The R Project SVN R

Rev

Rev 85908 | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 85908 Rev 90027
Line 15... Line 15...
15 
}
 15 
}
16 
\arguments{
 16 
\arguments{
17 
  \item{paths}{character vector of patterns for relative or absolute
 17 
  \item{paths}{character vector of patterns for relative or absolute
18 
    filepaths.  Missing values will be ignored.}
 18 
    filepaths.  Missing values will be ignored.}
19 
  \item{dirmark}{logical: should matches to directories from patterns
 19 
  \item{dirmark}{logical: should matches to directories from patterns
20 
    that do not already end in \code{/}
 20 
    that do not already end in \samp{/}
21 
#ifdef windows
 21 
#ifdef windows
22 
    or \code{\\}
 22 
    or \samp{\\}
23 
#endif
 23 
#endif
24 
    have a slash appended?  May not be supported on all platforms.}
 24 
    have a slash appended?  May not be supported on all platforms.}
25 
}
 25 
}
26 
\details{
 26 
\details{
27 
  This expands tilde (see \link{tilde expansion}) and wildcards in file paths.
27 
  This expands tilde (see \link{tilde expansion}) and wildcards in file paths.
Line 30... Line 30...
30 
  system's documentation on the \code{glob} system call.  There is a
 30 
  system's documentation on the \code{glob} system call.  There is a
31 
  POSIX 1003.2 standard (see
 31 
  POSIX 1003.2 standard (see
32 
  \url{https://pubs.opengroup.org/onlinepubs/9699919799/functions/glob.html})
 32 
  \url{https://pubs.opengroup.org/onlinepubs/9699919799/functions/glob.html})
33 
  but some OSes will go beyond this.
 33 
  but some OSes will go beyond this.
34 
 
 34 
 
35 
  All systems should interpret \code{*} (match zero or more characters),
 35 
  All systems should interpret \samp{*} (match zero or more characters),
36 
  \code{?} (match a single character) and (probably) \code{[} (begin a
 36 
  \samp{?} (match a single character) and (probably) \samp{[} (begin a
37 
  character class or range).  The handling of paths
 37 
  character class or range).  The handling of paths
38 
  ending with a separator is system-dependent.  On a POSIX-2008
 38 
  ending with a separator is system-dependent.  On a POSIX-2008
39 
  compliant OS they will match directories (only), but as they are not
 39 
  compliant OS they will match directories (only), but as they are not
40 
  valid filepaths on Windows, they match nothing there.  (Earlier POSIX
 40 
  valid filepaths on Windows, they match nothing there.  (Earlier POSIX
41 
  standards allowed them to match files.)
 41 
  standards allowed them to match files.)
42 
  % section A.4.12, https://pubs.opengroup.org/onlinepubs/9699919799/xrat/V4_xbd_chap04.html
 42 
  % section A.4.12, https://pubs.opengroup.org/onlinepubs/9699919799/xrat/V4_xbd_chap04.html
43 
 
 43 
 
44 
  The rest of these details are indicative (and based on the POSIX
 44 
  The rest of these details are indicative (and based on the POSIX
45 
  standard).
 45 
  standard).
46 
 
 46 
 
47 
  If a filename starts with \code{.} this may need to be matched
 47 
  If a filename starts with \samp{.} this may need to be matched
48 
  explicitly: for example \code{Sys.glob("*.RData")} may or may not
 48 
  explicitly: for example \code{Sys.glob("*.RData")} may or may not
49 
  match \file{.RData} but will not usually match \file{.aa.RData}.  Note
 49 
  match \file{.RData} but will not usually match \file{.aa.RData}.  Note
50 
  that this is platform-dependent: e.g.\sspace{}on Solaris
 50 
  that this is platform-dependent: e.g.\sspace{}on Solaris
51 
  \code{Sys.glob("*.*")} matches \file{.} and \file{..}.
 51 
  \code{Sys.glob("*.*")} matches \file{.} and \file{..}.
52 
 
 52 
 
53 
  \code{[} begins a character class.  If the first character in
 53 
  \samp{[} begins a character class.  If the first character in
54 
  \code{[...]} is not \code{!}, this is a character class which matches
 54 
  \samp{[...]} is not \samp{!}, this is a character class which matches
55 
  a single character against any of the characters specified.  The class
 55 
  a single character against any of the characters specified.  The class
56 
  cannot be empty, so \code{]} can be included provided it is first.  If
 56 
  cannot be empty, so \samp{]} can be included provided it is first.  If
57 
  the first character is \code{!}, the character class matches a single
 57 
  the first character is \samp{!}, the character class matches a single
58 
  character which is \emph{none} of the specified characters.  Whether
 58 
  character which is \emph{none} of the specified characters.  Whether
59 
  \code{.} in a character class matches a leading \code{.} in the
 59 
  \samp{.} in a character class matches a leading \samp{.} in the
60 
  filename is OS-dependent.
 60 
  filename is OS-dependent.
61 
 
 61 
 
62 
  Character classes can include ranges such as \code{[A-Z]}: include
 62 
  Character classes can include ranges such as \samp{[A-Z]}: include
63 
  \code{-} as a character by having it first or last in a class.  (The
 63 
  \samp{-} as a character by having it first or last in a class.  (The
64 
  interpretation of ranges should be locale-specific, so the example is
 64 
  interpretation of ranges should be locale-specific, so the example is
65 
  not a good idea in an Estonian locale.)
 65 
  not a good idea in an Estonian locale.)
66 
 
 66 
 
67 
  One can remove the special meaning of \code{?}, \code{*}  and
 67 
  One can remove the special meaning of \samp{?}, \samp{*}  and
68 
  \code{[} by preceding them by a backslash (except within a
 68 
  \samp{[} by preceding them by a backslash (except within a
69 
  character class).
 69 
  character class).
70 
#endif
 70 
#endif
71 
#ifdef windows
 71 
#ifdef windows
72 
  The \code{glob} system call is not part of Windows, and we supply a
 72 
  The \code{glob} system call is not part of Windows, and we supply a
73 
  partial emulation for wildcards expansion.
 73 
  partial emulation for wildcards expansion.
74 
 
 74 
 
75 
  Wildcards are \code{*} (match zero or more characters) and \code{?} (match
 75 
  Wildcards are \samp{*} (match zero or more characters) and \samp{?} (match
76 
  a single character).  If a filename starts with \code{.} this must be
 76 
  a single character).  If a filename starts with \samp{.} this must be
77 
  matched explicitly (on Windows, but note that this is platform-dependent).
 77 
  matched explicitly (on Windows, but note that this is platform-dependent).
78 
 
 78 
 
79 
  \code{[} begins a character class.  If the first character in
 79 
  \samp{[} begins a character class.  If the first character in
80 
  \code{[...]} is not \code{!}, this is a character class which matches
 80 
  \samp{[...]} is not \samp{!}, this is a character class which matches
81 
  a single character against any of the characters specified.  The class
 81 
  a single character against any of the characters specified.  The class
82 
  cannot be empty, so \code{]} can be included provided it is first.  If
 82 
  cannot be empty, so \samp{]} can be included provided it is first.  If
83 
  the first character is \code{!}, the character class matches a single
 83 
  the first character is \samp{!}, the character class matches a single
84 
  character which is \emph{none} of the specified characters.  Whether
 84 
  character which is \emph{none} of the specified characters.  Whether
85 
  \code{.} in a character class matches a leading \code{.} in the
 85 
  \samp{.} in a character class matches a leading \samp{.} in the
86 
  filename is OS-dependent.
 86 
  filename is OS-dependent.
87 
 
 87 
 
88 
  Character classes can include ranges such as \code{[A-Z]}: include
 88 
  Character classes can include ranges such as \samp{[A-Z]}: include
89 
  \code{-} as a character by having it first or last in a class.  (In
 89 
  \samp{-} as a character by having it first or last in a class.  (In
90 
  the current implementation ranges are in the numeric order of Unicode
 90 
  the current implementation ranges are in the numeric order of Unicode
91 
  code points.)
 91 
  code points.)
92 
 
 92 
 
93 
  One can remove the special meaning of \code{?}, \code{*}  and
 93 
  One can remove the special meaning of \samp{?}, \samp{*}  and
94 
  \code{[} by preceding them by a backslash (except within a
 94 
  \samp{[} by preceding them by a backslash (except within a
95 
  character class).  Note that on Windows \code{?} and \code{*} are not
 95 
  character class).  Note that on Windows \samp{?} and \samp{*} are not
96 
  valid in file names, so this is mainly for consistency with other
 96 
  valid in file names, so this is mainly for consistency with other
97 
  platforms.
 97 
  platforms.
98 
 
 98 
 
99 
  File paths in Windows are interpreted with separator \code{\\} or
 99 
  File paths in Windows are interpreted with separator \samp{\\} or
100 
  \code{/}.  Paths with a drive but relative (such as \code{c:foo\\bar})
 100 
  \samp{/}.  Paths with a drive but relative (such as \samp{c:foo\\bar})
101 
  are tricky, but an attempt is made to handle them correctly.  An
 101 
  are tricky, but an attempt is made to handle them correctly.  An
102 
  attempt is made to handle \abbr{UNC} paths starting with a double
 102 
  attempt is made to handle \abbr{UNC} paths starting with a double
103 
  backslash. UTF-8-encoded paths not valid in the current locale can be
 103 
  backslash. UTF-8-encoded paths not valid in the current locale can be
104 
  used.
 104 
  used.
105 
#endif
 105 
#endif