whiskels / notifierbot Goto Github PK

Current realization is to annotate method with @ Secured annotation and works only if one method in the class is annotated.
Would be better to add security as an optional attribute of @ BotCommand annotation to make it more understandable (handler-level security)

Implement slack reporter that will create daily/monthly reports about upcoming customer birthdays.

Data is loaded from Google Spreadsheets to inmemory storage daily.

Requests should be routed via proxy that provides static IP address to pass firewall

No tests right now

Classes are currently distributed between packages based on their functionality (domain, service, repository etc). It would be easier to make changes to the code if packages were created based on the features implemented (for example: payments package will include entity, dto, repository and service)

Implement spring event listeners for telegram bot and slack webhook handler

Refactor code, update logic for no data picture, change to map<double, list>

Usage of in-built Spring ability to activate different profiles based on file name makes it easier to support configs.

Application that is deployed to Heroku is restarted every 24 hours - current realization of cache doesn't work.

Persistent cache supported by Heroku is needed

Change from manual deploying via console to automatic deploy on push to master branch

Change implementation of slack reporting from day before event to on event.
If current day is a working day and there is an event - send report. If current day is Friday - send report if there are events in range [Friday;Sunday]. There is no need to check national holidays in current realisation

Ex.:
Was: report on Friday if there is a birthday on Monday;
Will be: report on Monday if there is a birthday on Monday

HR managers would like to have notifications three days before the employee event, so that they can prepare a gift

Implement additional feature - audition of slack payload execution.

• Refactoring - generic audition mechanism (data loading/payload execution). Use generic repositories, refactor JpaRepositories into Repositories with partial implementation of crud methods, maybe interface projections for DTO)
• Persistence of execution results and metadata in db
• Admin 'audit view' chat commands for Telegram

Current realization: data used by reporters may be outdated if called in the morning before an update occurred. For payment report this leads to an empty report.

Proposed realization: add additional method that returns date of data retrieval. This will let us get a report for payments in telegram in the afternoon and then get the same report in the morning of the next day

New spreadsheet will be used - need to update configs and code.

For the sake of clearer use of the bot it would be better to use reply keyboard

Due to nature of the values the formats in customer birthday might differ:
dd.MM, dd.MM., dd/MM, dd.MM.yyyy, dd/MM/yyyy

Additional telegram command similar to /admin_slack that will let admin to call DataLoader instances directly via Telegram message

Migrate from Travis as free plan ended and merge all prs

Implement birthday reporter for HR.

• last day of the month - all birthdays and work anniversaries for the next month
• day before birthday/anniversary - message with notification about upcoming event
• second week of the month - birthdays and anniversaries for the rest of the month

Should also return user roles

This will make a single point where server hour offset is applied and minimize the chances for any errors

Maybe it would be better not to have strict queries for financial operations like the one used to get revenue dto, but to use specifications to have more flexibility.
But there is a downside - this will lead to additional conversion from retrieved financial operation to dto in code

• deserialize JSON dates to LocalDate
  • For dd-MM - https://mkyong.com/java8/java-8-how-to-parse-date-with-dd-mmm-02-jan-without-year/
• check if it is possible to kill LocalDatePersistenceConverter when saving data to db

https://stackoverflow.com/questions/31882306/how-to-configure-travis-ci-to-build-pull-requests-merges-to-master-w-o-redunda

Bug is related to the stage of formation of revenue report. The pipeline for it is as follows:

• When scheduler calls method that updates data in DB:
  • Receive and filter financial operations from external API
  • Save filtered events to DB if there are any
• When scheduler calls method that creates report for slack:
  • Receive last update date and corresponding data from database

If there is a period of time where "data update" scheduled job returns no data to load to database, then a series of identical reports will be sent to slack, though it is expected to see one report with data which is followed by subsequent reports about no data.

Possible fix: change logic in DataProvider implementation to check whether current date (with timezone) is the same as the date of last update

The database used in the project is not going to be changed frequently, but some improvements may require changes. As this project uses several databases based on current environment (dev, prod) - it would be better to use database migration tool.

After recent vulnerabilities were discovered it would be nice to bump dependencies

Slack provides rich message design API so it would be better to split report creation logic between the modules in order to be able to create fantastic-looking messages for slack

It would be better to have a profile with database initialization using dummy values and third-party API links being stubs to let anyone to try the bot by just adding bot credentials

Implement admin command description and additional Helper that returns admin panel in telegram