private Ptg parseNumber() {
String number2 = null;
String exponent = null;
String number1 = GetNum();
if (look == '.') {
GetChar();
number2 = GetNum();
}
if (look == 'E') {
GetChar();
String sign = "";
if (look == '+') {
GetChar();
} else if (look == '-') {
GetChar();
sign = "-";
}
String number = GetNum();
if (number == null) {
throw expected("Integer");
}
exponent = sign + number;
}
if (number1 == null && number2 == null) {
throw expected("Integer");
}
return getNumberPtgFromString(number1, number2, exponent);
}
private ParseNode parseRangeExpression() {
ParseNode result = parseRangeable();
boolean hasRange = false;
while (look == ':') {
int pos = _pointer;
GetChar();
ParseNode nextPart = parseRangeable();
// Note - no range simplification here. An expr like "A1:B2:C3:D4:E5" should be
// grouped into area ref pairs like: "(A1:B2):(C3:D4):E5"
// Furthermore, Excel doesn't seem to simplify
// expressions like "Sheet1!A1:Sheet1:B2" into "Sheet1!A1:B2"
checkValidRangeOperand("LHS", pos, result);
checkValidRangeOperand("RHS", pos, nextPart);
ParseNode[] children = {
result, nextPart,
};
result = new ParseNode(RangePtg.instance, children);
hasRange = true;
}
if (hasRange) {
return augmentWithMemPtg(result);
}
return result;
}
private String parseStringLiteral() {
Match('"');
StringBuffer token = new StringBuffer();
while (true) {
if (look == '"') {
GetChar();
if (look != '"') {
break;
}
}
token.append(look);
GetChar();
}
return token.toString();
}
/** Get a Number */
private String GetNum() {
StringBuffer value = new StringBuffer();
while (IsDigit(this.look)) {
value.append(this.look);
GetChar();
}
return value.length() == 0 ? null : value.toString();
}
/**
* Note - caller should reset {@link #_pointer} upon <code>null</code> result
*
* @return The sheet name as an identifier <code>null</code> if '!' is not found in the right
* place
*/
private SheetIdentifier parseSheetName() {
String bookName;
if (look == '[') {
StringBuilder sb = new StringBuilder();
GetChar();
while (look != ']') {
sb.append(look);
GetChar();
}
GetChar();
bookName = sb.toString();
} else {
bookName = null;
}
if (look == '\'') {
StringBuffer sb = new StringBuffer();
Match('\'');
boolean done = look == '\'';
while (!done) {
sb.append(look);
GetChar();
if (look == '\'') {
Match('\'');
done = look != '\'';
}
}
Identifier iden = new Identifier(sb.toString(), true);
// quoted identifier - can't concatenate anything more
SkipWhite();
if (look == '!') {
GetChar();
return new SheetIdentifier(bookName, iden, _book);
}
return null;
}
// unquoted sheet names must start with underscore or a letter
if (look == '_' || Character.isLetter(look)) {
StringBuilder sb = new StringBuilder();
// can concatenate idens with dots
while (isUnquotedSheetNameChar(look)) {
sb.append(look);
GetChar();
}
SkipWhite();
if (look == '!') {
GetChar();
return new SheetIdentifier(bookName, new Identifier(sb.toString(), false), _book);
}
return null;
}
return null;
}
/**
* Parses simple factors that are not primitive ranges or range components i.e. '!', ':'(and equiv
* '...') do not appear Examples
*
* <pre>
* my.named...range.
* foo.bar(123.456, "abc")
* 123.456
* "abc"
* true
* </pre>
*/
private ParseNode parseNonRange(int savePointer) {
resetPointer(savePointer);
if (Character.isDigit(look)) {
return new ParseNode(parseNumber());
}
if (look == '"') {
return new ParseNode(new StringPtg(parseStringLiteral()));
}
// from now on we can only be dealing with non-quoted identifiers
// which will either be named ranges or functions
StringBuilder sb = new StringBuilder();
// defined names may begin with a letter or underscore
if (!Character.isLetter(look) && look != '_') {
throw expected("number, string, or defined name");
}
while (isValidDefinedNameChar(look)) {
sb.append(look);
GetChar();
}
SkipWhite();
String name = sb.toString();
if (look == '(') {
return function(name);
}
if (name.equalsIgnoreCase("TRUE") || name.equalsIgnoreCase("FALSE")) {
return new ParseNode(BoolPtg.valueOf(name.equalsIgnoreCase("TRUE")));
}
if (_book == null) {
// Only test cases omit the book (expecting it not to be needed)
throw new IllegalStateException("Need book to evaluate name '" + name + "'");
}
// 20101115, [email protected]: shall provide a temporary defined named record
// EvaluationName evalName = _book.getName(name, _sheetIndex);
EvaluationName evalName = _book.getOrCreateName(name, _sheetIndex);
if (evalName == null) {
// 20101112, [email protected]: shall return #NAME? error
// throw new FormulaParseException("Specified named range '"
// + name + "' does not exist in the current workbook.");
return new ParseNode(ErrPtg.NAME_INVALID);
}
// 20101115, [email protected]: unnecessary check
// if (evalName.isRange()) {
// return new ParseNode(evalName.createPtg());
// }
// TODO - what about NameX ?
// throw new FormulaParseException("Specified name '"
// + name + "' is not a range as expected.");
return new ParseNode(evalName.createPtg());
}
/** API call to execute the parsing of the formula */
private void parse() {
_pointer = 0;
GetChar();
_rootNode = unionExpression();
if (_pointer <= _formulaLength) {
String msg =
"Unused input ["
+ _formulaString.substring(_pointer - 1)
+ "] after attempting to parse the formula ["
+ _formulaString
+ "]";
throw new FormulaParseException(msg);
}
}
private String parseUnquotedIdentifier() {
if (look == '\'') {
throw expected("unquoted identifier");
}
StringBuilder sb = new StringBuilder();
while (Character.isLetterOrDigit(look) || look == '.') {
sb.append(look);
GetChar();
}
if (sb.length() < 1) {
return null;
}
return sb.toString();
}
private ParseNode unionExpression() {
ParseNode result = comparisonExpression();
boolean hasUnions = false;
while (true) {
SkipWhite();
switch (look) {
case ',':
GetChar();
hasUnions = true;
ParseNode other = comparisonExpression();
result = new ParseNode(UnionPtg.instance, result, other);
continue;
}
if (hasUnions) {
return augmentWithMemPtg(result);
}
return result;
}
}
/**
* Parses area refs (things which could be the operand of ':') and simple factors Examples
*
* <pre>
* A$1
* $A$1 : $B1
* A1 ....... C2
* Sheet1 !$A1
* a..b!A1
* 'my sheet'!A1
* .my.sheet!A1
* my.named..range.
* foo.bar(123.456, "abc")
* 123.456
* "abc"
* true
* </pre>
*/
private ParseNode parseRangeable() {
SkipWhite();
int savePointer = _pointer;
SheetIdentifier sheetIden = parseSheetName();
if (sheetIden == null) {
resetPointer(savePointer);
} else {
SkipWhite();
savePointer = _pointer;
}
SimpleRangePart part1 = parseSimpleRangePart();
if (part1 == null) {
if (sheetIden != null) {
if (look == '#') { // error ref like MySheet!#REF!
return new ParseNode(ErrPtg.valueOf(parseErrorLiteral()));
} else {
throw new FormulaParseException(
"Cell reference expected after sheet name at index " + _pointer + ".");
}
}
return parseNonRange(savePointer);
}
boolean whiteAfterPart1 = IsWhite(look);
if (whiteAfterPart1) {
SkipWhite();
}
if (look == ':') {
int colonPos = _pointer;
GetChar();
SkipWhite();
SimpleRangePart part2 = parseSimpleRangePart();
if (part2 != null && !part1.isCompatibleForArea(part2)) {
// second part is not compatible with an area ref e.g. S!A1:S!B2
// where S might be a sheet name (that looks like a column name)
part2 = null;
}
if (part2 == null) {
// second part is not compatible with an area ref e.g. A1:OFFSET(B2, 1, 2)
// reset and let caller use explicit range operator
resetPointer(colonPos);
if (!part1.isCell()) {
String prefix;
if (sheetIden == null) {
prefix = "";
} else {
prefix = "'" + sheetIden.getSheetIdentifier().getName() + '!';
}
throw new FormulaParseException(prefix + part1.getRep() + "' is not a proper reference.");
}
return createAreaRefParseNode(sheetIden, part1, part2);
}
return createAreaRefParseNode(sheetIden, part1, part2);
}
if (look == '.') {
GetChar();
int dotCount = 1;
while (look == '.') {
dotCount++;
GetChar();
}
boolean whiteBeforePart2 = IsWhite(look);
SkipWhite();
SimpleRangePart part2 = parseSimpleRangePart();
String part1And2 = _formulaString.substring(savePointer - 1, _pointer - 1);
if (part2 == null) {
if (sheetIden != null) {
throw new FormulaParseException(
"Complete area reference expected after sheet name at index " + _pointer + ".");
}
return parseNonRange(savePointer);
}
if (whiteAfterPart1 || whiteBeforePart2) {
if (part1.isRowOrColumn() || part2.isRowOrColumn()) {
// "A .. B" not valid syntax for "A:B"
// and there's no other valid expression that fits this grammar
throw new FormulaParseException(
"Dotted range (full row or column) expression '"
+ part1And2
+ "' must not contain whitespace.");
}
return createAreaRefParseNode(sheetIden, part1, part2);
}
if (dotCount == 1 && part1.isRow() && part2.isRow()) {
// actually, this is looking more like a number
return parseNonRange(savePointer);
}
if (part1.isRowOrColumn() || part2.isRowOrColumn()) {
if (dotCount != 2) {
// [email protected]: shall return #NAME?
// throw new FormulaParseException("Dotted range (full row or column) expression '" +
// part1And2
// + "' must have exactly 2 dots.");
return parseNonRange(savePointer);
}
}
return createAreaRefParseNode(sheetIden, part1, part2);
}
if (part1.isCell() && isValidCellReference(part1.getRep())) {
return createAreaRefParseNode(sheetIden, part1, null);
}
if (sheetIden != null) {
throw new FormulaParseException(
"Second part of cell reference expected after sheet name at index " + _pointer + ".");
}
return parseNonRange(savePointer);
}
/**
* Consumes the next input character if it is equal to the one specified otherwise throws an
* unchecked exception. This method does <b>not</b> consume whitespace (before or after the
* matched character).
*/
private void Match(char x) {
if (look != x) {
throw expected("'" + x + "'");
}
GetChar();
}
/** Skip Over Leading White Space */
private void SkipWhite() {
while (IsWhite(look)) {
GetChar();
}
}