The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

Apache::ConfigFile - Parse an Apache style httpd.conf configuration file

SYNOPSIS

    # 
    # Parse the standard Apache httpd.conf
    #
    use Apache::ConfigFile;
    my $ac = Apache::ConfigFile->read("/etc/apache/httpd.conf");

    # You can get at individual configuration commands using
    # the cmd_config() method:

    my $hostname = $ac->cmd_config('ServerName');
    my $doc_root = $ac->cmd_config('DocumentRoot');

    # Multiple values are returned as a list, meaning that you
    # can directly assign them to an array:

    my @perlmods = $ac->cmd_config('PerlModule');

    # And, you can use the cmd_config_hash() routine to get at
    # multiple settings where the first is a type of "key":

    my %ftypes   = $ac->cmd_config_hash('FileTypeSuffix');

    # Then, you can reset the context of the calls using the
    # cmd_context() method so that you are accessing the 
    # appropriate values. For example, if you had a context 
    # block like
    #
    #   <VirtualHost "10.1.1.2">
    #       ServerName "www.mydomain.com"
    #       DocumentRoot "/www/mydomain.com/htdocs"
    #   </VirtualHost>
    #
    # You would get to this definition via:

    my $vh = $ac->cmd_context(VirtualHost => '10.1.1.2');
    my $vhost_server_name = $vh->cmd_config('ServerName');
    my $vhost_doc_root    = $vh->cmd_config('DocumentRoot');

    # If you had multiple VirtualHost declarations for a
    # given IP (as would be the case if you're using 
    # NameVirtualHosts), you could cycle through them with:

    for my $vh ($ac->cmd_context(VirtualHost => '10.1.1.3')) {
        my $vhost_server_name = $vh->cmd_config('ServerName');
        my $vhost_doc_root    = $vh->cmd_config('DocumentRoot');
    } 

    # In fact, even better you can "search" for one by specifying
    # an additional set of criteria to cmd_config(). To just get
    # the VirtualHost "docs.mydomain.com", for example, try:
 
    my $docs_svr = $ac->cmd_context(VirtualHost => '10.1.1.3',
                                    ServerName  => 'docs.mydomain.com');
    my $docs_base_dir = $docs_svr->cmd_config('DocumentRoot');

    # In addition, this module will automatically autoload
    # directive-based functions, meaning you can get to
    # commonly-used commands by name:

    my $host = $ac->server_name;
    my $root = $ac->server_root;
    my $html = $ac->document_root;

    # You also get the mod_perl dir_config() command to get
    # at PerlSetVar declarations by name. So, the block:
    #
    #   <Location /myapp>
    #       SetHandler perl-script
    #       PerlHandler Apache::MyApp
    #       PerlSetVar MyAppRoot "/usr/myapp"
    #       PerlSetVar MyAppRefresh "30m"
    #   </Location>
    #
    # Would be accessed as:

    my $loc = $ac->cmd_context(Location => '/myapp');
    my $app_root = $loc->dir_config('MyAppRoot');
    my $app_refr = $loc->dir_config('MyAppRefresh');

    # Finally, you get two other utility methods. The first
    # will return the current data structure verbatim, and
    # the second one will return a dump which you can print
    # out or parse or whatever:

    my %config = $self->data;
    warn "DEBUG: ", $self->dump, "\n";

UNSUPPORTED

I am no longer supporting this module in any way, shape, or form, so it is truly provided "as-is". It has some edge-case bugs which will not be fixed. It will work on 95+% of Apache config files.

I believe some alternative modules, including Apache::Admin::Config and Config::ApacheFormat, are being actively supported still, although this may or may not be true by the time you read this.

In case you care, the main reason I wrote this was to support Apache-like config files as a general case. But it turns out the core httpd.conf is rife with special cases, and is just plain a pain in the ass.

If you would like to take over maintenance of this module, please contact me at nate@wiger.org

DESCRIPTION

This module parses the Apache httpd.conf, or any compatible config file, and provides methods for you to access the values from the config file. The above examples show basic usage of this module, which boils down to reading a given config file and then using the cmd_config() and cmd_context() functions to access its information.

By default, the config file is parsed more or less "verbatim", meaning that directives are case-sensitive, variables are not interpolated, and so forth. These features can be changed by options given to the read() function (see below).

The read() function is the constructor, which reads in a configuration file and returns an object with methods that can be used to access directives from the file. The simplest usage is something like this:

    use Apache::ConfigFile;
    my $ac = Apache::ConfigFile->read("/path/to/httpd.conf");

Which would parse the Apache httpd.conf file and give you back an $ac object with the following methods:

cmd_config()

Used to access individual configuration commands

cmd_context()

Used to change the context of the commands you're accessing

dir_config()

Used to access values set via the PerlSetVar command (like mod_perl)

For more examples of standard Apache usage, you should read the "SYNOPSIS" above or skip down to the "FUNCTIONS".

In addition to reading an Apache config file, this module provides some options that allow the Apache syntax to be extended. This is useful if you're writing your own application and want to use a config file resembling Apache's.

    use Apache::ConfigFile;
    my $ac = Apache::ConfigFile->read(
                    file => "/path/to/httpd.conf",
                    ignore_case  => 1,
                    expand_vars  => 1,
                    fix_booleans => 1
             );

These options would allow us to write a custom config file looking like this:

    BaseDir    "/export"
    ImageDir   "$BaseDir/images"
    BuildDir   "$BaseDir/images"

    <Release "sw7">
        OfficialName "Software Update 7"
        BuildPath "$BuildDir/sw7/REL"         
        Platforms Solaris Linux IRIX HP-UX
        Supported Yes
    </Release>

Then, you would be able to access it as follows:

    use Apache::ConfigFile;
    my $swcfg = Apache::ConfigFile->read("releases.conf");

    # Note that case does not matter
    my $rel = $swcfg->cmd_context(release => 'sw7');
    my $ofn = $rel->cmd_config('bUiLdPaTh');
    
    # This is autoloading + fix_booleans
    unless ($rel->supported) {
        die "Sorry, that release is not supported";
    } 

There are several things to note. First, all our cmd_ functions are now case-insensitive, since we turned on the ignore_case flag (which is off by default). Second, notice a couple things about our unless statement. Since we specified fix_booleans, the words "Yes", "True", and "On" will be converted to 1 (true), and "No", "False", and "Off" will become 0 (false). As such, we can use these directives in boolean statements throughout our code.

In addition, since this module provides autoloading so that all config commands are turned into functions, you can access values directly, as shown by the statement $rel->supported. This statement is equivalent to the longer $rel->cmd_config('supported').

Finally, if you just wish to manually navigate the data structure (which is a huge hash of hashes of arrays) without using the accessor functions, you can return the thing verbatim:

    my %conf = $ac->data;
    print "Release is $conf{'release'}\n";

However, note that the internal representation is subject to change, so using the accessor functions is recommended.

FUNCTIONS

read(filename)

read(file => filename, opt => val, opt => val)

The read() function reads the configuration file specified and returns an object with methods to access its directives. read() has two calling forms. In the simplest version, you just specify a filename, and a new Apache::ConfigFile object is returned. Or, if you want to specify options, you specify each one as a key/value pair. For example:

   # keep default options
   my $ac = Apache::ConfigFile->read("httpd.conf");

   # override the case sensitivity and boolean translation
   my $ac = Apache::ConfigFile->read(file => "httpd.conf",
                                     ignore_case  => 1,
                                     fix_booleans => 1);

The list of valid options is:

file

Path to configuration file. If not provided then /usr/local/apache/conf/httpd.conf is used by default.

ignore_case

If set to 1, then all directives will be case-insensitive and stored in lowercase. Defaults to 0.

fix_booleans

If set to 1, then the words "Yes", "True", and "On" will be converted to 1 (true), and "No", "False", and "Off" will become 0 (false). This allows you to easily use these types of directives in if statements. Defaults to 0.

expand_vars

If set to 1, then you can reuse variables that you have defined elsewhere in the config file by prefixing them with a $. For example:

    BaseDir   "/export"
    HomeDir   "$BaseDir/home"

Currently, you can only reuse variables defined at the very top-level. Variables defined within context blocks of any kind cannot be reused.

raise_error

If set to 1, any type of error becomes fatal. Defaults to 0.

inherit_from

If set, then context blocks inherit from the specified default context. For example, say you have the blocks:

    <Category kitchen>
        Name "Soup Kitchen"
        Email "soup@kitchen.com"
        Access all
    </Category>

    <Category tomato_soup>
        Name "Tomato Soup"
    </Category>

If you then specified:

    ->read(..., inherit_from => 'kitchen');

Then all those variables that are not seen in the tomato_soup block would be filled in based on their values from the kitchen block. So, tomato_soup would inherit Email and Access from kitchen, but would provide its own Name.

Note: In order for this to work, the block providing the inherited items MUST appear first, as shown above.

root_directive

If set this specifies a directive other than RootDirective for relative path resolutions. For example:

    ApplicationRoot /usr/local/etc

    my $ac = Apache::ConfigFile->read(
                     file => "/usr/local/etc/app.config",
                     root_directive => 'ApplicationRoot'
             );

This will cause /usr/local/etc to be added to relative paths for includes, etc. With this additional behavior, the term ServerRoot, as used elsewhere in this document, comes to mean any directive that is specified via this option. Also note that the default value of this option is 'ServerRoot'.

server_root

This explicitly sets the ServerRoot for relative path resolutions for includes, etc. This option overrides any ServerRoot values found in the config file.

cmd_config(directive)

This is the meat-and-potatoes of the module; the method that lets you access configuration directives from your file. Examples:

    my $server_name = $ac->cmd_config('ServerName');
    my $doc_root = $ac->cmd_config('DocumentRoot');

This is a fairly straightforward function. You just give it the name of the directive you wish to access and you get its value back. Each time you call it, you will get the value for the next available instance of that variable. If called in a scalar context, you will just get the first value, assumed to be the "key".

What this means is that if you have a config file like this:

    ErrorDocument 404 /errors/404.cgi
    ErrorDocument 500 /errors/500.cgi

To get each line you would use a while loop:

    while (my @line = $ac->cmd_config('ErrorDocument')) {
        print "For error $line[0] we're using $line[1]\n";
    }

Which should print:

    For error 404 we're using /errors/404.cgi
    For error 500 we're using /errors/500.cgi

If you want more flexibility, read the following two functions.

cmd_config_array(directive)

This returns the entire data structure for a given directive as an array of arrays. So, you could get all the ErrorDocument configs by saying:

    my @errors = $ac->cmd_config_array('ErrorDocument');

Then, you would have to iterate over these yourself, since each element is an array reference:

    for my $e (@errors) {
        print "Code is $e->[0] and script is $e->[1]\n";
    }

Which should print:

   Code is 404 and script is /errors/404.cgi 
   Code is 500 and script is /errors/500.cgi 

Assuming the same configuration as above.

cmd_config_hash(directive)

This is perhaps the most useful form. It returns a set of key/value pairs where the key is the first element and the value is the rest of the line. This is great for handling FileTypeSuffix or AddHandler lines, for example:

    my %handler = $ac->cmd_config_hash('AddHandler');

This would return a hash where the keys would be the first field, such as cgi-script or server-parsed, and value is the remaining line as an array reference.

As such, you could access a specific one as:

    print "Suffixes for CGI scripts are: @{$handler{cgi-script}}\n";

Which should print out something like this:

    Suffixes for CGI scripts are: .cgi .pl

Note that you had to derefence the value inside of a @{} since the value is an array reference. This is so that you can get a list of values reliably. For example:

    my %handler = $ac->cmd_config_hash('AddHandler');
    my @cgi_suffixes   = @{$handler{cgi-script}};
    my @shtml_suffixed = @{$handler{server-parsed}};

That way you get the proper values even in the case of embedded whitespace. In addition, it allows you to define your own complex directives:

    # Format: key "Real Name" option1 option2 option3
    CustomField lname "Last Name" 
    CustomField ctry  "Country" US CA MX JP Other

Then in your code:

    my %custom_field = $ac->cmd_config_hash('CustomField');
    while(my($key, $val) = each %custom_field) {
        my $label = shift(@$val) || ucfirst($key);
        # see if we have any options remaining
        if (@$val) {
            # have options; create select list
            print qq($label: <select name="$key">\n");
            for my $opt (@$val) {
                print qq(<option value="$opt">$opt</option>\n);
            }
            print qq(</select>\n); 
        } else {
            # no options; text field
            print qq($label: <input name="$key" type="text type="text"">\n);
        }
    }

That way you could use an Apache style config file to setup a custom form based application.

cmd_context(context => specification)

You use this command to change the current context of what you are looking at. When you start, you are looking at the very top-level of the config file. However, you may want to look at a specific virtual host or directory. You can do so with this command.

    my $vhost = $ac->cmd_context(VirtualHost => '10.1.1.2');
    my $server_name = $vhost->cmd_config('ServerName');
    my $doc_root    = $vhost->cmd_config('DocumentRoot');

You'll notice that the cmd_context() call returns an object will all the same methods, but the data structure now starts from that point down. The context has been altered so that you are looking at the <VirtualHost "10.1.1.2">. block. As such, any commands that you do will affect that part of the configuration.

In some cases, you may have multiple definitions for a certain context level. One example is VirtualHost blocks if you're using NameVirtualHosts. You have two options. First, you could cycle through all of them in sequence:

    for my $vhost ($ac->cmd_context(VirtualHost => '10.1.1.2')) {
        # ... do stuff ...
    }

However, you may not know what you're looking for. In this case, if you can iterate through all of them with something like this:

    for my $vhost ($ac->cmd_context('VirtualHost')) {
        # ... do stuff ...
    }

Since you didn't specify a specific block, the special var _ will be set with the text tag for that block. Printing it out will reveal which VirtualHost (or whatever) you're in:

    print $vhost->cmd_config('_');  # "10.1.1.2"

Conversely, you may know exactly which one you're looking for. If so, you can specify one additional "search" parameter. For example, if you want the superfoo server, you could say:

    my $sf = $ac->cmd_context(VirtualHost => '10.1.1.2',
                              ServerName  => 'superfoo');

And this would look for a context block that looked something like this:

    <VirtualHost "10.1.1.2">
        ServerName "superfoo"
        # ... more config options ...
    </VirtualHost>

you can easily access nested configurations as well. If you had a configuration like this:

    <Location "/upload">
        SetHandler perl-script
        PerlHandler Apache::MyUploadModule
        PerlSetVar MyUploadModuleMaxsize "5M"
        PerlSetVar MyUploadModuleTimeout "300s"
        <Limit>
            require user "bob"
            require user "jim"
        </Limit>
    </Location>

And you wanted to find out what the valid users were who could access this page, you would navigate it like so:

    my $loc = $ac->cmd_context(Location => '/upload');
    my $lim = $loc->cmd_context('Limit');
    my @users = $lim->cmd_config('require');

Or, more succintly:

    my @users = $ac->cmd_context(Location => '/upload')
                   ->cmd_context(Limit => '')->cmd_config('require');

Since cmd_context() returns an object pointing to the next context, you can chain calls together to get to a deeply nested level.

dir_config()

This routine is provided for mod_perl compatibility. It allows you to access configuration commands specified via the PerlSetVar directive. So, assuming the above example, you could access the settings for MyUploadModule like so:

    my $upload = $ac->cmd_context(Location => '/upload');

    my $maxsize = $upload->dir_config('MyUploadModuleMaxsize');
    my $timeout = $upload->dir_config('MyUploadModuleTimeout');

The idea is to provide an interface which walks and talks roughly like Apache actually would.

data()

This returns the entire data structure under the current context verabatim. So, you could get all the values for a VirtualHost with:

    my $vh = $ac->cmd_context(VirtualHost => '10.1.1.4');
    my %vhost = $vh->data;

If you specified ignore_case, then all the keys will be lowercase; otherwise, they will be in whatever case they are in the config file.

dump()

This returns a dump of the current data structure in string form. So for debugging purposes you can dump the config with something like this:

    warn "DUMP: ", $ac->dump, "\n";

reread()

You can use this function to reread the configuration file. For example, maybe you want your application to reread its config if it receives a SIGHUP:

    $SIG{HUP} = \&handler;
    sub handler {
        my $sig = shift;
        if ($sig eq 'HUP') {
            # reread our config file on kill -HUP
            $config->reread;
        }
    }

The above would handle a SIGHUP by rereading the config file.

write([file])

This writes the configuration out to disk. If no file is specified, then the one passed to read() is used. This method is currently under development and does not work. Patches welcome.

autoloaded calls

In addition to the above, you can also access values by calling a function named for the config command directly:

    my $server_name = $ac->cmd_config('ServerName');

Is the same as:

    my $server_name = $ac->server_name;

Underscores in the function name are taken as a place to put an uppercase letter. So these are all equivalent:

    my $doc_root = $ac->cmd_config('DocumentRoot');
    my $doc_root = $ac->DocumentRoot;   # looks silly
    my $doc_root = $ac->document_root;

Note, though, that the following would not work unless you had set the ignore_case option:

    my $doc_root = $ac->documentroot;   # won't work

This is because it will look for the directive Documentroot, which probably doesn't exist.

ALIASES

When I initially wrote this module, I tried to follow the internal Apache API pretty closely. However, for those unfamiliar with Apache these method names probably make little sense. As such, the following function aliases are provided

directive

Same as cmd_config()

directive_array

Same as cmd_config_array()

directive_hash

Same as cmd_config_hash()

section

Same as cmd_context()

So this code:

    my $vh = $ac->cmd_context(VirtualHost => '10.1.1.2');
    my $vhost_server_name = $vh->cmd_config('ServerName');
    my $vhost_doc_root    = $vh->cmd_config('DocumentRoot');
    my %error_handlers    = $ac->cmd_config_hash('ErrorDocument');

Could be rewritten as the following and work exactly the same:

    my $vh = $ac->section(VirtualHost => '10.1.1.2');
    my $vhost_server_name = $vh->directive('ServerName');
    my $vhost_doc_root    = $vh->directive('DocumentRoot');
    my %error_handlers    = $ac->directive_hash('ErrorDocument');

These will always be supported so feel free to use them.

NOTES

Currently LogFormat and any other directive with embedded quotes, even if escaped, are not handled correctly. I know there is a fix for it but I have a mental block and can't figure it out. Help!

Currently, the order of context blocks is not maintained. So, if you define two blocks:

    <Directory "/">
        Options +MultiViews
    </Directory>

    <Directory "/var/apache/htdocs">
        Options +ExecCGI
    </Directory>

There will be no way for you to tell the order in which these were defined. Normally this should not matter, since the idea of a context section is to create a logical entity. However, patches to overcome this limitation are welcomed.

This module has only been tested and used on UNIX platforms. It may or may not be broke elsewhere.

VERSION

$Id: ConfigFile.pm,v 1.23 2003/10/09 18:24:41 nwiger Exp $

AUTHOR

Copyright (c) 1999-2003, Nathan Wiger <nate@wiger.org>. All Rights Reserved.

This module is free software; you may copy this under the terms of the GNU General Public License, or the Artistic License, copies of which should have accompanied your Perl kit.