/cli-search

Primary LanguageGo

CLI Search

This CLI program will take json files as input and allow you to search any fields via CLI.

Installation

  1. Install go

  2. Clone the project into your desired directory:

$ git clone https://github.com/willshen8/cli-search.git cli-search
  1. cd into the cloned directory:
$ cd cli-search
  1. Build the executable

A Makefile is already created for you and you can simple run

make build

and then an executable called cli-search is created at the root level of your cloned directory.

Run the program

  1. To start search run the following command:
$ ./cli-search query organisation _id 101

where the command follow the convention of

$ ./cli-search query <table> <field> <value>
  1. Configuration files: The 3 JSON files are stored in the config directory by default. To specify the data source you wish to use, use the following command:
$ ./cli-search config <organisations.json> <users.json> <tickets.json>

The order of the files matters.

  1. To list all the available fields in a table:
$ ./cli-search list <table>
  1. To get help, type
$ ./cli-search --help

and you'll receive the following help window:

Welcome to cli-search!

Flags:
  --help  Show context-sensitive help (also try --help-long and --help-man).

Commands:
  help [<command>...]
    Show help.

  config <organisation json file> <users json file> <tickets json file>
    Config the data source files by specifying the files you want to use.

  list <table>
    List all available data fields in a table.

  query <table> <field> [<value>]
    Search a specific field in a table.

Note: Optional fields are denoted by |. For example value is an optional argument, and when left blank all values will be populated back as results window.

  1. To run all test files
$ make test
  1. To check for code coverage
$ make coverage
  1. To all everything in one go, e.g. build -> lint -> test -> coverage
$ make all

Design Decisions

  1. Data structure:

Database is best for searching data as we all know, and to resemble database characteristics without building one. I have considered few options and the best one appears to be a map of map of map.

map is native data structure in Go, and is a key-value pair structure.

  • The first level map contains the name of the table organisation, user and ticket, and each key points to another map that contains the field/row names.
  • The second level map's key values are the ID field of the data.
  • The third level map basically contains the underlying data in each data record.

Go's map are implemented using hash-map, which means it offers the benefits of O(1) for lookups, and O(n) in the worst case.

  1. Extensibility

As we are using map, adding additional tables is simply a matter of insert into the root map, and then define the relationships in entity.go file.

  1. Database entity relationships
  • Organisation table has a foreign key of organization_id which is part of user and ticket table.
  • Usertable contains two foreign keys: submitter_id and assignee_id.
  • Ticket table contains no foreign keys.
  1. Use of CLI Building Library - Kingping 2 This library is simple to use and does lot of error handling when user is specifying command line arguments. It is chosen over other popular library such as Cobra and urfave/cli for its better architecture overall.

Assumptions

  1. external_id field appears to be linked to external system and bears no relationship with other tables.

Limitations

  1. The size of the database is depend on the hardware or the max size of the heap, as it is implemented using map.
  2. Use the config command option to specify the source data files, the order matters. Due to time constraint I have not implemented checks to ensure the right files are specified.
  3. As map is used, the order of the fields when printing data is not guaranteed, this can be easily achieved by using a look up table/slice. But due to time constraint, it is treated as a non-important feature and not implemented.