Commit | Line | Data |
---|---|---|
c07a80fd | 1 | package FileCache; |
2 | ||
414ef3ea | 3 | our $VERSION = '1.07'; |
b75c8c73 | 4 | |
c07a80fd | 5 | =head1 NAME |
6 | ||
7 | FileCache - keep more files open than the system permits | |
8 | ||
9 | =head1 SYNOPSIS | |
10 | ||
c14fc35a JH |
11 | use FileCache; |
12 | # or | |
13 | use FileCache maxopen => 16; | |
14 | ||
ba1df86b JP |
15 | cacheout $mode, $path; |
16 | # or | |
c07a80fd | 17 | cacheout $path; |
18 | print $path @data; | |
19 | ||
ba1df86b JP |
20 | $fh = cacheout $mode, $path; |
21 | # or | |
22 | $fh = cacheout $path; | |
23 | print $fh @data; | |
c14fc35a | 24 | |
c07a80fd | 25 | =head1 DESCRIPTION |
26 | ||
27 | The C<cacheout> function will make sure that there's a filehandle open | |
c14fc35a | 28 | for reading or writing available as the pathname you give it. It |
ba1df86b JP |
29 | automatically closes and re-opens files if you exceed your system's |
30 | maximum number of file descriptors, or the suggested maximum I<maxopen>. | |
c07a80fd | 31 | |
c14fc35a | 32 | =over |
7c21b9ea | 33 | |
c14fc35a | 34 | =item cacheout EXPR |
7c21b9ea | 35 | |
c14fc35a JH |
36 | The 1-argument form of cacheout will open a file for writing (C<< '>' >>) |
37 | on it's first use, and appending (C<<< '>>' >>>) thereafter. | |
38 | ||
ba1df86b JP |
39 | Returns EXPR on success for convenience. You may neglect the |
40 | return value and manipulate EXPR as the filehandle directly if you prefer. | |
41 | ||
c14fc35a JH |
42 | =item cacheout MODE, EXPR |
43 | ||
44 | The 2-argument form of cacheout will use the supplied mode for the initial | |
45 | and subsequent openings. Most valid modes for 3-argument C<open> are supported | |
46 | namely; C<< '>' >>, C<< '+>' >>, C<< '<' >>, C<< '<+' >>, C<<< '>>' >>>, | |
47 | C< '|-' > and C< '-|' > | |
48 | ||
841bcc4d RGS |
49 | To pass supplemental arguments to a program opened with C< '|-' > or C< '-|' > |
50 | append them to the command string as you would system EXPR. | |
51 | ||
ba1df86b JP |
52 | Returns EXPR on success for convenience. You may neglect the |
53 | return value and manipulate EXPR as the filehandle directly if you prefer. | |
7c21b9ea | 54 | |
00baac8f RGS |
55 | =back |
56 | ||
ba1df86b | 57 | =head1 CAVEATS |
7c21b9ea | 58 | |
dfe3554a JP |
59 | While it is permissible to C<close> a FileCache managed file, |
60 | do not do so if you are calling C<FileCache::cacheout> from a package other | |
61 | than which it was imported, or with another module which overrides C<close>. | |
62 | If you must, use C<FileCache::cacheout_close>. | |
63 | ||
533968fe AB |
64 | Although FileCache can be used with piped opens ('-|' or '|-') doing so is |
65 | strongly discouraged. If FileCache finds it necessary to close and then reopen | |
66 | a pipe, the command at the far end of the pipe will be reexecuted - the results | |
67 | of performing IO on FileCache'd pipes is unlikely to be what you expect. The | |
68 | ability to use FileCache on pipes may be removed in a future release. | |
69 | ||
70 | FileCache does not store the current file offset if it finds it necessary to | |
71 | close a file. When the file is reopened, the offset will be as specified by the | |
72 | original C<open> file mode. This could be construed to be a bug. | |
73 | ||
c07a80fd | 74 | =head1 BUGS |
75 | ||
76 | F<sys/param.h> lies with its C<NOFILE> define on some systems, | |
ba1df86b JP |
77 | so you may have to set I<maxopen> yourself. |
78 | ||
c07a80fd | 79 | =cut |
80 | ||
dfe3554a | 81 | require 5.006; |
c07a80fd | 82 | use Carp; |
7c21b9ea | 83 | use strict; |
c14fc35a | 84 | no strict 'refs'; |
c9463f45 | 85 | |
c14fc35a JH |
86 | # These are not C<my> for legacy reasons. |
87 | # Previous versions requested the user set $cacheout_maxopen by hand. | |
88 | # Some authors fiddled with %saw to overcome the clobber on initial open. | |
c9463f45 MS |
89 | use vars qw(%saw $cacheout_maxopen); |
90 | $cacheout_maxopen = 16; | |
91 | ||
92 | use base 'Exporter'; | |
93 | our @EXPORT = qw[cacheout cacheout_close]; | |
94 | ||
95 | ||
7c21b9ea JP |
96 | my %isopen; |
97 | my $cacheout_seq = 0; | |
98 | ||
c14fc35a JH |
99 | sub import { |
100 | my ($pkg,%args) = @_; | |
02c473a9 | 101 | |
02c473a9 JB |
102 | # Use Exporter. %args are for us, not Exporter. |
103 | # Make sure to up export_to_level, or we will import into ourselves, | |
104 | # rather than our calling package; | |
02c473a9 JB |
105 | |
106 | __PACKAGE__->export_to_level(1); | |
107 | Exporter::import( $pkg ); | |
ba1df86b | 108 | |
c14fc35a | 109 | # Truth is okay here because setting maxopen to 0 would be bad |
ba1df86b | 110 | return $cacheout_maxopen = $args{maxopen} if $args{maxopen}; |
c9463f45 MS |
111 | |
112 | # XXX This code is crazy. Why is it a one element foreach loop? | |
113 | # Why is it using $param both as a filename and filehandle? | |
ba1df86b JP |
114 | foreach my $param ( '/usr/include/sys/param.h' ){ |
115 | if (open($param, '<', $param)) { | |
116 | local ($_, $.); | |
117 | while (<$param>) { | |
118 | if( /^\s*#\s*define\s+NOFILE\s+(\d+)/ ){ | |
119 | $cacheout_maxopen = $1 - 4; | |
120 | close($param); | |
121 | last; | |
122 | } | |
123 | } | |
124 | close $param; | |
c14fc35a | 125 | } |
c14fc35a JH |
126 | } |
127 | $cacheout_maxopen ||= 16; | |
128 | } | |
129 | ||
c07a80fd | 130 | # Open in their package. |
c07a80fd | 131 | sub cacheout_open { |
ba1df86b | 132 | return open(*{caller(1) . '::' . $_[1]}, $_[0], $_[1]) && $_[1]; |
c07a80fd | 133 | } |
134 | ||
c14fc35a | 135 | # Close in their package. |
c07a80fd | 136 | sub cacheout_close { |
ba1df86b JP |
137 | # Short-circuit in case the filehandle disappeared |
138 | my $pkg = caller($_[1]||0); | |
414ef3ea | 139 | defined fileno(*{$pkg . '::' . $_[0]}) && |
ba1df86b JP |
140 | CORE::close(*{$pkg . '::' . $_[0]}); |
141 | delete $isopen{$_[0]}; | |
c07a80fd | 142 | } |
143 | ||
144 | # But only this sub name is visible to them. | |
c07a80fd | 145 | sub cacheout { |
ba1df86b JP |
146 | my($mode, $file, $class, $ret, $ref, $narg); |
147 | croak "Not enough arguments for cacheout" unless $narg = scalar @_; | |
148 | croak "Too many arguments for cacheout" if $narg > 2; | |
c14fc35a | 149 | |
ba1df86b JP |
150 | ($mode, $file) = @_; |
151 | ($file, $mode) = ($mode, $file) if $narg == 1; | |
152 | croak "Invalid mode for cacheout" if $mode && | |
153 | ( $mode !~ /^\s*(?:>>|\+?>|\+?<|\|\-|)|\-\|\s*$/ ); | |
841bcc4d | 154 | |
ba1df86b | 155 | # Mode changed? |
8ac28360 | 156 | if( $isopen{$file} && ($mode||'>') ne $isopen{$file}->[1] ){ |
ba1df86b JP |
157 | &cacheout_close($file, 1); |
158 | } | |
c9463f45 | 159 | |
ba1df86b JP |
160 | if( $isopen{$file}) { |
161 | $ret = $file; | |
162 | $isopen{$file}->[0]++; | |
163 | } | |
164 | else{ | |
c14fc35a | 165 | if( scalar keys(%isopen) > $cacheout_maxopen -1 ) { |
ba1df86b JP |
166 | my @lru = sort{ $isopen{$a}->[0] <=> $isopen{$b}->[0] } keys(%isopen); |
167 | $cacheout_seq = 0; | |
168 | $isopen{$_}->[0] = $cacheout_seq++ for | |
169 | splice(@lru, int($cacheout_maxopen / 3)||$cacheout_maxopen); | |
170 | &cacheout_close($_, 1) for @lru; | |
c14fc35a | 171 | } |
ba1df86b JP |
172 | |
173 | unless( $ref ){ | |
174 | $mode ||= $saw{$file} ? '>>' : ($saw{$file}=1, '>'); | |
175 | } | |
176 | #XXX should we just return the value from cacheout_open, no croak? | |
177 | $ret = cacheout_open($mode, $file) or croak("Can't create $file: $!"); | |
c9463f45 | 178 | |
ba1df86b | 179 | $isopen{$file} = [++$cacheout_seq, $mode]; |
c07a80fd | 180 | } |
ba1df86b | 181 | return $ret; |
c07a80fd | 182 | } |
c07a80fd | 183 | 1; |