NAME Module::Generic - Generic Module to inherit from SYNOPSIS package MyModule; BEGIN { use strict; use warnings; use parent qw( Module::Generic ); }; sub init { my $self = shift( @_ ); # Requires parameters provided to have their equivalent method $self->{_init_strict_use_sub} = 1; # Smartly accepts key-value pairs as list or hash reference $self->SUPER::init( @_ ); # This won't be affected by parameters provided during instantiation $self->{_private_param} = 'some value'; return( $self ); } sub active { return( shift->_set_get_boolean( 'active', @_ ) ); } sub address { return( shift->_set_get_object( 'address', 'My::Address', @_ ) ); } sub age { return( shift->_set_get_number( 'age', @_ ) ); } sub name { return( shift->_set_get_scalar( 'name', @_ ) ); } sub uuid { return( shift->_set_get_uuid( 'uuid', @_ ) ); } sub remote_addr { return( shift->_set_get_ip( 'remote_addr', @_ ) ); } sub discount { return( shift->_set_get_class_array( 'discount', { amount => { type => 'number' }, discount => { type => 'object', class => 'My::Discount' }, }, @_ ) ); } sub settings { return( shift->_set_get_class( 'settings', { # Will create a Module::Generic::Array array object of objects of class MY::Item items => { type => 'object_array_object', class => 'My::Item' }, notify => { type => 'boolean' }, resumes_at => { type => 'datetime' }, timeout => { type => 'integer' }, customer => { definition => { billing_address => { package => "My::Address", type => "object" }, email => { type => "scalar" }, name => { type => "scalar" }, shipping_address => { package => "My::Address", type => "object" }, }, type => "class", }, }, @_ ) ); } VERSION v0.30.0 DESCRIPTION Module::Generic as its name says it all, is a generic module to inherit from. It is designed to provide a useful framework and speed up coding and debugging. It contains standard and support methods that may be superseded by your module. It also contains an AUTOLOAD transforming any hash object key into dynamic methods and also recognize the dynamic routine a la AutoLoader. The reason is that while "AutoLoader" provides the user with a convenient AUTOLOAD, I wanted a way to also keep the functionnality of Module::Generic AUTOLOAD that were not included in "AutoLoader". So the only solution was a merger. METHODS import import() is used for the AutoLoader mechanism and hence is not a public method. It is just mentionned here for info only. new new will create a new object for the package, pass any argument it might receive to the special standard routine init that *must* exist. Then it returns what returns "init". To protect object inner content from sneaking by third party, you can declare the package global variable *OBJECT_PERMS* and give it a Unix permission, but only 1 digit. It will then work just like Unix permission. That is, if permission is 7, then only the module who generated the object may read/write content of the object. However, if you set 5, the, other may look into the content of the object, but may not modify it. 7, as you would have guessed, allow other to modify the content of an object. If *OBJECT_PERMS* is not defined, permissions system is not activated and hence anyone may access and possibly modify the content of your object. If the module runs under mod_perl, and assuming you have set the variable "GlobalRequest" in your Apache configuration, it is recognised and a clean up registered routine is declared to Apache to clean up the content of the object. This methods calls "init", which does all the work of setting object properties and calling methods to that effect. as_hash This will recursively transform the object into an hash suitable to be encoded in json. It does this by calling each method of the object and build an hash reference with the method name as the key and the method returned value as the value. If the method returned value is an object, it will call its "as_hash" method if it supports it. It returns the hash reference built clear Alias for "clear_error" clear_error Clear all error from the object and from the available global variable $ERROR. This is a handy method to use at the beginning of other methods of calling package, so the end user may do a test such as: $obj->some_method( 'some arguments' ); die( $obj->error() ) if( $obj->error() ); # some_method() would then contain something like: sub some_method { my $self = shift( @_ ); ## Clear all previous error, so we may set our own later one eventually $self->clear_error(); # ... } This way the end user may be sure that if "$obj-"error()> returns true something wrong has occured. clone Clone the current object if it is of type hash or array reference. It returns an error if the type is neither. It returns the clone. colour_close The marker to be used to set the closing of a command line colour sequence. Defaults to ">" colour_closest Provided with a colour, this returns the closest standard one supported by terminal. A colour provided can be a colour name, or a 9 digits rgb value or an hexadecimal value colour_format Provided with a hash reference of parameters, this will return a string properly formatted to display colours on the command line. Parameters are: *text* or *message* This is the text to be formatted in colour. *bgcolour* or *bgcolor* or *bg_colour* or *bg_color* The value for the background colour. *colour* or *color* or *fg_colour* or *fg_color* or *fgcolour* or *fgcolor* The value for the foreground colour. Valid value can be a colour name, an rgb value like 255255255, a rgb annotation like "rgb(255, 255, 255)" or a rgba annotation like "rgba(255,255,255,0.5)" A colour can be preceded by the words "light" or "bright" to provide slightly lighter colour where supported. Similarly, if an rgba value is provided, and the opacity is less than 1, this is equivalent to using the keyword "light" It returns the text properly formatted to be outputted in a terminal. *style* The possible values are: *bold*, *italic*, *underline*, *blink*, *reverse*, *conceal*, *strike* colour_open The marker to be used to set the opening of a command line colour sequence. Defaults to "<" colour_parse Provided with a string, this will parse the string for colour formatting. Formatting can be encapsulated in another formatting, and can be expressed in 2 different ways. For example: $self->colour_parse( "And {style => 'i|b', color => green}what about{/} {style => 'blink', color => yellow}me{/} ?" ); would result with the words "what about" in italic, bold and green colour and the word "me" in yellow colour blinking (if supported). Another way is: $self->colour_parse( "And {bold light red on white}what about{/} {underline yellow}me too{/} ?" ); would return a string with the words "what about" in light red bold text on a white background, and the words "me too" in yellow with an underline. $self->colour_parse( "Hello {bold red on white}everyone! This is {underline rgb(0,0,255)}embedded{/}{/} text..." ); would return a string with the words "everyone! This is" in bold red characters on white background and the word "embedded" in underline blue color The idea for this syntax, not the code, is taken from Term::ANSIColor colour_to_rgb Convert a human colour keyword like "red", "green" into a rgb equivalent. coloured Provided with a colouring preference expressed as the first argument as string, and followed by 1 or more arguments that are concatenated to form the text string to format. For example: print( $o->coloured( 'bold white on red', "Hello it's me!\n" ) ); A colour can be expressed as a rgb, such as : print( $o->coloured( 'underline rgb( 0, 0, 255 ) on white', "Hello everyone!" ), "\n" ); rgb can also be rgba with the last decimal, normally an opacity used here to set light color if the value is less than 1. For example : print( $o->coloured( 'underline rgba(255, 0, 0, 0.5)', "Hello everyone!" ), "\n" ); debug Set or get the debug level. This takes and return an integer. Based on the value, "message" will or will not print out messages. For example : $self->debug( 2 ); $self->message( 2, "Debugging message here." ); Since 2 used in "message" is equal to the debug value, the debugging message is printed. If the debug value is switched to 1, the message will be silenced. deserialise my $ref = $self->deserialise( %hash_of_options ); my $ref = $self->deserialise( $hash_reference_of_options ); my $ref = $self->deserialise( $serialised_data, %hash_of_options ); my $ref = $self->deserialise( $serialised_data, $hash_reference_of_options ); This method use a specified serialiser class and deserialise the given data either directly from a specified file or being provided, and returns the perl data. The serialisers currently supported are: CBOR::Free, CBOR::XS, JSON, Sereal and Storable::Improved (or the legacy Storable). They are not required by Module::Generic, so you must install them yourself. If the serialiser chosen is not installed, this will set an errr and return "undef". It takes an hash or hash reference of options. You can also provide the data to deserialise as the first argument followed by an hash or hash reference of options. It can then: * retrieve data directly from File * retrieve data from a file handle (only with Storable) * Return the deserialised data The supported options are: * "base64" Thise can be set to a true value like 1, or to your preferred base64 encoder/decoder, or to an array reference containing 2 code references, the first one for encoding and the second one for decoding. If this is set simply to a true value, "deserialise" will call "_has_base64" to find out any installed base64 modules. Currently the ones supported are: Crypt::Misc and MIME::Base64. Of course, you need to have one of those modules installed first before it can be used. If this option is set and no appropriate module could be found, "deserialise" will return an error. * "data" Data to be deserialised. * "file" Provides a file path from which to read the serialised data. * "io" A file handle. This is used when the serialiser is Storable to call its function "store_fd" in Storable::Improved and "fd_retrieve" in Storable::Improved * *lock* Boolean. If true, this will lock the file before reading from it. This works only in conjonction with *file* and the serialiser Storable::Improved * "serialiser" Specify the class name of the serialiser to use. Supported serialiser can either be "CBOR" or CBOR::XS, Sereal and Storable If the serialiser is CBOR::XS the following additional options are supported: "max_depth", "max_size", "allow_unknown", "allow_sharing", "allow_cycles", "forbid_objects", "pack_strings", "text_keys", "text_strings", "validate_utf8", "filter" See CBOR::XS for detail on those options. If the serialiser is Sereal, the following additional options are supported: "refuse_snappy", "refuse_objects", "no_bless_objects", "validate_utf8", "max_recursion_depth", "max_num_hash_entries", "max_num_array_entries", "max_string_length", "max_uncompressed_size", "incremental", "alias_smallint", "alias_varint_under", "use_undef", "set_readonly", "set_readonly_scalars" See Sereal for detail on those options. If an error occurs, this sets an error and return "undef" deserialize Alias for "deserialise" dump Provided with some data, this will return a string representation of the data formatted by Data::Printer dump_hex Returns an hexadecimal dump of the data provided. This requires the module Devel::Hexdump and will return "undef" and set an "error" if not found. dump_print Provided with a file to write to and some data, this will format the string representation of the data using Data::Printer and save it to the given file. dumper Provided with some data, and optionally an hash reference of parameters as last argument, this will create a string representation of the data using Data::Dumper and return it. This sets Data::Dumper to be terse, to indent, to use "qq" and optionally to not exceed a maximum *depth* if it is provided in the argument hash reference. dumpto Alias for "dumpto_dumper" printer Same as "dumper", but using Data::Printer to format the data. dumpto_printer Same as "dump_print" above that is an alias of this method. dumpto_dumper Same as "dumpto_printer" above, but using Data::Dumper errno Sets or gets an error number. error Provided with a list of strings or an hash reference of parameters and this will set the current error issuing a Module::Generic::Exception object, call "warn" in perlfunc, or "$r-"warn> under Apache2 modperl, and returns undef() or an empty list in list context: if( $some_condition ) { return( $self->error( "Some error." ) ); } Note that you do not have to worry about a trailing line feed sequence. "error" takes care of it. The script calling your module could write calls to your module methods like this: my $cust_name = $object->customer->name || die( "Got an error in file ", $object->error->file, " at line ", $object->error->line, ": ", $object->error->trace, "\n" ); # or simply: my $cust_name = $object->customer->name || die( "Got an error: ", $object->error, "\n" ); If you want to use an hash reference instead, you can pass the following parameters. Any other parameters will be passed to the exception class. *class* The package name or class to use to instantiate the error object. By default, it will use Module::Generic::Exception class or the one specified with the object property "_exception_class" $self->do_something_bad || return( $self->error({ code => 500, message => "Oopsie", class => "My::NoWayException", }) ); my $exception = $self->error; # an My::NoWayException object Note, however, that if the class specified cannot be loaded for some reason, "error" in Module::Generic will die since this would be an error within another error. *message* The error message. Note also that by calling "error" it will not clear the current error. For that you have to call "clear_error" explicitly. Also, when an error is set, the global variable *ERROR* in the inheriting package is set accordingly. This is especially usefull, when your initiating an object and that an error occured. At that time, since the object could not be initiated, the end user can not use the object to get the error message, and then can get it using the global module variable *ERROR*, for example: my $obj = Some::Package->new || die( $Some::Package::ERROR, "\n" ); If the caller has disabled warnings using the pragma "no warnings", "error" will respect it and not call warn. Calling warn can also be silenced if the object has a property *quiet* set to true. The error message can be split in multiple argument. "error" will concatenate each argument to form a complete string. An argument can even be a reference to a sub routine and will get called to get the resulting string, unless the object property *_msg_no_exec_sub* is set to false. This can switched off with the method "noexec" If perl runs under Apache2 modperl, and an error handler is set with "error_handler", this will call the error handler with the error string. If an Apache2 modperl log handler has been set, this will also be called to log the error. If the object property *fatal* is set to true, this will call die instead of "warn" in perlfunc. Last, but not least since "error" returns undef in scalar context or an empty list in list context, if the method that triggered the error is chained, it would normally generate a perl error that the following method cannot be called on an undefined value. To solve this, when an object is expected, "error" returns a special object from module Module::Generic::Null that will enable all the chained methods to be performed and return the error when requested to. For example: my $o = My::Package->new; my $total $o->get_customer(10)->products->total || die( $o->error, "\n" ); Assuming this method here "get_customer" returns an error, the chaining will continue, but produce nothing and ultimately returns undef. error_handler Sets or gets a code reference that will be called to handle errors that have been triggered when calling "error" errors Used by error() to store the error sent to him for history. It returns an array of all error that have occured in lsit context, and the last error in scalar context. errstr Set/get the error string, period. It does not produce any warning like error would do. fatal Boolean. If enabled, any error will call "die" in perlfunc instead of returning "undef" in perlfunc and setting an error. Defaults to false. You can enable it in your own package by initialising it in your own "init" method like this: sub init { my $self = shift( @_ ); $self->{fatal} = 1; return( $self->SUPER::init( @_ ) ); } get Uset to get an object data key value: $obj->set( 'verbose' => 1, 'debug' => 0 ); ## ... my $verbose = $obj->get( 'verbose' ); my @vals = $obj->get( qw( verbose debug ) ); print( $out "Verbose level is $vals[ 0 ] and debug level is $vals[ 1 ]\n" ); This is no more needed, as it has been more conveniently bypassed by the AUTOLOAD generic routine with which you may say: $obj->verbose( 1 ); $obj->debug( 0 ); ## ... my $verbose = $obj->verbose(); Much better, no? init This is the "new" package object initializer. It is called by "new" and is used to set up any parameter provided in a hash like fashion: my $obj My::Module->new( 'verbose' => 1, 'debug' => 0 ); You may want to superseed "init" to have it suit your needs. "init" needs to returns the object it received in the first place or an error if something went wrong, such as: sub init { my $self = shift( @_ ); my $dbh = DB::Object->connect() || return( $self->error( "Unable to connect to database server." ) ); $self->{dbh} = $dbh; return( $self ); } In this example, using "error" will set the global variable $ERROR that will contain the error, so user can say: my $obj = My::Module->new() || die( $My::Module::ERROR ); If the global variable *VERBOSE*, *DEBUG*, *VERSION* are defined in the module, and that they do not exist as an object key, they will be set automatically and accordingly to those global variable. The supported data type of the object generated by the "new" method may either be a hash reference or a glob reference. Those supported data types may very well be extended to an array reference in a near future. When provided with an hash reference, and when object property *_init_strict_use_sub* is set to true, "init" will call each method corresponding to the key name and pass it the key value and it will set an error and skip it if the corresponding method does not exist. Otherwise, it calls each corresponding method and pass it whatever value was provided and check for that method return value. If the return value is "undef" in perlfunc and the value provided is not itself "undef", then it issues a warning and return the "error" that is assumed having being set by that method. Otherwise if the object property *_init_strict* is set to true, it will check the object property matching the hash key for the default value type and set an error and return undef if it does not match. Foe example, "init" in your module could be like this: sub init { my $self = shift( @_ ); $self->{_init_strict} = 1; $self->{products} = []; return( $self->SUPER::init( @_ ) ); } Then, if init is called like this: $object->init({ products => $some_string_but_not_array }) || die( $object->error, "\n" ); This would cause your script to die, because "products" value is a string and not an array reference. Otherwise, if none of those special object properties are set, the init will create an object property matching the key of the hash and set its value accordingly. For example : sub init { my $self = shift( @_ ); return( $self->SUPER::init( @_ ) ); } Then, if init is called like this: $object->init( products => $array_ref, first_name => 'John', last_name => 'Doe' }); The object would then contain the properties *products*, *first_name* and *last_name* and can be accessed as methods, such as : my $fname = $object->first_name; You can also alter the way "init" process the parameters received using the following properties you can set in your own "init" method, for example: sub init { my $self = shift( @_ ); # Set the order in which the parameters are processed, because some methods may rely on other methods' value $self->{_init_params_order} [qw( method1 method2 )]; # Enable strict sub, which means the corresponding method must exist for the parameter provided $self->{_init_strict_use_sub} = 1; # Set the class name of the exception to use in error() # Here My::Package::Exception should inherit from Module::Generic::Exception or some other Exception package $self->{_exception_class} = 'My::Package::Exception'; $self->SUPER::init( @_ ) || return( $self->pass_error ); return( $self ); } You can also specify a default exception class that will be used by "error" to create exception object, by setting the object property "_exception_class": sub init { my $self = shift( @_ ); $self->{name} = 'default_name'; # For any key-value pairs to be matched by a corresponding method $self->{_init_strict_use_sub} = 1; $self->{_exception_class} = 'My::Exception'; return( $self->SUPER::init( @_ ) ); } log_handler Provided a reference to a sub routine or an anonymous sub routine, this will set the handler that is called by "message" It returns the current value set. message message() is used to display verbose/debug output. It will display something to the extend that either *verbose* or *debug* are toggled on. If so, all debugging message will be prepended by " ## " by default or the prefix string specified with the *prefix* option, to highlight the fact that this is a debugging message. Addionally, if a number is provided as first argument to message(), it will be treated as the minimum required level of debugness. So, if the current debug state level is not equal or superior to the one provided as first argument, the message will not be displayed. For example: # Set debugness to 3 $obj->debug( 3 ); # This message will not be printed $obj->message( 4, "Some detailed debugging stuff that we might not want." ); # This will be displayed $obj->message( 2, "Some more common message we want the user to see." ); Now, why debug is used and not verbose level? Well, because mostly, the verbose level needs only to be true, that is equal to 1 to be efficient. You do not really need to have a verbose level greater than 1. However, the debug level usually may have various level. Also, the text provided can be separated by comma, and even be a code reference, such as: $self->message( 2, "I have found", "something weird here:", sub{ $self->dumper( $data ) } ); If the object has a property *_msg_no_exec_sub* set to true, then a code reference will not be called and instead be added to the string as is. This can be done simply like this: $self->noexec->message( 2, "I have found", "something weird here:", sub{ $self->dumper( $data ) } ); "message" also takes an optional hash reference as the last parameter with the following recognised options: *caller_info* This is a boolean value, which is true by default. When true, this will prepend the debug message with information about the caller of "message" *level* An integer. Debugging level. *message* The text of the debugging message. This is optional since this can be provided as first or consecutive arguments like in a list as demonstrated in the example above. This allows you to do something like this: $self->message( 2, { message => "Some debug message here", prefix => ">>" }); or $self->message( { message => "Some debug message here", prefix => ">>", level => 2 }); *no_encoding* Boolean value. If true and when the debugging is set to be printed to a file, this will not set the binmode to "utf-8" *prefix* By default this is set to "##". This value is used as the prefix used in debugging output. *type* Type of debugging message_check This is called by "message" Provided with a list of arguments, this method will check if the first argument is an integer and find out if a debug message should be printed out or not. It returns the list of arguments as an array reference. message_color Alias for "message_colour" message_colour This is the same as "message", except this will check for colour formatting, which "message" does not do. For example: $self->message_colour( 3, "And {bold light white on red}what about{/} {underline green}me again{/} ?" ); "message_colour" can also be called as message_color See also "colour_format" and "colour_parse" message_frame Return the optional hash reference of parameters, if any, that can be provided as the last argument to "message" messagef This works like "sprintf" in perlfunc, so provided with a format and a list of arguments, this print out the message. For example : $self->messagef( 1, "Customer name is %s", $cust->name ); Where 1 is the debug level set with "debug" messagef_colour This method is same as "message_colour" and messagef combined. It enables to pass sprintf-like parameters while enabling colours. message_log This is called from "message". Provided with a message to log, this will check if "message_log_io" returns a valid file handler, presumably to log file, and if so print the message to it. If no file handle is set, this returns undef, other it returns the value from "$io-"print> message_log_io Set or get the message log file handle. If set, "message_log" will use it to print messages received from "message" If no argument is provided bu your module has a global variable "LOG_DEBUG" set to true and global variable "DEB_LOG" set presumably to the file path of a log file, then this attempts to open in write mode the log file. It returns the current log file handle, if any. new_array Instantiate a new Module::Generic::Array object. If any arguments are provided, it will pass it to "new" in Module::Generic::Array and return the object. new_datetime Provided with some optional arguments and this will instantiate a new Module::Generic::DateTime object, passing it whatever argument was provided. Example: my $dt = DateTime->now( time_zone => 'Asia/Tokyo' ); # Returns a new Module::Generic::DateTime object my $d = $o->new_datetime( $dt ); # Returns a new Module::Generic::DateTime object with DateTime initiated automatically # to now with time zone set by default to UTC my $d = $o->new_datetime; new_file Instantiate a new Module::Generic::File object. If any arguments are provided, it will pass it to "new" in Module::Generic::File and return the object. new_glob This method is called instead of "new" in your package for GLOB type module. It will set an hash of options provided and call "init" and return the newly instantiated object upon success, or "undef" upon error. new_hash Instantiate a new Module::Generic::Hash object. If any arguments are provided, it will pass it to "new" in Module::Generic::Hash and return the object. new_json This method tries to load JSON and create a new object. By default it enables the following JSON object properties: "allow_blessed" in JSON "allow_nonref" in JSON "convert_blessed" in JSON "relaxed" in JSON new_null Returns a null value based on the expectations of the caller and thus without breaking the caller's call flow. If the caller wants an hash reference, it returns an empty hash reference. If the caller wants an array reference, it returns an empty array reference. If the caller wants a code reference, it returns an anonymous subroutine that returns "undef" or an empty list. If the caller is calling another method right after, this means this is an object context and "new_null" will instantiate a new Module::Generic::Null object. If any arguments were provided to "new_null", they will be passed along to "new" in Module::Generic::Null and the new object will be returned. In any other context, "undef" is returned or an empty list. Without using "new_null", if you return simply undef, like: my $val = $object->return_false->[0]; sub return_false{ return } The above would trigger an error that the value returned by "return_false" is not an array reference. Instead of checking on the recipient end what kind of returned value was returned, the caller only need to check if it is defined or not, no matter the context in which it is called. For example: my $this = My::Object->new; my $val = $this->call1; # return undef) # object context $val = $this->call1->call_again; # $val is undefined # hash reference context $val = $this->call1->fake->{name}; # $val is undefined # array reference context $val = $this->call1->fake->[0]; # $val is undefined # code reference context $val = $this->call1->fake->(); # $val is undefined # scalar reference context $val = ${$this->call1->fake}; # $val is undefined # simple scalar $val = $this->call1->fake; # $val is undefined package My::Object; use parent qw( Module::Generic ); sub call1 { return( shift->call2 ); } sub call2 { return( shift->new_null ); } sub call_again { my $self = shift( @_ ); print( "Got here in call_again\n" ); return( $self ); } This technique is also used by "error" to set an error object and return undef but still allow chaining beyond the error. See "error" and Module::Generic::Exception for more information. new_number Instantiate a new Module::Generic::Number object. If any arguments are provided, it will pass it to "new" in Module::Generic::Number and return the object. new_scalar Instantiate a new Module::Generic::Scalar object. If any arguments are provided, it will pass it to "new" in Module::Generic::Scalar and return the object. new_tempdir Returns a new temporary directory by calling "tempdir" in Module::Generic::File new_tempfile Returns a new temporary directory by calling "tempfile" in Module::Generic::File new_version Provided with a version and this will return a new version object. If the value provided is not a suitable version, this will set an error and return "undef" noexec Sets the module property *_msg_no_exec_sub* to true, so that any call to "message" whose arguments include a reference to a sub routine, will not try to execute the code. For example, imagine you have a sub routine such as: sub hello { return( "Hello !" ); } And in your code, you write: $self->message( 2, "Someone said: ", \&hello ); If *_msg_no_exec_sub* is set to false (by default), then the above would print out the following message: Someone said Hello ! But if *_msg_no_exec_sub* is set to true, then the same would rather produce the following : Someone said CODE(0x7f9103801700) pass_error Provided with an error, typically a Module::Generic::Exception object, but it could be anything as long as it is an object, hopefully an exception object, this will set the error value to the error provided, and without issuing any new warning nor creating a new Module::Generic::Exception object. It makes it possible to pass the error along so the caller can retrieve it later. This is typically used by a method calling another one in another module that produced an error. For example : sub getCustomerInfo { my $self = shift( @_ ); # Maybe a LWP::UserAgent sub class? my $client = $self->lwp_client_object; my $res = $client->get( $remote_api_endpoint ) || return( $self->pass_error( $client->error ) ); } Then : my $client_info = $object->getCustomerInfo || die( $object->error, "\n" ); Which would return the http client error that has been passed along You can optionally provide an hash of parameters as the last argument, such as: return( $self->pass_error( $obj->error, { class => 'My::Exception', code => 400 } ) ); Supported options are: "class" The name of a class name to re-bless the error object provided. "code" The error code to set in the error object being passed. quiet Set or get the object property *quiet* to true or false. If this is true, no warning will be issued when "error" is called. save Provided with some data and a file path, or alternatively an hash reference of options with the properties *data*, *encoding* and *file*, this will write to the given file the provided *data* using the encoding *encoding*. This is designed to simplify the tedious task of write to files. If it cannot open the file in write mode, or cannot print to it, this will set an error and return undef. Otherwise this returns the size of the file in bytes. serialise This method use a specified serialiser class and serialise the given data either by returning it or by saving it directly to a given file. The serialisers currently supported are: CBOR::Free, CBOR::XS, JSON, Sereal and Storable::Improved (or the legacy version Storable). They are not required by Module::Generic, so you must install them yourself. If the serialiser chosen is not installed, this will set an errr and return "undef". This method takes some data and an optional hash or hash reference of parameters. It can then: * save data directly to File * save data to a file handle (only with Storable::Improved / Storable) * Return the serialised data The supported parameters are: * *append* Boolean. If true, the serialised data will be appended to the given file. This works only in conjonction with *file* * *base64* Thise can be set to a true value like 1, or to your preferred base64 encoder/decoder, or to an array reference containing 2 code references, the first one for encoding and the second one for decoding. If this is set simply to a true value, "serialise" will call "_has_base64" to find out any installed base64 modules. Currently the ones supported are: Crypt::Misc and MIME::Base64. Of course, you need to have one of those modules installed first before it can be used. If this option is set and no appropriate module could be found, "serialise" will return an error. * *file* String. A file path where to store the serialised data. * *io* A file handle. This is used when the serialiser is Storable::Improved / Storable to call its function "store_fd" in Storable::Improved and "fd_retrieve" in Storable::Improved * *lock* Boolean. If true, this will lock the file before writing to it. This works only in conjonction with *file* and the serialiser Storable::Improved * *serialiser* or *serializer* A string being the class of the serialiser to use. This can be only either Sereal or Storable::Improved Additionally the following options are supported and passed through directly for each serialiser: * CBOR::Free: "canonical", "string_encode_mode", "preserve_references", "scalar_references" * CBOR: "max_depth", "max_size", "allow_unknown", "allow_sharing", "allow_cycles", "forbid_objects", "pack_strings", "text_keys", "text_strings", "validate_utf8", "filter" * JSON: "allow_blessed" "allow_nonref" "allow_unknown" "allow_tags" "ascii" "boolean_values" "canonical" "convert_blessed" "filter_json_object" "filter_json_single_key_object" "indent" "latin1" "max_depth" "max_size" "pretty" "relaxed" "space_after" "space_before" "utf8" * "encode" in Sereal::Decoder if the serialiser is Sereal: "aliased_dedupe_strings", "canonical", "canonical_refs", "compress", "compress_level", "compress_threshold", "croak_on_bless", "dedupe_strings", "freeze_callbacks", "max_recursion_depth", "no_bless_objects", "no_shared_hashkeys", "protocol_version", "snappy", "snappy_incr", "snappy_threshold", "sort_keys", "stringify_unknown", "undef_unknown", "use_protocol_v1", "warn_unknown" * Storable::Improved / Storable: no option available If an error occurs, this sets an error and return "undef" serialize Alias for "serialise" set set() sets object inner data type and takes arguments in a hash like fashion: $obj->set( 'verbose' => 1, 'debug' => 0 ); subclasses Provided with a *CLASS* value, this method try to guess all the existing sub classes of the provided *CLASS*. If *CLASS* is not provided, the class into which was blessed the calling object will be used instead. It returns an array of subclasses in list context and a reference to an array of those subclasses in scalar context. If an error occured, undef is returned and an error is set accordingly. The latter can be retrieved using the error method. true Returns a "true" variable from Module::Generic::Boolean false Returns a "false" variable from Module::Generic::Boolean verbose Set or get the verbosity level with an integer. will This will try to find out if an object supports a given method call and returns the code reference to it or undef if none is found. AUTOLOAD The special AUTOLOAD() routine is called by perl when no matching routine was found in the module. AUTOLOAD() will then try hard to process the request. For example, let's assue we have a routine foo. It will first, check if an equivalent entry of the routine name that was called exist in the hash reference of the object. If there is and that more than one argument were passed to this non existing routine, those arguments will be stored as a reference to an array as a value of the key in the object. Otherwise the single argument will simply be stored as the value of the key of the object. Then, if called in list context, it will return a array if the value of the key entry was an array reference, or a hash list if the value of the key entry was a hash reference, or finally the value of the key entry. If this non existing routine that was called is actually defined, the routine will be redeclared and the arguments passed to it. If this fails too, it will try to check for an AutoLoadable file in "auto/PackageName/routine_name.al" If the filed exists, it will be required, the routine name linked into the package name space and finally called with the arguments. If the require process failed or if the AutoLoadable routine file did not exist, AUTOLOAD() will check if the special routine EXTRA_AUTOLOAD() exists in the module. If it does, it will call it and pass it the arguments. Otherwise, AUTOLOAD will die with a message explaining that the called routine did not exist and could not be found in the current class. SUPPORT METHODS Those methods are designed to be called from the package inheriting from Module::Generic to perform various function and speed up development. __create_class Provided with an object property name and an hash reference representing a dictionary and this will produce a dynamically created class/module. If a property *_class* exists in the dictionary, it will be used as the class/package name, otherwise a name will be derived from the calling object class and the object property name. For example, in your module : sub products { return( 'products', shift->_set_get_class( { name => { type => 'scalar' }, customer => { type => 'object', class => 'My::Customer' }, orders => { type => 'array_as_object' }, active => { type => 'boolean' }, created => { type => 'datetime' }, metadata => { type => 'hash' }, stock => { type => 'number' }, url => { type => 'uri' }, }, @_ ) ); } Then calling your module method products such as : my $prod = $object->products({ name => 'Cool product', customer => { first_name => 'John', last_name => 'Doe', email => 'john.doe@example.com' }, orders => [qw( 123 987 456 654 )], active => 1, metadata => { transaction_id => 123, api_call_id => 456 }, stock => 10, uri => 'https://example.com/p/20' }); Using the resulting object $prod, we can access this dynamically created class/module such as : printf( <name, $prod->orders->length, $prod->customer->last_name,, $prod->url->path ) Product name: %s No of orders: %d Customer name: %s Product page path: %s EOT __instantiate_object Provided with an object property name, and a class/package name, this will attempt to load the module if it is not already loaded. It does so using "load_class" in Class::Load. Once loaded, it will init an object passing it the other arguments received. It returns the object instantiated upon success or undef and sets an "error" This is a support method used by "_instantiate_object" _instantiate_object This does the same thing as "__instantiate_object" and the purpose is for this method to be potentially superseded in your own module. In your own module, you would call "__instantiate_object" _can Provided with a value and a method name, and this will return true if the value provided is an object that "can" in UNIVERSAL perform the method specified, or false otherwise. You can also provide an array of method names to check instead of just a method name. In that case, all method names provided must be supported by the object otherwise it will return false. This makes it more convenient to write: if( $self->_can( $obj, 'some_method' ) ) { # ... } or if( $self->_can( $obj, [qw(some_method other_method )] ) ) { # ... } than to write: if( Scalar::Util::bless( $obj ) && $obj->can( 'some_method' ) { # ... } _get_args_as_array Provided with arguments and this support method will return the arguments provided as an array reference irrespective of whether they were initially provided as array reference or a simple array. For example: my $array = $self->_get_args_as_array(qw( those are arguments )); # returns an array reference containing: 'those', 'are', 'arguments' my $array = $self->_get_args_as_array( [qw( those are arguments )] ); # same result as previous example my $array = $self->_get_args_as_array(); # no args provided # returns an empty array reference _get_args_as_hash Provided with arguments and this support method will return the arguments provided as hash reference irrespective of whether they were initially provided as hash reference or a simple hash. In list context, this returns an hash reference and an array reference containing the order of the properties provided. For example: my $ref = $self->_get_args_as_hash( first => 'John', last => 'Doe' ); # returns hash reference { first => 'John', last => 'Doe' } my $ref = $self->_get_args_as_hash({ first => 'John', last => 'Doe' }); # same result as previous example my $ref = $self->_get_args_as_hash(); # no args provided # returns an empty hash reference my( $ref, $keys ) = $self->_get_args_as_hash( first => 'John', last => 'Doe' ); In the last example, $keys is an array object containing the list of properties passed an in the order they were provided, i.e. "first" and "last". If the properties were provided as an hash reference, the $keys returned will be the sorted list of properties, such as: my( $ref, $keys ) = $self->_get_args_as_hash({ last => 'Doe', first => 'John' }); Here, $keys will be sorted and contain the properties in their alphabetical order. However, this will return empty: my $ref = $self->_get_args_as_hash( { age => 42, city => 'Tokyo' }, some_other => 'parameter' ); This returns an empty hash reference, because although the first parameter is an hash reference, there is more than one parameter. As of version v0.24.0, this utility method allows for more advanced use and permits embedding parameters among arguments, remove them from the list and return them. For example: Assuming @_ contains: "foo bar debug 4 baz" my $ref = $self->_get_args_as_hash( @_, args_list => [qw( debug )] ); This will set $ref with "debug" only. Even the special parameter "args_list" does not have to be at the end and could be anywhere: my $ref = $self->_get_args_as_hash( 'foo', 'bar', args_list => [qw( debug )], 'debug', 4, 'baz' ); If you want to modify @_,because you need its content without any params, pass @_ as an array reference. my $ref = $self->_get_args_as_hash( \@_, args_list => [qw( debug )] ); say "@_"; $ref is an hash reference that would contain "debug" and @_ only contains "foo bar baz" You can also simply pass @_ as a reference to simply save memory. Assuming @_ is "foo bar baz 3 debug 4" my $ref = $self->_get_args_as_hash( \@_ ); This would set $ref to be an hash reference with keys "foo baz debug" _get_stack_trace This will return a Devel::StackTrace object initiated with the following options set: *indent* 1 This will set an initial indent tab *skip_frames* 1 This is set to 1 so this very method is not included in the frames stack _has_base64 Provided with a value and this returns an array reference containing 2 code references: one for encoding and one for decoding. Value provided can be a simple true value, such as 1, and then "_has_base64" will check if Crypt::Misc and MIME::Base64 are installed on the system and will use in priority MIME::Base64 The value provided can also be an array reference already containing 2 code references, and in such case, that value is simply returned. Nothing more is done. Finally, the value provided can be a module class name. "_has_base64" knows only of Crypt::Misc and MIME::Base64, so if you want to use any other one, arrange yourself to pass to "_has_base64" an array reference of 2 code references as explained above. _implement_freeze_thaw Provided with a list of package names and this method will implement in each of them the subroutines necessary to handle Storable::Improved (or the legacy Storable), CBOR and Sereal serialisation. In effect, it will check that the subroutines "FREEZE", "THAW", "STORABLE_freeze" and "STORABLE_thaw" exists or sets up simple ones if they are not defined. This works for packages that use hash-based objects. However, you need to make sure there is no specific package requirements, and if there is, you might need to customise those subroutines by yourself. _is_a Provided with an object and a package name and this will return true if the object is a blessed object from this package name (or a sub package of it), or false if not. The value of this is to reduce the burden of having to check whether the object actually exists, i.e. is not null or undef, if it is an object and if it is from that class. This allows to do it in just one method call like this: if( $self->_is_a( $obj, 'My::Package' ) ) { # Do something } Of course, if you are sure the object is actually an object, then you can directly do: if( $obj->isa( 'My::Package' ) ) { # Do something } _is_class_loadable Takes a module name and an optional version number and this will check if the module exist and can be loaded by looking at the @INC and using version to compare required version and existing version. It returns true if the module can be loaded or false otherwise. _is_class_loaded Provided with a class/package name, this returns true if the module is already loaded or false otherwise. It performs this test by checking if the module is already in %INC. _is_array Provided with some data, this checks if the data is of type array, even if it is an object. This uses "reftype" in Scalar::Util to achieve that purpose. So for example, an object such as : package My::Module; sub new { return( bless( [] => ( ref( $_[0] ) || $_[0] ) ) ); } This would produce an object like : My::Module=ARRAY(0x7f8f3b035c20) When checked with "_is_array" this, would return true just like an ordinary array. If you would use : ref( $object ); It would rather return the module package name: "My::Module" _is_class_loadable Provided with a package name, a.k.a. a class, and an optional version and this will endeavour to check if that class is installed and if a version is provided, if it is greater or equal to the version provided. If the module is not already loaded and a version was provided, it uses Module::Metadata to get that module version. It returns true if the module can be loaded or false otherwise. If an error occurred, it sets an error and returns "undef", so be sure to check whether the return value is defined. _is_class_loaded Provided with a package name, a.k.a. a class, and this returns true if the class has already been loaded or false otherwise. If you are running under mod_perl, this method will use "loaded" in Apache2::Module to find out, otherwise, it will simply check if the class exists in %INC _is_code Provided with some value, possibly, undefined, and this returns true if it is a "CODE", such as a subroutine reference or an anonymous subroutine, or false otherwise. _is_glob Provided with some value, possibly, undefined, and this returns true if it is a filehandle, or false otherwise. _is_hash Same as "_is_array", but for hash reference. _is_integer Returns true if the value provided is an integer, or false otherwise. A valid value includes an integer starting with "+" or "-" _is_ip Returns true if the given IP has a syntax compliant with IPv4 or IPv6 including CIDR notation or not, false otherwise. For this method to work, you need to have installed Regexp::Common::net _is_number Returns true if the provided value looks like a number, false otherwise. _is_object Provided with some data, this checks if the data is an object. It uses "blessed" in Scalar::Util to achieve that purpose. _is_scalar Provided with some data, this checks if the data is of type scalar reference, e.g. "SCALAR(0x7fc0d3b7cea0)", even if it is an object. _is_uuid Provided with a non-zero length value and this will check if it looks like a valid "UUID", i.e. a unique universal ID, and upon successful validation will set the value and return its representation as a Module::Generic::Scalar object. An empty string or "undef" can be provided and will not be checked. _load_class $self->_load_class( 'My::Module' ) || die( $self->error ); $self->_load_class( 'My::Module', qw( :some_tags SOME_CONSTANTS_TO_IMPORT ) ) || die( $self->error ); $self->_load_class( 'My::Module', qw( :some_tags SOME_CONSTANTS_TO_IMPORT ), { version => 'v1.2.3', caller => 'Its::Me' } ) || die( $self->error ); $self->_load_class( 'My::Module', { no_import => 1 } ) || die( $self->error ); Provided with a class/package name, some optional list of semantics to import, and, as the last parameter, an optional hash reference of options and this will attempt to load the module. This uses "use" in perlfunc, no external module. Upon success, it returns the package name loaded. It traps any error with an eval and return "undef" in perlfunc if an error occurred and sets an "error" accordingly. Possible options are: *caller* The package name of the caller. If this is not provided, it will default to the value provided with "caller" in perlfunc *no_import* Set to a true value and this will prevent the loaded module from importing anything into your namespace. This is the equivalent of doing: use My::Module (); *version* The minimum version for this class to load. This value is passed directly to "use" in perlfunc _load_classes This will load multiple classes by providing it an array reference of class name to load and an optional hash or hash reference of options, similar to those provided to "_load_class" If one of those classes failed to load, it will return immediately after setting an "error". _lvalue This provides a generic lvalue method that can be used both in assign context or lvalue context. As of version 0.29.6, this is an alias for "_set_get_callback", which provides more extensive features. _obj2h This ensures the module object is an hash reference, such as when the module object is based on a file handle for example. This permits Module::Generic to work no matter what is the underlying data type blessed into an object. _parse_timestamp Provided with a string representing a date or datetime, and this will try to parse it and return a DateTime object. It will also create a DateTime::Format::Strptime to preserve the original date/datetime string representation and assign it to the DateTime object. So when the DateTime object is stringified, it displays the same string that was originally parsed. Supported formats are: "2019-10-03 19-44+0000" or "2019-10-03 19:44:01+0000" Found in GNU PO files for example. "2019-06-19 23:23:57.000000000+0900" Found in PostgreSQL "2019-06-20T11:08:27" Matching ISO8601 format "2019-06-20 02:03:14" Found in SQLite "2019-06-20 11:04:01" Found in MySQL "Sun, 06 Oct 2019 06:41:11 GMT" Standard HTTP dates "12 March 2001 17:07:30 JST" "12-March-2001 17:07:30 JST" "12/March/2001 17:07:30 JST" "12 March 2001 17:07" "12 March 2001 17:07 JST" "12 March 2001 17:07:30+0900" "12 March 2001 17:07:30 +0900" "Monday, 12 March 2001 17:07:30 JST" "Monday, 12 Mar 2001 17:07:30 JST" "03/Feb/1994:00:00:00 0000" "2019-06-20" "2019/06/20" 2016.04.22 "2014, Feb 17" "17 Feb, 2014" "February 17, 2009" "15 July 2021" 22.04.2016 "22-04-2016" "17. 3. 2018." "17.III.2020" "17. III. 2018." 20030613 "2021年7月14日" Japanese regular date using occidental years "令和3年7月14日" Japanese regular date using Japanese era years Unix timestamp possibly followed by a dot and milliseconds Relative date to current date and time Example: -5Y - 5 years +2M + 2 months +3D + 3 days -2h - 2 hours -4m - 4 minutes -10s - 10 seconds 'now' The word now will set the return value to the current date and time _set_get sub name { return( shift->_set_get( 'name', @_ ) ); } Provided with an object property name and some value and this will set or get that value for that property. However, if the value stored is an array and is called in list context, it will return the array as a list and not the array reference. Same thing for an hash reference. It will return an hash in list context. In scalar context, it returns whatever the value is, such as array reference, hash reference or string, etc. _set_get_array Provided with an object property name and some data and this will store the data as an array reference. It returns the current value stored, such as an array reference notwithstanding it is called in list or scalar context. Example : sub products { return( shift->_set_get_array( 'products', @_ ) ); } _set_get_array_as_object Provided with an object property name and some data and this will store the data as an object of Module::Generic::Array If this is called with no data set, an object is created with no data inside and returned Example : # In your module sub products { return( shift->_set_get_array_as_object( 'products', @_ ) ); } And using your method: printf( "There are %d products\n", $object->products->length ); $object->products->push( $new_product ); Alternatively, you can pass an hash reference instead of an object property to provide callbacks that will be called upon addition or removal of value. This hash reference can contain the following properties: field The object property name callbacks An hash reference of operation type ("add" or "remove") to callback subroutine name or code reference pairs. For example: sub children { return( shift->set_get_array_as_object({ field => 'children', callbacks => { add => '_some_add_callback', remove => 'som_remove_callback', }, }), @_ ); } The value of the callback can be either a subroutine name or a code reference. _set_get_boolean sub is_true { return( shift->_set_get_boolean( 'is_true', @_ ) ); } Provided with an object property name and some data and this will store the data as a boolean value. If the data provided is a JSON::PP::Boolean or Module::Generic::Boolean object, the data is stored as is. If the data is a scalar reference, its referenced value is check and "true" in Module::Generic::Boolean or "false" in Module::Generic::Boolean is set accordingly. If the data is a string with value of "true" or "val" "true" in Module::Generic::Boolean or "false" in Module::Generic::Boolean is set accordingly. Otherwise the data provided is checked if it is a true value or not and "true" in Module::Generic::Boolean or "false" in Module::Generic::Boolean is set accordingly. If no value is provided, and the object property has already been set, this performs the same checks as above and returns either a JSON::PP::Boolean or a Module::Generic::Boolean object. Alternatively, you can pass an hash reference instead of an object property to provide callbacks that will be called upon addition or removal of value. This hash reference can contain the following properties: field The object property name callbacks An hash reference of operation type ("add" or "remove") to callback subroutine name or code reference pairs. For example: sub is_valid { return( shift->set_get_boolean({ field => 'is_valid', callbacks => { add => '_some_add_callback', remove => 'som_remove_callback', }, }), @_ ); } The value of the callback can be either a subroutine name or a code reference. _set_get_callback sub name : lvalue { return( shift->_set_get_callback({ get => sub { my $self = shift( @_ ); # The context hash is available with $_ if( $_->{list} ) { return( @{$self->{name}} ); } else { return( $self->{name} ); } }, set => sub { my $self = shift( @_ ); $self->message( 1, "Got here for 'name' in setter callback" ); return( $self->{name} = shift( @_ ) ); }, field => 'name' }, @_ ) ); } # ^^^^ # Don't forget the @_ ! Then, it can be called indifferently as: my $rv = $obj->name( 'John' ); # $rv is John $rv = $obj->name; # $rv is John $obj->name = 'Peter'; $rv = $obj->name; # $rv is Peter $obj->colours( qw( orange blue ) ); my @colours = $obj->colours; # returns a list of colours orange and blue my $colour = $obj->colours; # $colour is 'orange' Given an hash reference of parameters, and this support method will call the accessor "get" callback or mutator "set" callback depending on whether any arguments were provided. This support method supports "lvalue" methods as described in "Lvalue subroutines" in perlfunc It is similar as Sentinel, but on steroid, since it handles exception, and provides context, which is often critical. If a fatal exception occurs in a callback, it is trapped using try-catch block and an error object is set and "undef" is returned. However if an error occurs while operating in an "lvalue" assigned context, such as: $obj->name = 'Peter'; Then, to check if there was an error, you could do: if( $obj->error ) { # Do something here } If the "fatal" option is set to true, then it would simply die instead. Supported options are: * "fatal" Boolean. If true, this will result in any exception becoming fatal and thus die. * "field" The name of the object field for which this helper method is used. This is optional. * "get" The accessor subroutine reference or anonymous subroutine that will handle retrieving data. This is a mandatory option and this support method will die if this is not provided. It will be passed the current object, and return whatever is returned in list context, or in any other context, the first value that this callback would return. Also the special variable $_ will be available and contain the call context. * "set" The mutator subroutine reference or anonymous subroutine that will handle storing data. This is an optional option. This means you can set only an accessor "get" callback without specifying a mutator "set" callback. It will be passed the current object, and the list of arguments. If the method is used as a regular method, as opposed to an lvalue subroutine, then multiple arguments may be passed: $obj->colours( qw( blue orange ) ); but, if used as an "lvalue" method, of course, only one argument will be available: $obj->name = 'John'; Also the special variable $_ will be available and contain the call context. The value returned is passed back to the caller. The "context" provided with the special variable $_ inside the callback may have the following properties: * "assign" This is true when the call context is an "lvalue" subroutine to which a value is being assigned, such as: $obj->name = 'John'; * "boolean" This is true when the call context is a boolean, such as: if( $obj->active ) { # Do something } * "code" This is true when the call context is a code reference, such as: $obj->my_callback->(); * "count" Contains the number of arguments expected by the caller. This is especially interesting when in list context. * "glob" This is true when the call context is a glob. * "hash" This is true when the call context is an hash reference, such as: $obj->meta({ client_id => 1234567 }); my $id = $obj->meta->{client_id}; * "list" This is true when the call context is a list, such as: my @colours = $obj->colours; * "lvalue" This is true when the call context is an "lvalue" subroutine, such as: $obj->name = 'John'; * "object" This is true when the call context is an object, such as: $obj->something->another_method(); * "refscalar" my $name = ${$obj->name}; * "rvalue" This is true when the call context is from the right-hand side. my $name = $obj->name; * "scalar" This is true when the call context is a scalar: my $name = $obj->name; say $name; # John * "void" This is true when the call context is void, such as: $obj->pointless(); See also Want for more on this context-rich information. _set_get_class Given an object property name, a dynamic class fiels definition hash (dictionary), and optional arguments, this special method will create perl packages on the fly by calling the support method "__create_class" For example, consider the following: #!/usr/local/bin/perl BEGIN { use strict; use Data::Dumper; }; { my $o = MyClass->new( debug => 3 ); $o->setup->age( 42 ); print( "Age is: ", $o->setup->age, "\n" ); print( "Setup object is: ", $o->setup, "\n" ); $o->setup->billing->interval( 'month' ); print( "Billing interval is: ", $o->setup->billing->interval, "\n" ); print( "Billing object is: ", $o->setup->billing, "\n" ); $o->setup->rgb( 255, 122, 100 ); print( "rgb: ", join( ', ', @{$o->setup->rgb} ), "\n" ); exit( 0 ); } package MyClass; BEGIN { use strict; use lib './lib'; use parent qw( Module::Generic ); }; sub setup { return( shift->_set_get_class( 'setup', { name => { type => 'scalar' }, age => { type => 'number' }, metadata => { type => 'hash' }, rgb => { type => 'array' }, url => { type => 'uri' }, online => { type => 'boolean' }, created => { type => 'datetime' }, billing => { type => 'class', definition => { interval => { type => 'scalar' }, frequency => { type => 'number' }, nickname => { type => 'scalar' }, }} }) ); } 1; __END__ This will yield: Age is: 42 Setup object is: MyClass::Setup=HASH(0x7fa805abcb20) Billing interval is: month Billing object is: MyClass::Setup::Billing=HASH(0x7fa804ec3f40) rgb: 255, 122, 100 The advantage of this over _set_get_hash_as_object is that here one controls what fields / method are supported and with which data type. _set_get_class_array Provided with an object property name, a dictionary to create a dynamic class with "__create_class" and an array reference of hash references and this will create an array of object, each one matching a set of data provided in the array reference. So for example, imagine you had a method such as below in your module : sub products { return( shift->_set_get_class_array( 'products', { name => { type => 'scalar' }, customer => { type => 'object', class => 'My::Customer' }, orders => { type => 'array_as_object' }, active => { type => 'boolean' }, created => { type => 'datetime' }, metadata => { type => 'hash' }, stock => { type => 'number' }, url => { type => 'uri' }, }, @_ ) ); } Then your script would call this method like this : $object->products([ { name => 'Cool product', customer => { first_name => 'John', last_name => 'Doe', email => 'john.doe@example.com' }, active => 1, stock => 10, created => '2020-04-12T07:10:30' }, { name => 'Awesome tool', customer => { first_name => 'Mary', last_name => 'Donald', email => 'm.donald@example.com' }, active => 1, stock => 15, created => '2020-05-12T15:20:10' }, ]); And this would store an array reference containing 2 objects with the above data. _set_get_code Provided with an object property name and some code reference and this stores and retrieve the current value. It returns under and set an error if the provided value is not a code reference. _set_get_datetime sub created_on { return( shift->_set_get_datetime( 'created_on', @_ ) ); } Provided with an object property name and asome date or datetime string and this will attempt to parse it and save it as a DateTime object. If the data is a 10 digits integer, this will treat it as a unix timestamp. Parsing also recognise special word such as "now" The created DateTime object is associated a DateTime::Format::Strptime object which enables the DateTime object to be stringified as a unix timestamp using local time stamp, whatever it is. Even if there is no value set, and this method is called in chain, it returns a Module::Generic::Null whose purpose is to enable chaining without doing anything meaningful. For example, assuming the property *created* of your object is not set yet, but in your script you call it like this: $object->created->iso8601 Of course, the value of "iso8601" will be empty since this is a fake method produced by Module::Generic::Null. The return value of a method should always be checked. _set_get_file sub file { return( shift->_set_get_file( 'file', @_ ) ); } Provided with an object property name and a file and this will store the given file as a Module::Generic::File object. It returns "undef" and set an error if the provided value is not a proper file. Note that the files does not need to exist and it can also be a directory or a symbolic link or any other file on the system. _set_get_glob sub handle { return( shift->_set_get_glob( 'handle', @_ ) ); } Provided with an object property name and a glob (file handle) and this will store the given glob. It returns "undef" and set an error if the provided value is not a glob. _set_get_hash sub metadata { return( shift->_set_get_hash( 'metadata', @_ ) ); } Provided with an object property name and an hash reference and this set the property name with this hash reference. You can even pass it an associative array, and it will be saved as a hash reference, such as : $object->metadata( transaction_id => 123, customer_id => 456 ); my $hash = $object->metadata; _set_get_hash_as_mix_object sub metadata { return( shift->_set_get_hash_as_mix_object( 'metadata', @_ ) ); } Provided with an object property name, and an optional hash reference and this returns a Module::Generic::Hash object, which allows to manipulate the hash just like any regular hash, but it provides on top object oriented method described in details in Module::Generic::Hash. This is different from "_set_get_hash_as_object" below whose keys and values are accessed as dynamic methods and method arguments. _set_get_hash_as_object Provided with an object property name, an optional class name and an hash reference and this does the same as in "_set_get_hash", except it will create a class/package dynamically with a method for each of the hash keys, so that you can call the hash keys as method. Also it does this recursively while handling looping, in which case, it will reuse the object previously created, and also it takes care of adapting the hash key to a proper field name, so something like "99more-options" would become "more_options". If the value itself is a hash, it processes it recursively transforming "99more-options" to a proper package name "MoreOptions" prepended by $class_name provided as argument or whatever upper package was used in recursion processing. For example in your module : sub metadata { return( shift->_set_get_hash_as_object( 'metadata', @_ ) ); } Then populating the data : $object->metadata({ first_name => 'John', last_name => 'Doe', email => 'john.doe@example.com', }); printf( "Customer name is %s\n", $object->metadata->last_name ); _set_get_ip sub ip { return( shift->_set_get_ip( 'ip', @_ ) ); } This helper method takes a value and check if it is a valid IP address using "_is_ip". If "undef" or zero-byte value is provided, it will merely accept it, as it can be used to reset the value by the caller. If a value is successfully set, it returns a Module::Generic::Scalar object representing the string passed. From there you can pass the result to Net::IP in your own code, assuming you have that module installed. _set_get_lvalue This is now an alias for "_set_get_callback" _set_get_number Provided with an object property name and a number, and this will create a Module::Generic::Number object and return it. As of version v0.13.0 it also works as a lvalue method. See perlsub In your module: package MyObject; use parent qw( Module::Generic ); sub level : lvalue { return( shift->_set_get_number( 'level', @_ ) ); } In the script using module "MyObject": my $obj = MyObject->new; $obj->level = 3; # level is now 3 # or $obj->level( 4 ) # level is now 4 print( "Level is: ", $obj->level, "\n" ); # Level is 4 print( "Is it an odd number: ", $obj->level->is_odd ? 'yes' : 'no', "\n" ); # Is it an od number: no $obj->level++; # level is now 5 Alternatively, you can pass an hash reference instead of an object property to provide callbacks that will be called upon addition or removal of value. This hash reference can contain the following properties: * "callbacks" An hash reference of operation type ("add" or "remove") to callback subroutine name or code reference pairs. * "field" The object property name * "undef_ok" If this is set to a true value, this support method will allow undef to be set. Default to false, which means an undefined value passed will be ignored. For example: sub length { return( shift->set_get_number({ field => 'length', callbacks => { add => '_some_add_callback', remove => 'som_remove_callback', }, }), @_ ); } The value of the callback can be either a subroutine name or a code reference. _set_get_number_or_object Provided with an object property name and a number or an object and this call the value using "_set_get_number" or "_set_get_object" respectively _set_get_object sub myobject { return( shift->_set_get_object({ field => 'myobject', no_init => 1 }, My::Class, @_ ) ); } sub myobject { return( shift->_set_get_object( 'myobject', My::Class, @_ ) ); } Provided with an object property name, a class/package name and some data and this will initiate a new object of the given class passing it the data. The property name can also be an hash reference that will be used to provide more granular settings: "field" The actual property name "no_init" Boolean that, when set, instruct to not instantiate a class object if one is not instantiated yet. If you pass an undefined value, it will set the property as undefined, removing whatever was set before. You can also provide an existing object of the given class. "_set_get_object" will check the object provided does belong to the specified class or it will set an error and return undef. It returns the object currently set, if any. _set_get_object_lvalue Same as "_set_get_object_without_init" but with the possibility of setting the object value as an lvalue method: $o->my_property = $my_object; _set_get_object_without_init Sets or gets an object, but countrary to "_set_get_object" this method will not try to instantiate the object. _set_get_object_array2 Provided with an object property name, a class/package name and some array reference itself containing array references each containing hash references or objects, and this will create an array of array of objects. _set_get_object_array Provided with an object property name and a class/package name and similar to "_set_get_object_array2" this will create an array reference of objects. _set_get_object_array_object Provided with an object property name, a class/package name and some data and this will create an array of object similar to "_set_get_object_array", except the array produced is a Module::Generic::Array _set_get_object_variant Provided with an object property name, a class/package name and some data, and depending whether the data provided is an hash reference or an array reference, this will either instantiate an object for the given hash reference or an array of objects with the hash references in the given array. This means the value stored for the object property will vary between an hash or array reference. _set_get_scalar sub name { return( shift->_set_get_scalar( 'name', @_ ) ); } Provided with an object property name, and a string, possibly a number or anything really and this will set the property value accordingly. Very straightforward. It returns the currently value stored. _set_get_scalar_as_object Provided with an object property name, and a string or a scalar reference and this stores it as an object of Module::Generic::Scalar If there is already an object set for this property, the value provided will be assigned to it using "set" in Module::Generic::Scalar If it is called and not value is set yet, this will instantiate a Module::Generic::Scalar object with no value. So a call to this method can safely be chained to access the Module::Generic::Scalar methods. For example : sub name { return( shift->_set_get_scalar_as_object( 'name', @_ ) ); } Then, calling it : $object->name( 'John Doe' ); Getting the value : my $cust_name = $object->name; print( "Nothing set yet.\n" ) if( !$cust_name->length ); Alternatively, you can pass an hash reference instead of an object property to provide callbacks that will be called upon addition or removal of value. This hash reference can contain the following properties: field The object property name callbacks An hash reference of operation type ("add" or "remove") to callback subroutine name or code reference pairs. For example: sub name { return( shift->set_get_scalar_as_object({ field => 'name', callbacks => { add => '_some_add_callback', remove => 'som_remove_callback', }, }), @_ ); } The value of the callback can be either a subroutine name or a code reference. _set_get_scalar_or_object Provided with an object property name, and a class/package name and this stores the value as an object calling "_set_get_object" if the value is an object of class *class* or as a string calling "_set_get_scalar" If no value has been set yet, this returns a Module::Generic::Null object to enable chaining. _set_get_uri sub uri { return( shift->_set_get_uri( 'uri', @_ ) ); } sub uri { return( shift->_set_get_uri( { field => 'uri', class => 'URI::Fast' }, @_ ) ); } Provided with an object property name, and an uri and this creates an URI object and sets the property value accordingly. Alternatively, the property name can be an hash with the following properties: *field* The object property name *class* The URI class to use. By default, URI, but you could also use URI::Fast, or other class of your choice. That class will be loaded, if it is not loaded already. It accepts an URI object (or any other URI class object), an uri or urn string, or an absolute path, i.e. a string starting with "/". It returns the current value, if any, so the return value could be undef, thus it cannot be chained. Maybe it should return a Module::Generic::Null object ? _set_get_uuid Provided with an object, a property name, and an UUID (Universal Unique Identifier) and this stores it as an object of Module::Generic::Scalar. If an empty or undefined value is provided, it will be stored as is. However, if there is no value and this method is called in object context, such as in chaining, this will return a special Module::Generic::Null object that prevents perl error that whatever method follows was called on an undefined value. _set_get_version sub version { return( shift->_set_get_version( 'version', @_ ) ); } # or sub version : lvalue { return( shift->_set_get_version( 'version', @_ ) ); } # or sub version : lvalue { return( shift->_set_get_version( { field => 'version', class => 'Perl::Version' }, @_ ) ); } Provided with an object, a property name, and a version string and this stores it as an object of version by default. Alternatively, the property name can be an hash with the following properties: *field* The object property name *class* The version class to use. By default, version, but you could also use Perl::Version, or other class of your choice. That class will be loaded, if it is not loaded already. The value can also be assigned as an lvalue. Assuming you have a method "version" that implements "_set_get_version": $obj->version = $version; would work, but of course also: $obj->version( $version ); The value can be a legitimate version string, or a version object matching the "class" to be used, which is by default version. If it is a string, it will be made an object of the class specified using "parse" if that class supports it, or by simply calling "new". When called in get mode, it will convert any value pre-set, if any, into a version object of the specified class if the value is not an object of that class already, and return it, or else it will return an empty string or undef whatever you will have set in your object for this property. _to_array_object Provided with arguments or not, and this will return a Module::Generic::Array object of those data. my $array = $self->_to_array_object( qw( Hello world ) ); # Becomes an array object of 'Hello' and 'world' my $array = $self->_to_array_object( [qw( Hello world )] ); # Becomes an array object of 'Hello' and 'world' _warnings_is_enabled Called with the class object or providing another class object as argument, and this returns true if warnings are enabled for the given class, false otherwise. Example: $self->_warnings_is_enabled(); # Providing another class object $self->_warnings_is_enabled( $other_object ); _warnings_is_registered Called with the class object or providing another class object as argument, and this returns true if warnings are registered for the given class, false otherwise. This is useful, because calling "warnings::enabled()" to check if warnings are enabled for a given class when that class has not registered for warnings using the pragma "use warnings::register" will produce an error "Unknown warnings category". Example: $self->_warnings_is_registered(); # Providing another class object $self->_warnings_is_registered( $other_object ); __dbh if your module has the global variables "DB_DSN", this will create a database handler using DBI It will also use the following global variables in your module to set the database object: "DB_RAISE_ERROR", "DB_AUTO_COMMIT", "DB_PRINT_ERROR", "DB_SHOW_ERROR_STATEMENT", "DB_CLIENT_ENCODING", "DB_SERVER_PREPARE" If "DB_SERVER_PREPARE" is provided and true, "pg_server_prepare" will be set to true in the database handler. It returns the database handler object. DEBUG Return the value of your global variable *DEBUG*, if any. VERBOSE Return the value of your global variable *VERBOSE*, if any. ERROR & EXCEPTION HANDLING This module has been developed on the idea that only the main part of the application should control the flow and trigger exit. Thus, this module and all the others in this distribution do not die, but rather set and error and return undef. So you should always check for the return value. Error triggered are transformed into an Module::Generic::Exception object, or any exception class that is specified by the object property "_exception_class". For example: sub init { my $self = shift( @_ ); $self->SUPER::init( @_ ) || return( $self->pass_error ); $self->{_exception_class} = 'My::Exception'; return( $self ); } Those error objects can then be retrieved by calling "error" If, however, you wanted errors triggered to be fatal, you can set the object property "fatal" to a true value and/or set your package global variable $FATAL_ERROR to true. When "error" is called with an error, it will "die" in perlfunc with the error object rather than merely returning "undef". For example: package My::Module; BEGIN { use strict; use warnings; use parent qw( Module::Generic ); our $VERSION = 'v0.1.0'; our $FATAL_ERROR = 1; }; sub init { my $self = shift( @_ ); $self->{fatal} = 1; $self->SUPER::init( @_ ) || return( $self->pass_error ); $self->{_exception_class} = 'My::Exception'; return( $self ); } To catch fatal error you can use a "try-catch" block such as implemented by Nice::Try. Since perl version 5.33.7 you can use the try-catch block using an experimental feature "use feature 'try';", but this does not support "catch" by exception class. SERIALISATION The modules in the Module::Generic distribution all supports Storable::Improved (or the legacy Storable), Sereal and CBOR serialisation, by implementing the methods "FREEZE", "THAW", "STORABLE_freeze", "STORABLE_thaw" Even the IO modules like Module::Generic::File::IO and Module::Generic::Scalar::IO can be serialised and deserialised if the methods "FREEZE" and "THAW" are used. By design the methods "STORABLE_freeze" and "STORABLE_thaw" are not implemented in those modules because it would trigger a Storable exception "Unexpected object type (8) in store_hook()". Instead it is strongly encouraged you use the improved Storable::Improved which addresses and mitigate those issues. For serialisation with Sereal, make sure to instantiate the Sereal encoder with the "freeze_callbacks" option set to true, otherwise, "Sereal" will not use the "FREEZE" and "THAW" methods. See "FREEZE/THAW CALLBACK MECHANISM" in Sereal::Encoder for more information. For CBOR, it is recommended to use the option "allow_sharing" to enable the reuse of references, such as: my $cbor = CBOR::XS->new->allow_sharing; Also, if you use the option "allow_tags" with JSON, then all of those modules will work too, since this option enables support for the "FREEZE" and "THAW" methods. SEE ALSO Module::Generic::Exception, Module::Generic::Array, Module::Generic::Scalar, Module::Generic::Boolean, Module::Generic::Number, Module::Generic::Null, Module::Generic::Dynamic and Module::Generic::Tie, Module::Generic::File, Module::Generic::Finfo, Module::Generic::SharedMem, Module::Generic::Scalar::IO Number::Format, Class::Load, Scalar::Util AUTHOR Jacques Deguest COPYRIGHT & LICENSE Copyright (c) 2000-2020 DEGUEST Pte. Ltd. You can use, copy, modify and redistribute this package and associated files under the same terms as Perl itself.