Update of /cvsroot/htmlparser/htmlparser/src/org/htmlparser/lexer
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv3286/lexer
Modified Files:
Lexer.java
Log Message:
Fix bug #880283 Character ">" erroneously inserted by Lexer.
Some jsp tags are now handled in a separate jsp parse in the lexer.
Jsp tags embedded as attributes are still not handled.
Refer to bug #772700 Jsp Tags are not parsed correctly when in quoted attributes,
which is now reversed (i.e. in quotes are OK, outside of quotes causes problems),
but this points out a deficiency in the data structure holding tag contents (attribute lists)
that doesn't provide for tags within attributes.
Index: Lexer.java
===================================================================
RCS file: /cvsroot/htmlparser/htmlparser/src/org/htmlparser/lexer/Lexer.java,v
retrieving revision 1.24
retrieving revision 1.25
diff -C2 -d -r1.24 -r1.25
*** Lexer.java 2 Jan 2004 16:24:53 -0000 1.24
--- Lexer.java 24 Jan 2004 17:13:43 -0000 1.25
***************
*** 267,270 ****
--- 267,275 ----
if (0 == ch)
ret = makeString (probe);
+ else if ('%' == ch)
+ {
+ probe.retreat ();
+ ret = parseJsp (probe);
+ }
else if ('/' == ch || '%' == ch || Character.isLetter (ch))
{
***************
*** 974,977 ****
--- 979,1128 ----
}
+ /**
+ * Parse a java server page node.
+ * Scan characters until "%>" is encountered, or the input stream is
+ * exhausted, in which case <code>null</code> is returned.
+ * @param cursor The position at which to start scanning.
+ */
+ protected Node parseJsp (Cursor cursor)
+ throws
+ ParserException
+ {
+ boolean done;
+ char ch;
+ int state;
+ Vector attributes;
+ int code;
+ Node ret;
+
+ done = false;
+ state = 0;
+ code = 0;
+ attributes = new Vector ();
+ // <%xyz%>
+ // 012223d
+ // <%=xyz%>
+ // 0122223d
+ // <%@xyz%d
+ // 0122223d
+ while (!done)
+ {
+ ch = mPage.getCharacter (cursor);
+ switch (state)
+ {
+ case 0: // prior to the percent
+ switch (ch)
+ {
+ case '%': // <%
+ state = 1;
+ break;
+ // case 0: // <\0
+ // case '>': // <>
+ default:
+ done = true;
+ break;
+ }
+ break;
+ case 1: // prior to the optional qualifier
+ switch (ch)
+ {
+ case 0: // <%\0
+ case '>': // <%>
+ done = true;
+ break;
+ case '=': // <%=
+ case '@': // <%@
+ code = cursor.getPosition ();
+ attributes.addElement (new PageAttribute (mPage, mCursor.getPosition () + 1, code, -1, -1, (char)0));
+ state = 2;
+ break;
+ default: // <%x
+ code = cursor.getPosition () - 1;
+ attributes.addElement (new PageAttribute (mPage, mCursor.getPosition () + 1, code, -1, -1, (char)0));
+ state = 2;
+ break;
+ }
+ break;
+ case 2: // prior to the closing percent
+ switch (ch)
+ {
+ case 0: // <%x\0
+ case '>': // <%x>
+ done = true;
+ break;
+ case '\'':
+ case '"':// <%???"
+ state = ch;
+ break;
+ case '%': // <%???%
+ state = 3;
+ break;
+ default: // <%???x
+ break;
+ }
+ break;
+ case 3:
+ switch (ch)
+ {
+ case 0: // <%x??%\0
+ done = true;
+ break;
+ case '>':
+ state = 4;
+ done = true;
+ break;
+ default: // <%???%x
+ state = 2;
+ break;
+ }
+ break;
+ case '"':
+ switch (ch)
+ {
+ case 0: // <%x??"\0
+ done = true;
+ break;
+ case '"':
+ state = 2;
+ break;
+ default: // <%???'??x
+ break;
+ }
+ break;
+ case '\'':
+ switch (ch)
+ {
+ case 0: // <%x??'\0
+ done = true;
+ break;
+ case '\'':
+ state = 2;
+ break;
+ default: // <%???"??x
+ break;
+ }
+ break;
+ default:
+ throw new IllegalStateException ("how the fuck did we get in state " + state);
+ }
+ }
+
+ if (4 == state) // normal exit
+ {
+ if (0 != code)
+ {
+ state = cursor.getPosition () - 2; // reuse state
+ attributes.addElement (new PageAttribute (mPage, code, state, -1, -1, (char)0));
+ attributes.addElement (new PageAttribute (mPage, state, state + 1, -1, -1, (char)0));
+ }
+ else
+ throw new IllegalStateException ("jsp with no code!");
+ }
+ else
+ return (parseString (cursor, true)); // hmmm, true?
+
+ return (makeTag (cursor, attributes));
+ }
+
//
// NodeFactory interface
|
