ps2-census 0.13.0

Daybreak Games Census API client for Planetside 2 (UNOFFICIAL)

ps2-census

ps2-census is a low-level client for Daybreak's Planetside 2 Census API written in Python >= 3.8.

• ps2-census
  • Installation
  • Full examples
  • Query building
    • Basic query
    • Simple join
    • Nesting joins
      • Deeply nested join
      • Lateraly nested joins
    • Tree
  • Factories
  • Common enums
  • Event Stream
    • Usage
  • Next steps
  • Similar projects
  • Development
    • Environment
    • Technical details

Features:

• Build queries through method chaining
• Join collections and nest them through method chaining
• Get raw (deserialized) responses as Python dictionaries
• Access common enums directly
• Subscribe to event streams

By default the example service ID is used; however it is not recommended for production. You should get your own service ID from the webside below and supply it to the client whenever needed.

More information about the Census API is available on the official Census documentation here.

Iridar's blog is also recommended to understand the quirks of this particular API as this knowledge is necessary to use ps2-census properly.

Installation

pip install ps2-census

Full examples

Before diving in the details of the ps2-census client, full examples are available in the examples folder.

They currently include:

• all_about_trac_5.py (and the output in all_about_trac_5.json): building and executing a query that fetches pretty much everything that's to know about the TR TRAC-5 carbine
• character_death_event_stream.py (and the output in caracter_death_event_stream.ndjson): subscribing to all character death events on the SolTech server, receive at most 20 events and print them, then gracefuly disconnect

Query building

Queries are made on collections with optional joins on other collections as well as various commands that alter the output.

A Census API collection is analog to a relation in a generic relational database system, and a Census API join is analog to a join between these relations.

Basic query

To build a query, instantiate the Query class with a Collection (and your service ID, though it will be omitted in next examples for conciseness):

from ps2_census import Collection, Query

query: Query = Query(Collection.ITEM, service_id=YOUR_SERVICE_ID)

Chain methods to alter the query further:

query: Query = (
    Query(Collection.ITEM)
    .lang("en")
    .sort(("item_id", 1), ("faction_id", -1))
    .limit(30)
    [...]
)

Available methods are:

• filter(field: str, value: Union[str, int], modifier: Optional[SearchModifier]): filter the query on a field; ps2_census.SearchModifier contains all the modifiers made available by the Census API (SearchModifier.CONTAINS, SearchModifier.LESS_OR_EQUAL, ...)
• show(*args: str): only return the provided fields in results
• hide(*args: str): do not return the provided fields in results
• sort(*args: Tuple[str, Literal[1, -1]]): sort the results by field, either in increasing or decreasing order
• has(*args: str): only return results which have the specified fields
• case(arg: bool): whether filter()s are case sensive or not; default to True
• limit(arg: int): limit the return to at most arg results; required in tendem with start() for queries having too large of a result and therefore fail; defaults to 1
• limit_per_db(arg: int): limit the return to at most arg * databases count results; useful when querying the Collection.CHARACTER collection whose objects are randomly distributed among multiple databases in order to have more predictable results
• start(arg: int): start with the argth object within the results of the query
• include_null(arg: bool): whether to include keys with None values in results; defaults to False
• lang(arg: str): only keep the supplied language for internationalized strings
• join(arg: Join): perform a collection join; see the following documentation for additional details
• tree(arg: Tree): rearrange lists of data into trees; see the following documentation for additional details
• timing(arg: bool): show query timing information
• exact_match_first(arg: bool): when using filter()s with SearchModifiers, put exact matches at the top of the results list disregarding sort()s
• distinct(arg: str): get the distinct values for a certain field
• retry(arg: bool): retry queries at most one time before failing; defaults to True

Execute the query in one of the 2 ways made available by the Census API:

• .count() to get the items count

query.count()
> {'count': 21048}

• .get() to get the results

query.get()
> {'item_list': [{...}, {...}, ...], 'returned': 30}

count() and get() calls are when the query is actually sent to the Census API endpoint. They will raise status exceptions if appropriate.

Simple join

In order to perform joins instantiate the Join class with a Collection, add any additional chained methods to it, and pass it to the Query object via query.join():

from ps2_census import Collection, Join, Query

query: Query = (
    Query(Collection.ITEM)
    .join(
        Join(Collection.WEAPON_DATASHEET)
        .outer(0)
        .on("item_id")
        .to("item_id")
        .inject_at("weapon_datasheet")
        [...]
    )
)

Available Join methods are:

• nest(other: Join): nest another join within this one; see the following documentation for additional details
• on(arg: str): specify the field on this collection (the Query or parent Join) to join on; if not provided will default to this collection's ID ({this_type}_id)
• to(arg: str): specify the field on the other collection (the Join's) to join to; if not provided will default to the on value
• list(arg: Literal[0, 1]): whether the joined data is a list (and therefore will result in a list of objects) or not; 1 if it is a list, 0 if not; default to 0
• show(*args: str): only keep the provided fields in results
• hide(*args: str): do not keep the provided fields in results
• inject_at(arg: str): the field name where the joined data will be injected in the parent element (Query result item or parent Join element)
• terms(*args: Tuple[str, Union[str, int]]): filter the join result by conditions (eg. terms(("faction_id", 1), ("skill_set_id", 129)))
• outer(arg: Literam[0, 1]): whether the join will perform an outer join (include non-matches) of an inner join (exclude non-matches); 1 for outer, 0 for inner; defaults to 1

Multiple joins can be performed one after another on the same Query and the trees will be merged in the result:

query: Query = (
    Query(Collection.ITEM)
    .join(
        Join(Collection.WEAPON_DATASHEET)
    )
    .join(
        Join(Collection.ITEM_TO_WEAPON)
    )
)

Nesting joins

Joins can be nested one into another both deeply and lateraly. Nesting is done by reference: each Join instance contains a list of references to its nested Joins.

Note: be careful about modifying the Join instances after their creation if you assign them to locals for convenience, as shown in the following examples. As nesting is done by reference and the whole structure is "compiled" (into a querystring) on the call to query.join(), if any modification is done to an instance between its creation and the resolution, it will indirectly impact the result of any other parent and/or nested Join. Make use of the below described factories to make clean copies of your instances as necessary.

Deeply nested join

Deeply nested join are necessary in order to access data structures deeper in the collections tree. To deeply nest joins, instantiate the Join class multiple times and combine them through join1.nest(join2.nest(join3)) where join3 is nested within join2 and join2 is nested within join1:

from ps2_census import Collection, Join, Query

item_to_weapon_join: Join = (
    Join(Collection.ITEM_TO_WEAPON)
    .on("item_id")
    .to("item_id")
    .inject_at("item_to_weapon")
)

weapon_join: Join = (
    Join(Collection.WEAPON)
    .on("weapon_id")
    .to("weapon_id")
    .inject_at("weapon")
)

weapon_to_fire_group_join: Join = (
    Join(Collection.WEAPON_TO_FIRE_GROUP)
    .on("weapon_id")
    .to("weapon_id")
    .inject_at("weapon_to_fire_group")
)

query: Query = (
    Query(Collection.ITEM)
    .filter("item_type_id", ItemType.WEAPON)
    .join(
        item_to_weapon_join.nest(
            weapon_join.nest(
                weapon_to_fire_group_join
            )
        )
    )
)

Lateraly nested joins

Lateraly nested joins are necessary in order to access data structures at the same depth in the collections tree. To laterally nest joins, instantiate the Join class multiple times and combine them through join1.nest(join2).nest(join3) where join2 and join3 are nested within join1:

from ps2_census import Collection, Join, Query

fire_group_to_fire_mode_join: Join = (
    Join(Collection.FIRE_GROUP_TO_FIRE_MODE)
    .outer(0)
    .on("fire_group_id")
    .to("fire_group_id")
    .list(1)
    .inject_at("fire_group_to_fire_modes")
)

fire_mode_join: Join = (
    Join(Collection.FIRE_MODE_2)
    .outer(0)
    .on("fire_mode_id")
    .to("fire_mode_id")
    .inject_at("fire_mode")
)

fire_mode_to_projectile_join: Join = (
    Join(Collection.FIRE_MODE_TO_PROJECTILE)
    .outer(0)
    .on("fire_mode_id")
    .to("fire_mode_id")
    .inject_at("fire_mode_to_projectile")
)

player_state_group_join: Join = (
    Join(Collection.PLAYER_STATE_GROUP_2)
    .outer(0)
    .on("player_state_group_id")
    .to("player_state_group_id")
    .list(1)
    .inject_at("player_state_groups")
)

query: Query = (
    Query(Collection.WEAPON_TO_FIRE_GROUP)
    .join(
        fire_group_to_fire_mode_join
        .nest(
            fire_mode_join
            .nest(fire_mode_to_projectile_join)
            .nest(player_state_group_join)
        )
    )
)

Tree

Trees are also built using their own class, Tree, then passed to the Query object:

from ps2_census import Collection, Query, Tree

query: Query = (
    Query(Collection.ITEM)
    .tree(
        Tree("name.en")
        .prefix("en_name_")
        [...]
    )
)

This will return a dictionary of items with their english name prefixed by en_name_ as keys and the objects themselves as values, instead of a flat list of items.

Available methods are:

• list(arg: Literal[0, 1]): 0 if tree data is not a list, 1 if it is a list; defaults to 0
• prefix(arg: str): prefix to add to the field value
• start(arg: str): where the tree starts; defaults to the root (root list objects will be formatted as a tree)

Factories

In order to ease the project-wide definition of Query, Join and Tree objects, each of these classes has a get_factory() method.

Upon invocation it saves a copy of the current object and returns a callable that, upon each invocation, returns a fresh copy of that saved state.

query: Query = Query(Collection.ABILITY).has("someField").case(False)

query_factory: Callable[[], Query] = query.get_factory()

query_copy: Query = query_factory()

assert query == query_copy
> True

query = query.timing(True)

assert query != query_copy
> True

query_copy = query_copy.lang("en")

another_copy = query_factory()

assert query == another_copy
> True

assert query_copy != another_copy
> True

Factories obtained in this way are useful to define Query, Join or Trees somewhere in your code and store their factory callables, with the guarantee that you cannot modify the shared definition anywhere.

my_module.py

from ps2_census import Query, Collection

my_query_factory: Callable[[], Query] = (
    Query(Collection.ABILITY)
    .lang("en")
    .get_factory()
)

main.py

from my_module import my_query_factory

query: Query = my_query_factory()
query.start(100)  # does not modify any shared object

Common enums

Census API data uses a lot of integer enumerations that are collections themselves.

For example the faction_id key in items from Collection.ITEM is an integer that represents a specific fation, refering to Collection.FACTION: Vanu is 1, NC is 2 etc.

In order to reduce the amount of necessary joins, which are arguably the most complex part of Census queries, some common enumerations are provided in ps2_census.enums as Python enum.IntEnum classes, including:

• ArmorFacing
• FacilityType
• Faction
• FireModeType
• ItemType
• ItemCategory
• MetagameEventState
• PlayerState
• ProjectileFlightType
• ResistType
• ResourceType
• RewardType
• TargetType
• Vehicle
• World

These typically do not change often and ps2-census will be updated whenever there is such a change.

They can be used just for reference, but also in queries for filtering.

See the following example for filtering weapon items only using ps2_census.enums.ItemType:

from ps2_census.enums import ItemType
query = (
    Query(Collection.ITEM)
    .filter("item_type_id", ItemType.WEAPON)
)

Event Stream

ps2-census offers a client that handles connection to the WebSocket endpoint, subscription to various streams and reception of the events.

Note: because the client uses the websockets library, we need to handle async calls.

Usage

First you need to connect to the WebSocket endpoint; to do this, instantiate the EventStream class:

from ps2_census import EventStream

stream: EventStream = await EventStream(service_id=YOUR_SERVICE_ID)

Then, subscribe to events:

from ps2_census import CharacterEvent, WorldEvent, EventStreamWorld, GenericCharacter

await stream.subscribe(
    worlds=[EventStreamWorld.SOLTECH, EventStreamWorld.BRIGGS],
    events=[CharacterEvent.DEATH, WorldEvent.CONTINENT_LOCK],
    characters=[GenericCharacter.ALL, "1234", "5678"],
    logical_and_characters_with_worlds=True
)

Where:

• worlds is a list of EventStreamWorld objects. Use [EventStreamWorld.ALL] for all worlds
• events is a list of CharacterEvent, WorldEvent or GenericEvent objects. Use GenericEvent.ALL to get all events (character and world)
• characters is a list of character IDs as strings or the special GenericCharacter.ALL to subscribe to all characters
• logical_and_characters_with_worlds is True if you want to match all events concerning the characters AND the worlds; default is False, so it matches all events concerning the characters OR the worlds

You can perform multiple subscriptions one after another on the same EventStream object; they are additively merged.

Finally, you need to handle received events from your subscription:

await stream.receive()

This simple example put together (you might want to develop it further to do more than simply print events, handle graceful deconnection, etc):

import asyncio

from ps2_census import CharacterEvent, WorldEvent, EventStream, EventStreamWorld, GenericCharacter

async def main():
    stream: EventStream = await EventStream()

    await stream.subscribe(
        worlds=[EventStreamWorld.SOLTECH, EventStreamWorld.BRIGGS],
        events=[CharacterEvent.DEATH, WorldEvent.CONTINENT_LOCK],
        characters=[GenericCharacter.ALL, "1234", "5678"],
        logical_and_characters_with_worlds=True
    )

    while True:
        print(await stream.receive())

asyncio.run(main())

More information about the Planetside2 Census event stream can be found at here.

Next steps

• Improve the EventStream client to make it slightly higher-level (handle disconnections, filter events, etc)

Similar projects

• leonhard-s' auraxium

Development

Environment

In order to develop ps2-census:

• Setup a virtual environment with python 3.8
• Install poetry
• Install dependencies with poetry install
• Run tests with pytest
• Update dependencies with poetry update

Technical details

All interaction with the Census API (except the WebSocket events stream part) is done through query parameters. Query bodies are always empty and all information is carried in the query URL string.

ps2-census' objective is to wrap the building of these (often very long and complex) querystrings and provide a comfortable, structured programmatic interface.

To achieve that, we make use of the Query class which:

• on initialization sets up basic query elements such as endpoint, service ID, collection and namespace,
• at each "command" method call adds query parameters to its parameters dictionary attribute,
• at get() or count() build the actual URL and execute the query using the requests library.

Collection joins are more complicated as they are not built as proper query parameters; inside the final query, each call to query.join() creates a new query parameter, eg. ...&c:join={join_string}&... with the join_string being written in a specific format.

Hence the Join class which:

• on initialization sets up the collection,
• at each method call adds items to its items dictionary attribute,
• at each nest() call stores a reference to the nested Join instance in its nested_joins list attribute,
• at __str__() builds the join_string that will become a main query's parameter value, in the specific format and using the items elements as well as recursively calling __str__() on nested_joins items.

Therefore, the join_string is not actually computed before a call to a Join instance's __str__() method, and it is recursively computed across nested joins.

The event streams part is more conventional, as ps2-census just needs to connect to a WebSocket endpoint, issue subscription commands in a custom but simple format and then read messages from the stream.