/**
* Parse a string value and covert it to its proper data type.
*
* @param value
* @return The parsed value.
* @throws ParsingException Thrown if not Advisory is available and there was an error parsing.
*/
@SuppressWarnings("unchecked")
public static <T> T parseValue(String value) throws ParsingException {
final T result;
assert value != null;
/*
* todo Restructure the whole class so I've got parse and parseArray as
* static methods and the share code properly.
*/
if ("null".equals(value)) {
result = null;
} else {
final List<Object> list = new ArrayList<>();
final Text t = new Text();
t.append(value);
if (any(t, list) && t.isEof()) {
result = (T) list.get(0);
} else {
error("Count not parse value: " + value);
result = null;
}
}
return result;
}
private static boolean number(Text t, DataType requiredType, List<Object> valueList) {
// posNumber
// : cardinality
// | '0b' BinaryDigits+ [zZ]?
// | '0x' HexDigit+ [zZ]?
// | '-'? Pint? ( '.' Digit+ ) ('e' Pint) [fF]
// | '-'? Pint [zZfF]?
// ;
final boolean result;
assert !t.isEof() : "Should have been checked by caller";
if (binary(t, valueList, requiredType) || hex(t, valueList, requiredType)) {
// ?todo I could allow negatives here
// ( binary | hex | '0' )
result = true;
} else {
// ( Int DecPart | Int | DecPart ) ( 'e' Int )? [zZfF]?
final int start = t.cursor();
if ((t.consumeInt() && (decPart(t) || true) && (exponent(t) || true))
|| (t.consume('-') || true) && decPart(t) && (exponent(t) || true)) {
final String string = t.getString(start);
final Object value = deriveType(string, requiredType, t, 10);
valueList.add(value);
result = true;
} else {
result = false;
}
}
return result;
}
/**
* Parse a value.
*
* @param tb Source.
* @param type The type of the array, may be any..
* @param valueList Result added here.
* @return Return true if a value has been added to the valueList and the cursor advanced or false
* and the cursor is as it was.
*/
private static boolean value(Text t, DataType type, List<Object> valueList) {
boolean result;
if (t.isEof()) {
result = false;
} else if (t.consume("null")) {
valueList.add(null);
result = true;
} else {
switch (type) {
case z:
case Z:
case f:
case F:
result = number(t, type, valueList);
break;
case text:
result = text(t, valueList);
break;
case identifier:
result = identifier(t, valueList);
break;
case path:
result = path(t, valueList);
break;
case datetime:
result = datetime(t, valueList);
break;
case date:
result = date(t, valueList);
break;
case time:
result = time(t, valueList);
break;
case bool:
result = bool(t, valueList);
break;
case cardinality:
result = cardinality(t, valueList);
break;
case any:
result = any(t, valueList);
break;
default:
throw new UnexpectedException("value: " + type);
}
}
return result;
}
/**
* Given the parsing context and what numerical data type is expected convert a string to the
* correct type. Note no attempt is made to let the magnitude of the number influence our choice.
*
* @param string The string to convert to a number. E.g. "123.3e2". If it contains a '.' or an 'e'
* then the type must either be f or F.
* @param requiredType Either z, Z, f, F or any.
* @param tb The source. The cursor will be at the end of the number but any type specifier will
* not have been consumed. If there is one then we'll eat it.
* @return The derived type.
* @throws ParsingException If there is a clash of types.
*/
private static Object deriveType(String string, DataType requiredType, Text t, int radix) {
final Object result;
// Figure out the correct type...
final DataType derivedType;
if (t.isEof()) {
if (requiredType == DataType.any) {
if (string.indexOf('.') >= 0 || string.indexOf('e') >= 0) {
derivedType = DataType.f;
} else {
derivedType = DataType.z;
}
} else {
derivedType = requiredType;
}
} else {
final char c = t.peek();
if (c == 'z' || c == 'Z' || c == 'f' || c == 'F') {
t.consume(c);
derivedType = DataType.valueOf(String.valueOf(c));
if (!(requiredType == DataType.any || requiredType == derivedType)) {
throw new ParsingException("Incompatible type: " + string + c);
}
} else {
if (requiredType == DataType.any) {
if (string.indexOf('.') >= 0 || string.indexOf('e') >= 0) {
derivedType = DataType.f;
} else {
derivedType = DataType.z;
}
} else {
derivedType = requiredType;
}
}
}
switch (derivedType) {
case z:
result = new Long(Long.parseLong(string, radix));
break;
case Z:
result = new BigInteger(string, radix);
break;
case f:
result = new Double(string);
break;
case F:
result = new BigDecimal(string);
break;
// $CASES-OMITTED$
default:
throw new UnexpectedException("toType: " + derivedType);
}
return result;
}
private static boolean decPart(Text t) {
final boolean result;
// '.' Digits+
final int save = t.cursor();
if (t.consume('.') && t.consumeAscii(Text.ASCII_0_9)) {
result = true;
} else {
t.setCursor(save);
result = false;
}
return result;
}
private static boolean exponent(Text t) {
final boolean result;
// 'e' Int
final int save = t.cursor();
if (t.consume('e') && t.consumeInt()) {
result = true;
} else {
t.setCursor(save);
result = false;
}
return result;
}
private static boolean identifier(Text t, List<Object> valueList) {
final boolean result;
final int start = t.cursor();
if (Identifier.consume(t)) {
final String string = t.getString(start);
valueList.add(new Identifier(string));
result = true;
} else {
result = false;
}
return result;
}
private static boolean bool(Text t, List<Object> valueList) {
final boolean result;
if (t.consume("true")) {
valueList.add(Boolean.TRUE);
result = true;
} else if (t.consume("false")) {
valueList.add(Boolean.FALSE);
result = true;
} else {
result = false;
}
return result;
}
private static boolean date(Text t, List<Object> valueList) {
final boolean result;
final int start = t.cursor();
if (t.consume('@') && date(t)) {
final String string = t.getString(start + 1); // Jump the'@'
final LocalDate value = DateU.parseStandardDate(string);
valueList.add(value);
result = true;
} else {
result = false;
}
return result;
}
private static boolean text(Text t, List<Object> valueList) {
final boolean result;
final int start = t.cursor();
if (t.consumeString()) {
// +/- 1s: drop the quotes
final String textValue = unescape(t, start + 1, t.cursor() - 1);
valueList.add(textValue);
result = true;
} else {
// ?todo Could easily detect unterminated string and throw here
result = false;
}
return result;
}
private static boolean binary(Text t, List<Object> valueList, DataType requiredType) {
final boolean result;
// '0b' [01]+ [zZfF]?
final int save = t.cursor();
if (t.consume("0b") && t.consumeAscii(Text.ASCII_0_1)) {
final String string = t.getString(save + 2);
final Object value = deriveType(string, requiredType, t, 2);
valueList.add(value);
result = true;
} else {
t.setCursor(save);
result = false;
}
return result;
}
/**
* Convert a value to an Oak markup representation.
*
* @param value The value to represent.
* @param tb The buffer in which to write the markup.
*/
public void toMarkup(Object value, Text t) {
if (value == null) {
t.append("null");
} else {
switch (this) {
case z:
case Z:
case f:
case F:
case bool:
t.append(value.toString());
break;
case cardinality:
((Cardinality) value).toString(t);
break;
case text:
escapeForOak((String) value, true, t);
break;
case identifier:
t.append(value.toString());
break;
case path:
((Path) value).toString(t);
break;
case datetime:
t.append('@');
t.append(DateU.formatStandardDatetime((LocalDateTime) value));
break;
case date:
t.append('@');
t.append(DateU.formatStandardDate((LocalDate) value));
break;
case time:
t.append('@');
t.append(DateU.formatStandardTime((LocalTime) value));
break;
case any:
default:
throw new UnexpectedException("asString: " + this);
}
}
}
public static void toString(Object value, Text t) {
if (value == null) {
t.append("null");
} else {
final Class<? extends Object> clazz = value.getClass();
final DataType type = getDataType(clazz);
if (type == null) {
throw new RuntimeException("Invalid type: " + clazz.getName());
}
type.toMarkup(value, t);
}
}
/**
* Convert a potentially escaped string to text.
*
* @param t The source string.
* @param start Starting offset.
* @param end Ending offset.
* @return Return the un-escaped string which may be zero-length but not null.
*/
private static String unescape(Text t, int start, int end) {
final Text result = new Text();
/*
* We know that there's a starting and ending quote that's been removed
* from the string but otherwise we can't trust the contents.
*/
for (int i = start; i < end; i++) {
final char c = t.charAt(i);
final char escaped;
if (c == '\\') {
if (i < end - 1) {
i++;
final char next = t.charAt(i);
if (next == 't') {
escaped = '\t';
} else if (next == 'n') {
escaped = '\n';
} else if (next == '"') {
escaped = '"';
} else if (next == '\\') {
escaped = '\\';
} else {
error("Invalid escape: '\\" + next + '\'');
escaped = next;
}
} else {
error("Unterminated string");
escaped = c;
}
} else {
escaped = c;
}
result.append(escaped);
}
return result.toString();
}
/**
* There's a bug in this as it requires seconds in the string whereas the time() function does
* not. It seems to be in the Java parser.
*
* @param tb Source to parse.
* @return A Temporal or null if none found.
*/
@Nullable
private static Temporal temporal(Text t) {
Temporal result;
final int save = t.cursor();
t.consume('@');
final int start = save + 1;
// yyyy/MM/dd HH:mm:ss
try {
if (date(t)) {
final int save2 = t.cursor();
t.ws();
if (time(t)) {
final String string = t.getString(start);
result = DateU.parseStandardDatetime(string);
} else {
t.setCursor(save2);
final String string = t.getString(start);
result = DateU.parseStandardDate(string);
}
} else if (time(t)) {
final String string = t.getString(start);
result = DateU.parseStandardTime(string);
} else {
result = null;
t.setCursor(save);
}
} catch (final DateTimeParseException e) {
result = null;
t.setCursor(save);
}
return result;
}
private static boolean time(Text t) {
final boolean result;
// HH ':' mm ( : ss )?
// Try for HH:mm...
int reset = t.cursor();
if (t.consumeAscii(Text.ASCII_0_9) && t.consume(':') && t.consumeAscii(Text.ASCII_0_9)) {
result = true;
reset = t.cursor();
// Try for :ss...
if (t.consume(':') && t.consumeAscii(Text.ASCII_0_9)) {
reset = t.cursor();
}
} else {
result = false;
}
t.setCursor(reset);
return result;
}
/**
* Escape a string for Oak. If the string is null then "null" is returned otherwise the string is
* returned with [tn\"] escaped \r discarded and the whole thing surrounded in quotes if
* requested.
*
* @param string The String to process.
* @param quote Double quotes are added if true.
* @param result Where to place the processed String.
*/
public static void escapeForOak(String string, boolean quote, Text result) {
if (string == null) {
result.append("null");
} else {
if (quote) {
result.append('"');
}
final char[] ca = string.toCharArray();
for (final char c : ca) {
switch (c) {
case '\t':
result.append("\\t");
break;
case '\n':
result.append("\\n");
break;
case '\"':
result.append("\\\"");
break;
case '\\':
result.append("\\\\");
break;
default:
if (c < ' ' || c >= '~') {
// Must be four digits
result.append("\\u" + Integer.toHexString(c | 0x10000).substring(1));
} else {
result.append(c);
}
}
}
if (quote) {
result.append('"');
}
}
}
/**
* Convert a list of values to an Oak markup String, e.g. "[ true, null, false ]".
*
* @param values The list of values
* @return An Oak markup representation of the values.
* @see #parse(String)
*/
public String toString(List<Object> values) {
assert values != null : "Arrays can't be null, only empty";
final Text t = new Text(true);
t.append('[');
if (values.size() > 0) {
t.space();
for (final Object value : values) {
t.delimit();
toMarkup(value, t);
}
t.space();
}
t.append(']');
return t.toString();
}
private static boolean date(Text t) {
final boolean result;
// yyyy '/' MM '/' dd
final int save = t.cursor();
if (t.consumeAscii(Text.ASCII_0_9)
&& t.consume('/')
&& t.consumeAscii(Text.ASCII_0_9)
&& t.consume('/')
&& t.consumeAscii(Text.ASCII_0_9)) {
result = true;
} else {
result = false;
t.setCursor(save);
}
return result;
}
private static boolean any(Text t, List<Object> valueList) {
final boolean result;
// value
// : Text
// | Path
// | Date | Time | DateTime
// | 'null'
// | 'true' | 'false'
// | z | Z | f | F
// | Identifier
// | Cardinality
// ;
/*
* Try and parse a value, we don't know the type so use the first
* character as an indicator to jump to the right type.
*
* We know we're not EOF and "null" has been dealt with by the caller.
*/
final char c = t.peek();
if (c == '"') { // Text
result = text(t, valueList);
} else if (c == '`') { // Path
result = path(t, valueList);
} else if (c == '@') { // Temporal
result =
datetime(t, valueList) // Must be first
|| date(t, valueList)
|| time(t, valueList);
} else if (c >= '0' && c <= '9' || c == '.' || c == '-') {
result = cardinality(t, valueList) || number(t, DataType.any, valueList);
} else {
result =
bool(t, valueList) // Must be first
|| identifier(t, valueList);
}
return result;
}
private static List<Object> elementList(Text t, DataType type) {
final List<Object> result = new ArrayList<>();
// elementList: ( WS? value ( WS? ',' WS? value )* WS? )?
if (t.ws() && value(t, type, result)) {
while (true) {
final int save = t.cursor();
if (t.ws() && t.consume(',') && t.ws() && value(t, type, result)) {
continue;
}
t.setCursor(save);
break;
}
}
return result;
}
/**
* Parse a string into a list of data values. White space is supported only in between the [square
* brackets].
*
* @param string A string representation of an Oak property array, e.g. [ true, null, false ]
* @param type The type of the array, may be 'any'.
* @return A potentially empty but not null list of values.
*/
@SuppressWarnings("unchecked")
public static <T> List<T> parseArray(String string, DataType type) {
final List<?> result;
final Text t = new Text();
t.append(string);
if (t.consume('[')
&& (result = elementList(t, type)) != null
&& (t.ws() && t.consume(']'))
&& t.isEof()) {
// OK
} else {
throw new ParsingException("Invalid array: " + string);
}
return (List<T>) result;
}
/**
* Convert a value to an Oak markup representation.
*
* @param value The value to represent.
* @return E.g. `/Some/Path`
* @see #parse(String)
*/
public final String toString(Object value) {
final Text t = new Text();
toMarkup(value, t);
return t.toString();
}
/**
* In Oak it is possible to embed a tab character as an ASCII 8 or as a \t but internally we store
* this in ASCII. The same goes for newlines and other tables (see the Oak reference). Here we
* convert external format to internal format.
*
* @param source The text to process, may be null. E.g. "Hello\tworld" (with the quotes)
* @return Return an internal format string.
*/
@Nullable
public static String textToInternalFormat(@Nullable String source) throws ParsingException {
final String result;
assert source == null
|| source.length() >= 2
&& source.charAt(0) == '"'
&& source.charAt(source.length() - 1) == '"';
if (source == null) {
result = null;
} else {
final Text t = new Text();
final char[] ca = source.toCharArray();
final int length = ca.length - 1; // 1 because remove quotes
for (int i = 1; i < length; i++) { // 1 because remove quotes
final char c = ca[i];
if (c == '\\') {
if (i == length - 1) {
error("Invalid text " + source + ", escape at end of line");
} else {
i++;
final char next = ca[i];
if (next == 't') {
t.append('\t');
} else if (next == 'n') {
t.append('\n');
} else if (next == '"') {
t.append('"');
} else if (next == '\\') {
t.append('\\');
} else if (next == 'u') {
// Unicode, 1-4 hex characters...
i++;
final Text hex = new Text();
hex.append(source);
hex.setCursor(i);
final int start = i;
if (hex.consumeAscii(Text.ASCII_0_F)) {
final int end = start + Math.min(4, hex.cursor() - start);
final String string = hex.getString(start, end);
final char u = (char) Integer.parseInt(string, 16);
t.append(u);
i = end - 1;
} else {
error("Invalid text " + source + ", incorrect unicode");
}
} else {
error("Invalid text: \\" + next);
}
}
} else {
t.append(c);
}
}
result = t.toString();
}
return result;
}
