Commit | Line | Data |
---|---|---|
c07a80fd | 1 | package FileCache; |
2 | ||
841bcc4d | 3 | our $VERSION = 1.04; |
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 | |
ba1df86b | 55 | =head1 CAVEATS |
7c21b9ea | 56 | |
dfe3554a JP |
57 | While it is permissible to C<close> a FileCache managed file, |
58 | do not do so if you are calling C<FileCache::cacheout> from a package other | |
59 | than which it was imported, or with another module which overrides C<close>. | |
60 | If you must, use C<FileCache::cacheout_close>. | |
61 | ||
c07a80fd | 62 | =head1 BUGS |
63 | ||
64 | F<sys/param.h> lies with its C<NOFILE> define on some systems, | |
ba1df86b JP |
65 | so you may have to set I<maxopen> yourself. |
66 | ||
67 | =head1 NOTES | |
68 | ||
841bcc4d RGS |
69 | FileCache installs localized signal handlers for CHLD (a.k.a. CLD) and PIPE |
70 | to handle deceased children from 2-arg C<cacheout> with C<'|-'> or C<'-|'> | |
71 | I<expediently>. The children would otherwise be reaped eventually, unless you | |
72 | terminated before repeatedly calling cacheout. | |
c07a80fd | 73 | |
74 | =cut | |
75 | ||
dfe3554a | 76 | require 5.006; |
c07a80fd | 77 | use Carp; |
42bff5bd | 78 | use Config; |
7c21b9ea | 79 | use strict; |
c14fc35a | 80 | no strict 'refs'; |
c14fc35a JH |
81 | # These are not C<my> for legacy reasons. |
82 | # Previous versions requested the user set $cacheout_maxopen by hand. | |
83 | # Some authors fiddled with %saw to overcome the clobber on initial open. | |
ba1df86b | 84 | use vars qw(%saw $cacheout_maxopen); |
7c21b9ea JP |
85 | my %isopen; |
86 | my $cacheout_seq = 0; | |
87 | ||
c14fc35a JH |
88 | sub import { |
89 | my ($pkg,%args) = @_; | |
ba1df86b JP |
90 | $pkg = caller(1); |
91 | *{$pkg.'::cacheout'} = \&cacheout; | |
92 | *{$pkg.'::close'} = \&cacheout_close; | |
93 | ||
c14fc35a | 94 | # Truth is okay here because setting maxopen to 0 would be bad |
ba1df86b JP |
95 | return $cacheout_maxopen = $args{maxopen} if $args{maxopen}; |
96 | foreach my $param ( '/usr/include/sys/param.h' ){ | |
97 | if (open($param, '<', $param)) { | |
98 | local ($_, $.); | |
99 | while (<$param>) { | |
100 | if( /^\s*#\s*define\s+NOFILE\s+(\d+)/ ){ | |
101 | $cacheout_maxopen = $1 - 4; | |
102 | close($param); | |
103 | last; | |
104 | } | |
105 | } | |
106 | close $param; | |
c14fc35a | 107 | } |
c14fc35a JH |
108 | } |
109 | $cacheout_maxopen ||= 16; | |
110 | } | |
111 | ||
c07a80fd | 112 | # Open in their package. |
c07a80fd | 113 | sub cacheout_open { |
841bcc4d RGS |
114 | # Reap our children |
115 | local $SIG{CLD} ||= 'IGNORE'if $Config{sig_name} =~ /\bCLD\b/; | |
116 | local $SIG{CHLD} ||= 'IGNORE'if $Config{sig_name} =~ /\bCHLD\b/; | |
117 | local $SIG{PIPE} ||= 'IGNORE'if $Config{sig_name} =~ /\bPIPE\b/; | |
118 | ||
ba1df86b | 119 | return open(*{caller(1) . '::' . $_[1]}, $_[0], $_[1]) && $_[1]; |
c07a80fd | 120 | } |
121 | ||
c14fc35a | 122 | # Close in their package. |
c07a80fd | 123 | sub cacheout_close { |
ba1df86b JP |
124 | # Short-circuit in case the filehandle disappeared |
125 | my $pkg = caller($_[1]||0); | |
126 | fileno(*{$pkg . '::' . $_[0]}) && | |
127 | CORE::close(*{$pkg . '::' . $_[0]}); | |
128 | delete $isopen{$_[0]}; | |
c07a80fd | 129 | } |
130 | ||
131 | # But only this sub name is visible to them. | |
c07a80fd | 132 | sub cacheout { |
ba1df86b JP |
133 | my($mode, $file, $class, $ret, $ref, $narg); |
134 | croak "Not enough arguments for cacheout" unless $narg = scalar @_; | |
135 | croak "Too many arguments for cacheout" if $narg > 2; | |
c14fc35a | 136 | |
ba1df86b JP |
137 | ($mode, $file) = @_; |
138 | ($file, $mode) = ($mode, $file) if $narg == 1; | |
139 | croak "Invalid mode for cacheout" if $mode && | |
140 | ( $mode !~ /^\s*(?:>>|\+?>|\+?<|\|\-|)|\-\|\s*$/ ); | |
841bcc4d | 141 | |
ba1df86b JP |
142 | # Mode changed? |
143 | if( $isopen{$file} && ($mode||'>') ne $isopen{$file}->[2] ){ | |
144 | &cacheout_close($file, 1); | |
145 | } | |
146 | ||
147 | if( $isopen{$file}) { | |
148 | $ret = $file; | |
149 | $isopen{$file}->[0]++; | |
150 | } | |
151 | else{ | |
c14fc35a | 152 | if( scalar keys(%isopen) > $cacheout_maxopen -1 ) { |
ba1df86b JP |
153 | my @lru = sort{ $isopen{$a}->[0] <=> $isopen{$b}->[0] } keys(%isopen); |
154 | $cacheout_seq = 0; | |
155 | $isopen{$_}->[0] = $cacheout_seq++ for | |
156 | splice(@lru, int($cacheout_maxopen / 3)||$cacheout_maxopen); | |
157 | &cacheout_close($_, 1) for @lru; | |
c14fc35a | 158 | } |
ba1df86b JP |
159 | |
160 | unless( $ref ){ | |
161 | $mode ||= $saw{$file} ? '>>' : ($saw{$file}=1, '>'); | |
162 | } | |
163 | #XXX should we just return the value from cacheout_open, no croak? | |
164 | $ret = cacheout_open($mode, $file) or croak("Can't create $file: $!"); | |
165 | ||
166 | $isopen{$file} = [++$cacheout_seq, $mode]; | |
c07a80fd | 167 | } |
ba1df86b | 168 | return $ret; |
c07a80fd | 169 | } |
c07a80fd | 170 | 1; |