/** Parse the document, calling back the handler */
void parse() throws IOException, SchemaParseException {
char c;
StringBuffer curWord = new StringBuffer();
int wordContext = WORD_NO_CONTEXT;
int level = 0;
while ((c = this.getNext()) != '\0') {
switch (c) {
case '<':
// Check if we have <! or <--
char c1 = this.getNext();
if (c1 == '!') {
// Check if the next word is reserved
if (wordContext != WORD_NO_CONTEXT && wordContext != WORD_COMMENT) {
System.err.println(
"Error: found a '<!' sequence within another '<!' sequence"); // NOI18N
throw new SchemaParseException(
"Warning: found a '<!' sequence within another '<!' sequence"); // NOI18N
}
if (wordContext != WORD_COMMENT) wordContext = WORD_CHECK;
} else if (c1 == '?') {
wordContext = WORD_PI;
} else {
curWord.append(c);
curWord.append(c1);
}
break;
case '>':
// Might be the end of a comment or <!element
switch (wordContext) {
case WORD_NO_CONTEXT:
// System.err.println("Error: Found '>' without '<!'");// NOI18N
throw new SchemaParseException("Error: Found '>' without '<!'"); // NOI18N
case WORD_PI:
String word = curWord.toString();
int len = word.length();
if (word.endsWith("?")) { // NOI18N
// Ignore any PI
curWord.delete(0, len);
wordContext = WORD_NO_CONTEXT;
} else curWord.append(c);
break;
case WORD_COMMENT:
word = curWord.toString();
len = word.length();
if (word.endsWith("--")) { // NOI18N
this.handler.endElement();
curWord.delete(0, len);
wordContext = WORD_NO_CONTEXT;
} else curWord.append(c);
break;
case WORD_ENTITY:
wordContext = WORD_NO_CONTEXT;
break;
default:
wordContext = this.processWord(curWord, wordContext);
this.handler.endElement();
wordContext = WORD_NO_CONTEXT;
}
break;
case '(':
if (wordContext == WORD_ELEMENT || wordContext == WORD_ATTLIST) {
wordContext = this.processWord(curWord, wordContext);
this.handler.startGroupElements();
} else curWord.append(c);
break;
case ')':
wordContext = this.processWord(curWord, wordContext);
if (wordContext == WORD_ELEMENT || wordContext == WORD_ATTLIST) {
int instance = this.getInstanceValue(this.peekNext());
// Get rid of the extra character
if (instance != Common.TYPE_1) this.getNext();
this.handler.endGroupElements(instance);
} else curWord.append(c);
break;
case '|':
wordContext = this.processWord(curWord, wordContext);
if (wordContext == WORD_ELEMENT || wordContext == WORD_ATTLIST) this.handler.character(c);
else curWord.append(c);
break;
case '\n':
case '\r':
case '\t':
case ' ':
case ',':
wordContext = this.processWord(curWord, wordContext);
break;
//
default:
curWord.append(c);
}
}
if (wordContext != WORD_NO_CONTEXT) System.out.println("Warning: unexpected EOF"); // NOI18N
}
/** Find out the type of the current word */
private int processWord(StringBuffer curWord, int wordContext) throws SchemaParseException {
String word = curWord.toString();
int len = word.length();
if (len > 0) {
// We have some word to play with
switch (wordContext) {
case WORD_CHECK:
if (word.startsWith("--")) { // NOI18N
if (len > 2) word = curWord.substring(2);
else word = ""; // NOI18N
this.handler.startElement(word, word, Common.COMMENT);
wordContext = WORD_COMMENT;
} else if (word.equals("ELEMENT")) // NOI18N
wordContext = WORD_ELEMENT1;
else if (word.equals("ATTLIST")) // NOI18N
wordContext = WORD_ATTLIST1;
else if (word.equals("ENTITY")) // NOI18N
wordContext = WORD_ENTITY1;
else {
// System.err.println("Error: found an unknown '<!' sequence (" + word + ")"); // NOI18N
throw new SchemaParseException(
"Error: found an unknown '<!' sequence (" + word + ")"); // NOI18N
}
break;
case WORD_COMMENT:
this.handler.element(word, word, 0);
break;
case WORD_ELEMENT1:
this.handler.startElement(word, word, Common.ELEMENT);
wordContext = WORD_ELEMENT;
break;
case WORD_ATTLIST1:
this.handler.startElement(word, word, Common.ATTLIST);
wordContext = WORD_ATTLIST;
break;
case WORD_ENTITY1:
wordContext = WORD_ENTITY;
break;
case WORD_ENTITY:
break;
case WORD_ELEMENT:
case WORD_ATTLIST:
// Find out the instance value (*, ? or +)
int instance = this.getInstanceValue(word.charAt(len - 1));
// Get rid of the extra character
if (instance != Common.TYPE_1) word = curWord.substring(0, len - 1);
try {
this.handler.element(word, word, instance);
} catch (MissingEndOfEltException e) {
if (wordContext == WORD_ATTLIST) {
//
// The TreeBuilder is done with the previous
// attribute and would expect an end of ATTLIST
// declaration.
// We might have several attributes declared on the
// same ATTLIST declaration.
// Let's continue assuming so, the TreeBuilder
// checks the attribute semantic and will throw
// if this is not the case.
//
this.handler.startElement(e.propName, e.propName, Common.ATTLIST);
this.handler.element(word, word, instance);
}
}
break;
default:
}
curWord.delete(0, len);
}
return wordContext;
}