Commit | Line | Data |
c07a80fd |
1 | package FileCache; |
2 | |
ba1df86b |
3 | our $VERSION = 1.03; |
b75c8c73 |
4 | |
c07a80fd |
5 | =head1 NAME |
6 | |
7 | FileCache - keep more files open than the system permits |
8 | |
9 | =head1 SYNOPSIS |
10 | |
c14fc35a |
11 | use FileCache; |
12 | # or |
13 | use FileCache maxopen => 16; |
14 | |
ba1df86b |
15 | cacheout $mode, $path; |
16 | # or |
c07a80fd |
17 | cacheout $path; |
18 | print $path @data; |
19 | |
ba1df86b |
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 |
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 |
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 |
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 |
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 | |
ba1df86b |
49 | Returns EXPR on success for convenience. You may neglect the |
50 | return value and manipulate EXPR as the filehandle directly if you prefer. |
7c21b9ea |
51 | |
ba1df86b |
52 | =head1 CAVEATS |
7c21b9ea |
53 | |
dfe3554a |
54 | While it is permissible to C<close> a FileCache managed file, |
55 | do not do so if you are calling C<FileCache::cacheout> from a package other |
56 | than which it was imported, or with another module which overrides C<close>. |
57 | If you must, use C<FileCache::cacheout_close>. |
58 | |
c07a80fd |
59 | =head1 BUGS |
60 | |
61 | F<sys/param.h> lies with its C<NOFILE> define on some systems, |
ba1df86b |
62 | so you may have to set I<maxopen> yourself. |
63 | |
64 | =head1 NOTES |
65 | |
66 | FileCache installs signal handlers for CHLD (a.k.a. CLD) and PIPE in the |
67 | calling package to handle deceased children from 2-arg C<cacheout> with C<'|-'> |
68 | or C<'-|'> I<expediently>. The children would otherwise be reaped eventually, |
69 | unless you terminated before repeatedly calling cacheout. |
c07a80fd |
70 | |
71 | =cut |
72 | |
dfe3554a |
73 | require 5.006; |
c07a80fd |
74 | use Carp; |
42bff5bd |
75 | use Config; |
7c21b9ea |
76 | use strict; |
c14fc35a |
77 | no strict 'refs'; |
c14fc35a |
78 | # These are not C<my> for legacy reasons. |
79 | # Previous versions requested the user set $cacheout_maxopen by hand. |
80 | # Some authors fiddled with %saw to overcome the clobber on initial open. |
ba1df86b |
81 | use vars qw(%saw $cacheout_maxopen); |
7c21b9ea |
82 | my %isopen; |
83 | my $cacheout_seq = 0; |
84 | |
c14fc35a |
85 | sub import { |
86 | my ($pkg,%args) = @_; |
ba1df86b |
87 | $pkg = caller(1); |
88 | *{$pkg.'::cacheout'} = \&cacheout; |
89 | *{$pkg.'::close'} = \&cacheout_close; |
90 | |
91 | # Reap our children |
42bff5bd |
92 | ${"$pkg\::SIG"}{'CLD'} = 'IGNORE' if $Config{sig_name} =~ /\bCLD\b/; |
93 | ${"$pkg\::SIG"}{'CHLD'} = 'IGNORE' if $Config{sig_name} =~ /\bCHLD\b/; |
94 | ${"$pkg\::SIG"}{'PIPE'} = 'IGNORE' if $Config{sig_name} =~ /\bPIPE\b/; |
c14fc35a |
95 | |
96 | # Truth is okay here because setting maxopen to 0 would be bad |
ba1df86b |
97 | return $cacheout_maxopen = $args{maxopen} if $args{maxopen}; |
98 | foreach my $param ( '/usr/include/sys/param.h' ){ |
99 | if (open($param, '<', $param)) { |
100 | local ($_, $.); |
101 | while (<$param>) { |
102 | if( /^\s*#\s*define\s+NOFILE\s+(\d+)/ ){ |
103 | $cacheout_maxopen = $1 - 4; |
104 | close($param); |
105 | last; |
106 | } |
107 | } |
108 | close $param; |
c14fc35a |
109 | } |
c14fc35a |
110 | } |
111 | $cacheout_maxopen ||= 16; |
112 | } |
113 | |
c07a80fd |
114 | # Open in their package. |
c07a80fd |
115 | sub cacheout_open { |
ba1df86b |
116 | return open(*{caller(1) . '::' . $_[1]}, $_[0], $_[1]) && $_[1]; |
c07a80fd |
117 | } |
118 | |
c14fc35a |
119 | # Close in their package. |
c07a80fd |
120 | sub cacheout_close { |
ba1df86b |
121 | # Short-circuit in case the filehandle disappeared |
122 | my $pkg = caller($_[1]||0); |
123 | fileno(*{$pkg . '::' . $_[0]}) && |
124 | CORE::close(*{$pkg . '::' . $_[0]}); |
125 | delete $isopen{$_[0]}; |
c07a80fd |
126 | } |
127 | |
128 | # But only this sub name is visible to them. |
c07a80fd |
129 | sub cacheout { |
ba1df86b |
130 | my($mode, $file, $class, $ret, $ref, $narg); |
131 | croak "Not enough arguments for cacheout" unless $narg = scalar @_; |
132 | croak "Too many arguments for cacheout" if $narg > 2; |
c14fc35a |
133 | |
ba1df86b |
134 | ($mode, $file) = @_; |
135 | ($file, $mode) = ($mode, $file) if $narg == 1; |
136 | croak "Invalid mode for cacheout" if $mode && |
137 | ( $mode !~ /^\s*(?:>>|\+?>|\+?<|\|\-|)|\-\|\s*$/ ); |
138 | |
139 | # Mode changed? |
140 | if( $isopen{$file} && ($mode||'>') ne $isopen{$file}->[2] ){ |
141 | &cacheout_close($file, 1); |
142 | } |
143 | |
144 | if( $isopen{$file}) { |
145 | $ret = $file; |
146 | $isopen{$file}->[0]++; |
147 | } |
148 | else{ |
c14fc35a |
149 | if( scalar keys(%isopen) > $cacheout_maxopen -1 ) { |
ba1df86b |
150 | my @lru = sort{ $isopen{$a}->[0] <=> $isopen{$b}->[0] } keys(%isopen); |
151 | $cacheout_seq = 0; |
152 | $isopen{$_}->[0] = $cacheout_seq++ for |
153 | splice(@lru, int($cacheout_maxopen / 3)||$cacheout_maxopen); |
154 | &cacheout_close($_, 1) for @lru; |
c14fc35a |
155 | } |
ba1df86b |
156 | |
157 | unless( $ref ){ |
158 | $mode ||= $saw{$file} ? '>>' : ($saw{$file}=1, '>'); |
159 | } |
160 | #XXX should we just return the value from cacheout_open, no croak? |
161 | $ret = cacheout_open($mode, $file) or croak("Can't create $file: $!"); |
162 | |
163 | $isopen{$file} = [++$cacheout_seq, $mode]; |
c07a80fd |
164 | } |
ba1df86b |
165 | return $ret; |
c07a80fd |
166 | } |
c07a80fd |
167 | 1; |