|
1 | | -# ffc-server-python-sdk |
| 1 | +# ffc-server-python-sdk |
| 2 | + |
| 3 | +## Introduction |
| 4 | + |
| 5 | +This is the Python Server SDK for the feature management platform [featureflag.co](https://featureflag.co/). It is |
| 6 | +intended for use in a multiple-users python server applications. |
| 7 | + |
| 8 | +This SDK has two main purposes: |
| 9 | + |
| 10 | +- Store the available feature flags and evaluate the feature flags by given user in the server side SDK |
| 11 | +- Sends feature flags usage, and custom events for the insights and A/B/n testing. |
| 12 | + |
| 13 | +## Data synchonization |
| 14 | + |
| 15 | +We use websocket to make the local data synchronized with the server, and then store them in the memory by default. |
| 16 | +Whenever there is any changes to a feature flag or his related data, the changes would be pushed to the SDK, the average |
| 17 | +synchronization time is less than **100** ms. Be aware the websocket connection can be interrupted by any error or |
| 18 | +internet interruption, but it would be restored automatically right after the problem is gone. |
| 19 | + |
| 20 | +## Offline mode support |
| 21 | + |
| 22 | +In the offline mode, SDK DOES not exchange any data with [featureflag.co](https://featureflag.co/), this mode is only use for internal test for instance. |
| 23 | + |
| 24 | +To open the offline mode: |
| 25 | +```python |
| 26 | +config = Config(env_secret, offline=True) |
| 27 | +``` |
| 28 | + |
| 29 | +## Evaluation of a feature flag |
| 30 | + |
| 31 | +SDK will initialize all the related data(feature flags, segments etc.) in the bootstrapping and receive the data updates |
| 32 | +in real time, as mentioned in the above |
| 33 | + |
| 34 | +After initialization, the SDK has all the feature flags in the memory and all evaluation is done locally and synchronously, the average evaluation time is < **10** ms. |
| 35 | + |
| 36 | +## Installation |
| 37 | +install the sdk in using pip, this version of the SDK is compatible with Python 3.6 through 3.9. |
| 38 | + |
| 39 | +``` |
| 40 | +pip install ffc-server-python-sdk |
| 41 | +``` |
| 42 | + |
| 43 | +## SDK |
| 44 | + |
| 45 | +Applications SHOULD instantiate a single instance for the lifetime of the application. In the case where an application |
| 46 | +needs to evaluate feature flags from different environments, you may create multiple clients, but they should still be |
| 47 | +retained for the lifetime of the application rather than created per request or per thread. |
| 48 | + |
| 49 | +### Bootstrapping |
| 50 | + |
| 51 | +The bootstrapping is in fact the call of constructor of `FFCClient`, in which the SDK will be initialized, using |
| 52 | +streaming from [featureflag.co](https://featureflag.co/). |
| 53 | + |
| 54 | +The constructor will return when it successfully connects, or when the timeout(default: 15 seconds) expires, whichever comes first. If it has not succeeded in connecting when the timeout elapses, you will receive the client in an uninitialized state where feature flags will return default values; it will still continue trying to connect in the background unless there has been a network error or you close the client(using `stop()`). You can detect whether initialization has succeeded by calling `initialize()`. |
| 55 | + |
| 56 | +The best way to use the SDK as a singleton, first make sure you have called `ffcclient.set_config()` at startup time. Then `ffcclient.get()` will return the same shared `ffcclient.client.FFCClient` instance each time. The client will be initialized if it runs first time. |
| 57 | +```python |
| 58 | +from ffcclient.config import Config |
| 59 | +from ffcclient import get, set_config |
| 60 | + |
| 61 | +set_config(Config(env_secret)) |
| 62 | +client = get() |
| 63 | + |
| 64 | +if client.initialize: |
| 65 | + # your code |
| 66 | + |
| 67 | +``` |
| 68 | +You can also manage your `ffcclient.client.FFCClient`, the SDK will be initialized if you call `ffcclient.client.FFCClient` constructor. |
| 69 | +```python |
| 70 | +from ffcclient.config import Config |
| 71 | +from ffcclient.client import FFCClient |
| 72 | + |
| 73 | +client = FFCClient(Config(env_secret), start_wait=15) |
| 74 | + |
| 75 | +if client.initialize: |
| 76 | + # your code |
| 77 | + |
| 78 | +``` |
| 79 | +If you prefer to have the constructor return immediately, and then wait for initialization to finish at some other |
| 80 | +point, you can use `ffcclient.client.FFCClient.update_status_provider` object, which provides an asynchronous way, as follows: |
| 81 | + |
| 82 | +``` python |
| 83 | +from ffcclient.config import Config |
| 84 | +from ffcclient.client import FFCClient |
| 85 | + |
| 86 | +client = FFCClient(Config(env_secret), start_wait=0) |
| 87 | +if client._update_status_provider.wait_for_OKState(): |
| 88 | + # your code |
| 89 | + |
| 90 | +``` |
| 91 | + |
| 92 | + |
| 93 | +### Evaluation |
| 94 | + |
| 95 | +SDK calculates the value of a feature flag for a given user, and returns a flag vlaue/an object that describes the way |
| 96 | +that the value was determined. |
| 97 | + |
| 98 | +`User`: A dictionary of attributes that can affect flag evaluation, usually corresponding to a user of your application. |
| 99 | +This object contains built-in properties(`key`, `name`, `email` and `country`). The only mandatory property is the key, |
| 100 | +which must uniquely identify each user; this could be a username or email address for authenticated users, or a ID for anonymous users. |
| 101 | + |
| 102 | +All other built-in properties are optional, it's strongly recommended to set **name** in order to search your user quickly |
| 103 | +You may also define custom properties with arbitrary names and values. |
| 104 | + |
| 105 | +```python |
| 106 | +if client.initialize: |
| 107 | + user = {'key': user_key, 'name': user_name} |
| 108 | + flag_value = client.variation(flag_key, user) |
| 109 | + # your if/else code according to flag value |
| 110 | + |
| 111 | +``` |
| 112 | +If evaluation called before SDK client initialized or you set the wrong flag key or user for the evaluation, SDK will return |
| 113 | +the default value you set. The `FlagState` will explain the details of the last evaluation including error raison. |
| 114 | + |
| 115 | +SDK support the String, Boolean, and Number as the return type of flag values, see pydoc for more details. |
| 116 | + |
| 117 | +### Experiments (A/B/n Testing) |
| 118 | +We support automatic experiments for pageviews and clicks, you just need to set your experiment on our SaaS platform, then you should be able to see the result in near real time after the experiment is started. |
| 119 | + |
| 120 | +In case you need more control over the experiment data sent to our server, we offer a method to send custom event. |
| 121 | +```python |
| 122 | +client.track_metric(user, event_name, numeric_value); |
| 123 | +``` |
| 124 | +**numeric_value** is not mandatory, the default value is **1**. |
| 125 | + |
| 126 | +Make sure `track_metric` is called after the related feature flag is evaluated by simply calling `variation` or `variation_detail` |
| 127 | +otherwise, the custom event won't be included into the experiment result. |
0 commit comments