Scroll to navigation

Config::Std(3pm) User Contributed Perl Documentation Config::Std(3pm)

NAME

Config::Std - Load and save configuration files in a standard format

VERSION

This document describes Config::Std version 0.903

SYNOPSIS

    use Config::Std;
    # Load named config file into specified hash...
    read_config 'demo2.cfg' => my %config;
    # Extract the value of a key/value pair from a specified section...
    $config_value = $config{Section_label}{key};
    # Change (or create) the value of a key/value pair...
    $config{Other_section_label}{other_key} = $new_val;
    # Update the config file from which this hash was loaded...
    write_config %config;
    # Write the config information to another file as well...
    write_config %config, $other_file_name;

DESCRIPTION

This module implements yet another damn configuration-file system.

The configuration language is deliberately simple and limited, and the module works hard to preserve as much information (section order, comments, etc.) as possible when a configuration file is updated.

The whole point of Config::Std is to encourage use of one standard layout and syntax in config files. Damian says "I could have gotten away with it, I would have only allowed one separator. But it proved impossible to choose between ":" and "=" (half the people I asked wanted one, half wanted the other)." Providing round-trip file re-write is the spoonful of sugar to help the medicine go down. The supported syntax is within the general INI file family

See Chapter 19 of "Perl Best Practices" (O'Reilly, 2005) for more detail on the rationale for this approach.

Configuration language

The configuration language is a slight extension of the Windows INI format.

Comments

A comment starts with a "#" character (Perl-style) or a ";" character (INI-style), and runs to the end of the same line:

    # This is a comment
    ; Ywis, eke hight thilke

Comments can be placed almost anywhere in a configuration file, except inside a section label, or in the key or value of a configuration variable:

    # Valid comment
    [ # Not a comment, just a weird section label ]
    ; Valid comment
    key: value  ; Not a comment, just part of the value

NOTE BENE -- that last is a BAD EXAMPLE of what is NOT supported. This module supports full-line comments only, not on same line with semantic content.

Sections

A configuration file consists of one or more sections, each of which is introduced by a label in square brackets:

    [SECTION1]        # Almost anything is a valid section label
    [SECTION 2]       # Internal whitespace is allowed (except newlines)
    [%^$%^&!!!]       # The label doesn't have to be alphanumeric
    [ETC. ETC. AS MANY AS YOU WANT]

The only restriction on section labels is that they must be by themselves on a single line (except for any surrounding whitespace or trailing comments), and they cannot contain the character "]".

Every line after a given section label until the next section label (or the end of the config file) belongs to the given section label. If no section label is currently in effect, the current section has an empty label. In other words, there is an implicit:

    []                # Label is the empty string

at the start of each config file.

Configuration variables

Each non-empty line within a section must consist of the specification of a configuration variable. Each such variable consists of a key and a string value. For example:

    name: George
     age: 47
    his weight! : 185

The key consists of every character (including internal whitespace) from the start of the line until the key/value separator. So, the previous example declares three keys: 'name', 'age', and 'his weight!'.

Note that whitespace before and after the key is removed. This makes it easier to format keys cleanly:

           name : George
            age : 47
    his weight! : 185

The key/value separator can be either a colon (as above) or an equals sign, like so:

           name= George
            age=  47
    his weight! = 185

Both types of separators can be used in the same file, but neither can be used as part of a key. Newlines are not allowed in keys either.

When writing out a config file, Config::Std tries to preserve whichever separator was used in the original data (if that data was read in). New data (created by code not parsed by "read_config") is written back with a colon as its default separator, unless you specify the only other separator value '=' when the module is loaded:

    use Config::Std { def_sep => '=' };

Note that this does not change read-in parsing, does not change punctuation for values that were parsed, and will not allow values other than '=' or ':'.

Everything from the first non-whitespace character after the separator, up to the end of the line, is treated as the value for the config variable. So all of the above examples define the same three values: 'George', '47', and '185'.

In other words, any whitespace immediately surrounding the separator character is part of the separator, not part of the key or value.

Note that you can't put a comment on the same line as a configuration variable. The "# etc." is simply considered part of the value:

    [Delimiters]
    block delims:    { }
    string delims:   " "
    comment delims:  # \n

You can comment a config var on the preceding or succeeding line:

    [Delimiters]
    # Use braces to delimit blocks...
    block delims:    { }
    # Use double quotes to delimit strings
    string delims:   " "
    # Use octothorpe/newline to delimit comments
    comment delims:  # \n

Multi-line configuration values

A single value can be continued over two or more lines. If the line immediately after a configuration variable starts with the separator character used in the variable's definition, then the value of the variable continues on that line. For example:

    address: 742 Evergreen Terrace
           : Springfield
           : USA

The newlines then form part of the value, so the value specified in the previous example is: "742 Evergreen Terrace\nSpringfield\nUSA"

Note that the second and subsequent lines of a continued value are considered to start where the whitespace after the original separator finished, not where the whitespace after their own separator finishes. For example, if the previous example had been:

    address: 742 Evergreen Terrace
           :   Springfield
           :     USA

then the value would be:

    "742 Evergreen Terrace\n  Springfield\n    USA"

If a continuation line has less leading whitespace that the first line:

    address:   742 Evergreen Terrace
           :  Springfield
           : USA

it's treated as having no leading whitespace:

    "742 Evergreen Terrace\nSpringfield\nUSA"

Multi-part configuration values

If the particular key appears more than once in the same section, it is considered to be part of the same configuration variable. The value of that configuration value is then a list, containing all the individual values for each instance of the key. For example, given the definition:

    cast: Homer
    cast: Marge
    cast: Lisa
    cast: Bart
    cast: Maggie

the corresponding value of the 'cast' configuration variable is: "['Homer', 'Marge', 'Lisa', 'Bart', 'Maggie']"

Individual values in a multi-part list can also be multi-line (see above). For example, given:

    extras: Moe
          : (the bartender)
    extras: Smithers
          : (the dogsbody)

the value for the 'extras' config variable is: "["Moe\n(the bartender)", "Smithers\n(the dogsbody)"]"

Internal representation

Each section label in a configuration file becomes a top-level hash key when the configuration file is read in. The corresponding value is a nested hash reference.

Each configuration variable's key becomes a key in that nested hash reference. Each configuration variable's value becomes the corresponding value in that nested hash reference.

Single-line and multi-line values become strings. Multi-part values become references to arrays of strings.

For example, the following configuration file:

    # A simple key (just an identifier)...
    simple : simple value
    # A more complex key (with whitespace)...
    more complex key : more complex value
    # A new section...
    [MULTI-WHATEVERS]
    # A value spread over several lines...
    multi-line : this is line 1
               : this is line 2
               : this is line 3
    # Several values for the same key...
    multi-value: this is value 1
    multi-value: this is value 2
    multi-value: this is value 3

would be read into a hash whose internal structure looked like this:

    {
       # Default section...
       '' => {
          'simple'           => 'simple value',
          'more complex key' => 'more complex value',
       },
       # Named section...
       'MULTI-WHATEVERS' => {
            'multi-line'  => "this is line 1\nthis is line 2\nthis is line 3",
            'multi-value' => [ 'this is value 1',
                               'this is value 2',
                               'this is value 3'
                             ],
        }
    }

INTERFACE

The following subroutines are exported automatically whenever the module is loaded...

"read_config($filename => %config_hash)"
"read_config($filename => $config_hash_ref)"
"read_config($string_ref => %config_hash_or_ref)"
The read_config() subroutine takes two arguments: the filename of a configuration file, and a variable into which the contents of that configuration file are to be loaded.

If the variable is a hash, then the configuration sections and their key/value pairs are loaded into nested subhashes of the hash.

If the variable is a scalar with an undefined value, a reference to an anonymous hash is first assigned to that scalar, and that hash is then filled as described above.

The subroutine returns true on success, and throws an exception on failure.

If you pass a reference to the string as the first argument to read_config() it uses that string as the source of the config info. For example:

        use Config::Std;
        # here we load the config text to a scalar
        my $cfg = q{
        [Section 1]
        attr1 = at
        attr2 = bat
        [Section 2]
        attr3 = cat
        };
        # here we parse the config from that scalar by passing a reference to it.
        read_config( \$cfg, my %config );
        use Data::Dumper 'Dumper';
        warn Dumper [ \%config ];
    
"write_config(%config_hash => $filename)"
"write_config($config_hash_ref => $filename)"
The write_config() subroutine takes two arguments: the hash or hash reference containing the configuration data to be written out to disk, and an optional filename specifying which file it is to be written to.

The data hash must conform to the two-level structure described earlier: with top-level keys naming sections and their values being references to second-level hashes that store the keys and values of the configuartion variables. If the structure of the hash differs from this, an exception is thrown.

If a filename is also specified, the subroutine opens that file and writes to it. It no filename is specified, the subroutine uses the name of the file from which the hash was originally loaded using read_config(). It no filename is specified and the hash wasn't originally loaded using read_config(), an exception is thrown.

The subroutine returns true on success and throws and exception on failure.

If necessary (typically to avoid conflicts with other modules), you can have the module export its two subroutines with different names by loading it with the appropriate options:

    use Config::Std { read_config => 'get_ini', write_config => 'update_ini' };
    # and later...
    get_ini($filename => %config_hash);
    # and later still...
    update_ini(%config_hash);

You can also control how much spacing the module puts between single- line values when they are first written to a file, by using the "def_gap" option:

    # No empty line between single-line config values...
    use Config::Std { def_gap => 0 };
    # An empty line between all single-line config values...
    use Config::Std { def_gap => 1 };

Regardless of the value passed for "def_gap", new multi-line values are always written with an empty line above and below them. Likewise, values that were previously read in from a file are always written back with whatever spacing they originally had.

DIAGNOSTICS

You tried to read in a configuration file, but the file you specified didn't exist. Perhaps the filepath you specified was wrong. Or maybe your application didn't have permission to access the file you specified.
You tried to read in a configuration file, but the file you specified was being written by someone else (they had a file lock active on it). Either try again later, or work out who else is using the file.
You passed a scalar variable as the destination into read_config() was supposed to load a configuration file, but that variable already had a defined value, so read_config() couldn't autovivify a new hash for you. Did you mean to pass the subroutine a hash instead of a scalar?
You called "write_config" and passed it a hash containing a configuration variable whose value wasn't a single string, or a list of strings. The configuration file format supported by this module only supports those two data types as values. If you really need to store other kinds of data in a configuration file, you should consider using "Data::Dumper" or "YAML" instead.
You tried to calll write_config() with only a configuration hash, but that hash wasn't originally loaded using read_config(), so write_config() has no idea where to write it to. Either make sure the hash you're trying to save was originally loaded using read_config(), or else provide an explicit filename as the second argument to write_config().
You tried to update or create a configuration file, but the file you specified could not be opened for writing (for the reason given in the parentheses). This is often caused by incorrect filepaths or lack of write permissions on a directory.
You tried to update or create a configuration file, but the file you specified was being written at the time by someone else (they had a file lock active on it). Either try again later, or work out who else is using the file.

CONFIGURATION AND ENVIRONMENT

Config::Std requires no configuration files or environment variables. (To do so would be disturbingly recursive.)

DEPENDENCIES

This module requires the Class::Std module (available from the CPAN)

INCOMPATIBILITIES

Those variants of INI file dialect supporting partial-line comment are incompatible. (This is the price of keeping comments when re-writing.)

BUGS AND LIMITATIONS

A daemon re-reading its config file has reported a memory leak.
This is a config file module. Tests written before "TAP" got parallel testing are unsafe with parallel testing, surprise! Settings are now included to force serial testing (until we refactor all tests to use temp dirs?).

If using an older Perl < 5.21.1, and Module.PL, and getting out-of-sequence test failures installing this module, either update Test::Harness~'>= 3.31' or export HARNESS_OPTIONS=j1 (or force/no-test, or use Build.PL and/or perl-5.22.0 or newer instead).

If you attempt to load read_config() and write_config() at runtime with "require", you can not rely upon the prototype to convert a regular hash to a reference. To work around this, you must explicitly pass a reference to the config hash.

    require Config::Std;
    Config::Std->import;
    my %config;
    read_config($file, \%config);
    write_config(\%config, $file);
    
If the config file being read contains carriage returns and line feeds at the end of each line rather than just line feeds (i.e. the standard Windows file format, when read on a machine expecting POSIX file format), Config::Std emits an error with embedded newline.

Workaround is match file line-endings to locale.

This will be fixed in 1.000.

A comment before the first section is not always retained on write-back, if the '' default section is empty.

Please report any bugs or feature requests to "bug-config-std@rt.cpan.org", or through the web interface at <http://rt.cpan.org>.

AUTHOR

Damian Conway "<DCONWAY@cpan.org>" Maintainers Bill Ricker "<BRICKER@cpan.org>" Tom Metro "<tmetro@cpan.org>"

LICENCE AND COPYRIGHT

Copyright (c) 2005, Damian Conway "<DCONWAY@cpan.org>". Copyright (c) 2011,2014,2017, D.Conway, W.Ricker "<BRICKER@cpan.org>" All rights reserved.

This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

DISCLAIMER OF WARRANTY

BECAUSE THIS SOFTWARE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE SOFTWARE, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE SOFTWARE "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE SOFTWARE IS WITH YOU. SHOULD THE SOFTWARE PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR, OR CORRECTION.

IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE SOFTWARE AS PERMITTED BY THE ABOVE LICENCE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE SOFTWARE (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE SOFTWARE TO OPERATE WITH ANY OTHER SOFTWARE), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

2024-03-05 perl v5.38.2