X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fhtml_parser.cpp;h=36411950093c29c0a8d54a9d0744509ebd5dd156;hb=96a9d0598df0722dc31117fd4559bb4b23a2225e;hp=1436553f691048fa1b9cd66905ea2fa1c4d0ed5a;hpb=897639233e3a6232d039666ba38b393bf7ac0ef0;p=metaproxy-moved-to-github.git

diff --git a/src/html_parser.cpp b/src/html_parser.cpp
index 1436553..3641195 100644
--- a/src/html_parser.cpp
+++ b/src/html_parser.cpp
@@ -37,14 +37,12 @@ namespace metaproxy_1 {
                      const char *text_start, const char *text_end);
         int tagEnd(HTMLParserEvent &event,
                    const char *tag, int tag_len, const char *cp);
-        int tagStart(HTMLParserEvent &event,
-                     int *tag_len, const char *cp, const char which);
         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);
+                          const char **value, int *val_len, int *tr);
         Rep();
         ~Rep();
         int m_verbose;
@@ -81,6 +79,11 @@ void mp::HTMLParser::parse(mp::HTMLParserEvent & event, const char *str) const
     m_p->parse_str(event, str);
 }
 
+static int isAlpha(int c)
+{
+    return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z');
+}
+
 static int skipSpace(const char *cp)
 {
     int i = 0;
@@ -99,8 +102,10 @@ static int skipName(const char *cp)
 
 int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
                                        const char *cp, int *attr_len,
-                                       const char **value, int *val_len)
+                                       const char **value, int *val_len,
+                                       int *tr)
 {
+    int v0, v1;
     int i = skipName(cp);
     *attr_len = i;
     *value = NULL;
@@ -109,14 +114,13 @@ 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] == '\'')
         {
-            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])
@@ -124,6 +128,7 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
         }
         else
         {
+            *tr = 0;
             v0 = i;
             while (cp[i] && !strchr(SPACECHR ">", cp[i]))
                 i++;
@@ -131,8 +136,8 @@ int mp::HTMLParser::Rep::skipAttribute(HTMLParserEvent &event,
         }
         *value = cp + v0;
         *val_len = v1 - v0;
+        i += skipSpace(cp + i);
     }
-    i += skipSpace(cp + i);
     return i;
 }
 
@@ -147,63 +152,19 @@ int mp::HTMLParser::Rep::tagAttrs(HTMLParserEvent &event,
         int attr_len;
         const char *value;
         int val_len;
-        int nor = skipAttribute(event, cp+i, &attr_len, &value, &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)
-	{
-            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);
-	}
-        else
-        {
-            i++;
-        }
-    }
-    return i;
-}
 
-int mp::HTMLParser::Rep::tagStart(HTMLParserEvent &event,
-                                  int *tag_len,
-                                  const char *cp, const char which)
-{
-    int i;
-    switch (which)
-    {
-    case '/':
-        i = skipName(cp);
-        *tag_len = i;
-        if (m_verbose)
-            printf("------ tag close %.*s\n", i, cp);
-        event.closeTag(cp, i);
-        break;
-    case '!':
-        for (i = 0; cp[i] && cp[i] != '>'; i++)
-            ;
-        *tag_len = i;
-        event.openTagStart(cp, i);
+        x[0] = tr;
+        x[1] = 0;
         if (m_verbose)
-            printf("------ dtd %.*s\n", i, cp);
-        break;
-    case '?':
-        for (i = 0; cp[i] && cp[i] != '>'; i++)
-            ;
-        *tag_len = i;
-        event.openTagStart(cp, i);
-        if (m_verbose)
-            printf("------ pi %.*s\n", i, cp);
-        break;
-    default:
-        i = skipName(cp);
-        *tag_len = i;
-        if (m_verbose)
-            printf("------ tag open %.*s\n", i, cp);
-        event.openTagStart(cp, i);
-
-        i += tagAttrs(event, cp, i, cp + i);
-
-        break;
+            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;
 }
@@ -216,7 +177,11 @@ 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;
@@ -224,6 +189,9 @@ 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++;
     }
@@ -245,34 +213,82 @@ void mp::HTMLParser::Rep::tagText(HTMLParserEvent &event,
 void mp::HTMLParser::Rep::parse_str(HTMLParserEvent &event, const char *cp)
 {
     const char *text_start = cp;
-    const char *text_end = cp;
     while (*cp)
     {
-        if (cp[0] == '<' && cp[1])  //tag?
+        if (*cp++ != '<')
+            continue;
+
+        if (*cp == '!')
         {
-            char which = cp[1];
-            if (which == '/')
-                cp++;
-            if (!strchr(SPACECHR, cp[1])) //valid tag starts
+            int i;
+            tagText(event, text_start, cp - 1);
+            if (cp[1] == '-' && cp[2] == '-')
             {
-                int i = 0;
-                int tag_len;
-
-                tagText(event, text_start, text_end); //flush any text
-                cp++;
-                i += tagStart(event, &tag_len, cp, which);
-                i += tagEnd(event, cp, tag_len, cp + i);
-                cp += i;
-                text_start = cp;
-                text_end = cp;
-                continue;
+                for (i = 3; cp[i]; i++)
+                    if (cp[i] == '-' && cp[i+1] == '-' && cp[i+2] == '>')
+                    {
+                        i+= 2;
+                        event.openTagStart(cp, i);
+                        break;
+                    }
             }
+            else
+            {
+                for (i = 1; cp[i] && cp[i] != '>'; i++)
+                    ;
+                event.openTagStart(cp, i);
+            }
+            if (m_verbose)
+                printf("------ dtd %.*s\n", i, cp);
+            i += tagEnd(event, cp, i, cp + i);
+            cp += i;
+            text_start = cp;
+        }
+        else if (*cp == '?')
+        {
+            int i;
+            tagText(event, text_start, cp - 1);
+            for (i = 1; cp[i] && cp[i] != '>'; i++)
+                ;
+            event.openTagStart(cp, i);
+            if (m_verbose)
+                printf("------ pi %.*s\n", i, cp);
+            i += tagEnd(event, cp, i, cp + i);
+            cp += i;
+            text_start = cp;
+        }
+        else if (*cp == '/' && isAlpha(cp[1]))
+        {
+            int i;
+            tagText(event, text_start, cp - 1);
+
+            i = skipName(++cp);
+            event.closeTag(cp, i);
+            if (m_verbose)
+                printf("------ tag close %.*s\n", i, cp);
+            i += tagEnd(event, cp, i, cp + i);
+            cp += i;
+            text_start = cp;
+        }
+        else if (isAlpha(*cp))
+        {
+            int i, j;
+            tagText(event, text_start, cp - 1);
+            i = skipName(cp);
+            event.openTagStart(cp, i);
+            if (m_verbose)
+                printf("------ tag open %.*s\n", i, cp);
+            j = tagAttrs(event, cp, i, cp + i);
+            j += tagEnd(event, cp, i, cp + i + j);
+            cp += i + j;
+            text_start = cp;
         }
-        //text
-        cp++;
-        text_end = cp;
     }
-    tagText(event, text_start, text_end); //flush any text
+    tagText(event, text_start, cp);
+}
+
+mp::HTMLParserEvent::~HTMLParserEvent()
+{
 }
 
 /*