projects / metaproxy-moved-to-github.git / commitdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | patch | inline | side by side (parent: c68b4f1)
Event.text for whitespace in <x/ > tags
authorAdam Dickmeiss <adam@indexdata.dk>
Thu, 27 Jun 2013 12:23:02 +0000 (14:23 +0200)
committerAdam Dickmeiss <adam@indexdata.dk>
Thu, 27 Jun 2013 12:23:02 +0000 (14:23 +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 47b2e14..1436553 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);
         Rep();
         ~Rep();
         int m_verbose;
@@ -94,8 +97,9 @@ 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 i = skipName(cp);
     *attr_len = i;
@@ -143,7 +147,7 @@ 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 nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len);
         i += nor;
        if (nor)
        {
@@ -209,10 +213,13 @@ 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] == '>')
diff --git a/src/test_html_parser.cpp b/src/test_html_parser.cpp
index 5659152..cdb266c 100644 (file)
--- a/src/test_html_parser.cpp
+++ b/src/test_html_parser.cpp
@@ -67,18 +67,17 @@ public:
     }
 };
 
-
 BOOST_AUTO_TEST_CASE( test_html_parser_1 )
 {
     try
     {
         mp::HTMLParser hp;
-        const char* html = 
+        const char* html =
             "<html><body><a t1=v1 t2='v2' t3=\"v3\">some text</a>"
             "<hr><table ></table  ><a href=\"x\"/></body></html>";
-        const char* expected = 
+        const char* expected =
             "<html><body><a t1=\"v1\" t2=\"v2\" t3=\"v3\">some text</a>"
-            "<hr><table></table><a href=\"x\"/></body></html>";
+            "<hr><table></table  ><a href=\"x\"/></body></html>";
         MyEvent e;
         hp.set_verbose(1);
         hp.parse(e, html);
Z39.50/Web Router