eolexe / campaigner Goto Github PK

The Test Process

Preparation

Generate 1000 campaigns with the “Campaign generator”, X=50,Y=10,Z=1000, the output file is downloadable.
Import the above data to the server app with data struct or database to make the searching provision ready.

• Case A:
  POST an user to “http://localhost:3000/search” to verify the search result

• Case B:
  Use “wrk” (github.com/wg/wrk) to do the performance test
  wrk -c 64 -d 10s http://localhost:3000/search_auto

Endpoint:

[POST] http://myhost:3000/search

User story:

Create an API for campaign searching with user info.
The request will POST a user info, and return the searching result.

Input:

Input should be the same as output of the User Generator endpoint: #2

Definitions:

• Targeted User: If all targets of the campaign can be found in an user's profile, and the user’s profile attribute value can be found in the list of the campaign target attri_list.

Tasks:

• create endpoint
• reuse matching business logic

Endpoint:

[GET] http://myhost:3000/user

User story:

Create a API for user generation.
This endpoint will return a user info defined as following.
There’s a counter for request (starts with 0), each request will increase the counter by 1.
Create a function to generate the following user info when it receive a request.

	"user": “u”+counter,
	"profile":[
		"attr_A" : "A" + random(200)
		"attr_B" : "B" + random(200)
		...
	]

The length of the profile will be increased by 1 everytime when a new user is generated. The attribute name and value contained in the profile need to follow the alphabet order, such as [“attr_A” : “A1”], [“attr_B” : “B23”], etc. When the profile length reaches its maximum length of 26, it gets reset to 1 again, which means it only contains “attr_A”.

Output examples:

The 1st user

{
	"user": “u1”,
	"profile":[
		"attr_A" : "A1"
	]
}

The 3rd user

{
	"user": “u3”,
	"profile":[
		"attr_A" : "A23"
		"attr_B" : "B132"
		"attr_C" : "C45"
	]
}

The 27th user

{
	"user": “u27”,
	"profile":[
		"attr_A" : "A109"
	]
}

The 28th user

{
	"user": “u28”,
	"profile":[
		"attr_A" : "A109"
		"attr_B" : "B87"
	]
}

Endpoint:

[GET] http://myhost:3000/campaign?x={number}&y={number}&z={number}

Params:

The generator has parameters: X (X <= 100) , Y (Y =< 26), Z (Z <= 10000)

User Story:

Generate Z number of campaigns, each campaign has a target list (length is random and less than Y).
The target has an attribution list (length is random and less than X).
Each campaign has an offer price, which is randomly generated.

Output:

Here is a sample output JSON file:

[
	{
		"compaign_name": "campaign870",
		"price":3.25
		"target_list" : [
			{"target":"attr_A", "attr_list":["A0",...,"A99"]},
			{"target":"attr_B", "attr_list":["B0",...,"B36"]},
			...
		]
	},
	...
]

Tasks:

• Create a campaign generator logic
• Cover generator logic with unit tests

https://github.com/shopspring/decimal

Endpoint:

GET http://myhost:3000/search_auto

User Story:

The request for the search API will trigger the function to create a user in step 2 and pass the user to the search function, if the user is a “Targeted User”, then Find the the campaign that has the HIGHEST price, otherwise return “none”.

Endpoint:

[POST] http://myhost:3000/import_camp

User story:

Create a API to import the campaign data (JSON file) generated by step 1 (#1) , then save them in some data struct or database.

Notes:

• Use redis as persistence store

Tasks:

• Base project skeleton
• Makefile and docker infrastructure

Create a matcher logic that will search through the data persisted in the application using #3 and return the campaign that will best fit to the input data.

Input data of the matcher should be the user profile data, same as generated here #2

User story

Assume there’s 2 campaigns:

{
		"compaign_name": "campaign1",
		"price":0.25
		"target_list" : [
			{"target":"attr_A", "attr_list":["A0",...,"A9"]},
			{"target":"attr_B", "attr_list":["B0",...,"B16"]},
		]
},
{
		"compaign_name": "campaign2",
		"price":0.35
		"target_list" : [
			{"target":"attr_A", "attr_list":["A0",...,"A19"]},
		]
}

To help you understanding the abstract model, let’s assume some campaign targeting {attr_A = “country”, attr_list=[“US”,“UK”,...]}, {attr_B= “hobby”, attr_list = [“football”,“fishing”,“sports”]}.

Matching cases:

• add composer + docker
• use redis as datasource

gracefull shutdown?