CHI::Driver::File -- File-based cache using one file per entry in a multi-level directory structure
use CHI;
my $cache = CHI->new(driver => 'File', root_dir => '/path/to/cache/root', depth => 3);
This cache driver stores data on the filesystem, so that it can be shared between processes on a single machine, or even on multiple machines if using NFS.
Each item is stored in its own file. By default, during a set, a temporary file is created and then atomically renamed to the proper file. While not the most efficient, it eliminates the need for locking (with multiple overlapping sets, the last one "wins") and makes this cache usable in environments like NFS where locking might normally be undesirable.
By default, the base filename is the key itself, with unsafe characters replaced with an escape sequence similar to URI escaping. The filename length is capped at 255 characters, which is the maximum for most Unix systems, so gets/sets for keys that escape to longer than 255 characters will fail. You can also use a digest of the key (e.g. MD5, SHA) for the base filename by specifying "key_digest".
The files are evenly distributed within a multi-level directory structure with a customizable depth, to minimize the time needed to search for a given entry.
When using this driver, the following options can be passed to CHI->new() in addition to the CHI.
The location in the filesystem that will hold the root of the cache. Defaults to a directory called 'chi-driver-file' under the OS default temp directory (e.g. '/tmp' on UNIX). This directory will be created as needed on the first cache set.
Permissions mode to use when creating directories. Defaults to 0775.
Permissions mode to use when creating files, modified by the current umask. Defaults to 0666.
Extension to append to filename. Default is ".dat".
The number of subdirectories deep to place cache files. Defaults to 2. This should be large enough that no leaf directory has more than a few hundred files. Each non-leaf directory contains up to 16 subdirectories (0-9, A-F).
Digest algorithm to use on the key before storing - e.g. "MD5", "SHA-1", or "SHA-256".
Can be a Digest object, or a string or hashref which will passed to Digest->new(). You will need to ensure Digest is installed to use these options. Also, "get_keys" in CHI is currently not supported when a digest is used, this will hopefully be fixed at a later date.
By default, no digest is performed and the entire key is used in the filename, after escaping unsafe characters.
Returns the full path to the cache file representing $key, whether or not that entry exists. Returns the empty list if a valid path cannot be computed, for example if the key is too long.
Returns the full path to the directory representing this cache's namespace, whether or not it has any entries.
By default, during a set, a temporary file is created and then atomically renamed to the proper file. This eliminates the need for locking. You can subclass and override method generate_temporary_filename to either change the path of the temporary filename, or skip the temporary file and rename altogether by having it return undef.
Jonathan Swartz
Copyright (C) 2007 Jonathan Swartz.
This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
