SSH_REGEX_SYNTAX_SSH
-------------------------------------------------------------------------------

ESCAPE CHARACTER: tilde (~)

MAIN SYNTAXES AND META CHARACTERS THAT NEED TO BE ESCAPED IF MEANT LITERALLY

Here `E' (and 'F') denotes an arbitrary subexpression.

  ~ - the escape character

  ( - start a capturing subexpression

  ) - end a capturing subexpression

  { - start anonymous, non-capturing subexpression

  } - end anonymous, non-capturing subexpression

  E|F - disjunction, match (inclusively) either E or F. 
        E is preferred if both match.

  E* - Kleene star, match E zero or more times
  E*? - Kleene star, but match non-greedily (lazy match)

  E+ - closure, match E one or more times
  E+? - closure, but match non-greedily (lazy match)

  E? - option, match E optionally once
  E?? - option, but match non-greedily (lazy match)

  . - match ANY character, including possibly the NUL byte and the
      newline characters

  E/n/ - match E exactly n times

  E/n,/ or
  E/n,0/ - match E n or more times

  E/,n/ or
  E/0,n/ - match E at most n times

  E/n,m/ - match E no less than n times and no more than m times

  E/n,/?
  E/n,0/?
  E/,n/?
  E/0,n/?
  E/n,m/? - the lazy versions

  [ - start a character set, see below

  >C - one-character lookahead. `C' must be either a literal character
       or parse as a character set. Match the empty string anywhere
       provided that the next character is `C' or belongs to it.

  <C - one-character lookback. As above, but examine the previous
       character instead of the next character.

  $ - match the empty string at the end of the input

  ^ - match the empty string at the start of the input

ESCAPED TOKENS

  ~0n..n - the literal byte with octal value n..n

  ~0 - the NUL byte

  ~[1-9]..x - the literal byte with decimal value [1-9]..x

  ~xn..n or
  ~0xn..n - the literal byte with hexadecimal value n..n

  ~< - match the empty string at the beginning of a word

  ~> - match the empty string at the end of a word

  ~b - match the empty string at a word boundary

  ~B - match the empty string provided it is not at a word boundary

  ~d - match any digit, equivalent to [0:9]

  ~D - match any character except a digit

  ~s - match a whitespace character (matches space, newline, line feed,
         carriage return, tab and vertical tab).

  ~S - match a non-whitespace character

  ~w - match a word-consituent character, equivalent to [a:zA:Z0:9-]

  ~W - match a non-word-consituent character

  ~a - literal alarm character
  ~e - literal escape character 
  ~f - literal line feed
  ~n - literal new line, equivalent to C's \n so can be more than one
       character long
  ~r - literal carriage return
  ~t - literal tab

  All other escaped characters denote the literal character itself.

CHARACTER SETS

  A character set starts with [ and ends at non-escaped ] that is not
  part of a POSIX character set specifier and that does not follow
  immediately after [.

  The following characters have special meaning and need to be escaped
  if meant literally:

  : - range operator, except immediately after [ where it loses its
      special meaning

  - (minus sign) - until next +, the characters, ranges and sets will
                   be subtracted from the current set instead of
                   adding.  If appears as the first character after [,
                   start subtracting from a set containing all
                   characters instead of the empty set

  + - until next -, the characters, ranges and sets will be added to
      tue curent set. This is the default

  [:alnum:] - characters for which `isalnum' returns true (see ctype.h)
  [:alpha:] - ditto, but for `isalpha'
  [:cntrl:] - ditto, but for `iscntrl'
  [:digit:] - ditto, but for `isdigit'
  [:graph:] - ditto, but for `isgraph'
  [:lower:] - ditto, but for `islower'
  [:print:] - ditto, but for `isprint'
  [:punct:] - ditto, but for `ispunct'
  [:space:] - ditto, but for `isspace'
  [:upper:] - ditto, but for `isupper'
  [:xdigit:] - ditto, but for `isxdigit'

  Example:

    [[:xdigit:]-a:e]

  is typically equivalent to [0123456789ABCDEFf].

  It is also possible to include the predefined escaped character sets
  into a newly defined one, so

    [~d~s]

  matches digits and whitespace characters.

  Also, escape sequences resulting in literals work inside character
  sets.

SSH_REGEX_SYNTAX_EGREP
-------------------------------------------------------------------------------

ESCAPE CHARACTER: backslash (\)

MAIN SYNTAXES AND META CHARACTERS THAT NEED TO BE ESCAPED IF MEANT LITERALLY

Here `E' (and 'F') denotes an arbitrary subexpression.

  \ - the escape character

  ( - start a capturing subexpression

  ) - end a capturing subexpression

  E|F - disjunction, match (inclusively) either E or F. 
        E is preferred if both match.

  E* - Kleene star, match E zero or more times

  E+ - closure, match E one or more times

  E? - option, match E optionally once

  . - match any character except for newline characters (\n, \f, \r) and
      and the NUL byte

  E{n} - match E exactly n times

  E{n,} or
  E{n,0} - match E n or more times

  E{,n} or
  E{0,n} - match E at most n times

  E{n,m} - match E no less than n times and no more than m times

  [ - start a character set, see below

  $ - match the empty string at the end of the input

  ^ - match the empty string at the start of the input

ESCAPED TOKENS

  \0n..n - the literal byte with octal value n..n

  \0 - the NUL byte

  \[1-9]..x - the literal byte with decimal value [1-9]..x

  \xn..n or
  \0xn..n - the literal byte with hexadecimal value n..n

  \< - match the empty string at the beginning of a word

  \> - match the empty string at the end of a word

  \b - match the empty string at a word boundary

  \B - match the empty string provided it is not at a word boundary

  \w - match a word-consituent character, equivalent to [a:zA:Z0:9-]

  \W - match a non-word-consituent character

  \a - literal alarm character
  \e - literal escape character 
  \f - literal line feed
  \n - literal new line, equivalent to C's \n so can be more than one
       character long
  \r - literal carriage return
  \t - literal tab

  All other escaped characters denote the literal character itself.

CHARACTER SETS
 
  A character set starts with [ and ends at non-escaped ] that is not
  part of a POSIX character set specifier and that does not follow
  immediately after [.

  The following characters have special meaning and need to be escaped
  if meant literally:

  - (minus sign) - range operator, except immediately after [ where
		   it loses its special meaning

  ^ - immediately after the starting [, denotes complement: the whole
      character set will be complemented. Otherwise literal ^.

  [:alnum:] - characters for which `isalnum' returns true (see ctype.h)
  [:alpha:] - ditto, but for `isalpha'
  [:cntrl:] - ditto, but for `iscntrl'
  [:digit:] - ditto, but for `isdigit'
  [:graph:] - ditto, but for `isgraph'
  [:lower:] - ditto, but for `islower'
  [:print:] - ditto, but for `isprint'
  [:punct:] - ditto, but for `ispunct'
  [:space:] - ditto, but for `isspace'
  [:upper:] - ditto, but for `isupper'
  [:xdigit:] - ditto, but for `isxdigit'

  Example:

    [[:xdigit:]XY]

  is typically equivalent to [0123456789ABCDEFabcdefXY].

  It is also possible to include the predefined escaped character sets
  into a newly defined one, so

    [\d\s]

  matches digits and whitespace characters.

  Also, escape sequences resulting in literals work inside character
  sets.

SSH_REGEX_SYNTAX_ZSH_FILEGLOB
-------------------------------------------------------------------------------

ESCAPE CHARACTER: backslash (\)

MAIN SYNTAXES AND META CHARACTERS THAT NEED TO BE ESCAPED IF MEANT LITERALLY

  *   - match any string consisting of zero or more characters that can
      	be any characters except for slash (/). However, if the previous
      	character is slash (/) or `*' is tried to match at the beginning of
      	the string, do not allow dot (.).
       
      	That is, `*' functions as normal `*' in UNIX shell fileglobs.

  ?   - match any single character except for slash (/). However, do not
      	match dot (.) if at the beginning of the string or if the
      	previous character is slash (/).
      
      	That is, `?' functions as normal `?' in UNIX shell fileglobs
      	(at least ZSH. I'm not sure about discarding the dot).

  **/ - match any sequence of characters that is either empty or ends
  	in a slash, however so that the substring `/.' is not
  	allowed. This mimics ZSH's genious **/ construct. (Observe that
  	`**' is equivalent to `*'.)

  E#  - Act as Kleene star, match E zero or more times

  E## - closure, match E one or more times

  (   - start a capturing subexpression

  )   - end a capturing subexpression

  E|F - disjunction, match (inclusively) either E or F. 
        E is preferred if both match.

  [   - start a character set a'la egrep syntax except that ! is
        treated the same as ^.
