dmarkham / enumer Goto Github PK

First of all i'm a happy user of the enumer :) But i've an situation in which i need a bit of help.

When my enum is in code that is only build when certain constraints are met. (See: here) i'd like to have the constraints also in the generated code. If the build constraints not get exported the compile would generated errors as the generate file is build without the source containing the enum base definition.

I wrote a small PR in which i added a command line argument to add the build constraints. This works fine.

I was thinking whether it would be possible to deduct the build tags from the original source file and copy them into the generated one. But could not think of an easy way to that.

Note the comment option does not work as it adds a space between // and the first char of the comment

Now we have two libraries with similar features but not the same. And when I add annotations like

//go:generate enumer -type=From -json -transform=snake

Binary can be from any of them and result can be different. Is there any way to make sure that code will be generated only by your binary? For example, add different binary name for ensure or some other tricks that provide incompatibility with original version?

Title says it all :)

I propose enhancing enumer to allow specifying different transformation methods for different marshaling types.

Our team currently implements custom GraphQL and JSON marshaling methods based on UPPER_SNAKE_CASE and snake_case representations, respectively. It'd be extremely helpful if enumer can generate both for us while keeping the current representations.

I suppose the feature will accept a transformation method via the value of each marshaling flag as shown below.

-gqlgen=snake-upper -json=snake

If this sounds like a good feature to add, I am ready and willing to contribute to its implementation.

It would be useful to perform string -> enum conversion in a case-insensitive fashion. While I could convert my string to lowercase and have all of my values lowercase as well, some of them are acronyms which looks ugly if not written with capital letters.

Sometime there is already a String() method.

in my case, i'm converting iota to 1 character string and they includes both lower and upper version of some alphabets. i've read #21 , so maybe add a flag to enable case sensitive in <Type>String(s string).

We need this in order to be a better drop in replacement. We are currently just missing this feature!

https://github.com/golang/tools/blob/4fd33079060a7ef7bffe3a1c2369620f4f21ad06/cmd/stringer/stringer.go#L84

Hi! I have a use-case where I want to use enumer to map postgres enums with the -sql flag. There may be conflicts in my enums, hence I cannot declare multiple enums in the same package.

Example:

//go:generate ...
type EnumA int
const (
    Test EnumA = iota
    Cool EnumA
)

//go:generate ...
type EnumB int
const (
    VeryCool EnumB = iota
    Cool EnumB // this will cause issues!
)

It would be neat if we could instead declare the values as strings, e.g.:

//go:generate ...
type EnumA string
const (
    ATest EnumA = "TEST"
    ACool EnumA = "COOL"
)

//go:generate ...
type EnumB string
const (
    BVeryCool EnumB = "VERY_COOL"
    BCool EnumB = "COOL"
)

When using the -linecomment=true I think that operations that change the text should not apply, for example -transform= -trimprefix= -addprefix= as you are saying how do you want it to look like:

  -linecomment
        use line comment text as printed text when present

We had an error due to that (but was our fault) as we had a type like:

package enumer

type DeltaType int

//go:generate enumer -type=DeltaType -transform=snake -output=delta_type_string.go -linecomment=true
const (
	Common    DeltaType = iota + 1 // =
	LeftOnly                       // -
	RightOnly                      // +
	Unknown                        // ?
)

It was a copy paste error but it ended up having a .String() that always returned "" as the linecomment was not "valid" to snake (which could potentially rise errors if a type is gonna be transformed to just ""? 🤔 ).

If we want to change this behavior I'm open to do a PR which will also solve #27 😄

I sometimes want to convert an int item specified by enum to a string, e.g. to specify RESTFullAPI request parameters, etc.
It would be nice to have an option to generate a ToString() method.

Hi @dmarkham

While digging in the previous PR of the project I noticed that you had made some work on BSON support back in 2019 but did not finish it.

If I was to open a PR to make a -bson flag again would you be willing to merge it ?
I understood that you did not like the fact that the generated code would be tied to the external mongodb package but I don't think there is any alternative way to implement a BSON marshaller for now.

This can be prevented by adding a tools.go file:

//go:build tools

package tools

import (
	// Protect this entry in go.mod from being removed by go mod tidy.
	_ "github.com/dmarkham/enumer"
)

Should probably go into README.md.

we can optimize IsAXXX method to use switch statement instead of for loop.
with this, we dont need stringBelongsMethodSet

Firstly, I'd like to thank you for great work, I found enumer really helpful!

I am not sure about proper solution, however I think that may be doable by another flag/parameter.

In perfect scenario it would either implement validator interface, or return validator.ValidationErrors.
If not, maybe there is possibility to generate helper validator functions to bind for API models.

hi, just got an issue with go.18 beta2, reproduce steps:

1. build enumer from source with go 1.18beta2
2. use enumer from step1 to generate code
3. got error: enumer: internal error: package "bytes" without types was imported from ...."

enumer unit test also failed on go1.18

In my project, I wrote a test to fully cover enumer-generated code.
I would be happy/willing to take that test and add a flag to enumer so that enumer will generate tests too.

Let me know.

See https://github.com/muir/xop-go/blob/main/internal/enumer_test.go

Hi team,

I was trying to do some unhappy path test case using
//go:generate enumer -type=BusinessType -json -transform upper

And when I test a "does not belong to BusinessType values" case:
I entered ABC, and expect to get "ABC does not belong to BusinessType values",
but what I get is "abc does not belong to BusinessType values"

I think it's because the generated function BusinessTypeString transformed it into lower case, it's not impacting the code logic, but my test case becomes a little weird:

Input: "ABC"
Output: "abc does not belong to BusinessType values"

The question is, is there any way to generate without the toLower() transform in function BusinessTypeString?

Many thanks!

Best regards,

Hi, to improve interoperability with ent and support its enum type-safety, I'd propose to add the Values() []string method (from the EnumValues interface) to the generated enum types, next to String() string.
This feature should likely be opt-in, so it would be beneficial to add a flag for it. I'd propose -values (or -valuesFunc) flag.

I've seen, there is already a generated ***Strings() []string-func, which would just need to be wrapped by the Values() []string method in question.

@dmarkham Would you be open for a PR for this?

An Example:

func (Pill) Values() []string {
  return PillStrings()
}

P.S.: I understand there's a potential for confusion, due to the pre-existing PillValues() []Pill function. But to have typesafety with enums in ent, this proposal is the only way to go from what I see.