projects / metaproxy-moved-to-github.git / commitdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: c271eb1)
HTML parser stops eating tag when <-char is met
authorAdam Dickmeiss <adam@indexdata.dk>
Mon, 19 Aug 2013 11:40:04 +0000 (13:40 +0200)
committerAdam Dickmeiss <adam@indexdata.dk>
Mon, 19 Aug 2013 11:40:04 +0000 (13:40 +0200)
src/html_parser.cpp patch | blob | history
src/test_html_parser.cpp patch | blob | history

diff --git a/src/html_parser.cpp b/src/html_parser.cpp
index 3641195..fba5595 100644 (file)
--- a/src/html_parser.cpp
+++ b/src/html_parser.cpp
@@ -95,7 +95,7 @@ static int skipSpace(const char *cp)
 static int skipName(const char *cp)
 {
     int i;
-    for (i = 0; cp[i] && !strchr(SPACECHR "/>=", cp[i]); i++)
+    for (i = 0; cp[i] && !strchr(SPACECHR "/>=<", cp[i]); i++)
         ;
     return i;
 }
@@ -146,7 +146,7 @@ int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
                                   const char *cp)
 {
     int i = skipSpace(cp);
-    while (cp[i] && cp[i] != '>' && cp[i] != '/')
+    while (cp[i] && cp[i] != '>' && cp[i] != '/' && cp[i] != '<')
     {
         const char *attr_name = cp + i;
         int attr_len;
@@ -174,7 +174,7 @@ int mp::HTMLParser::Rep::tagEnd(HTMLParserEvent &event,
 {
     int i = 0;
     int close_it = 0;
-    for (; cp[i] && cp[i] != '/' && cp[i] != '>'; i++)
+    for (; cp[i] && cp[i] != '/' && cp[i] != '>' && cp[i] != '<'; i++)
         ;
     if (i > 0)
     {
diff --git a/src/test_html_parser.cpp b/src/test_html_parser.cpp
index 3604436..6e0ea4c 100644 (file)
--- a/src/test_html_parser.cpp
+++ b/src/test_html_parser.cpp
@@ -241,6 +241,36 @@ BOOST_AUTO_TEST_CASE( test_html_parser_5 )
     }
 }
 
+BOOST_AUTO_TEST_CASE( test_html_parser_6 )
+{
+    try
+    {
+        mp::HTMLParser hp;
+        const char* html =
+            "<html><script><x;</script></html>";
+
+        const char* expected = html;
+        MyEvent e;
+        hp.set_verbose(0);
+        hp.parse(e, html);
+
+        BOOST_CHECK_EQUAL(std::string(expected), e.out);
+        if (std::string(expected) != e.out)
+        {
+            std::cout << "Expected" << std::endl;
+            std::cout << expected << std::endl;
+            std::cout << "Got" << std::endl;
+            std::cout << e.out << std::endl;
+        }
+    }
+    catch (std::exception & e)
+    {
+        std::cout << e.what();
+        std::cout << std::endl;
+        BOOST_CHECK (false);
+    }
+}
+
 
 /*
  * Local variables:
Z39.50/Web Router