This is a solution for the backend devchallenge 2023, see the task description below.
docker compose up
After successful start the app should log something like that:
devchallenge1_jkhdfbnsk3 | [info] welcome to sbt 1.9.6 (GraalVM Community Java 17.0.8)
devchallenge1_jkhdfbnsk3 | [info] loading project definition from /root/project
devchallenge1_jkhdfbnsk3 | [info] loading settings for project root from build.sbt ...
devchallenge1_jkhdfbnsk3 | [info] set current project to devchallenge1scala (in build file:/root/)
devchallenge1_jkhdfbnsk3 | [info] running pkg.Main
May need to wait a few extra seconds for the HTTP server to start.
Request examples:
curl -X POST 'http://localhost:8080/api/v1/sheet1/CeLl3' -d '{"value":"123"}'
{"value":"123","result":"123"}
# http-server -p 8000
curl -X POST 'http://localhost:8080/api/v1/sheet1/a1/subscribe' -d '{"webhook_url":"https://localhost:8000/a1"}'
curl -X POST 'http://localhost:8080/api/v1/sheet1/a1' -d '{"value":"123"}'
{"value":"123","result":"123"}
# http-server gets an update: POST /a1 {"value":"123","result":"123"}
After the container was run once using docker compose up, the container would be created and unit tests could be run
using this command:
docker exec -it devchallenge1_jkhdfbnsk3 sbt test
Performance characteristics:
- write complexity: tree traversals are N+M
- where N is the number of cells using this cell recursively
- and M is the number of cells this cell refers to recursively
- read complexity: 1
Corner cases covered:
- very long formulas, like a million entries of
1+1+1+... - very long cell dependency chains, also in order of a million is quite ok
- a lot of performance tests
- more cases are described in unit tests in
src/test/scaladir
Future improvements:
- it may be possible to change the core data structure for storing data to find better balance between write-heavy and read-heavy load
- an optimiser like relational databases have can be introduced, gathering query statistics and analysing data distribution
We all know there is no better software in the world than Excel The powerful idea behind the cells and formulas allows many of us to understand programming. Today is your time to pay respect to spreadsheets and implement backend API for this fantastic tool.
As a user, I want to have API service with exact endpoints:
POST /api/v1/:sheet_id/:cell_id accept params {"value": "1"} implements UPSERT strategy (update or insert) for both sheet_id and cell_id
1) 201 if the value is OK
2) 422 if the value is not OK e.g. new value leads to dependent formula ERROR compilation
Examples:
POST /api/v1/devchallenge-xx/var1 with {"value:": "0"}
Response: {"value:": "0", "result": "0"}
POST /api/v1/devchallenge-xx/var1 with {"value:": "1"}
Response: {"value:": "1", "result": "1"}
POST /api/v1/devchallenge-xx/var2 with {"value": "2"}
Response: {"value:": "2", "result": "2"}
POST /api/v1/devchallenge-xx/var3 with {"value": "=var1+var2"}
Response: {"value": "=var1+var2", "result": "3"}
POST /api/v1/devchallenge-xx/var4 with {"value": "=var3+var4"}
Response: {"value": "=var3+var4", "result": "ERROR"}
GET /api/v1/:sheet_id/:cell_id
1) 200 if the value present
2) 404 if the value is missing
Examples:
GET /api/v1/devchallenge-xx/var1
Response: {"value": "1", result: "1"}
GET /api/v1/devchallenge-xx/var1
Response: {"value": "2", result: "2"}
GET /api/v1/devchallenge-xx/var3
Response: {"value": "=var1+var2", result: "3"}
GET /api/v1/:sheet_id
1) 200 if the sheet is present
2) 404 if the sheet is missing
Response:
{
"var1": {"value": "1", "result": "1"},
"var2": {"value": "2", "result": "2"},
"var3": {"value": "=var1+var2", "result": "3"}
}
- Add to existing spreadsheet backend support of the following functions:
SUM
"=SUM(1, 2)" => {"value:": "=SUM(1, 2)", "result": "3"}
"=SUM(var1, var2)" => {"value:": "=SUM(var1, var2)", "result": "3"}
"=SUM(var1, 2, var3)" => {"value:": "=SUM(var3, 2, var3)", "result": "6"}
AVG
"=AVG(1, 2)" => {"value:": "=AVG(1, 2)", "result": "1.5"}"
"=AVG(var1, var2)" => {"value:": "=AVG(var1, var2)", "result": "1.5"}"
"=AVG(var1, 2, var3)" => {"value:": "=AVG(1, 2, var3)", "result": "2"}"
MIN
"=MIN(1, 2)" => {"value:": "=AVG(1, 2)", "result": "1"}"
"=MIN(var1, var2)" => {"value:": "=MIN(var1, var2)", "result": "1"}"
"=MIN(-1, var1, var2, var3)" => {"value:": "=MIN(-1, var1, var2, var3)", "result": "-1"}"
MAX
"=MAX(1, 2)" => {"value:": "=MAX(1, 2)", "result": "2"}"
"=MAX(var1, var2)" => {"value:": "=MAX(var1, var2)", "result": "2"}"
"=MAX(-1, var1, var2, var3)" => {"value:": "=MAX(-1, var1, var2, var3)", "result": "3"}"
- Implement subscriptions with webhooks
POST /api/v1/devchallenge-xx/var3/subscribe with {"webhook_url": "https://..."}
Response 201 with {"webhook_url": "https://"}
As a user, I expect to receive an HTTP POST request on the given webhook_url with the following payload:
{"value": "=var1+var2", "result": "3"}
whenever the result of the var3 is updated.
- Supports basic data types: string, integer, float
- Support basic math operations like +, -, /, * and () as well.
- :sheet_id - should be any URL-compatible text that represents the namespace and can be generated on the client
- :cell_id - should be any URL-compatible text that represents a cell (variable) and can be generated on the client
- :sheet_id and :cell_id are case-insensitive
- Be creative and cover as many corner cases as you can. Please do not forget to mention them in the README.md
- Data should be persisted and available between docker containers restarts
Upload the source on the platform in one file archive with the name in the format FileName.zip.
☝️ Please note that the .git directory should not be present in the archive.
☝️ Please note that the name of the archive and file names inside the archive should not contain your first or last
name. The size of the solution archive should not exceed 10 MB.
The Organizers and Judges reserve the right to disqualify the Participant's work if the work:
- Contains any reference to the Participant's name, surname, e-mail address, company, address, or other personal data;
- Completed in a different format than specified in the task;
- Performed with the help of third parties, and not by the Participant personally.
The archive should contain
- 'docker-compose.yml' file in the root, which starts a server with the given endpoints on port 8080, available from localhost as 'http://localhost:8080/'.
- README.md, where you wrote instructions on how to start service and tests and some thoughts about your choices during performing this task and the next steps to make your service better.
- Your test suite should also run inside docker with a single command line. Please leave instructions in the README.md.
October 5, at 00:00 (GMT+3) — after the time runs out, the possibility of uploading works to the platform will be
automatically blocked. The participants who have moved on to the Final will be announced on October 16, 2023.
☝️ Set your time zone in the settings on the DEV Challenge platform to more conveniently track the time of the deadline.
Technical assessment
- Result Correctness — 90 points.
- Following API Format — 40 points.
- Performance — 80 points.
- Code quality — 40 points.
- Test — 80 points.
- Corner cases — 54 points.
☝️ Participants in the Backend nomination will be divided into Hard and Lite levels in the Final.
Questions and clarifications regarding the content of tasks:
Slack channel: #nomination-backend.
☝️ Judges will ignore questions that do not relate to the tasks of the Championship.
Organizational questions:
Contact us via e-mail at hello@devchallenge.it or Slack
channel #02-ask-the-organizers.