A buff is a timed object, attached to a game entity. It is capable of modifying values, triggering code, or both.
It is a common design pattern in RPGs, particularly action games.
Features:
-`BuffHandler`: A buff handler to apply to your objects.
-`BaseBuff`: A buff class to extend from to create your own buffs.
-`BuffableProperty`: A sample property class to show how to automatically check modifiers.
-`CmdBuff`: A command which applies buffs.
-`samplebuffs.py`: Some sample buffs to learn from.
## Quick Start
Assign the handler to a property on the object, like so.
```python
@lazy_property
def buffs(self) -> BuffHandler:
return BuffHandler(self)
```
You may then call the handler to add or manipulate buffs like so: `object.buffs`. See **Using the Handler**.
### Customization
If you want to customize the handler, you can feed the constructor two arguments:
-`dbkey`: The string you wish to use as the attribute key for the buff database. Defaults to "buffs". This allows you to keep separate buff pools - for example, "buffs" and "perks".
-`autopause`: If you want this handler to automatically pause playtime buffs when its owning object is unpuppeted.
> **Note**: If you enable autopausing, you MUST initialize the property in your owning object's
> `at_init` hook. Otherwise, a hot reload can cause playtime buffs to not update properly
> on puppet/unpuppet. You have been warned!
Let's say you want another handler for an object, `perks`, which has a separate database and
respects playtime buffs. You'd assign this new property as so:
Call the handler's `add` method. This requires a class reference, and also contains a number of
optional arguments to customize the buff's duration, stacks, and so on. You can also store any arbitrary value
in the buff's cache by passing a dictionary through the `to_cache` optional argument. This will not overwrite the normal
values on the cache.
```python
self.buffs.add(StrengthBuff) # A single stack of StrengthBuff with normal duration
self.buffs.add(DexBuff, stacks=3, duration=60) # Three stacks of DexBuff, with a duration of 60 seconds
self.buffs.add(ReflectBuff, to_cache={'reflect': 0.5}) # A single stack of ReflectBuff, with an extra cache value
```
Two important attributes on the buff are checked when the buff is applied: `refresh` and `unique`.
-`refresh` (default: True) determines if a buff's timer is refreshed when it is reapplied.
-`unique` (default: True) determines if this buff is unique; that is, only one of it exists on the object.
The combination of these two booleans creates one of three kinds of keys:
-`Unique is True, Refresh is True/False`: The buff's default key.
-`Unique is False, Refresh is True`: The default key mixed with the applier's dbref. This makes the buff "unique-per-player", so you can refresh through reapplication.
-`Unique is False, Refresh is False`: The default key mixed with a randomized number.
### Get Buffs
The handler has several getter methods which return instanced buffs. You won't need to use these for basic functionality, but if you want to manipulate
buffs after application, they are very useful. The handler's `check`/`trigger` methods utilize some of these getters, while others are just for developer convenience.
`get(key)` is the most basic getter. It returns a single buff instance, or `None` if the buff doesn't exist on the handler. It is also the only getter
that returns a single buff instance, rather than a dictionary.
> **Note**: The handler method `has(buff)` allows you to check if a matching key (if a string) or buff class (if a class) is present on the handler cache, without actually instantiating the buff. You should use this method for basic "is this buff present?" checks.
Group getters, listed below, return a dictionary of values in the format `{buffkey: instance}`. If you want to iterate over all of these buffs,
you should do so via the `dict.values()` method.
-`get_all()` returns all buffs on this handler. You can also use the `handler.all` property.
-`get_by_type(BuffClass)` returns buffs of the specified type.
-`get_by_stat(stat)` returns buffs with a `Mod` object of the specified `stat` string in their `mods` list.
-`get_by_trigger(string)` returns buffs with the specified string in their `triggers` list.
-`get_by_source(Object)` returns buffs applied by the specified `source` object.
-`get_by_cachevalue(key, value)` returns buffs with the matching `key: value` pair in their cache. `value` is optional.
All group getters besides `get_all()` can "slice" an existing dictionary through the optional `to_filter` argument.
```python
dict1 = handler.get_by_type(Burned) # This finds all "Burned" buffs on the handler
dict2 = handler.get_by_source(self, to_filter=dict1) # This filters dict1 to find buffs with the matching source
```
> **Note**: Most of these getters also have an associated handler property. For example, `handler.effects` returns all buffs that can be triggered, which
> is then iterated over by the `get_by_trigger` method.
### Remove Buffs
There are also a number of remover methods. Generally speaking, these follow the same format as the getters.
-`remove(key)` removes the buff with the specified key.
-`clear()` removes all buffs.
-`remove_by_type(BuffClass)` removes buffs of the specified type.
-`remove_by_stat(stat)` removes buffs with a `Mod` object of the specified `stat` string in their `mods` list.
-`remove_by_trigger(string)` removes buffs with the specified string in their `triggers` list.
-`remove_by_source(Object)` removes buffs applied by the specified source
-`remove_by_cachevalue(key, value)` removes buffs with the matching `key: value` pair in their cache. `value` is optional.
You can also remove a buff by calling the instance's `remove` helper method. You can do this on the dictionaries returned by the
getters listed above.
```python
to_remove = handler.get_by_trigger(trigger) # Finds all buffs with the specified trigger
for buff in to_remove.values(): # Removes all buffs in the to_remove dictionary via helper methods
buff.remove()
```
### Check Modifiers
Call the handler `check(value, stat)` method when you want to see the modified value.
This will return the `value`, modified by any relevant buffs on the handler's owner (identified by
the `stat` string).
For example, let's say you want to modify how much damage you take. That might look something like this:
This method calls the `at_pre_check` and `at_post_check` methods at the relevant points in the process. You can use to this make
buffs that are reactive to being checked; for example, removing themselves, altering their values, or interacting with the game state.
> **Note**: You can also trigger relevant buffs at the same time as you check them by ensuring the optional argument `trigger` is True in the `check` method.
Multiply/divide modifiers in this buff system are additive by default. This means that two +50% modifiers will equal a +100% modifier. But what if you want to apply mods multiplicatively?
First, you should carefully consider if you truly want multiplicative modifiers. Here's some things to consider.
- They are unintuitive to the average user, as two +50% damage buffs equal +125% instead of +100%.
- They lead to "power explosion", where stacking buffs in the right way can turn characters into unstoppable forces
Doing purely-additive multipliers allows you to better control the balance of your game. Conversely, doing multiplicative multipliers enables very fun build-crafting where smart usage of buffs and skills can turn you into a one-shot powerhouse. Each has its place.
The best design practice for multiplicative buffs is to divide your multipliers into "tiers", where each tier is applied separately. You can easily do this with multiple `check` calls.
```python
damage = damage
damage = handler.check(damage, 'damage')
damage = handler.check(damage, 'empower')
damage = handler.check(damage, 'radiant')
damage = handler.check(damage, 'overpower')
```
#### Buff Strength Priority (Advanced)
Sometimes you only want to apply the strongest modifier to a stat. This is supported by the optional `strongest` bool arg in the handler's check method
Call the handler's `trigger(string)` method when you want an event call. This will call the `at_trigger` hook method on all buffs with the relevant trigger `string`.
For example, let's say you want to trigger a buff to "detonate" when you hit your target with an attack.
You'd write a buff that might look like this:
```python
class Detonate(BaseBuff):
...
triggers = ['take_damage']
def at_trigger(self, trigger, *args, **kwargs)
self.owner.take_damage(100)
self.remove()
```
And then call `handler.trigger('take_damage')` in the method you use to take damage.
> **Note** You could also do this through mods and `at_post_check` if you like, depending on how to want to add the damage.
### Ticking
Ticking buffs are slightly special. They are similar to trigger buffs in that they run code, but instead of
doing so on an event trigger, they do so on a periodic tick. A common use case for a buff like this is a poison,
or a heal over time.
```python
class Poison(BaseBuff):
...
tickrate = 5
def at_tick(self, initial=True, *args, **kwargs):
_dmg = self.dmg * self.stacks
if not initial:
self.owner.location.msg_contents(
"Poison courses through {actor}'s body, dealing {damage} damage.".format(
actor=self.owner.named, damage=_dmg
)
)
```
To make a buff ticking, ensure the `tickrate` is 1 or higher, and it has code in its `at_tick`
method. Once you add it to the handler, it starts ticking!
> **Note**: Ticking buffs always tick on initial application, when `initial` is `True`. If you don't want your hook to fire at that time,
> make sure to check the value of `initial` in your `at_tick` method.
### Context
Every important handler method optionally accepts a `context` dictionary.
Context is an important concept for this handler. Every method which checks, triggers, or ticks a buff passes this
dictionary (default: empty) to the buff hook methods as keyword arguments (`**kwargs`). It is used for nothing else. This allows you to make those
methods "event-aware" by storing relevant data in the dictionary you feed to the method.
For example, let's say you want a "thorns" buff which damages enemies that attack you. Let's take our `take_damage` method
There are two helper methods on the handler that allow you to get useful buff information back.
-`view`: Returns a dictionary of tuples in the format `{buffkey: (buff.name, buff.flavor)}`. Finds all buffs by default, but optionally accepts a dictionary of buffs to filter as well. Useful for basic buff readouts.
-`view_modifiers(stat)`: Returns a nested dictionary of information on modifiers that affect the specified stat. The first layer is the modifier type (`add/mult/div`) and the second layer is the value type (`total/strongest`). Does not return the buffs that cause these modifiers, just the modifiers themselves (akin to using `handler.check` but without actually modifying a value). Useful for stat sheets.
You can also create your own custom viewing methods through the various handler getters, which will always return the entire buff object.
-`ticknum`: How many ticks the buff has gone through
-`timeleft`: How much time is remaining on the buff
-`ticking`/`stacking`: If this buff ticks/stacks (checks `tickrate` and `maxstacks`)
#### Buff Cache (Advanced)
Buffs always store some useful mutable information about themselves in the cache (what is stored on the owning object's database attribute). A buff's cache corresponds to `{buffkey: buffcache}`, where `buffcache` is a dictionary containing __at least__ the information below:
-`ref` (class): The buff class path we use to construct the buff.
-`start` (float): The timestamp of when the buff was applied.
-`source` (Object): If specified; this allows you to track who or what applied the buff.
-`prevtick` (float): The timestamp of the previous tick.
-`duration` (float): The cached duration. This can vary from the class duration, depending on if the duration has been modified (paused, extended, shortened, etc).
-`tickrate` (float): The buff's tick rate. Cannot go below 0. Altering the tickrate on an applied buff will not cause it to start ticking if it wasn't ticking before. (`pause` and `unpause` to start/stop ticking on existing buffs)
Sometimes you will want to dynamically update a buff's cache at runtime, such as changing a tickrate in a hook method, or altering a buff's duration.
You can do so by using the interface `buff.cachekey`. As long as the attribute name matches a key in the cache dictionary, it will update the stored
cache with the new value.
If there is no matching key, it will do nothing. If you wish to add a new key to the cache, you must use the `buff.update_cache(dict)` method,
which will properly update the cache (including adding new keys) using the dictionary provided.
> **Example**: You want to increase a buff's duration by 30 seconds. You use `buff.duration += 30`. This new duration is now reflected on both the instance and the cache.
The buff cache can also store arbitrary information. To do so, pass a dictionary through the handler `add` method (`handler.add(BuffClass, to_cache=dict)`),
set the `cache` dictionary attribute on your buff class, or use the aforementioned `buff.update_cache(dict)` method.
> **Example**: You store `damage` as a value in the buff cache and use it for your poison buff. You want to increase it over time, so you use `buff.damage += 1` in the tick method.
Mods are stored in the `mods` list attribute. Buffs which have one or more Mod objects in them can modify stats. You can use the handler method to check all
mods of a specific stat string and apply their modifications to the value; however, you are encouraged to use `check` in a getter/setter, for easy access.
Mod objects consist of only four values, assigned by the constructor in this order:
-`stat`: The stat you want to modify. When `check` is called, this string is used to find all the mods that are to be collected.
-`mod`: The modifier. Defaults are `add` (addition/subtraction), `mult` (multiply), and `div` (divide). Modifiers are calculated additively (see `_calculate_mods` for more)
# Finds our "modgen" cache value, and generates a mod from it
modgen = list(self.cache.get("modgen"))
if modgen:
self.mods = [Mod(*modgen)]
```
### Triggers
Buffs which have one or more strings in the `triggers` attribute can be triggered by events.
When the handler's `trigger` method is called, it searches all buffs on the handler for any with a matchingtrigger,
then calls their `at_trigger` hooks. Buffs can have multiple triggers, and you can tell which trigger was used by
the `trigger` argument in the hook.
```python
class AmplifyBuff(BaseBuff):
triggers = ['damage', 'heal']
def at_trigger(self, trigger, **kwargs):
if trigger == 'damage': print('Damage trigger called!')
if trigger == 'heal': print('Heal trigger called!')
```
### Ticking
A buff which ticks isn't much different than one which triggers. You're still executing arbitrary hooks on
the buff class. To tick, the buff must have a `tickrate` of 1 or higher.
```python
class Poison(BaseBuff):
...
# this buff will tick 6 times between application and cleanup.
duration = 30
tickrate = 5
def at_tick(self, initial, **kwargs):
self.owner.take_damage(10)
```
> **Note**: The buff always ticks once when applied. For this **first tick only**, `initial` will be True in the `at_tick` hook method. `initial` will be False on subsequent ticks.
Ticks utilize a persistent delay, so they should be pickleable. As long as you are not adding new properties to your buff class, this shouldn't be a concern.
If you **are** adding new properties, try to ensure they do not end up with a circular code path to their object or handler, as this will cause pickling errors.
### Extras
Buffs have a grab-bag of extra functionality to let you add complexity to your designs.
#### Conditionals
You can restrict whether or not the buff will `check`, `trigger`, or `tick` through defining the `conditional` hook. As long
as it returns a "truthy" value, the buff will apply itself. This is useful for making buffs dependent on game state - for
example, if you want a buff that makes the player take more damage when they are on fire:
