Refactor: compiled regex part of Replace object

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

diff --git a/src/html_parser.cpp b/src/html_parser.cpp
index ddbbbe0..8c4426d 100644 (file)
--- a/src/html_parser.cpp
+++ b/src/html_parser.cpp
@@ -102,6 +102,7 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
                                        const char **value, int *val_len,
                                        int *tr)
 {
+    int v0, v1;
     int i = skipName(cp);
     *attr_len = i;
     *value = NULL;
@@ -110,7 +111,6 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
     i += skipSpace(cp + i);
     if (cp[i] == '=')
     {
-        int v0, v1;
         i++;
         i += skipSpace(cp + i);
         if (cp[i] == '\"' || cp[i] == '\'')
@@ -133,8 +133,8 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
         }
         *value = cp + v0;
         *val_len = v1 - v0;
+        i += skipSpace(cp + i);
     }
-    i += skipSpace(cp + i);
     return i;
 }
 
@@ -150,22 +150,18 @@ int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
         const char *value;
         int val_len;
         int tr;
+        char x[2];
         int nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len, &tr);
+        if (!nor)
+            break;
         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, x);
-       }
-        else
-        {
-            i++;
-        }
+
+        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, x);
     }
     return i;
 }
@@ -222,7 +218,11 @@ int mp::HTMLParser::Rep::tagEnd(HTMLParserEvent &event,
     for (; cp[i] && cp[i] != '/' && cp[i] != '>'; i++)
         ;
     if (i > 0)
+    {
+        if (m_verbose)
+            printf("------ text %.*s\n", i, cp);
         event.text(cp, i);
+    }
     if (cp[i] == '/')
     {
         close_it = 1;
@@ -230,6 +230,9 @@ int mp::HTMLParser::Rep::tagEnd(HTMLParserEvent &event,
     }
     if (cp[i] == '>')
     {
+        if (m_verbose)
+            printf("------ any tag %s %.*s\n",
+                   close_it ? " close" : "end", tag_len, tag);
         event.anyTagEnd(tag, tag_len, close_it);
         i++;
     }