
+    bool html_context,
+    std::list<boost::regex> &skip_list)
+{
+    bool replace = false;
+    std::string res;
+    const char *cp = content.c_str();
+    const char *cp0 = cp;
+    while (*cp)
     {
-        subst = true;
-        if (it->attr.length() > 0)
+        if (html_context && !strncmp(cp, "&quot;", 6))
         {
-            subst = false;
-            std::vector<std::string> attr;
-            boost::split(attr, it->attr, boost::is_any_of(","));
-            size_t i;
-            for (i = 0; i < attr.size(); i++)
+            cp += 6;
+            res.append(cp0, cp - cp0);
+            cp0 = cp;
+            while (*cp)
             {
-                if (attr[i].compare("#text") == 0)
-                {
-                    subst = true;
-                }
+                if (!strncmp(cp, "&quot;", 6))
+                    break;
+                if (*cp == '\n')
+                    break;
+                cp++;
             }
+            if (!*cp)
+                break;
+            std::string s(cp0, cp - cp0);
+            if (ruleptr->test_patterns(vars, s, true, skip_list))
+                replace = true;
+            cp0 = cp;
+            res.append(s);
         }
+        else if (*cp == '"' || *cp == '\'')
+        {
+            int m = *cp;
+            cp++;
+            res.append(cp0, cp - cp0);
+            cp0 = cp;
+            while (*cp)
+            {
+                if (cp[-1] != '\\' && *cp == m)
+                    break;
+                if (*cp == '\n')
+                    break;
+                cp++;
+            }
+            if (!*cp)
+                break;
+            std::string s(cp0, cp - cp0);
+            if (ruleptr->test_patterns(vars, s, true, skip_list))
+                replace = true;
+            cp0 = cp;
+            res.append(s);
+        }
+        else if (*cp == '/' && cp[1] == '/')
+        {
+            while (cp[1] && cp[1] != '\n')
+                cp++;
+        }
+        cp++;
     }
-    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());
+    res.append(cp0, cp - cp0);
+    content = res;
+    return replace;
 }
 
-
-/**
- * Tests pattern from the vector in order and executes recipe on
- the first match.
- */
-const std::string yf::HttpRewrite::Rule::test_patterns(
-        std::map<std::string, std::string> & vars,
-        const std::string & txt) const
+bool yf::HttpRewrite::Within::exec(
+    std::map<std::string, std::string> & vars,
+    std::string & txt, bool anchor,
+    std::list<boost::regex> & skip_list) const
 {
-    std::list<Replace>::const_iterator it = replace_list.begin();
-
-    for (; it != replace_list.end(); it++)
+    if (type == "quoted-literal")
     {
-        std::string out = it->search_replace(vars, txt);
-        if (!out.empty()) return out;
+        return embed_quoted_literal(txt, vars, rule, true, skip_list);
+    }
+    else
+    {
+        return rule->test_patterns(vars, txt, anchor, skip_list);
     }
-    return "";
 }
 
-const std::string yf::HttpRewrite::Replace::search_replace(
-        std::map<std::string, std::string> & vars,
-        const std::string & txt) const
+bool yf::HttpRewrite::Rule::test_patterns(
+    std::map<std::string, std::string> & vars,
+    std::string & txt, bool anchor,
+    std::list<boost::regex> & skip_list )
 {
-    //exec regex against value
-    boost::regex re(regex);
-    boost::smatch what;
+    bool replaces = false;
+    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<Replace>::iterator bit = replace_list.end();
+        boost::smatch bwhat;
+        bool match_one = false;
+        {
+            std::list<Replace>::iterator it = replace_list.begin();
+            for (; it != replace_list.end(); it++)
+            {
+                if (it->start_anchor && !first)
+                    continue;
+                boost::smatch what;
+                if (regex_search(start, end, what, it->re))
+                {
+                    if (!match_one || what[0].first < bwhat[0].first)
+                    {
+                        bwhat = what;
+                        bit = it;
+                    }
+                    match_one = true;
+                }
+            }
+            if (!match_one)
+                break;
+        }
+        first = false;
+        replaces = true;
         size_t i;
-        for (i = 1; i < what.size(); ++i)
+        for (i = 1; i < bwhat.size(); ++i)
         {
             //check if the group is named
-            std::map<int, std::string>::const_iterator it
-                = group_index.find(i);
-            if (it != group_index.end())
+            std::map<int, std::string>::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] = bwhat[i];
             }
 
         }
+        // Compare against skip_list
+        bool skipthis = false;
+        std::list<boost::regex>::iterator si = skip_list.begin();
+        for ( ; si != skip_list.end(); si++) {
+            if ( boost::regex_search(bwhat.str(0), *si) )
+            {
+                skipthis = true;
+                break;
+            }
+        }
         //prepare replacement string
-        std::string rvalue = 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);
-        out.append(rvalue);
-        start = what[0].second; //move search forward
+        std::string rvalue = bit->sub_vars(vars);
+        out.append(start, bwhat[0].first);
+        if ( skipthis )
+        {
+            yaz_log(YLOG_LOG,"! Not rewriting '%s', skiplist match",
+                    bwhat.str(0).c_str() );
+            out.append(bwhat.str(0).c_str());
+        }
+        else
+        {
+            yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
+                    bwhat.str(0).c_str(), rvalue.c_str());
+            out.append(rvalue);
+        }
+        start = bwhat[0].second; //move search forward
     }
-    //if we had a match cat the last part
-    if (start != txt.begin())
-        out.append(start, end);
-    return out;
+    out.append(start, end);
+    txt = out;
+    return replaces;
 }
 
-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)
     {
@@ -522,11 +686,11 @@ void yf::HttpRewrite::Replace::parse_groups()
         }
         esc = false;
     }
-    regex = res;
+    re = res;
 }
 
-std::string yf::HttpRewrite::Replace::sub_vars (
-        const std::map<std::string, std::string> & vars) const
+std::string yf::HttpRewrite::Replace::sub_vars(
+    const std::map<std::string, std::string> & vars) const
 {
     std::string out;
     bool esc = false;
@@ -579,6 +743,111 @@ yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 {
 }
 
+void yf::HttpRewrite::Content::parse(
+    int verbose,
+    std::string &content,
+    std::map<std::string, std::string> &vars,
+    std::list<boost::regex> & skip_list ) const
+{
+    if (type == "html")
+    {
+        HTMLParser parser;
+        Event ev(this, vars, skip_list);
+
+        parser.set_verbose(verbose);
+
+        parser.parse(ev, content.c_str());
+        content = ev.result();
+    }
+    if (type == "quoted-literal")
+    {
+        quoted_literal(content, vars, skip_list);
+    }
+}
+
+void yf::HttpRewrite::Content::quoted_literal(
+    std::string &content,
+    std::map<std::string, std::string> &vars,
+    std::list<boost::regex> & skip_list ) const
+{
+    std::list<Within>::const_iterator it = within_list.begin();
+    if (it != within_list.end())
+        embed_quoted_literal(content, vars, it->rule, false, skip_list);
+}
+
+void yf::HttpRewrite::Content::configure(
+    const xmlNode *ptr, std::map<std::string, RulePtr > &rules)
+{
+    for (; ptr; ptr = ptr->next)
+    {
+        if (ptr->type != XML_ELEMENT_NODE)
+            continue;
+        if (!strcmp((const char *) ptr->name, "within"))
+        {
+            static const char *names[7] =
+                { "header", "attr", "tag", "rule", "reqline", "type", 0 };
+            std::string values[6];
+            mp::xml::parse_attr(ptr, names, values);
+            Within w;
+            if (values[0].length() > 0)
+                w.header.assign(values[0], boost::regex_constants::icase);
+            if (values[1].length() > 0)
+                w.attr.assign(values[1], boost::regex_constants::icase);
+            if (values[2].length() > 0)
+                w.tag.assign(values[2], boost::regex_constants::icase);
+
+            std::vector<std::string> rulenames;
+            boost::split(rulenames, values[3], boost::is_any_of(","));
+            if (rulenames.size() == 0)
+            {
+                throw mp::filter::FilterException
+                    ("Empty rule in '" + values[3] +
+                     "' in http_rewrite filter");
+            }
+            else if (rulenames.size() == 1)
+            {
+                std::map<std::string,RulePtr>::const_iterator it =
+                    rules.find(rulenames[0]);
+                if (it == rules.end())
+                    throw mp::filter::FilterException
+                        ("Reference to non-existing rule '" + rulenames[0] +
+                         "' in http_rewrite filter");
+                w.rule = it->second;
+
+            }
+            else
+            {
+                RulePtr rule(new Rule);
+                size_t i;
+                for (i = 0; i < rulenames.size(); i++)
+                {
+                    std::map<std::string,RulePtr>::const_iterator it =
+                        rules.find(rulenames[i]);
+                    if (it == rules.end())
+                        throw mp::filter::FilterException
+                            ("Reference to non-existing rule '" + rulenames[i] +
+                             "' in http_rewrite filter");
+                    RulePtr subRule = it->second;
+                    std::list<Replace>::iterator rit =
+                        subRule->replace_list.begin();
+                    for (; rit != subRule->replace_list.end(); rit++)
+                        rule->replace_list.push_back(*rit);
+                }
+                w.rule = rule;
+            }
+            w.reqline = values[4] == "1";
+            w.type = values[5];
+            if (w.type.empty() || w.type == "quoted-literal")
+                ;
+            else
+                throw mp::filter::FilterException
+                    ("within type must be quoted-literal or none in "
+                     " in http_rewrite filter");
+            within_list.push_back(w);
+        }
+    }
+}
+
 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 {
     static const char *names[2] = { "verbose", 0 };
@@ -608,11 +877,12 @@ 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
@@ -622,10 +892,12 @@ 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
@@ -635,25 +907,26 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
             }
             rules[values[0]] = rule;
         }
-        else if (!strcmp((const char *) ptr->name, "within"))
+        else if (!strcmp((const char *) ptr->name, "content"))
         {
-            static const char *names[6] =
-                { "header", "attr", "tag", "rule", "reqline", 0 };
-            std::string values[5];
+            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<std::string,RulePtr>::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);
+            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.assign(values[1], boost::regex::icase);
+            c.configure(ptr->children, rules);
+            phase.content_list.push_back(c);
         }
         else
         {