HTMLParserEvent, attributes takes quoting sep

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

diff --git a/src/html_parser.cpp b/src/html_parser.cpp
index 694df69..ddbbbe0 100644 (file)
--- a/src/html_parser.cpp
+++ b/src/html_parser.cpp
@@ -42,6 +42,9 @@ namespace metaproxy_1 {
         int tagAttrs(HTMLParserEvent &event,
                      const char *name, int len,
                      const char *cp);
+        int skipAttribute(HTMLParserEvent &event,
+                          const char *cp, int *attr_len,
+                          const char **value, int *val_len, int *tr);
         Rep();
         ~Rep();
         int m_verbose;
@@ -67,6 +70,12 @@ mp::HTMLParser::~HTMLParser()
 {
 }
 
+void mp::HTMLParser::set_verbose(int v)
+{
+    m_p->m_verbose = v;
+}
+
+
 void mp::HTMLParser::parse(mp::HTMLParserEvent & event, const char *str) const
 {
     m_p->parse_str(event, str);
@@ -88,8 +97,10 @@ static int skipName(const char *cp)
     return i;
 }
 
-static int skipAttribute(const char *cp, int *attr_len,
-                         const char **value, int *val_len)
+int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
+                                       const char *cp, int *attr_len,
+                                       const char **value, int *val_len,
+                                       int *tr)
 {
     int i = skipName(cp);
     *attr_len = i;
@@ -104,9 +115,9 @@ static int skipAttribute(const char *cp, int *attr_len,
         i += skipSpace(cp + i);
         if (cp[i] == '\"' || cp[i] == '\'')
         {
-            char tr = cp[i];
+            *tr = cp[i];
             v0 = ++i;
-            while (cp[i] != tr && cp[i])
+            while (cp[i] != *tr && cp[i])
                 i++;
             v1 = i;
             if (cp[i])
@@ -114,6 +125,7 @@ static int skipAttribute(const char *cp, int *attr_len,
         }
         else
         {
+            *tr = 0;
             v0 = i;
             while (cp[i] && !strchr(SPACECHR ">", cp[i]))
                 i++;
@@ -137,14 +149,18 @@ int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
         int attr_len;
         const char *value;
         int val_len;
-        int nor = skipAttribute(cp+i, &attr_len, &value, &val_len);
+        int tr;
+        int nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len, &tr);
         i += nor;
        if (nor)
        {
+            char x[2];
+            x[0] = tr;
+            x[1] = 0;
             if (m_verbose)
                 printf ("------ attr %.*s=%.*s\n", attr_len, attr_name,
                         val_len, value);
-            event.attribute(name, len, attr_name, attr_len, value, val_len);
+            event.attribute(name, len, attr_name, attr_len, value, val_len, x);
        }
         else
         {
@@ -203,10 +219,13 @@ int mp::HTMLParser::Rep::tagEnd(HTMLParserEvent &event,
 {
     int i = 0;
     int close_it = 0;
-    while (cp[i] && cp[i] != '>')
+    for (; cp[i] && cp[i] != '/' && cp[i] != '>'; i++)
+        ;
+    if (i > 0)
+        event.text(cp, i);
+    if (cp[i] == '/')
     {
-        if (cp[i] == '/')
-            close_it = 1;
+        close_it = 1;
         i++;
     }
     if (cp[i] == '>')