SYNOPSIS
use Data::ModeMerge;
my $hash1 = { a=>1, c=>1, d=>{ da =>[1]} };
my $hash2 = { a=>2, "-c"=>2, d=>{"+da"=>[2]} };
# if you want Data::ModeMerge to behave like many other merging
# modules (e.g. Hash::Merge or Data::Merger), turn off modes
# (prefix) parsing and options key parsing.
my $mm = Data::ModeMerge->new(config => {parse_prefix=>0, options_key=>undef});
my $res = $mm->merge($hash1, $hash2);
die $res->{error} if $res->{error};
# $res->{result} -> { a=>2, c=>1, "-c"=>2, d=>{da=>[1], "+da"=>[2]} }
# otherwise Data::ModeMerge will parse prefix as well as options
# key
my $res = $mm->merge($hash1, $hash2);
die $res->{error} if $res->{error};
# $res->{result} -> { a=>2, c=>-1, d=>{da=>[1,2]} }
$res = $merge({ a =>1, { a2 =>1, ""=>{parse_prefix=>0}},
{".a"=>2, {".a2"=>2 }});
# $res->{result} -> { a=>12, {a2=>1, ".a2"=>2} }, parse_prefix is turned off in just the subhash
# procedural interface
my $res = mode_merge($hash1, $hash2, {allow_destroy_hash=>0});
DESCRIPTION
There are already several modules on CPAN to do recursive data
structure merging, like Data::Merger and Hash::Merge. Data::ModeMerge
differs in that it offers merging "modes" and "options". It provides
greater flexibility on what the result of a merge between two data
should/can be. This module may or may not be what you need.
One application of this module is in handling configuration. Often
there are multiple levels of configuration, e.g. in your typical Unix
command-line program there are system-wide config file in /etc,
per-user config file under ~/, and command-line options. It's
convenient programatically to load each of those in a hash and then
merge system-wide hash with the per-user hash, and then merge the
result with the command-line hash to get the a single hash as the final
configuration. Your program can from there on deal with this just one
hash instead of three.
In a typical merging process between two hashes (left-side and
right-side), when there is a conflicting key, then the right-side key
will override the left-side. This is usually the desired behaviour in
our said program as the system-wide config is there to provide
defaults, and the per-user config (and the command-line arguments)
allow a user to override those defaults.
But suppose that the user wants to unset a certain configuration
setting that is defined by the system-wide config? She can't do that
unless she edits the system-wide config (in which she might need admin
rights), or the program allows the user to disregard the system-wide
config. The latter is usually what's implemented by many Unix programs,
e.g. the -noconfig command-line option in mplayer. But this has two
drawbacks: a slightly added complexity in the program (need to provide
a special, extra comand-line option) and the user loses all the default
settings in the system-wide config. What she needed in the first place
was to just unset a single setting (a single key-value pair of the
hash).
Data::ModeMerge comes to the rescue. It provides a so-called DELETE
mode.
mode_merge({foo=>1, bar=>2}, {"!foo"=>undef, bar=>3, baz=>1});
will result ini:
{bar=>3, baz=>1}
The ! prefix tells Data::ModeMerge to do a DELETE mode merging. So the
final result will lack the foo key.
On the other hand, what if the system admin wants to protect a certain
configuration setting from being overriden by the user or the
command-line? This is useful in a hosting or other retrictive
environment where we want to limit users' freedom to some levels. This
is possible via the KEEP mode merging.
mode_merge({"^bar"=>2, "^baz"=>1}, {bar=>3, "!baz"=>0, qux=>7});
will result in:
{"^bar"=>2, "^baz"=>1, qux=>7}
effectively protecting bar and baz from being overriden/deleted/etc.
Aside from the two mentioned modes, there are also a few others
available by default: ADD (prefix +), CONCAT (prefix .), SUBTRACT
(prefix -), as well as the plain ol' NORMAL/override (optional prefix
*).
You can add other modes by writing a mode handler module.
You can change the default prefixes for each mode if you want. You can
disable each mode individually.
You can default to always using a certain mode, like the NORMAL mode,
and ignore all the prefixes, in which case Data::ModeMerge will behave
like most other merge modules.
There are a few other options like whether or not the right side is
allowed a "change the structure" of the left side (e.g. replacing a
scalar with an array/hash, destroying an existing array/hash with
scalar), maximum length of scalar/array/hash, etc.
You can change default mode, prefixes, disable/enable modes, etc on a
per-hash basis using the so-called options key. See the OPTIONS KEY
section for more details.
This module can handle (though not all possible cases)
circular/recursive references.
MERGING PREFIXES AND YOUR DATA
Merging with this module means you need to be careful when your hash
keys might contain one of the mode prefixes characters by accident,
because it will trigger the wrong merge mode and moreover the prefix
characters will be stripped from the final result (unless you configure
the module not to do so).
A rather common case is when you have regexes in your hash keys.
Regexes often begins with ^, which coincidentally is a prefix for the
KEEP mode. Or perhaps you have dot filenames as hash keys, where it
clashes with the CONCAT mode. Or perhaps shell wildcards, where * is
also used as the prefix for NORMAL mode.
To avoid clashes, you can either:
* exclude the keys using
exclude_merge/include_merge/exclude_parse/include_parse config
settings
* turn off some modes which you don't want via the disable_modes
config
* change the prefix for that mode so that it doesn't clash with your
data via the set_prefix config
* disable prefix parsing altogether via setting parse_prefix config
to 0
You can do this via the configuration, or on a per-hash basis, using
the options key.
See Data::ModeMerge::Config for more details on configuration.
OPTIONS KEY
Aside from merging mode prefixes, you also need to watch out if your
hash contains a "" (empty string) key, because by default this is the
key used for options key.
Options key are used to specify configuration on a per-hash basis.
If your hash keys might contain "" keys which are not meant to be an
options key, you can either:
* change the name of the key for options key, via setting options_key
config to another string.
* turn off options key mechanism, by setting options_key config to
undef.
See Data::ModeMerge::Config for more details about options key.
MERGING MODES
NORMAL (optional '*' prefix on left/right side)
mode_merge({ a =>11, b=>12}, { b =>22, c=>23}); # {a=>11, b=>22, c=>23}
mode_merge({"*a"=>11, b=>12}, {"*b"=>22, c=>23}); # {a=>11, b=>22, c=>23}
ADD ('+' prefix on the right side)
mode_merge({i=>3}, {"+i"=>4, "+j"=>1}); # {i=>7, j=>1}
mode_merge({a=>[1]}, {"+a"=>[2, 3]}); # {a=>[1, 2, 3]}
Additive merge on hashes will be treated like a normal merge.
CONCAT ('.' prefix on the right side)
mode_merge({i=>3}, {".i"=>4, ".j"=>1}); # {i=>34, j=>1}
Concative merge on arrays will be treated like additive merge.
SUBTRACT ('-' prefix on the right side)
mode_merge({i=>3}, {"-i"=>4}); # {i=>-1}
mode_merge({a=>["a","b","c"]}, {"-a"=>["b"]}); # {a=>["a","c"]}
Subtractive merge on hashes behaves like a normal merge, except that
each key on the right-side hash without any prefix will be assumed to
have a DELETE prefix, i.e.:
mode_merge({h=>{a=>1, b=>1}}, {-h=>{a=>2, "+b"=>2, c=>2}})
is equivalent to:
mode_merge({h=>{a=>1, b=>1}}, {h=>{"!a"=>2, "+b"=>2, "!c"=>2}})
and will merge to become:
{h=>{b=>3}}
DELETE ('!' prefix on the right side)
mode_merge({x=>WHATEVER}, {"!x"=>WHATEVER}); # {}
KEEP ('^' prefix on the left/right side)
If you add '^' prefix on the left side, it will be protected from being
replaced/deleted/etc.
mode_merge({'^x'=>WHATEVER1}, {"x"=>WHATEVER2}); # {x=>WHATEVER1}
For hashes, KEEP mode means that all keys on the left side will not be
replaced/modified/deleted, *but* you can still add more keys from the
right side hash.
mode_merge({a=>1, b=>2, c=>3},
{a=>4, '^c'=>1, d=>5},
{default_mode=>'KEEP'});
# {a=>1, b=>2, c=>3, d=>5}
Multiple prefixes on the right side is allowed, where the merging will
be done by precedence level (highest first):
mode_merge({a=>[1,2]}, {'-a'=>[1], '+a'=>[10]}); # {a=>[2,10]}
but not on the left side:
mode_merge({a=>1, '^a'=>2}, {a=>3}); # error!
Precedence levels (from highest to lowest):
KEEP
NORMAL
SUBTRACT
CONCAT ADD
DELETE
FUNCTIONS
mode_merge($l, $r[, $config_vars])
A non-OO wrapper for merge() method. Exported by default. See merge
method for more details.
ATTRIBUTES
config
A hashref for config. See Data::ModeMerge::Config.
modes
combine_rules
path
errors
mem
cur_mem_key
METHODS
For typical usage, you only need merge().
push_error($errmsg)
Used by mode handlers to push error when doing merge. End users
normally should not need this.
register_mode($name_or_package_or_obj)
Register a mode. Will die if mode with the same name already exists.
check_prefix($hash_key)
Check whether hash key has prefix for certain mode. Return the name of
the mode, or undef if no prefix is detected.
check_prefix_on_hash($hash)
This is like check_prefix but performed on every key of the specified
hash. Return true if any of the key contain a merge prefix.
add_prefix($hash_key, $mode)
Return hash key with added prefix with specified mode. Log merge error
if mode is unknown or is disabled.
remove_prefix($hash_key)
Return hash key will any prefix removed.
remove_prefix_on_hash($hash)
This is like remove_prefix but performed on every key of the specified
hash. Return the same hash but with prefixes removed.
merge($l, $r)
Merge two nested data structures. Returns the result hash: {
success=>0|1, error=>'...', result=>..., backup=>... }. The 'error' key
is set to contain an error message if there is an error. The merge
result is in the 'result' key. The 'backup' key contains replaced
elements from the original hash/array.
CREATING AND USING YOUR OWN MODE
Let's say you want to add a mode named FOO. It will have the prefix
'?'.
Create the mode handler class, e.g. Data::ModeMerge::Mode::FOO. It's
probably best to subclass from Data::ModeMerge::Mode::Base. The class
must implement name(), precedence_level(), default_prefix(),
default_prefix_re(), and
merge_{SCALAR,ARRAY,HASH}_{SCALAR,ARRAY,HASH}(). For more details, see
the source code of Base.pm and one of the mode handlers (e.g.
NORMAL.pm).
To use the mode, register it:
my $mm = Data::ModeMerge->new;
$mm->register_mode('FOO');
This will require Data::ModeMerge::Mode::FOO. After that, define the
operations against other modes:
# if there's FOO on the left and NORMAL on the right, what mode
# should the merge be done in (FOO), and what the mode should be
# after the merge? (NORMAL)
$mm->combine_rules->{"FOO+NORMAL"} = ["FOO", "NORMAL"];
# we don't define FOO+ADD
$mm->combine_rules->{"FOO+KEEP"} = ["KEEP", "KEEP"];
# and so on
FAQ
What is this module good for? Why would I want to use this module instead
of the other hash merge modules?
If you just need to (deeply) merge two hashes, chances are you do not
need this module. Use, for example, Hash::Merge, which is also flexible
enough because it allows you to set merging behaviour for merging
different types (e.g. SCALAR vs ARRAY).
You might need this module if your data is recursive/self-referencing
(which, last time I checked, is not handled well by Hash::Merge), or if
you want to be able to merge differently (i.e. apply different merging
modes) according to different prefixes on the key, or through special
key. In other words, you specify merging modes from inside the hash
itself.
I originally wrote Data::ModeMerge this for Data::Schema and
Config::Tree. I want to reuse the "parent" schema (or configuration) in
more ways other than just override conflicting keys. I also want to be
able to allow the parent to protect certain keys from being overriden.
I found these two features lacking in all merging modules that I've
evaluated prior to writing Data::ModeMerge.
SEE ALSO
Data::ModeMerge::Config
Other merging modules on CPAN: Data::Merger (from Data-Utilities),
Hash::Merge, Hash::Merge::Simple
Data::Schema and Config::Tree (among others, two modules which use
Data::ModeMerge)