private void expect(String string, Token... tokens) {
System.out.println("Tokenizing [" + string + "]");
ExprLexer tokenizer = new ExprLexer(string);
int expectedIndex = 0;
while (!tokenizer.isEndOfInput()) {
String expectedString;
Token expected;
if (expectedIndex < tokens.length) {
expected = tokens[expectedIndex];
expectedString = expected.toString();
} else {
expected = null;
expectedString = "<NOTHING>";
}
Token actual = tokenizer.next();
System.out.println(
String.format("Expected: %15s, got %s", expectedString, actual.toString()));
if (expected != null) {
assertEquals("tokenStart", expected.getTokenStart(), actual.getTokenStart());
assertEquals("text", expected.getString(), actual.getString());
assertEquals("type", expected.getType(), actual.getType());
if (!expected.equals(actual)) {
System.err.println("Unexpected result!");
throw new AssertionError();
}
}
expectedIndex++;
}
}
public final Enumeration Enumeration() throws ParseException {
Token t;
ASTStringNode name;
ASTStringNode name1;
ArrayList<ASTStringNode> name1List = new ArrayList<ASTStringNode>();
Token firstToken = token;
jj_consume_token(OB);
t = jj_consume_token(NAME);
name = new ASTStringNode(t.toString(), new WToken(token));
label_6:
while (true) {
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case 32:;
break;
default:
jj_la1[16] = jj_gen;
break label_6;
}
jj_consume_token(32);
t = jj_consume_token(NAME);
name1 = new ASTStringNode(t.toString(), new WToken(token));
name1List.add(name1);
}
jj_consume_token(CB);
{
if (true) return new Enumeration(name, name1List, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
public static void main(String args[]) {
Token tok;
try {
// Read from the command line for the filename
Lexer lex = new Lexer(args[0]);
while (true) {
tok = lex.nextToken();
// if token returns error token we print the line for error and exit
if (tok.getKind() == Tokens.Error) {
System.out.println(
"Error in Line " + lex.source.getLineno() + " ErrorValue: " + tok.toString());
System.exit(0);
}
// Printing the information about the line and the token
String p =
"Left: "
+ tok.getLeftPosition()
+ "| Right: "
+ tok.getRightPosition()
+ "| Type: "
+ TokenType.tokens.get(tok.getKind())
+ "| Value: ";
// if ((tok.getKind() == Tokens.Identifier) || (tok.getKind() == Tokens.INTeger))
p += tok.toString();
System.out.println(p + "| Line: " + lex.source.getLineno());
}
} catch (Exception e) {
}
}
public static void main(String[] args) {
String inFile = "C:\\Users\\Sukoon\\Documents\\NetBeansProjects\\cdproj\\src\\Sample.in";
String outFile = "C:\\Users\\Sukoon\\Desktop\\Sample.out";
if (args.length > 1) {
inFile = args[0];
outFile = args[1];
}
Lexer lexer = new Lexer(inFile);
try {
BufferedWriter writer = new BufferedWriter(new FileWriter(outFile));
Token t;
while ((t = lexer.nextToken()) != null) {
writer.write(t.toString());
writer.newLine();
}
writer.close();
System.out.println("Done tokenizing file: " + inFile);
System.out.println("Output written in file: " + outFile);
} catch (IOException e) {
e.printStackTrace();
}
}
public final AttListDecl AttListDecl() throws ParseException {
Token t;
ASTStringNode name;
AttribDef attribDef;
ArrayList<AttribDef> attribDefList = new ArrayList<AttribDef>();
Token firstToken = token;
jj_consume_token(34);
t = jj_consume_token(NAME);
name = new ASTStringNode(t.toString(), new WToken(token));
label_4:
while (true) {
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case NAME:;
break;
default:
jj_la1[11] = jj_gen;
break label_4;
}
attribDef = AttribDef();
attribDefList.add(attribDef);
}
jj_consume_token(PCB);
{
if (true) return new AttListDecl(name, attribDefList, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
public final DTD DTD() throws ParseException {
DTDEntry dTDEntry;
ArrayList<DTDEntry> dTDEntryList = new ArrayList<DTDEntry>();
Token t;
ASTStringNode eof;
Token firstToken = token;
label_1:
while (true) {
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case 31:
case 34:;
break;
default:
jj_la1[0] = jj_gen;
break label_1;
}
dTDEntry = DTDEntry();
dTDEntryList.add(dTDEntry);
}
t = jj_consume_token(0);
eof = new ASTStringNode(t.toString(), new WToken(token));
{
if (true) return new DTD(dTDEntryList, eof, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
/** Prints the content and annotations. */
public String toString() {
StringBuffer sb = new StringBuffer();
sb.append(" SENTENCE begin\n");
for (Token tok : tokens) sb.append(tok.toString());
sb.append(" SENTENCE end\n");
return sb.toString();
}
@Override
public String toString() {
StringBuilder sb = new StringBuilder();
for (Token t : tokens)
sb.append(t.toString()).append(" String:").append(t.getString()).append("\n");
return sb.toString();
}
public static final Valor PValorInteiro() throws ParseException {
Token token;
token = jj_consume_token(INTEGER_LITERAL);
{
if (true) return new ValorInteiro(Integer.parseInt(token.toString()));
}
throw new Error("Missing return statement in function");
}
public String toString() {
final StringBuilder result = new StringBuilder();
// overall format is (cat)token, or:
// (cat)'text'[startIdx,endIdx{wordCount}](seqNum.revisNum)
result.append("(").append(category).append(')').append(token.toString());
return result.toString();
}
public static final Valor PValorString() throws ParseException {
Token token;
token = jj_consume_token(STRING_LITERAL);
String tokenStr = token.toString();
tokenStr = tokenStr.substring(1, tokenStr.length() - 1);
{
if (true) return new ValorString(tokenStr);
}
throw new Error("Missing return statement in function");
}
public final AttribValue AttribValue() throws ParseException {
Token t;
ASTStringNode quotedstr;
Token firstToken = token;
t = jj_consume_token(QUOTEDSTR);
quotedstr = new ASTStringNode(t.toString(), new WToken(token));
{
if (true) return new AttribValue(quotedstr, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
public final Cp Cp() throws ParseException {
Token t;
ASTStringNode name;
Modifier modifier = null;
SeqOrChoice seqOrChoice;
Modifier modifier1 = null;
Token firstToken = token;
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case PCDATA:
jj_consume_token(PCDATA);
{
if (true) return new Cp1(firstToken.next, token);
}
break;
case NAME:
t = jj_consume_token(NAME);
name = new ASTStringNode(t.toString(), new WToken(token));
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case STAR:
case PLUS:
case Q:
modifier = Modifier();
break;
default:
jj_la1[7] = jj_gen;
;
}
{
if (true) return new Cp2(name, modifier, firstToken.next, token);
}
break;
case OB:
seqOrChoice = SeqOrChoice();
switch ((jj_ntk == -1) ? jj_ntk() : jj_ntk) {
case STAR:
case PLUS:
case Q:
modifier1 = Modifier();
break;
default:
jj_la1[8] = jj_gen;
;
}
{
if (true) return new Cp3(seqOrChoice, modifier1, firstToken.next, token);
}
break;
default:
jj_la1[9] = jj_gen;
jj_consume_token(-1);
throw new ParseException();
}
throw new Error("Missing return statement in function");
}
/** This function constructs the tree for debug purposes in a text view. */
public StringBuilder toString(StringBuilder prefix, boolean isTail, StringBuilder sb) {
if (right != null) {
right.toString(
new StringBuilder().append(prefix).append(isTail ? "│ " : " "), false, sb);
}
sb.append(prefix).append(isTail ? "└── " : "┌── ").append(token.toString()).append("\n");
if (left != null) {
left.toString(new StringBuilder().append(prefix).append(isTail ? " " : "│ "), true, sb);
}
return sb;
}
public static void main(String args[]) {
System.out.println("start.");
Lexer l = new Lexer();
while (true) {
try {
Token t = l.scan();
if (t.tag == Tag.NUM) {
System.out.println(
"Token : " + t.toString() + " , Tag:" + t.tag + " , value:" + ((Num) t).value);
} else if (t.tag == Tag.ID) {
System.out.println(
"Token : " + t.toString() + " , Tag:" + t.tag + " , lexeme:" + ((Word) t).lexeme);
} else {
System.out.println("Token : " + t.toString() + " , Tag:" + t.tag);
}
} catch (Exception e) {
e.printStackTrace();
}
}
}
/** Writes the data to file */
public void writeToFile(String path) {
try (DataOutputStream outputStream =
new DataOutputStream(new BufferedOutputStream(new FileOutputStream(path)))) {
// First write all unique tokens
Map<Token, Integer> tokenToId = new HashMap<>();
Set<Token> tokens = new HashSet<>();
for (NGram ngram : this.ngramModel.getNgrams().keySet()) {
if (ngram.length() == 1) {
tokens.add(ngram.at(0));
}
}
outputStream.writeInt(tokens.size());
int id = 0;
for (Token token : tokens) {
outputStream.writeUTF(token.toString());
tokenToId.put(token, id);
id++;
}
// Then the Good-Turing parameters
outputStream.writeInt(this.ngramModel.getGoodTuringEstimation().getTotal());
outputStream.writeDouble(this.ngramModel.getGoodTuringEstimation().getA());
outputStream.writeDouble(this.ngramModel.getGoodTuringEstimation().getB());
// Then the top ranked unigrams
outputStream.writeInt(this.ngramModel.topUnigrams().size());
for (NGram ngram : this.ngramModel.topUnigrams()) {
Token token = ngram.at(0);
outputStream.writeInt(tokenToId.get(token));
}
// Then the n-grams, where the token id points to the previous table
outputStream.writeInt(this.ngramModel.getNgrams().size());
for (Map.Entry<NGram, Integer> current : this.ngramModel.getNgrams().entrySet()) {
outputStream.writeInt(current.getKey().length());
for (int i = 0; i < current.getKey().length(); i++) {
Token token = current.getKey().at(i);
outputStream.writeInt(tokenToId.get(token));
}
outputStream.writeInt(current.getValue());
}
outputStream.flush();
} catch (Exception e) {
e.printStackTrace();
}
}
public final ElementDecl ElementDecl() throws ParseException {
Token t;
ASTStringNode name;
ContentSpec contentSpec;
Token firstToken = token;
jj_consume_token(31);
t = jj_consume_token(NAME);
name = new ASTStringNode(t.toString(), new WToken(token));
contentSpec = ContentSpec();
jj_consume_token(PCB);
{
if (true) return new ElementDecl(name, contentSpec, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
public final AttribDef AttribDef() throws ParseException {
Token t;
ASTStringNode name;
AttribType attribType;
DefaultDecl defaultDecl;
Token firstToken = token;
t = jj_consume_token(NAME);
name = new ASTStringNode(t.toString(), new WToken(token));
attribType = AttribType();
defaultDecl = DefaultDecl();
{
if (true) return new AttribDef(name, attribType, defaultDecl, firstToken.next, token);
}
throw new Error("Missing return statement in function");
}
public static String parse(
Lexer lexer,
HashMap<Nonterminal, HashMap<Terminal, Rule>> parsingTable,
Nonterminal startSymbol) {
Stack<Token> stack = new Stack<Token>();
ArrayList<Terminal> inputs = lexer.getTokens();
int count = 0;
stack.push(startSymbol);
while (!stack.isEmpty() && count < inputs.size()) {
Token top = stack.peek();
System.out.println(
"Testing top: " + top.toString() + ", next input: " + inputs.get(count).toString());
if (top instanceof Nonterminal) {
Nonterminal nextNonTerm = (Nonterminal) top;
Rule rules = parsingTable.get(nextNonTerm).get(inputs.get(count));
if (rules == null)
return "Invalid. No rule found for "
+ nextNonTerm.toString()
+ " and the input "
+ inputs.get(count);
stack.pop();
for (int i = rules.rule.length - 1; i > -1; i--) {
if (!rules.rule[i].equals(new Terminal(Terminal.TerminalType.EPSILON)))
stack.push(rules.rule[i]);
}
} else if (top instanceof Terminal) {
Terminal nextTerm = (Terminal) top;
Terminal nextInput = inputs.get(count);
if (nextTerm.equals(nextInput)) {
++count;
stack.pop();
} else return nextTerm + " and " + nextInput + " do not match: invalid";
}
System.out.println("Current stack: " + stack.toString());
}
if (stack.isEmpty()) return "Valid";
return "Invalid. Ran out of tokens early or ran over input count";
}
protected void onActivityResult(int requestCode, int resultCode, Intent intent) {
switch (requestCode) {
case REQUEST_PICK_TOKEN:
if (resultCode == RESULT_OK && intent != null) {
Uri name = intent.getData();
try {
// URI vs Uri -> Java/Android fail ...
File f = new File(new URI(name.toString()));
Token t = Token.fromFile(f);
SharedPreferences p = PreferenceManager.getDefaultSharedPreferences(getBaseContext());
SharedPreferences.Editor e = p.edit();
e.putString("token", t.toString());
e.commit();
Toast.makeText(this, "Token successfully imported.", Toast.LENGTH_LONG).show();
} catch (Exception e) {
e.printStackTrace();
}
}
break;
default:
Log.w(TAG, "Unknown activity " + intent.toString());
}
}
public void writeDictionary(TokenStream tokenStream, String docId, int docSize) {
count++;
while (tokenStream.hasNext()) {
Node node = new Node();
node.setDocId(docId);
Token token = tokenStream.next();
String tokenText = token.toString();
int indexPos = token.getIndexPos();
node.setIndexPos(indexPos);
node.setDocLength(docSize);
if (!tokenText.isEmpty()) {
node.setTokenText(tokenText);
placeList.add(node);
}
}
if (count == 20000) {
count = 0;
sortFiles();
placeList = new ArrayList<Node>();
initializeWriter();
}
}
public IllegalTokenNameException(Token token) {
super(token.toString());
}
/**
* toString
*
* @return regresa una representacion en cadena de el nodo.
*/
public String toString() {
return tok.toString();
}
public Token next() {
currentToken = internalNext();
if (debug) System.out.println(currentToken == null ? "(null)" : currentToken.toString());
return currentToken;
}
