public class FieldUtil extends Object
Modifier and Type | Field and Description |
---|---|
protected static String |
ID |
protected static String |
LENGTH_PREFIX |
Modifier and Type | Method and Description |
---|---|
static String |
decimalToString(BigDecimal value)
Creates a canonical string representation of the supplied
BigDecimal value, whereby all string representations are
lexicographically sortable. |
protected static org.apache.lucene.document.Field |
idField(String key) |
protected static org.apache.lucene.search.TermQuery |
idQuery(String key) |
protected static org.apache.lucene.index.Term |
idTerm(String key) |
static String |
lengthField(String propertyName)
Creates the name of the Lucene document field which will store the length of a property.
|
protected static String |
negate(String value)
Compute the "negated" string, which replaces the digits (0 becomes 9, 1 becomes 8, ... and 9 becomes 0).
|
protected static StringBuilder |
negate(StringBuilder value)
Compute the "negated" string, which replaces the digits (0 becomes 9, 1 becomes 8, ... and 9 becomes 0).
|
protected static void |
removeTralingZeros(StringBuilder sb)
Utility to remove the trailing 0's.
|
static BigDecimal |
stringToDecimal(String value)
Converts the canonical string representation of a
BigDecimal value into the object form. |
protected static final String ID
protected static final String LENGTH_PREFIX
public static String lengthField(String propertyName)
propertyName
- the name of the property; may not be nullnull
protected static org.apache.lucene.document.Field idField(String key)
protected static org.apache.lucene.search.TermQuery idQuery(String key)
protected static org.apache.lucene.index.Term idTerm(String key)
public static String decimalToString(BigDecimal value)
BigDecimal
value, whereby all string representations are
lexicographically sortable. This makes it possible to store the wide range of values that can be represented by BigDecimal,
while still enabling sorting and range queries.
This canonical form represents all decimal values using a prescribed format, which is based upon Steven Rowe's suggestion but with modifications to handle variable-length exponents (per his suggestion in the last sentence), use spaces between fields on where required (for minimal length), and utilize an optimized (e.g., shorter) form when the value is '0' or the exponent is '0'. Thus, this format contains only digits (e.g., '0'..'9') and the '-' and 'A' characters.
<significand-sign><exponent-sign><exponent-length> <exponent><significand>where:
BigDecimal.unscaledValue()
;value.precision() - value.scale() - 1
;<significand-sign>
is '-' if the significand is negative, '0' if equal to zero, or '1' if positive;
<exponent-sign>
is '-' if the exponent is negative, '0' if equal to zero, or '1' if positive; if
'0', then the <exponent-length>
and <exponent>
fields are not written;<exponent-length>
is the postive value representing the length of the <exponent>
,
and is not included when the <exponent-sign>
is '0';<exponent>
is the integer used to define the number of factors of 10 that are applied to the
significand, obtained by computing value.precision() - value.scale() - 1
;<significand>
is the part of the number containing the significant figures, and is a (big) integer
value obtained from the BigDecimal using BigDecimal.unscaledValue()
;<exponent-length>
field is negated such that each digit is replaced
with (base - digit - 1)
.
In the case of a negative significand, the <significand>
field is negated such that each digit is replaced
with (base - digit - 1)
and appended by 'A' (which is greater than all other digits) to ensure that
significands with greater precision are ordered before those that share significand prefixes but have lesser precision. In
this case, the <exponent-length>
and <exponent>
parts are also negated (unless they already
are).
Thus, the format for a negative BigDecimal value becomes:
-<reversed-exponent-sign><negated-exponent-length> <negated-exponent><significand><sentinel>where the
<sentinel>
is always 'A'. Note that the exponent length field is also negated.
Here are several examples that show BigDecimal values and their corresponding canonical string representation:
+5.E-3 => 1-8 65 +1.E-2 => 1-8 71 +1.0E-2 => 1-8 71 +1.0000E-2 => 1-8 71 +1.1E-2 => 1-8 711 +1.11E-2 => 1-8 7111 +1.2E-2 => 1-8 712 +5.E-2 => 1-8 75 +7.3E+2 => 111 273 +7.4E+2 => 111 274 +7.45E+2 => 111 2745 +8.7654E+3 => 111 387654Here is how a BigDecimal value of
zero
is represented:
0.0E0 => 0BigDecimal values with an exponent of '0' are represented as follows:
+1.2E0 => 1012 -1.2E0 => -087AAnd here are some negative value examples:
-8.7654E+3 => --8 612345A -7.45E+2 => --8 7254A -7.4E+2 => --8 725A -7.3E+2 => --8 726A -5.E-2 => -18 24A -1.2E-2 => -18 287A -1.11E-2 => -18 2888A -1.1E-2 => -18 288A -1.0000E-2 => -18 28A -1.0E-2 => -18 28A -1.E-2 => -18 28A -5.E-3 => -18 34A -5.E-4 => -18 44A
This canonical form is valid for all values of BigDecimal
.
value
- the value to be converted into its canonical form; may not be nullstringToDecimal(String)
public static BigDecimal stringToDecimal(String value)
BigDecimal
value into the object form.
See decimalToString(BigDecimal)
to documentation of the canonical form.
value
- the canonical string representation; may not be null or emptydecimalToString(BigDecimal)
protected static String negate(String value)
value
- the input string; may not be nullnegate(StringBuilder)
protected static StringBuilder negate(StringBuilder value)
value
- the input string; may not be nullnegate(String)
protected static void removeTralingZeros(StringBuilder sb)
sb
- the input string builder; may not be nullCopyright © 2008–2016 JBoss, a division of Red Hat. All rights reserved.