A regular expression matching engine using Non-deterministic Finite Automaton (NFA). This engine does not conform to the POSIX regular expression.

---

How to use

A. Standard way

 RegularExpression re = new RegularExpression(regex);
 if (re.matches(text)) { ... }
 

B. Capturing groups

 RegularExpression re = new RegularExpression(regex);
 Match match = new Match();
 if (re.matches(text, match)) {
     ... // You can refer captured texts with methods of the Match class.
 }
 

Case-insensitive matching

 RegularExpression re = new RegularExpression(regex, "i");
 if (re.matches(text) >= 0) { ...}
 

Options

You can specify options to RegularExpression(regex, options) or setPattern(regex, options). This options parameter consists of the following characters.

"i"

This option indicates case-insensitive matching.

"m"

^ and $ consider the EOL characters within the text.

"s"

. matches any one character.

"u"

Redefines \d \D \w \W \s \S \b \B \< \> as becoming to Unicode.

"w"

By this option, \b \B \< \> are processed with the method of 'Unicode Regular Expression Guidelines' Revision 4. When "w" and "u" are specified at the same time, \b \B \< \> are processed for the "w" option.

","

The parser treats a comma in a character class as a range separator. [a,b] matches a or , or b without this option. [a,b] matches a or b with this option.

"X"

By this option, the engine confoms to XML Schema: Regular Expression. The match() method does not do subsring matching but entire string matching.

---

Syntax

Differences from the Perl 5 regular expression There is 6-digit hexadecimal character representation (\vHHHHHH.) Supports subtraction, union, and intersection operations for character classes. Not supported: \ooo (Octal character representations), \G, \C, \lc, \ uc, \L, \U, \E, \Q, \N{name}, (?{code}), (??{code})

Meta characters are `. * + ? { [ ( ) | \ ^ $'.

• Character

  . (A period)

  Matches any one character except the following characters.

  LINE FEED (U+000A), CARRIAGE RETURN (U+000D), PARAGRAPH SEPARATOR (U+2029), LINE SEPARATOR (U+2028)

  This expression matches one code point in Unicode. It can match a pair of surrogates.

  When the "s" option is specified, it matches any character including the above four characters.

  \e \f \n \r \t

  Matches ESCAPE (U+001B), FORM FEED (U+000C), LINE FEED (U+000A), CARRIAGE RETURN (U+000D), HORIZONTAL TABULATION (U+0009)

  \cC

  Matches a control character. The C must be one of '@', 'A'-'Z', '[', '\', ']', '^', '_'. It matches a control character of which the character code is less than the character code of the C by 0x0040.

  For example, a \cJ matches a LINE FEED (U+000A), and a \c[ matches an ESCAPE (U+001B).

  a non-meta character

  Matches the character.

  \ + a meta character

  Matches the meta character.

  \xHH \x{HHHH}

  Matches a character of which code point is HH (Hexadecimal) in Unicode. You can write just 2 digits for \xHH, and variable length digits for \x{HHHH}.

  \vHHHHHH

  Matches a character of which code point is HHHHHH (Hexadecimal) in Unicode.

  \g

  Matches a grapheme.

  It is equivalent to (?[\p{ASSIGNED}]-[\p{M}\p{C}])?(?:\p{M}|[\x{094D}\x{09CD}\x{0A4D}\x{0ACD}\x{0B3D}\x{0BCD}\x{0C4D}\x{0CCD}\x{0D4D}\x{0E3A}\x{0F84}]\p{L}|[\x{1160}-\x{11A7}]|[\x{11A8}-\x{11FF}]|[\x{FF9E}\x{FF9F}])*

  \X

  Matches a combining character sequence. It is equivalent to (?:\PM\pM*)

• Character class

  + *

  [R₁R₂...R_n] (without "," option) + *

  [R₁,R₂,...,R_n] (with "," option)

  Positive character class. It matches a character in ranges.

  R_n:

  • A character (including \e \f \n \r \t \xHH \x{HHHH} \vHHHHHH)

    This range matches the character.

  • C₁-C₂

    This range matches a character which has a code point that is >= C₁'s code point and <= C₂'s code point. + *

  • A POSIX character class: [:alpha:] [:alnum:] [:ascii:] [:cntrl:] [:digit:] [:graph:] [:lower:] [:print:] [:punct:] [:space:] [:upper:] [:xdigit:], + * and negative POSIX character classes in Perl like [:^alpha:]

    ...

  • \d \D \s \S \w \W \p{name} \P{name}

    These expressions specifies the same ranges as the following expressions.

  Enumerated ranges are merged (union operation). [a-ec-z] is equivalent to [a-z]

  [^R₁R₂...R_n] (without a "," option)

  [^R₁,R₂,...,R_n] (with a "," option)

  Negative character class. It matches a character not in ranges.

  (?[ranges]op[ranges]op[ranges] ... ) (op is - or + or &.)

  Subtraction or union or intersection for character classes.

  For exmaple, (?[A-Z]-[CF]) is equivalent to [A-BD-EG-Z], and (?[0x00-0x7f]-[K]&[\p{Lu}]) is equivalent to [A-JL-Z].

  The result of this operations is a positive character class even if an expression includes any negative character classes. You have to take care on this in case-insensitive matching. For instance, (?[^b]) is equivalent to [\x00-ac-\x{10ffff}], which is equivalent to [^b] in case-sensitive matching. But, in case-insensitive matching, (?[^b]) matches any character because it includes 'B' and 'B' matches 'b' though [^b] is processed as [^Bb].

  [R₁R₂...-[R_nR_n+1...]] (with an "X" option)

  Character class subtraction for the XML Schema. You can use this syntax when you specify an "X" option.

  \d

  Equivalent to [0-9].

  When a "u" option is set, it is equivalent to \p{Nd}.

  \D

  Equivalent to [^0-9]

  When a "u" option is set, it is equivalent to \P{Nd}.

  \s

  Equivalent to [ \f\n\r\t]

  When a "u" option is set, it is equivalent to [ \f\n\r\t\p{Z}].

  \S

  Equivalent to [^ \f\n\r\t]

  When a "u" option is set, it is equivalent to [^ \f\n\r\t\p{Z}].

  \w

  Equivalent to [a-zA-Z0-9_]

  When a "u" option is set, it is equivalent to [\p{Lu}\p{Ll}\p{Lo}\p{Nd}_].

  \W

  Equivalent to [^a-zA-Z0-9_]

  When a "u" option is set, it is equivalent to [^\p{Lu}\p{Ll}\p{Lo}\p{Nd}_].

  \p{name}

  Matches one character in the specified General Category (the second field in UnicodeData.txt) or the specified Block. The following names are available:

  Unicode General Categories:

  L, M, N, Z, C, P, S, Lu, Ll, Lt, Lm, Lo, Mn, Me, Mc, Nd, Nl, No, Zs, Zl, Zp, Cc, Cf, Cn, Co, Cs, Pd, Ps, Pe, Pc, Po, Sm, Sc, Sk, So,

  (Currently the Cn category includes U+10000-U+10FFFF characters)

  Unicode Blocks:

  Basic Latin, Latin-1 Supplement, Latin Extended-A, Latin Extended-B, IPA Extensions, Spacing Modifier Letters, Combining Diacritical Marks, Greek, Cyrillic, Armenian, Hebrew, Arabic, Devanagari, Bengali, Gurmukhi, Gujarati, Oriya, Tamil, Telugu, Kannada, Malayalam, Thai, Lao, Tibetan, Georgian, Hangul Jamo, Latin Extended Additional, Greek Extended, General Punctuation, Superscripts and Subscripts, Currency Symbols, Combining Marks for Symbols, Letterlike Symbols, Number Forms, Arrows, Mathematical Operators, Miscellaneous Technical, Control Pictures, Optical Character Recognition, Enclosed Alphanumerics, Box Drawing, Block Elements, Geometric Shapes, Miscellaneous Symbols, Dingbats, CJK Symbols and Punctuation, Hiragana, Katakana, Bopomofo, Hangul Compatibility Jamo, Kanbun, Enclosed CJK Letters and Months, CJK Compatibility, CJK Unified Ideographs, Hangul Syllables, High Surrogates, High Private Use Surrogates, Low Surrogates, Private Use, CJK Compatibility Ideographs, Alphabetic Presentation Forms, Arabic Presentation Forms-A, Combining Half Marks, CJK Compatibility Forms, Small Form Variants, Arabic Presentation Forms-B, Specials, Halfwidth and Fullwidth Forms

  Others:

  ALL (Equivalent to [\u0000-\v10FFFF])

  ASSGINED (\p{ASSIGNED} is equivalent to \P{Cn})

  UNASSGINED (\p{UNASSIGNED} is equivalent to \p{Cn})

  \P{name}

  Matches one character not in the specified General Category or the specified Block.

• Selection and Quantifier

  X|Y

  ...

  X*

  Matches 0 or more X.

  X+

  Matches 1 or more X.

  X?

  Matches 0 or 1 X.

  X{number}

  Matches number times.

  X{min,}

  ...

  X{min,max}

  ...

  X*?

  X+?

  X??

  X{min,}?

  X{min,max}?

  Non-greedy matching.

• Grouping, Capturing, and Back-reference

  (?:X)

  Grouping. "foo+" matches "foo" or "foooo". If you want it matches "foofoo" or "foofoofoo", you have to write "(?:foo)+".

  (X)

  Grouping with capturing. It make a group and applications can know where in target text a group matched with methods of a Match instance after matches(String,Match). The 0th group means whole of this regular expression. The Nth gorup is the inside of the Nth left parenthesis.

  For instance, a regular expression is " *([^<:]*) +<([^>]*)> *" and target text is "From: TAMURA Kent <kent@trl.ibm.co.jp>":

  • Match.getCapturedText(0): " TAMURA Kent <kent@trl.ibm.co.jp>"
  • Match.getCapturedText(1): "TAMURA Kent"
  • Match.getCapturedText(2): "kent@trl.ibm.co.jp"

  \1 \2 \3 \4 \5 \6 \7 \8 \9

  (?>X)

  Independent expression group. ................

  (?options:X)

  (?options-options2:X)

  ............................

  The options or the options2 consists of 'i' 'm' 's' 'w'. Note that it can not contain 'u'.

  (?options)

  (?options-options2)

  ......

  These expressions must be at the beginning of a group.

• Anchor

  \A

  Matches the beginnig of the text.

  \Z

  Matches the end of the text, or before an EOL character at the end of the text, or CARRIAGE RETURN + LINE FEED at the end of the text.

  \z

  Matches the end of the text.

  ^

  Matches the beginning of the text. It is equivalent to \A.

  When a "m" option is set, it matches the beginning of the text, or after one of EOL characters ( LINE FEED (U+000A), CARRIAGE RETURN (U+000D), LINE SEPARATOR (U+2028), PARAGRAPH SEPARATOR (U+2029).)

  $

  Matches the end of the text, or before an EOL character at the end of the text, or CARRIAGE RETURN + LINE FEED at the end of the text.

  When a "m" option is set, it matches the end of the text, or before an EOL character.

  \b

  Matches word boundary. (See a "w" option)

  \B

  Matches non word boundary. (See a "w" option)

  \<

  Matches the beginning of a word. (See a "w" option)

  \>

  Matches the end of a word. (See a "w" option)

• Lookahead and lookbehind

  (?=X)

  Lookahead.

  (?!X)

  Negative lookahead.

  (?<=X)

  Lookbehind.

  (Note for text capturing......)

  (?<!X)

  Negative lookbehind.

• Misc.

  (?(condition)yes-pattern|no-pattern),

  (?(condition)yes-pattern)

  ......

  (?#comment)

  Comment. A comment string consists of characters except ')'. You can not write comments in character classes and before quantifiers.

---

BNF for the regular expression

 regex ::= ('(?' options ')')? term ('|' term)*
 term ::= factor+
 factor ::= anchors | atom (('*' | '+' | '?' | minmax ) '?'? )?
            | '(?#' [^)]* ')'
 minmax ::= '{' ([0-9]+ | [0-9]+ ',' | ',' [0-9]+ | [0-9]+ ',' [0-9]+) '}'
 atom ::= char | '.' | char-class | '(' regex ')' | '(?:' regex ')' | '\' [0-9]
          | '\w' | '\W' | '\d' | '\D' | '\s' | '\S' | category-block | '\X'
          | '(?>' regex ')' | '(?' options ':' regex ')'
          | '(?' ('(' [0-9] ')' | '(' anchors ')' | looks) term ('|' term)? ')'
 options ::= [imsw]* ('-' [imsw]+)?
 anchors ::= '^' | '$' | '\A' | '\Z' | '\z' | '\b' | '\B' | '\<' | '\>'
 looks ::= '(?=' regex ')'  | '(?!' regex ')'
           | '(?<=' regex ')' | '(?<!' regex ')'
 char ::= '\\' | '\' [efnrtv] | '\c' [@-_] | code-point | character-1
 category-block ::= '\' [pP] category-symbol-1
                    | ('\p{' | '\P{') (category-symbol | block-name
                                       | other-properties) '}'
 category-symbol-1 ::= 'L' | 'M' | 'N' | 'Z' | 'C' | 'P' | 'S'
 category-symbol ::= category-symbol-1 | 'Lu' | 'Ll' | 'Lt' | 'Lm' | Lo'
                     | 'Mn' | 'Me' | 'Mc' | 'Nd' | 'Nl' | 'No'
                     | 'Zs' | 'Zl' | 'Zp' | 'Cc' | 'Cf' | 'Cn' | 'Co' | 'Cs'
                     | 'Pd' | 'Ps' | 'Pe' | 'Pc' | 'Po'
                     | 'Sm' | 'Sc' | 'Sk' | 'So'
 block-name ::= (See above)
 other-properties ::= 'ALL' | 'ASSIGNED' | 'UNASSIGNED'
 character-1 ::= (any character except meta-characters)

 char-class ::= '[' ranges ']'
                | '(?[' ranges ']' ([-+&] '[' ranges ']')? ')'
 ranges ::= '^'? (range ','?)+
 range ::= '\d' | '\w' | '\s' | '\D' | '\W' | '\S' | category-block
           | range-char | range-char '-' range-char
 range-char ::= '\[' | '\]' | '\\' | '\' [,-efnrtv] | code-point | character-2
 code-point ::= '\x' hex-char hex-char
                | '\x{' hex-char+ '}'
                | '\v' hex-char hex-char hex-char hex-char hex-char hex-char
 hex-char ::= [0-9a-fA-F]
 character-2 ::= (any character except \[]-,)
 

---

TODO

• Unicode Technical Standard #18: Unicode Regular Expressions
  • 2.4 Canonical Equivalents
  • Level 3
• Parsing performance

---