pcre2syntax - Man Page

The full syntax and semantics of the regular expressions that are supported by PCRE2 are described in the pcre2pattern documentation. This document contains a quick-reference summary of the syntax.

 \x         where x is non-alphanumeric is a literal x
 \Q...\E    treat enclosed characters as literal

Note that white space inside \Q...\E is always treated as literal, even if PCRE2_EXTENDED is set, causing most other white space to be ignored.

With one exception, wherever brace characters { and } are required to enclose data for constructions such as \g{2} or \k{name}, space and/or horizontal tab characters that follow { or precede } are allowed and are ignored. In the case of quantifiers, they may also appear before or after the comma. The exception is \u{...} which is not Perl-compatible and is recognized only when PCRE2_EXTRA_ALT_BSUX is set. This is an ECMAScript compatibility feature, and follows ECMAScript's behaviour.

This table applies to ASCII and Unicode environments. An unrecognized escape sequence causes an error.

 \a         alarm, that is, the BEL character (hex 07)
 \cx        "control-x", where x is a non-control ASCII character
 \e         escape (hex 1B)
 \f         form feed (hex 0C)
 \n         newline (hex 0A)
 \r         carriage return (hex 0D)
 \t         tab (hex 09)
 \0dd       character with octal code 0dd
 \ddd       character with octal code ddd, or backreference
 \o{ddd..}  character with octal code ddd..
 \N{U+hh..} character with Unicode code point hh.. (Unicode mode only)
 \xhh       character with hex code hh
 \x{hh..}   character with hex code hh..

If PCRE2_ALT_BSUX or PCRE2_EXTRA_ALT_BSUX is set ("ALT_BSUX mode"), the following are also recognized:

 \U         the character "U"
 \uhhhh     character with hex code hhhh
 \u{hh..}   character with hex code hh.. but only for EXTRA_ALT_BSUX

When \x is not followed by {, from zero to two hexadecimal digits are read, but in ALT_BSUX mode \x must be followed by two hexadecimal digits to be recognized as a hexadecimal escape; otherwise it matches a literal "x". Likewise, if \u (in ALT_BSUX mode) is not followed by four hexadecimal digits or (in EXTRA_ALT_BSUX mode) a sequence of hex digits in curly brackets, it matches a literal "u".

Note that \0dd is always an octal code. The treatment of backslash followed by a non-zero digit is complicated; for details see the section "Non-printing characters" in the pcre2pattern documentation, where details of escape processing in EBCDIC environments are also given. \N{U+hh..} is synonymous with \x{hh..} in PCRE2 but is not supported in EBCDIC environments. Note that \N not followed by an opening curly bracket has a different meaning (see below).

 .          any character except newline;
              in dotall mode, any character whatsoever
 \C         one code unit, even in UTF mode (best avoided)
 \d         a decimal digit
 \D         a character that is not a decimal digit
 \h         a horizontal white space character
 \H         a character that is not a horizontal white space character
 \N         a character that is not a newline
 \p{xx}     a character with the xx property
 \P{xx}     a character without the xx property
 \R         a newline sequence
 \s         a white space character
 \S         a character that is not a white space character
 \v         a vertical white space character
 \V         a character that is not a vertical white space character
 \w         a "word" character
 \W         a "non-word" character
 \X         a Unicode extended grapheme cluster

\C is dangerous because it may leave the current matching point in the middle of a UTF-8 or UTF-16 character. The application can lock out the use of \C by setting the PCRE2_NEVER_BACKSLASH_C option. It is also possible to build PCRE2 with the use of \C permanently disabled.

By default, \d, \s, and \w match only ASCII characters, even in UTF-8 mode or in the 16-bit and 32-bit libraries. However, if locale-specific matching is happening, \s and \w may also match characters with code points in the range 128-255. If the PCRE2_UCP option is set, the behaviour of these escape sequences is changed to use Unicode properties and they match many more characters, but there are some option settings that can restrict individual sequences to matching only ASCII characters.

Property descriptions in \p and \P are matched caselessly; hyphens, underscores, and white space are ignored, in accordance with Unicode's "loose matching" rules.

 C          Other
 Cc         Control
 Cf         Format
 Cn         Unassigned
 Co         Private use
 Cs         Surrogate

 L          Letter
 Ll         Lower case letter
 Lm         Modifier letter
 Lo         Other letter
 Lt         Title case letter
 Lu         Upper case letter
 Lc         Ll, Lu, or Lt
 L&         Ll, Lu, or Lt

 M          Mark
 Mc         Spacing mark
 Me         Enclosing mark
 Mn         Non-spacing mark

 N          Number
 Nd         Decimal number
 Nl         Letter number
 No         Other number

 P          Punctuation
 Pc         Connector punctuation
 Pd         Dash punctuation
 Pe         Close punctuation
 Pf         Final punctuation
 Pi         Initial punctuation
 Po         Other punctuation
 Ps         Open punctuation

 S          Symbol
 Sc         Currency symbol
 Sk         Modifier symbol
 Sm         Mathematical symbol
 So         Other symbol

 Z          Separator
 Zl         Line separator
 Zp         Paragraph separator
 Zs         Space separator

 Xan        Alphanumeric: union of properties L and N
 Xps        POSIX space: property Z or tab, NL, VT, FF, CR
 Xsp        Perl space: property Z or tab, NL, VT, FF, CR
 Xuc        Universally-named character: one that can be
              represented by a Universal Character Name
 Xwd        Perl word: property Xan or underscore

Perl and POSIX space are now the same. Perl added VT to its space character set at release 5.18.

Unicode defines a number of binary properties, that is, properties whose only values are true or false. You can obtain a list of those that are recognized by \p and \P, along with their abbreviations, by running this command:

 pcre2test -LP

Many script names and their 4-letter abbreviations are recognized in \p{sc:...} or \p{scx:...} items, or on their own with \p (and also \P of course). You can obtain a list of these scripts by running this command:

 pcre2test -LS

 \p{Bidi_Class:<class>}   matches a character with the given class
 \p{BC:<class>}           matches a character with the given class

The recognized classes are:

 AL          Arabic letter
 AN          Arabic number
 B           paragraph separator
 BN          boundary neutral
 CS          common separator
 EN          European number
 ES          European separator
 ET          European terminator
 FSI         first strong isolate
 L           left-to-right
 LRE         left-to-right embedding
 LRI         left-to-right isolate
 LRO         left-to-right override
 NSM         non-spacing mark
 ON          other neutral
 PDF         pop directional format
 PDI         pop directional isolate
 R           right-to-left
 RLE         right-to-left embedding
 RLI         right-to-left isolate
 RLO         right-to-left override
 S           segment separator
 WS          which space

 [...]       positive character class
 [^...]      negative character class
 [x-y]       range (can be used for hex characters)
 [[:xxx:]]   positive POSIX named set
 [[:^xxx:]]  negative POSIX named set

 alnum       alphanumeric
 alpha       alphabetic
 ascii       0-127
 blank       space or tab
 cntrl       control character
 digit       decimal digit
 graph       printing, excluding space
 lower       lower case letter
 print       printing, including space
 punct       printing, excluding alphanumeric
 space       white space
 upper       upper case letter
 word        same as \w
 xdigit      hexadecimal digit

In PCRE2, POSIX character set names recognize only ASCII characters by default, but some of them use Unicode properties if PCRE2_UCP is set. You can use \Q...\E inside a character class.

 ?           0 or 1, greedy
 ?+          0 or 1, possessive
 ??          0 or 1, lazy
 *           0 or more, greedy
 *+          0 or more, possessive
 *?          0 or more, lazy
 +           1 or more, greedy
 ++          1 or more, possessive
 +?          1 or more, lazy
 {n}         exactly n
 {n,m}       at least n, no more than m, greedy
 {n,m}+      at least n, no more than m, possessive
 {n,m}?      at least n, no more than m, lazy
 {n,}        n or more, greedy
 {n,}+       n or more, possessive
 {n,}?       n or more, lazy
 {,m}        zero up to m, greedy
 {,m}+       zero up to m, possessive
 {,m}?       zero up to m, lazy

 \b          word boundary
 \B          not a word boundary
 ^           start of subject
               also after an internal newline in multiline mode
               (after any newline if PCRE2_ALT_CIRCUMFLEX is set)
 \A          start of subject
 $           end of subject
               also before newline at end of subject
               also before internal newline in multiline mode
 \Z          end of subject
               also before newline at end of subject
 \z          end of subject
 \G          first matching position in subject

 \K          set reported start of match

From release 10.38 \K is not permitted by default in lookaround assertions, for compatibility with Perl. However, if the PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK option is set, the previous behaviour is re-enabled. When this option is set, \K is honoured in positive assertions, but ignored in negative ones.

 expr|expr|expr...

 (...)           capture group
 (?<name>...)    named capture group (Perl)
 (?'name'...)    named capture group (Perl)
 (?P<name>...)   named capture group (Python)
 (?:...)         non-capture group
 (?|...)         non-capture group; reset group numbers for
                  capture groups in each alternative

In non-UTF modes, names may contain underscores and ASCII letters and digits; in UTF modes, any Unicode letters and Unicode decimal digits are permitted. In both cases, a name must not start with a digit.

 (?>...)         atomic non-capture group
 (*atomic:...)   atomic non-capture group

 (?#....)        comment (not nestable)

Changes of these options within a group are automatically cancelled at the end of the group.

 (?a)            all ASCII options
 (?aD)           restrict \d to ASCII in UCP mode
 (?aS)           restrict \s to ASCII in UCP mode
 (?aW)           restrict \w to ASCII in UCP mode
 (?aP)           restrict all POSIX classes to ASCII in UCP mode
 (?aT)           restrict POSIX digit classes to ASCII in UCP mode
 (?i)            caseless
 (?J)            allow duplicate named groups
 (?m)            multiline
 (?n)            no auto capture
 (?r)            restrict caseless to either ASCII or non-ASCII
 (?s)            single line (dotall)
 (?U)            default ungreedy (lazy)
 (?x)            ignore white space except in classes or \Q...\E
 (?xx)           as (?x) but also ignore space and tab in classes
 (?-...)         unset the given option(s)
 (?^)            unset imnrsx options

(?aP) implies (?aT) as well, though this has no additional effect. However, it means that (?-aP) is really (?-PT) which disables all ASCII restrictions for POSIX classes.

Unsetting x or xx unsets both. Several options may be set at once, and a mixture of setting and unsetting such as (?i-x) is allowed, but there may be only one hyphen. Setting (but no unsetting) is allowed after (?^ for example (?^in). An option setting may appear at the start of a non-capture group, for example (?i:...).

The following are recognized only at the very start of a pattern or after one of the newline or \R options with similar syntax. More than one of them may appear. For the first three, d is a decimal number.

 (*LIMIT_DEPTH=d) set the backtracking limit to d
 (*LIMIT_HEAP=d)  set the heap size limit to d * 1024 bytes
 (*LIMIT_MATCH=d) set the match limit to d
 (*NOTEMPTY)      set PCRE2_NOTEMPTY when matching
 (*NOTEMPTY_ATSTART) set PCRE2_NOTEMPTY_ATSTART when matching
 (*NO_AUTO_POSSESS) no auto-possessification (PCRE2_NO_AUTO_POSSESS)
 (*NO_DOTSTAR_ANCHOR) no .* anchoring (PCRE2_NO_DOTSTAR_ANCHOR)
 (*NO_JIT)       disable JIT optimization
 (*NO_START_OPT) no start-match optimization (PCRE2_NO_START_OPTIMIZE)
 (*UTF)          set appropriate UTF mode for the library in use
 (*UCP)          set PCRE2_UCP (use Unicode properties for \d etc)

Note that LIMIT_DEPTH, LIMIT_HEAP, and LIMIT_MATCH can only reduce the value of the limits set by the caller of pcre2_match() or pcre2_dfa_match(), not increase them. LIMIT_RECURSION is an obsolete synonym for LIMIT_DEPTH. The application can lock out the use of (*UTF) and (*UCP) by setting the PCRE2_NEVER_UTF or PCRE2_NEVER_UCP options, respectively, at compile time.

These are recognized only at the very start of the pattern or after option settings with a similar syntax.

 (*CR)           carriage return only
 (*LF)           linefeed only
 (*CRLF)         carriage return followed by linefeed
 (*ANYCRLF)      all three of the above
 (*ANY)          any Unicode newline sequence
 (*NUL)          the NUL character (binary zero)

These are recognized only at the very start of the pattern or after option setting with a similar syntax.

 (*BSR_ANYCRLF)  CR, LF, or CRLF
 (*BSR_UNICODE)  any Unicode newline sequence

 (?=...)                     )
 (*pla:...)                  ) positive lookahead
 (*positive_lookahead:...)   )

 (?!...)                     )
 (*nla:...)                  ) negative lookahead
 (*negative_lookahead:...)   )

 (?<=...)                    )
 (*plb:...)                  ) positive lookbehind
 (*positive_lookbehind:...)  )

 (?<!...)                    )
 (*nlb:...)                  ) negative lookbehind
 (*negative_lookbehind:...)  )

Each top-level branch of a lookbehind must have a limit for the number of characters it matches. If any branch can match a variable number of characters, the maximum for each branch is limited to a value set by the caller of pcre2_compile() or defaulted. The default is set when PCRE2 is built (ultimate default 255). If every branch matches a fixed number of characters, the limit for each branch is 65535 characters.

These assertions are specific to PCRE2 and are not Perl-compatible.

 (?*...)                                )
 (*napla:...)                           ) synonyms
 (*non_atomic_positive_lookahead:...)   )

 (?<*...)                               )
 (*naplb:...)                           ) synonyms
 (*non_atomic_positive_lookbehind:...)  )

 (*script_run:...)           ) script run, can be backtracked into
 (*sr:...)                   )

 (*atomic_script_run:...)    ) atomic script run
 (*asr:...)                  )

 \n              reference by number (can be ambiguous)
 \gn             reference by number
 \g{n}           reference by number
 \g+n            relative reference by number (PCRE2 extension)
 \g-n            relative reference by number
 \g{+n}          relative reference by number (PCRE2 extension)
 \g{-n}          relative reference by number
 \k<name>        reference by name (Perl)
 \k'name'        reference by name (Perl)
 \g{name}        reference by name (Perl)
 \k{name}        reference by name (.NET)
 (?P=name)       reference by name (Python)

 (?R)            recurse whole pattern
 (?n)            call subroutine by absolute number
 (?+n)           call subroutine by relative number
 (?-n)           call subroutine by relative number
 (?&name)        call subroutine by name (Perl)
 (?P>name)       call subroutine by name (Python)
 \g<name>        call subroutine by name (Oniguruma)
 \g'name'        call subroutine by name (Oniguruma)
 \g<n>           call subroutine by absolute number (Oniguruma)
 \g'n'           call subroutine by absolute number (Oniguruma)
 \g<+n>          call subroutine by relative number (PCRE2 extension)
 \g'+n'          call subroutine by relative number (PCRE2 extension)
 \g<-n>          call subroutine by relative number (PCRE2 extension)
 \g'-n'          call subroutine by relative number (PCRE2 extension)

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

 (?(n)               absolute reference condition
 (?(+n)              relative reference condition (PCRE2 extension)
 (?(-n)              relative reference condition (PCRE2 extension)
 (?(<name>)          named reference condition (Perl)
 (?('name')          named reference condition (Perl)
 (?(name)            named reference condition (PCRE2, deprecated)
 (?(R)               overall recursion condition
 (?(Rn)              specific numbered group recursion condition
 (?(R&name)          specific named group recursion condition
 (?(DEFINE)          define groups for reference
 (?(VERSION[>]=n.m)  test PCRE2 version
 (?(assert)          assertion condition

Note the ambiguity of (?(R) and (?(Rn) which might be named reference conditions or recursion tests. Such a condition is interpreted as a reference condition if the relevant named group exists.

All backtracking control verbs may be in the form (*VERB:NAME). For (*MARK) the name is mandatory, for the others it is optional. (*SKIP) changes its behaviour if :NAME is present. The others just set a name for passing back to the caller, but this is not a name that (*SKIP) can see. The following act immediately they are reached:

 (*ACCEPT)       force successful match
 (*FAIL)         force backtrack; synonym (*F)
 (*MARK:NAME)    set name to be passed back; synonym (*:NAME)

The following act only when a subsequent match failure causes a backtrack to reach them. They all force a match failure, but they differ in what happens afterwards. Those that advance the start-of-match point do so only if the pattern is not anchored.

 (*COMMIT)       overall failure, no advance of starting point
 (*PRUNE)        advance to next starting character
 (*SKIP)         advance to current matching position
 (*SKIP:NAME)    advance to position corresponding to an earlier
                 (*MARK:NAME); if not found, the (*SKIP) is ignored
 (*THEN)         local failure, backtrack to next alternation

The effect of one of these verbs in a group called as a subroutine is confined to the subroutine call.

 (?C)            callout (assumed number 0)
 (?Cn)           callout with numerical data n
 (?C"text")      callout with string data

The allowed string delimiters are ` ' " ^ % # $ (which are the same for the start and the end), and the starting delimiter { matched with the ending delimiter }. To encode the ending delimiter within the string, double it.

pcre2pattern(3), pcre2api(3), pcre2callout(3), pcre2matching(3), pcre2(3).

Philip Hazel
Retired from University Computing Service
Cambridge, England.

Last updated: 12 October 2023
Copyright (c) 1997-2023 University of Cambridge.

ctags-optlib(7), grep(1), pcre2grep(1), pcre2pattern(3), pdfgrep(1).