No-nest mode for script content in HTML parser MP-486

[metaproxy-moved-to-github.git] / src / filter_http_rewrite.cpp

diff --git a/src/filter_http_rewrite.cpp b/src/filter_http_rewrite.cpp
index 69556ca..0f7ea87 100644 (file)
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_rewrite.cpp
@@ -229,22 +229,11 @@ void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
             if (!it->header.empty() &&
                 regex_match(header->name, it->header))
             {
-                std::string sheader(header->name);
-                sheader += ": ";
-                sheader += header->value;
-
-                if (it->exec(vars, sheader, true))
+                // Match and replace only the header value
+                std::string hval(header->value);
+                if (it->exec(vars, hval, true))
                 {
-                    size_t pos = sheader.find(": ");
-                    if (pos == std::string::npos)
-                    {
-                        yaz_log(YLOG_LOG, "Header malformed during rewrite, ignoring");
-                        continue;
-                    }
-                    header->name = odr_strdup(
-                        o, sheader.substr(0, pos).c_str());
-                    header->value = odr_strdup(
-                        o, sheader.substr(pos + 2, std::string::npos).c_str());
+                    header->value = odr_strdup(o, hval.c_str());
                 }
             }
         }
@@ -260,6 +249,10 @@ void yf::HttpRewrite::Phase::rewrite_body(
 {
     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++)
     {
@@ -269,13 +262,18 @@ void yf::HttpRewrite::Phase::rewrite_body(
             && regex_match(content_type, cit->content_re))
             break;
     }
-    if (cit == content_list.end())
+    if (cit == content_list.end()) {
+        yaz_log(YLOG_LOG,"rewrite_body: No content rule matched %s, not rewriting",
+                content_type );  
         return;
+    }
 
     int i;
     for (i = 0; i < *content_len; i++)
-        if ((*content_buf)[i] == 0)
+        if ((*content_buf)[i] == 0) {
+            yaz_log(YLOG_LOG,"rewrite_body: Looks like binary stuff, not rewriting");
             return;  // binary content. skip
+        }
 
     std::string content(*content_buf, *content_len);
     cit->parse(m_verbose, content, vars);
@@ -715,13 +713,46 @@ void yf::HttpRewrite::Content::configure(
                 w.attr.assign(values[1], boost::regex_constants::icase);
             if (values[2].length() > 0)
                 w.tag.assign(values[2], boost::regex_constants::icase);
-            std::map<std::string,RulePtr>::const_iterator it =
-                rules.find(values[3]);
-            if (it == rules.end())
+
+            std::vector<std::string> rulenames;
+            boost::split(rulenames, values[3], boost::is_any_of(","));
+            if (rulenames.size() == 0)
+            {
                 throw mp::filter::FilterException
-                    ("Reference to non-existing rule '" + values[3] +
+                    ("Empty rule in '" + values[3] +
                      "' in http_rewrite filter");
-            w.rule = it->second;
+            }
+            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")
@@ -810,8 +841,8 @@ void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
             Content c;
 
             c.type = values[0];
-            // if (!values[1].empty())
-                c.content_re = values[1];
+            if (!values[1].empty())
+                c.content_re.assign(values[1], boost::regex::icase);
             c.configure(ptr->children, rules);
             phase.content_list.push_back(c);
         }