#enums #csv #codegen #command-line #variant-name #associated-constants

bin+lib csvenum

Generate code for enums with associated constants from a .csv in your command line

5 releases

0.1.4 Mar 21, 2024
0.1.3 Mar 16, 2024
0.1.2 Mar 14, 2024
0.1.1 Mar 13, 2024
0.1.0 Mar 13, 2024

#155 in Data structures

Download history 260/week @ 2024-03-10 247/week @ 2024-03-17 6/week @ 2024-03-24 41/week @ 2024-03-31

554 downloads per month

MIT/Apache

145KB
3K SLoC

csvenum

Docs Changelog

CLI to generate Rust enums with associated constants from a csv-table.

Why?

If you have ever needed to declare a lot of constants you will know that it may require looking at a dataset and copying the values to code.

So here's a code-gen to assist in creating enums with associated constants.

Just declare a table in a .csv, generate and you're done.

You can just grab a file like this, add type information and be good to go.

If you want to associate data with a pre-existing enum, you might want to check out strum instead.

Usage

Since this crate is meant to be a tool for speeding up Rust development, it is available as a cargo install.

Note that it is in active development, so please check back for updates and report any issues you may find.

cargo install csvenum

After installing, create a table like this in a .csv:

ENUM        , &str      , usize     ,  (f64 , f64)

Country     , ISO3      , Numeric   ,  Lat_Lon

Sweden      , SWE       , 752       ,  (60.128161 , 18.643501)

Vietnam     , VDR       , 704       ,  (14.058324 , 108.277199)

Brazil      , BRA       , 076       ,  (-14.235004 , -51.92528)

To generate the code pass the filename to the CLI. Here countries.csv.

csvenum countries.csv

The generated enum will be in a file called country.rs in the directory of the passed .csv.

In code you can now access the values like so:

    let countries = Country::get_all_variants();
    for country in countries {
        println!("{}", country);
        println!("{}", country.as_iso3());
    }    

Will output:

Sweden, ISO3 = SWE , Numeric = 752 , Lat_Lon = (60.128161 , 18.643501) 
SWE
Vietnam, ISO3 = VDR , Numeric = 704 , Lat_Lon = (14.058324 , 108.277199) 
VDR
Brazil, ISO3 = BRA , Numeric = 076 , Lat_Lon = (-14.235004 , -51.92528) 
BRA

See the table format, CLI options and the list of features below for details.

Output examples can be found here.

Table format

A table for code-gen with csvenum will always have the following shape.

  • First line: Specifies the types of the column values, starting with the word ENUM.

  • Second line: Specifies the name for the enum and the column names, referred to here as properties.

  • Third line and after: The data.

The data will be rows, starting with the variant name, followed by the values per property.

Example

ENUM,      &str,       usize,      [usize; 3]        <-- Column types

GPIOpin,    Address,    Value,      Numbers          <-- Enum name followed by the property names

PIN0,       0x00,       42,         [1,5,7]          <-- Variant name and associated values
PIN1,       0x02,       56,         [8,4,2]
PIN2,       0x04,       68,         [12,3,2]

Note that you can use commas in nested fields when they are enclosed with the appropiate symbol.

  • Strings/Regex : ""

  • Tuples : ()

  • Array : []

Most software will export fields with nested commas with quotes, so a tuple would be "(1,2)". This is perfectly fine for the parser.

Duplicate values in a column will be collected into an array that holds all corresponding variants.

For now tables are limited to only include constant values (i.e. types with constant initializers), but there are plans to provide OnceLock / Lazy implementations for others.

Also there is an arbitrary but reasonable limit on value nesting depth to avoid headaches.

Note that property names will have to follow valid Rust variable naming rules.

Bools can also be 0 or 1 in the table instead of just false or true.

CLI options

csvenum --help
Codegen for enums with associated constants

Usage: csvenum [OPTIONS] <FILENAME_CSV>

Arguments:
  <FILENAME_CSV>  Filename of the CSV file (required)

Options:
  -o, --outpath 
          Path to the output dir/file
  -s, --split-properties 
          Whether to split property declarations into separate files, defaults to: false [possible values: true, false]
  -v, --variant-str-fns 
          Generate variant as & from str fns , defaults to: true [possible values: true, false]
  -i, --impl-links 
          Pure conversion functions only or also impl links to them, defaults to: true [possible values: true, false]
  -h, --help
          Print help
  -V, --version
          Print version

Features

Generated Code:

  • (Always) - Declaration of the enum with the given variants and doc-strings that include all properties + values

  • (Option, true) - Variant name as and from str functions + std::fmt::Display impl that prints the name and all associated values to a string.

  • (Always) - Declares property values as constants and "as" and "from" conversion function between them. You can opt to split the properties into separate files.

  • (Option, true) - Generates an impl block for the enum that contains links to the property conversion functions, also generates a test module.

  • (Always) - Generates a get_all_variants function -> [MyEnum; N_variants]

Available Types:

  • Const types (initialized by const expressions):

      - numeric types    
    
      - &str
    
      - tuples 
    
      - arrays
    
  • Regex: - relies on the regex crate being present, cargo add regex

      - will create a const &str and a static `OnceLock<Regex>` 
    
      - "as"-method will return &Regex
    
      - "from"-method iterates over the associated regexes and returns the first match, i.e the matching variant
    
      - nested Regex, i.e. Regex in tuples or arrays are not planned
      
      - can't have duplicates for a given column/property 
    

Known Issues

  1. Trailing commas in .csv - could fix but shouldnt really have them to begin with.

  2. Invalid characters in variants and properties for Rust identifiers.

Future plans

  1. Dedicated Column for Ord implementation

  2. Generate FromStr impl to check all associated string constants.

  3. Provide OnceLock or similar wrappers for non-const statics.

  4. Option on data with missing values.

  5. Static HashMaps for larger datasets.

Please report any issue you find or suggestion you have to further improve this tool!

Dependencies

~3.5–5MB
~88K SLoC