HTML parser: elements starts with [A-Za-z]

[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 d4fdbcb..a918f44 100644 (file)
--- a/src/filter_http_rewrite.cpp
+++ b/src/filter_http_rewrite.cpp
@@ -40,12 +40,11 @@ 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;
-            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(std::string pattern);
@@ -56,7 +55,7 @@ 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, bool anchor);
         };
         class HttpRewrite::Within {
         public:
@@ -158,8 +157,12 @@ 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;
     }
 
@@ -170,7 +173,7 @@ void yf::HttpRewrite::Phase::rewrite_reqline (mp::odr & o,
         {
             RulePtr rule = it->rule;
             yaz_log(YLOG_LOG, "Proxy request URL is %s", path.c_str());
-            std::string npath = rule->test_patterns(vars, path);
+            std::string npath = rule->test_patterns(vars, path, true);
             if (!npath.empty())
             {
                 yaz_log(YLOG_LOG, "Rewritten request URL is %s", npath.c_str());
@@ -196,7 +199,7 @@ void yf::HttpRewrite::Phase::rewrite_headers(mp::odr & o,
                 sheader += header->value;
 
                 RulePtr rule = it->rule;
-                std::string out = rule->test_patterns(vars, sheader);
+                std::string out = rule->test_patterns(vars, sheader, true);
                 if (!out.empty())
                 {
                     size_t pos = out.find(": ");
@@ -343,7 +346,7 @@ void yf::HttpRewrite::Event::attribute(const char *tag, int tag_len,
         if (subst)
         {
             std::string input(value, val_len);
-            output = it->rule->test_patterns(m_vars, input);
+            output = it->rule->test_patterns(m_vars, input, true);
         }
         if (output.empty())
             wrbuf_write(m_w, value, val_len);
@@ -375,7 +378,7 @@ void yf::HttpRewrite::Event::text(const char *value, int len)
     if (it != m_phase->within_list.end())
     {
         std::string input(value, len);
-        output = it->rule->test_patterns(m_vars, input);
+        output = it->rule->test_patterns(m_vars, input, false);
     }
     if (output.empty())
         wrbuf_write(m_w, value, len);
@@ -383,56 +386,58 @@ 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, bool anchor)
 {
-    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
-{
-    boost::smatch what;
+    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();
+        {
+            std::string::const_iterator best_pos = txt.end();
+            std::list<Replace>::iterator it = replace_list.begin();
+            for (; it != replace_list.end(); it++)
+            {
+                if (it->start_anchor && !first)
+                    continue;
+                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;
+        }
+        first = false;
         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
-                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;
@@ -444,6 +449,7 @@ 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)
     {