DataSource Python SDK
Project description
datasource-python
Python SDK for all DataSource iterations
Installation
Install this package via pip:
pip install bocks_ds
Usage
The Client
Import this package and instantiate a client object.
import bocks_ds
ds_client = bocks_ds.Client("starfinder") # provide the name of the datasource to access
Error Example:
try:
bad_client = Client("bad_client")
except DSTargetError as e:
print(e) # The target 'bad_client' provided in Client initialization is not in available target names:\n['starfinder', 'pathfinder']
Fetching Data
There are currently only two options to consider when fetching data: the type and the parameters.
The "type" can be thought of as a data table, though we will discuss some complexities later on.
response = ds_client.armor.get(['name', 'price']) # 'armor' here is the query type
if response.status_code == 200: # optional status check
data = response.json() # '.json()' gives the API data output we came for
Note that the ds_client
will accept any value as an attribute, which it will then use to craft an API request.
Error Example:
try:
client.bad_name.get(['name']).json()
except DSQueryError as e:
print(e) # <Response 400> DataSource did not find table/field 'bad_name'.
Refining Your Request
The API allows for active filtering of strings and integers within your request. To do this, you'll need to set arguments immediately prior to making the request.
For data values that return strings, you may refine with the terms <value_name>_like
and <value_name>_is
. For "like" queries, sequencing is not considered and the search is not case sensitive. For "is" queries, only exact, case-sensitive matches will be returned.
For data values that return integers, you may refine with the terms <value_name>_min
, <value_name>_max
and <value_name>_equals
. These equate to "greater than or eqal to", "less than or equal to", and "equal to", respectively.
Finally, it is often valuable to select an item specifically by its ID
, which functions in a more direct manner than the queries for integers (see example below).
All arguments must be presented as a single dictionary, as seen in the examples below.
client.armor.set_arguments({"name_like":"basic"})
response = ds_client.armor.get(['name', 'price'])
client.armor.set_arguments({"price_min":200, "price_max":2000})
response = ds_client.armor.get(['name', 'price'])
client.armor.set_arguments({"id":1})
response = ds_client.armor.get(['name', 'price'])
Note that these requests do not stack, but you can place all terms into a single dictionary to futher refine results. If set_arguments
is called multiple times, the final call will overwrite previous calls.
query = {"name_like":"basic", "price_min":200, "price_max":2000}
client.armor.set_arguments(query)
response = ds_client.armor.get(['name', 'price'])
As a final note: all arguments are cleared when get
is called.
Error Example
try:
client.armor.set_arguments({"name_min":200})
erroneous_armor = client.armor.get(['name', 'price']) # Throws exception due to errors in response.json()
except DSQueryError as e:
print(e) # <Response 400> ['Unknown argument "name_min" on field "Query.armor". Did you mean "name_is", "name_like", "type_min", "bulk_min", or "level_min"?']
Fetching Nested Data
As mentioned previously, not all types match their data tables precisely. Some types include additional data from relationships with other tables. Documentation for these relationships is automatically generated in the API Documentation (such as at docs.sfdatasource.com).
In order to access nested data, it is required that you present a dictionary describing the data to fetch.
Where we previously provided a list of string field-names to query, we will now include a dictionary in that list, as seen in the examples below.
query = [
"name",
{
"effect_ranges": ["name", "description"],
}
]
all_spells = ds_client.spells.get(query)
This logic is recursive, so in the event that a relationship target has it's own relationships you may do the following:
query = [
"name",
{
"class_proficiencies": ['name'],
"class_features": ["name", {
"class_special_skills": ['name']
}],
}
]
spells = client.classes.get(query)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.