http_client: disables HTTP redirect by default
[metaproxy-moved-to-github.git] / src / filter_http_rewrite.cpp
index 492d97a..d417f46 100644 (file)
@@ -40,15 +40,13 @@ namespace metaproxy_1 {
     namespace filter {
         class HttpRewrite::Replace {
         public:
-            std::string regex;
+            boost::regex re;
+            boost::smatch what;
             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(
                 const std::map<std::string, std::string> & vars) const;
-            void parse_groups();
+            void parse_groups(std::string pattern);
         };
 
         class HttpRewrite::Rule {
@@ -56,7 +54,7 @@ namespace metaproxy_1 {
             std::list<Replace> replace_list;
             const std::string test_patterns(
                 std::map<std::string, std::string> & vars,
-                const std::string & txt) const;
+                const std::string & txt);
         };
         class HttpRewrite::Within {
         public:
@@ -85,7 +83,8 @@ namespace metaproxy_1 {
             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;
@@ -157,8 +156,12 @@ 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;
     }
 
@@ -304,7 +307,8 @@ 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();
     bool subst = false;
@@ -332,19 +336,23 @@ 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 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);
     }
-    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)
@@ -377,70 +385,65 @@ void yf::HttpRewrite::Event::text(const char *value, int len)
         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::Rule::test_patterns(
         std::map<std::string, std::string> & vars,
-        const std::string & txt) const
+        const std::string & txt)
 {
-    std::list<Replace>::const_iterator it = replace_list.begin();
-
-    for (; it != replace_list.end(); it++)
-    {
-        std::string out = it->search_replace(vars, txt);
-        if (!out.empty()) return out;
-    }
-    return "";
-}
-
-const std::string yf::HttpRewrite::Replace::search_replace(
-        std::map<std::string, std::string> & vars,
-        const std::string & txt) const
-{
-    //exec regex against value
-    boost::regex re(regex);
-    boost::smatch what;
+    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();
+        {
+            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->what[0].first < best_pos)
+                    {
+                        best_pos = it->what[0].first;
+                        bit = it;
+                    }
+                }
+            }
+            if (bit == replace_list.end())
+                break;
+        }
+
         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<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] = 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;
     yaz_log(YLOG_LOG, "Parsing groups from '%s'", str.c_str());
     for (size_t i = 0; i < str.size(); ++i)
@@ -496,7 +499,7 @@ void yf::HttpRewrite::Replace::parse_groups()
         }
         esc = false;
     }
-    regex = res;
+    re = res;
 }
 
 std::string yf::HttpRewrite::Replace::sub_vars(
@@ -582,11 +585,12 @@ 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
@@ -596,10 +600,12 @@ 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