Option attribute values for HTML parser
[metaproxy-moved-to-github.git] / src / filter_http_rewrite.cpp
index fd1b887..bf01aba 100644 (file)
@@ -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>
@@ -45,7 +46,7 @@ namespace metaproxy_1 {
             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();
         };
@@ -68,7 +69,9 @@ namespace metaproxy_1 {
 
         class HttpRewrite::Phase {
         public:
+            Phase();
             std::list<Within> within_list;
+            int m_verbose;
             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,
@@ -78,17 +81,17 @@ namespace metaproxy_1 {
                 std::map<std::string, std::string> & vars) const;
         };
         class HttpRewrite::Event : public HTMLParserEvent {
-            void openTagStart(const char *name);
-            void anyTagEnd(const char *name, int close_it);
-            void attribute(const char *tagName, 
-                           const char *name, 
-                           const char *value,
-                           int val_len);
-            void closeTag(const char *name);
+            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 *sep);
+            void closeTag(const char *tag, int tag_len);
             void text(const char *value, int len);
             const Phase *m_phase;
             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;
         public:
             Event(const Phase *p, std::map<std::string, std::string> &vars);
@@ -226,6 +229,9 @@ void yf::HttpRewrite::Phase::rewrite_body(mp::odr & o,
 
         HTMLParser parser;
         Event ev(this, vars);
+
+        parser.set_verbose(m_verbose);
+
         std::string buf(*content_buf, *content_len);
 
         parser.parse(ev, buf.c_str());
@@ -240,7 +246,6 @@ yf::HttpRewrite::Event::Event(const Phase *p,
     ) : m_phase(p), m_vars(vars)
 {
     m_w = wrbuf_alloc();
-    enabled_within = m_phase->within_list.end();
 }
 
 yf::HttpRewrite::Event::~Event()
@@ -253,36 +258,44 @@ const char *yf::HttpRewrite::Event::result()
     return wrbuf_cstr(m_w);
 }
 
-void yf::HttpRewrite::Event::openTagStart(const char *name)
+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_phase->within_list.begin();
+    for (; it != m_phase->within_list.end(); it++)
     {
-        std::list<Within>::const_iterator it =
-            m_phase->within_list.begin();
-        for (; it != m_phase->within_list.end(); it++)
+        if (it->tag.length() > 0 && yaz_strcasecmp(it->tag.c_str(),
+                                                   t.c_str()) == 0)
         {
-            if (it->tag.length() > 0 && it->tag.compare(name) == 0)
+            std::vector<std::string> attr;
+            boost::split(attr, it->attr, boost::is_any_of(","));
+            size_t i;
+            for (i = 0; i < attr.size(); i++)
             {
-                enabled_within = it;
+                if (attr[i].compare("#text") == 0)
+                {
+                    s_within.push(it);
+                    return;
+                }
             }
         }
     }
-    wrbuf_putc(m_w, '<');
-    wrbuf_puts(m_w, name);
 }
 
-void yf::HttpRewrite::Event::anyTagEnd(const char *name, int close_it)
+void yf::HttpRewrite::Event::anyTagEnd(const char *tag, int tag_len,
+                                       int close_it)
 {
     if (close_it)
     {
-        std::list<Within>::const_iterator it = enabled_within;
-        if (it != m_phase->within_list.end())
+        if (!s_within.empty())
         {
-            if (it->tag.compare(name) == 0)
-            {
-                enabled_within = m_phase->within_list.end();
-            }
+            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)
+                s_within.pop();
         }
     }
     if (close_it)
@@ -290,24 +303,28 @@ void yf::HttpRewrite::Event::anyTagEnd(const char *name, int close_it)
     wrbuf_putc(m_w, '>');
 }
 
-void yf::HttpRewrite::Event::attribute(const char *tagName,
-                                         const char *name,
-                                         const char *value,
-                                         int val_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 *sep)
 {
     std::list<Within>::const_iterator it = m_phase->within_list.begin();
     bool subst = false;
 
     for (; it != m_phase->within_list.end(); it++)
     {
-        if (it->tag.length() == 0 || it->tag.compare(tagName) == 0)
+        std::string t(tag, tag_len);
+        if (it->tag.length() == 0 ||
+            yaz_strcasecmp(it->tag.c_str(), t.c_str()) == 0)
         {
+            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") && attr[i].compare(name) == 0)
+                if (attr[i].compare("#text") &&
+                    yaz_strcasecmp(attr[i].c_str(), a.c_str()) == 0)
                     subst = true;
             }
         }
@@ -316,61 +333,46 @@ void yf::HttpRewrite::Event::attribute(const char *tagName,
     }
 
     wrbuf_putc(m_w, ' ');
-    wrbuf_puts(m_w, name);
-    wrbuf_puts(m_w, "=\"");
-
-    std::string output;
-    if (subst)
+    wrbuf_write(m_w, attr, attr_len);
+    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 *name)
+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())
     {
-        if (it->tag.compare(name) == 0)
-        {
-            enabled_within = m_phase->within_list.end();
-        }
+        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)
+            s_within.pop();
     }
     wrbuf_puts(m_w, "</");
-    wrbuf_puts(m_w, name);
+    wrbuf_write(m_w, tag, tag_len);
 }
 
 void yf::HttpRewrite::Event::text(const char *value, int len)
 {
-    std::list<Within>::const_iterator it = enabled_within;
-    bool subst = false;
-
-    if (it != m_phase->within_list.end())
-    {
-        subst = true;
-        if (it->attr.length() > 0)
-        {
-            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++)
-            {
-                if (attr[i].compare("#text") == 0)
-                {
-                    subst = true;
-                }
-            }
-        }
-    }
+    std::list<Within>::const_iterator it = m_phase->within_list.end();
+    if (!s_within.empty())
+        it = s_within.top();
     std::string output;
-    if (subst)
+    if (it != m_phase->within_list.end())
     {
         std::string input(value, len);
         output = it->rule->test_patterns(m_vars, input);
@@ -421,8 +423,7 @@ const std::string yf::HttpRewrite::Replace::search_replace(
                 = group_index.find(i);
             if (it != group_index.end())
             {   //it is
-                if (!what[i].str().empty())
-                    vars[it->second] = what[i];
+                vars[it->second] = what[i];
             }
 
         }
@@ -503,8 +504,8 @@ void yf::HttpRewrite::Replace::parse_groups()
     regex = 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;
@@ -553,9 +554,19 @@ std::string yf::HttpRewrite::Replace::sub_vars (
     return out;
 }
 
+yf::HttpRewrite::Phase::Phase() : m_verbose(0)
+{
+}
 
 void yf::HttpRewrite::configure_phase(const xmlNode *ptr, Phase &phase)
 {
+    static const char *names[2] = { "verbose", 0 };
+    std::string values[1];
+    values[0] = "0";
+    mp::xml::parse_attr(ptr, names, values);
+
+    phase.m_verbose = atoi(values[0].c_str());
+
     std::map<std::string, RulePtr > rules;
     for (ptr = ptr->children; ptr; ptr = ptr->next)
     {