X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_http_rewrite.cpp;h=5cab5decba138d6983ca8decc6422fe830abc3d0;hb=d89b94329401b771c440dc6aafd72b9309a8b0ad;hp=d417f46cead1f552277b757294e2315554dac812;hpb=0cc98fc161f86765524e7d53ac0fb6a2196cf9af;p=metaproxy-moved-to-github.git

diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp
index d417f46..5cab5de 100644
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_rewrite.cpp
@@ -40,8 +40,8 @@ namespace metaproxy_1 {
     namespace filter {
         class HttpRewrite::Replace {
         public:
+            bool start_anchor;
             boost::regex re;
-            boost::smatch what;
             std::string recipe;
             std::map<int, std::string> group_index;
             std::string sub_vars(
@@ -52,31 +52,47 @@ namespace metaproxy_1 {
         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);
+            bool test_patterns(
+                std::map<std::string, std::string> &vars,
+                std::string &txt, bool anchor);
         };
         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) 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) const;
+            void parse(int verbose, std::string &content,
+                       std::map<std::string, std::string> & vars) const;
+        };
         class HttpRewrite::Phase {
         public:
             Phase();
-            std::list<Within> within_list;
             int m_verbose;
+            std::list<Content> content_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) const;
         };
         class HttpRewrite::Event : public HTMLParserEvent {
             void openTagStart(const char *tag, int tag_len);
@@ -87,12 +103,12 @@ namespace metaproxy_1 {
                            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::stack<std::list<Within>::const_iterator> s_within;
             std::map<std::string, std::string> &m_vars;
         public:
-            Event(const Phase *p, std::map<std::string, std::string> &vars);
+            Event(const Content *p, std::map<std::string, std::string> &vars);
             ~Event();
             const char *result();
         };
@@ -123,7 +139,10 @@ 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();
@@ -135,8 +154,11 @@ 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;
     }
 }
@@ -165,18 +187,23 @@ void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
         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())
+            if (it->exec(vars, path, true))
             {
-                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());
             }
         }
 }
@@ -185,67 +212,77 @@ void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
         Z_HTTP_Header *headers,
         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);
+                if (it->exec(vars, hval, true))
                 {
-                    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) 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);
+    *content_buf = odr_strdup(o, content.c_str());
+    *content_len = strlen(*content_buf);
 }
 
-yf::HttpRewrite::Event::Event(const Phase *p,
+yf::HttpRewrite::Event::Event(const Content *p,
                               std::map<std::string, std::string> & vars
-    ) : m_phase(p), m_vars(vars)
+    ) : m_content(p), m_vars(vars)
 {
     m_w = wrbuf_alloc();
 }
@@ -266,22 +303,15 @@ void yf::HttpRewrite::Event::openTagStart(const char *tag, int tag_len)
     wrbuf_write(m_w, tag, tag_len);
 
     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++)
+    std::list<Within>::const_iterator it = m_content->within_list.begin();
+    for (; it != m_content->within_list.end(); it++)
     {
-        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::vector<std::string> attr;
-            boost::split(attr, it->attr, boost::is_any_of(","));
-            size_t i;
-            for (i = 0; i < attr.size(); i++)
+            if (!it->attr.empty() && regex_match("#text", it->attr))
             {
-                if (attr[i].compare("#text") == 0)
-                {
-                    s_within.push(it);
-                    return;
-                }
+                s_within.push(it);
+                return;
             }
         }
     }
@@ -296,7 +326,7 @@ void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
         {
             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)
+            if (regex_match(t, it->tag))
                 s_within.pop();
         }
     }
@@ -310,25 +340,17 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_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;
@@ -344,13 +366,12 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
         std::string output;
         if (subst)
         {
-            std::string input(value, val_len);
-            output = it->rule->test_patterns(m_vars, input);
+            std::string s(value, val_len);
+            it->exec(m_vars, s, true);
+            wrbuf_puts(m_w, s.c_str());
         }
-        if (output.empty())
-            wrbuf_write(m_w, value, val_len);
         else
-            wrbuf_puts(m_w, output.c_str());
+            wrbuf_write(m_w, value, val_len);
         wrbuf_puts(m_w, sep);
     }
 }
@@ -361,7 +382,7 @@ void yf::HttpRewrite::Event::closeTag(const char *tag, int tag_len)
     {
         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)
+        if (regex_match(t, it->tag))
             s_within.pop();
     }
     wrbuf_puts(m_w, "</");
@@ -370,25 +391,106 @@ 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 = m_phase->within_list.end();
+    std::list<Within>::const_iterator it = m_content->within_list.end();
     if (!s_within.empty())
         it = s_within.top();
-    std::string output;
-    if (it != m_phase->within_list.end())
+    if (it != m_content->within_list.end())
     {
-        std::string input(value, len);
-        output = it->rule->test_patterns(m_vars, input);
+        std::string s(value, len);
+        it->exec(m_vars, s, false);
+        wrbuf_puts(m_w, s.c_str());
     }
-    if (output.empty())
+    else
         wrbuf_write(m_w, value, len);
+}
+
+static bool embed_quoted_literal(
+    std::string &content,
+    std::map<std::string, std::string> &vars,
+    mp::filter::HttpRewrite::RulePtr ruleptr,
+    bool html_context)
+{
+    bool replace = false;
+    std::string res;
+    const char *cp = content.c_str();
+    const char *cp0 = cp;
+    while (*cp)
+    {
+        if (html_context && !strncmp(cp, "&quot;", 6))
+        {
+            cp += 6;
+            res.append(cp0, cp - cp0);
+            cp0 = cp;
+            while (*cp)
+            {
+                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))
+                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))
+                replace = true;
+            cp0 = cp;
+            res.append(s);
+        }
+        else if (*cp == '/' && cp[1] == '/')
+        {
+            while (cp[1] && cp[1] != '\n')
+                cp++;
+        }
+        cp++;
+    }
+    res.append(cp0, cp - cp0);
+    content = res;
+    return replace;
+}
+
+bool yf::HttpRewrite::Within::exec(
+    std::map<std::string, std::string> & vars,
+    std::string & txt, bool anchor) const
+{
+    if (type == "quoted-literal")
+    {
+        return embed_quoted_literal(txt, vars, rule, true);
+    }
     else
-        wrbuf_puts(m_w, output.c_str());
+    {
+        return rule->test_patterns(vars, txt, anchor);
+    }
 }
 
-const std::string yf::HttpRewrite::Rule::test_patterns(
-        std::map<std::string, std::string> & vars,
-        const std::string & txt)
+bool yf::HttpRewrite::Rule::test_patterns(
+    std::map<std::string, std::string> & vars,
+    std::string & txt, bool anchor)
 {
+    bool replaces = false;
+    bool first = anchor;
     std::string out;
     std::string::const_iterator start, end;
     start = txt.begin();
@@ -396,47 +498,53 @@ const std::string yf::HttpRewrite::Rule::test_patterns(
     while (1)
     {
         std::list<Replace>::iterator bit = replace_list.end();
+        boost::smatch bwhat;
+        bool match_one = false;
         {
-            std::string::const_iterator best_pos = txt.end();
             std::list<Replace>::iterator it = replace_list.begin();
             for (; it != replace_list.end(); it++)
             {
-                if (regex_search(start, end, it->what, it->re))
+                if (it->start_anchor && !first)
+                    continue;
+                boost::smatch what;
+                if (regex_search(start, end, what, it->re))
                 {
-                    if (it->what[0].first < best_pos)
+                    if (!match_one || what[0].first < bwhat[0].first)
                     {
-                        best_pos = it->what[0].first;
+                        bwhat = what;
                         bit = it;
                     }
+                    match_one = true;
                 }
             }
-            if (bit == replace_list.end())
+            if (!match_one)
                 break;
         }
-
+        first = false;
+        replaces = true;
         size_t i;
-        for (i = 1; i < bit->what.size(); ++i)
+        for (i = 1; i < bwhat.size(); ++i)
         {
             //check if the group is named
             std::map<int, std::string>::const_iterator git
                 = bit->group_index.find(i);
             if (git != bit->group_index.end())
             {   //it is
-                vars[git->second] = bit->what[i];
+                vars[git->second] = bwhat[i];
             }
 
         }
         //prepare replacement string
         std::string rvalue = bit->sub_vars(vars);
         yaz_log(YLOG_LOG, "! Rewritten '%s' to '%s'",
-                bit->what.str(0).c_str(), rvalue.c_str());
-        out.append(start, bit->what[0].first);
+                bwhat.str(0).c_str(), rvalue.c_str());
+        out.append(start, bwhat[0].first);
         out.append(rvalue);
-        start = bit->what[0].second; //move search forward
+        start = bwhat[0].second; //move search forward
     }
-    if (start != txt.begin())
-        out.append(start, end);
-    return out;
+    out.append(start, end);
+    txt = out;
+    return replaces;
 }
 
 void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
@@ -445,6 +553,7 @@ void yf::HttpRewrite::Replace::parse_groups(std::string pattern)
     bool esc = false;
     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)
     {
@@ -556,6 +665,109 @@ yf::HttpRewrite::Phase::Phase() : m_verbose(0)
 {
 }
 
+void yf::HttpRewrite::Content::parse(
+    int verbose,
+    std::string &content,
+    std::map<std::string, std::string> &vars) const
+{
+    if (type == "html")
+    {
+        HTMLParser parser;
+        Event ev(this, vars);
+
+        parser.set_verbose(verbose);
+
+        parser.parse(ev, content.c_str());
+        content = ev.result();
+    }
+    if (type == "quoted-literal")
+    {
+        quoted_literal(content, vars);
+    }
+}
+
+void yf::HttpRewrite::Content::quoted_literal(
+    std::string &content,
+    std::map<std::string, std::string> &vars) const
+{
+    std::list<Within>::const_iterator it = within_list.begin();
+    if (it != within_list.end())
+        embed_quoted_literal(content, vars, it->rule, false);
+}
+
+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 };
@@ -615,25 +827,26 @@ 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
         {