web-ridge / gqlgen-sqlboiler Goto Github PK

View Code? Open in Web Editor NEW

74.0 74.0 13.0 616 KB

This is a plugin for gqlgen to generate converts + filter queries and resolvers for sqlboiler

License: MIT License

Go 100.00%

gqlgen-sqlboiler's People

Contributors

Stargazers

Watchers

Forkers

tony alichherawalla alexswiss randallmlough huynhhung0 sjhitchner orible robinlindhout omirror kudinovkv kolomerot stephenafamo proorc

gqlgen-sqlboiler's Issues

failed to assign all query results to <type> slice

For the following schema and generated query mods, sqlboiler returns an error

SELECT * FROM "deal" WHERE (creator_id = $1 AND EXISTS(SELECT 1 FROM "user" WHERE (id = $1)));
[8875fd38-b75d-45a7-a5a1-cbee58c8716d 8875fd38-b75d-45a7-a5a1-cbee58c8716d]
{"level":"error","error":"models: failed to assign all query results to Deal slice: bind failed to execute query: pq: got 2 parameters but the statement requires 1","time":"2020-05-11T07:27:58-04:00","message":"Could not list deals"}

If comment subQueryMods := UserWhereToMods(m, !hasForeignKeyInRoot) part from function below it seem to start working

func UserWhereSubqueryToMods(m *graphql_models.UserWhere, foreignColumn string) []qm.QueryMod {
	if m == nil {
		return nil
	}
	var queryMods []qm.QueryMod
	queryMods = append(queryMods)

	// if foreign key exist so we can filter on ID in the root table instead of subquery
	hasForeignKeyInRoot := foreignColumn != ""
	if hasForeignKeyInRoot {
		queryMods = append(queryMods, IDFilterToMods(m.ID, foreignColumn)...)
	}

	subQueryMods := UserWhereToMods(m, !hasForeignKeyInRoot)
	if len(subQueryMods) > 0 {
		subQuery := models.Users(append(subQueryMods, qm.Select("1"))...)
		queryMods = appendSubQuery(queryMods, subQuery.Query)
	}
	return queryMods
}

type Deal {
    id: ID!
    creator: User!
    negotiation: Negotiation
    agreement: Agreement
    updatedAt: Int!
    createdAt: Int!
    agreements: [Agreement]
    appointments: [Appointment]
    negotiations: [Negotiation]
    userDeals: [UserDeal]
}

input DealFilter {
    search: String
    where: DealWhere
}

input DealWhere {
    id: IDFilter
    creator: UserWhere
    negotiation: NegotiationWhere
    agreement: AgreementWhere
    updatedAt: IntFilter
    createdAt: IntFilter
    agreements: AgreementWhere
    appointments: AppointmentWhere
    negotiations: NegotiationWhere
    userDeals: UserDealWhere
    or: DealWhere
    and: DealWhere
}

type Query {
    deals(filter: DealFilter): [Deal!]!
}

Allow custom scope resolvers for things like userId, organizationId, creatorId, tenantId

We now have some check for userId, organizationId, userOrganizationId which are used at webRidge.

This won't work for other thing like authorId, or creatorId. User from a GraphQL api is generally not allowed to put this field inside of the api, this field needs to be resolved based on auth context or something else.

We need to support his use case by deprecating the userId, organizationId, userOrganizationId fields https://github.com/web-ridge/gqlgen-sqlboiler/blob/master/resolver.gotpl#L110.

Instead user would be allowed to put a resolver config object inside the convert_plugin.go

Instead of

api.AddPlugin(gbgen.NewResolverPlugin(
			convertHelpersDir,
			sqlboilerDir,
			gqlgenModelDir,
			"github.com/yourauth/implementation" // leave empty if you don't have auth
		)),

Maybe something like this

api.AddPlugin(gbgen.NewResolverPlugin(
			convertHelpersDir,
			sqlboilerDir,
			gqlgenModelDir,
			&gqlgen.ResolverConfig{
				[]CustomImport{
					{
						Alias: "auth",
						Package: "github.com/yourauth/implementation",
					},
				},
				[]ResolveIDs{
					{
						BoilerColumnName: "userID",
						Resolver: "auth.UserIDFromContext",
						NullResolver: "auth.NullableUserIDFromContext",
					},
				}
		)),

How to do validation?

Since you are fighting my own battles (gqlgen and go) can I ask you how are you dealing with the hot topic of validations?

E.g. from your example:

input ImageCreateInput {
  postId: ID!
  views: Int
  originalUrl: String
}

how to validate that e.g. views is max = 30 or originalUrl != ""?

Filters for sub models don't generate the right subqueries

When I filter on

query{
  flowBlocks(filter:{
    where: {
      block:{
        title:{
          equalTo: "Wat wil je reinigen?"
        }
      }
    }
  }){
    id
    flow{
      title
    }
    block{
      id
      title
      organization{
        id
      }
    }
  }
}

It results in the following query:

SELECT * 
FROM   `flow_block` 
WHERE  `flow_block`.`organization_id` = 1 
       AND ( EXISTS(SELECT 1 
                    FROM   `block` 
                    WHERE  ( title = 'Wat wil je reinigen?' )) );

This results in all almost all flowblocks returning while it should only return flowblocks which contain a block with this title so the query should be:

SELECT * 
FROM   `flow_block` 
WHERE  `flow_block`.`organization_id` = 1 
       AND ( EXISTS(SELECT 1 
                    FROM   `block` 
                    WHERE  ( title = 'Wat wil je reinigen?' ) 
                           AND `flow_block`.`block_id` = `block`.`id`) );

Plugin Panics

Hey, trying out your library and experiencing a few issues. Looking for help debugging what's going on.

I'm assuming there is something strange with my DB schema (Postgres). Wondering if you can offer any advice on how to debug. Using this library will save me tons of time. Happy to dig in but thought you might see some obvious issues.

When I run the plugin I get the following panic

% go run convert_plugin.go                                                                                                     
[convert] get boiler models
[convert] get extra's from schema
[convert] get model with information
[WARN] boiler name not available for  RecipeBatch.Additives
[WARN] boiler name not available for  RecipeBatch.Fragrances
[WARN] boiler name not available for  RecipeBatch.Lipids
[WARN] boiler type not available for  Lye
[WARN] boiler name not available for  RecipeBatch.Lye
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0x130bcda]

goroutine 1 [running]:
github.com/web-ridge/gqlgen-sqlboiler/v2.getConvertConfig(0x0, 0x0, 0x0, 0xc00010d440, 0xc0008fb680, 0x1, 0x13edb77, 0x26, 0xc0008edfa0, 0x17, ...)
	/Users/steve/src/github.com/sjhitchner/gqlgen-sqlboiler/convert.go:863 +0x78a
github.com/web-ridge/gqlgen-sqlboiler/v2.enhanceModelsWithFields(0x0, 0x0, 0x0, 0xc00029c5c0, 0xc0000d0c60, 0xc000301400, 0x6c, 0x80)
	/Users/steve/src/github.com/sjhitchner/gqlgen-sqlboiler/convert.go:455 +0xa89
github.com/web-ridge/gqlgen-sqlboiler/v2.GetModelsWithInformation(0x0, 0x0, 0x0, 0xc0000d0c60, 0xc000358000, 0x12, 0x12, 0x0, 0x5, 0x8)
	/Users/steve/src/github.com/sjhitchner/gqlgen-sqlboiler/convert.go:158 +0xe2
github.com/web-ridge/gqlgen-sqlboiler/v2.(*ConvertPlugin).MutateConfig(0xc0001a4310, 0xc0000d0c60, 0xc0001a4310, 0x27870a8)
	/Users/steve/src/github.com/sjhitchner/gqlgen-sqlboiler/convert.go:193 +0x491
github.com/99designs/gqlgen/api.Generate(0xc0000d0c60, 0xc0000eff48, 0x2, 0x2, 0x1, 0x1)
	/Users/steve/go/pkg/mod/github.com/99designs/[email protected]/api/generate.go:65 +0x676
main.main()
	/Users/steve/src/github.com/sjhitchner/soapcalc/backend/convert_plugin.go:34 +0x255
exit status 2

There is panic is because field.BoilerField.Relationship.Name is nil. Should this ever happen?
https://github.com/web-ridge/gqlgen-sqlboiler/blob/main/convert.go#L863

I wrapped that line in a nil check to stop the panic but get the following error when I run the plugin again.

[resolver] generate file
error!!
resolvergen: convert.gotpl: template: convert.gotpl:27:19: executing "convert.gotpl" at <$.Backend.Directory>: can't evaluate field Backend in type *gqlgen_sqlboiler.ResolverBuild
exit status 3

This is the plugin configuration

	output := gbgen.Config{
		Directory:   "generated/graphql/helpers",
		PackageName: "helpers",
	}
	backend := gbgen.Config{
		Directory:   "generated/db/models",
		PackageName: "models",
	}
	frontend := gbgen.Config{
		Directory:   "generated/graphql/models",
		PackageName: "gmodels",
	}

I used the https://github.com/web-ridge/sqlboiler-graphql-schema library is generate my graphql schema and got a similar nil error
https://github.com/web-ridge/sqlboiler-graphql-schema/blob/main/main.go#L714
relationType := boilerField.Relationship.Name
Which I changed to
relationType := boilerField.RelationshipName

This change successfully generated a valid schema but needed a few edits to deal with pluralizations, and remove relationships that I didn't want to expose.

I could post the DB schema and graphql schema but the files are over 1000 lines.

Library is very cool. Glad to help out debugging!

func BooleanFilterToMods has empty body

func BooleanFilterToMods(m *graphql_models.BooleanFilter, column string) []qm.QueryMod {

	return nil
}

resolver.go fails to generate.

8:15PM DBG [resolver] generate file
8:15PM ERR could not parse golang file error="src.go:1:1: expected 'package', found 'EOF'"
error while trying generate resolver and converts
resolver-webridge: errors while writing template to E:/server/resolver.go writeError: ,
contentError: formatting: 38:14: unknown escape sequence, importError: E:/server/resolver.go:38:14: unknown escape sequence
exit status 3

And this is the contents of Resolver.go:

"
package

Unknown revision github.com/gertd/[email protected]

It looks like there is a typo in go.mod as v0.1.2 is the latest published version of github.com/gertd/go-pluralize

data assignment is ignoring cases with string type

In some cases function itself assigned instead of its result.
In other cases it just inserts coma
Some referenced functions do not actually exist in boilergql package

We have also failed some difficulties with referencing ID type. For some reason gqlgen-sqlboiler generates it as uint even though in models generated by sqlboiler it definitely specified as string
Samples of the issue

Allow user to override generated functions in resolver

Works the same as convert overrides looks in package for other resolver names, if they exist it will still generate the resolver but the overrided resolver will be commented out

Support edges/connections [WIP]

[Question] Supporting external record identifiers as strings that look up non primary key

I'll try to explain myself fully.

In my current code, I do the following when sending data to the client;

Here is the database model

Here is the external model being constructed from the database model

I use the unique UUID as the identifier when referring to it on the end client.

Example uID:
'90c1939d-3414-11eb-8ffe-0242ac120005'

Allow user to override generated functions in convert

~~Do a three-way-diff merge for changes and let user choose parts of code which should not take over generated code.~~

~~{defaultNameOfResolver}_custom.go~~
~~Every function which is stated there will be skipped in generation.~~

Example for one-to-many (Invoice/Order)

Do we have an example of how to manage a one-to-many model?

Example: an invoice with multiple rows. When I create or update an invoice the rows are saved or modified or deleted at the same time.

Add option to sort data

It would be nice to sort on columns maybe even more than 1 if you want it. And would be cool if it was type safe in graphql schema.

sort: {
by: [name]
type: ASC | DESC
}

Support gqlgen multiple .graphql files

https://gqlgen.com/recipes/migration-0.11/

Blocked by:
web-ridge/sqlboiler-graphql-schema#1

Congratulations and questions/proposals/fears

Your work is amazing!

It feels like every line of code is sweaty. 😄

A huge thank you for all this.

In a few hours I want to start a project with this.

Issues

Copying the code from the Readme for convert_plugin.go I have to write an additional argument for gbgen.NewConvertPlugin. What should I write there?

Proposals

I undertake to write documentation for every single thing I learn, day after day: explanations that maybe those who created it take for granted, recipes for routine things etc.
But we have to decide where to write them: one thing that doesn't steal more than 10 minutes of setup. A Gitbook on github pages?

Fears

How to write business logic after all the generation?

Example:

If I'm creating a Player for a Team I wanna check if Team has enough "space" before (in a simple DB transaction).

Is this eventuality considered? I think so, of course. I just don't know where to start. 😢
Have you ever needed to use something for tenants in the DB?

Something like https://github.com/influitive/apartment for Rails.

Quite simply adding WHERE tenant_id = [ID] to each DB query and the same for insert/update/delete queries.

Thanks again for this huge huge huge work! 😄

Add example code + video tutorial so user can see this library in action

Fix filtering for string id's

We need to fix filters when model has string primary id's instead of uints
#12 (comment)

Allow id's other than uint like int again since PostgresSQL does not support uint

https://stackoverflow.com/questions/20810134/why-unsigned-integer-is-not-available-in-postgresql

Panics when building outside of GOPATH

Piggy backing off my issue on your examples: web-ridge/gqlgen-sqlboiler-examples#2

and this issue: #3

Is this library able to work outside of the GOPATH while using modules?

I created a simple example that attempts to use postgres. Unfortunately, I haven't been able to get it to work at all. It panics everytime I go go run convert_plugin.go

go run convert_plugin.go
[convert] get boiler models
[convert] get extra's from schema
[convert] get model with information
[convert] render preload.gotpl
renderError template.gotpl: template: template.gotpl:26:3: executing "template.gotpl" at <reserveImport $.BackendModelsPath>: error calling reserveImport: empty ambient import
[convert] render convert.gotpl
renderError template.gotpl: template: template.gotpl:27:3: executing "template.gotpl" at <reserveImport $.BackendModelsPath>: error calling reserveImport: empty ambient import
[convert] render convert_input.gotpl
renderError template.gotpl: template: template.gotpl:27:3: executing "template.gotpl" at <reserveImport $.BackendModelsPath>: error calling reserveImport: empty ambient import
[convert] render filter.gotpl
renderError template.gotpl: template: template.gotpl:26:3: executing "template.gotpl" at <reserveImport $.BackendModelsPath>: error calling reserveImport: empty ambient import
panic: recursive or concurrent call to RenderToFile detected

Would love some insight here.

Example prpoject: https://github.com/randallmlough/gql-boil-psql

boilergql.GetPreloadModsWithLevel generated with invalid amount of args

mods := boilergql.GetPreloadModsWithLevel(ctx, AgreementPreloadMap, AgreementPayloadPreloadLevels.Agreement)

There is expected one more argument level

Different foreign key than table name

E.g. it understands if

organization{
   galleryItemId: 
}

galleryItem{
   id:
}

But not

organization{
   logoId: 
}

galleryItem{
   id:
}

IDFilterToMods converts ids to uint

type IDFilter struct {
	EqualTo    *string  `json:"equalTo"`
	NotEqualTo *string  `json:"notEqualTo"`
	In         []string `json:"in"`
	NotIn      []string `json:"notIn"`
}

func IDFilterToMods(m *graphql_models.IDFilter, column string) []qm.QueryMod {
	if m == nil {
		return nil
	}
	var queryMods []qm.QueryMod
	if m.EqualTo != nil {
		queryMods = append(queryMods, qmhelper.Where(column, qmhelper.EQ, boilergql.IDToBoiler(*m.EqualTo)))
	}
	if m.NotEqualTo != nil {
		queryMods = append(queryMods, qmhelper.Where(column, qmhelper.NEQ, boilergql.IDToBoiler(*m.NotEqualTo)))
	}
	if len(m.In) > 0 {
		queryMods = append(queryMods, qm.WhereIn(column+in, boilergql.IDsToBoilerInterfaces(m.In)...))
	}
	if len(m.NotIn) > 0 {
		queryMods = append(queryMods, qm.WhereIn(column+notIn, boilergql.IDsToBoilerInterfaces(m.NotIn)...))
	}

	return queryMods
}

expected

func IDFilterToMods(m *graphql_models.IDFilter, column string) []qm.QueryMod {
	if m == nil {
		return nil
	}
	var queryMods []qm.QueryMod

	if m.EqualTo != nil {
		queryMods = append(queryMods, qmhelper.Where(column, qmhelper.EQ, *m.EqualTo))
	}
	if m.NotEqualTo != nil {
		queryMods = append(queryMods, qmhelper.Where(column, qmhelper.NEQ, *m.NotEqualTo))
	}
	if len(m.In) > 0 {
		queryMods = append(queryMods, qm.WhereIn(column+in, boilergql.IDsToInterfaces(m.In)...))
	}
	if len(m.NotIn) > 0 {
		queryMods = append(queryMods, qm.WhereIn(column+notIn, boilergql.IDsToInterfaces(m.NotIn)...))
	}
	
	return queryMods
}

undefined conversion functions

undefined: boilergql.PointerIntToTimeTime

graph/helpers/convert.go:238:14: undefined: boilergql.PgeoNullPointToInt
graph/helpers/convert.go:451:14: undefined: boilergql.NullDotFloat32ToPointerFloat64
graph/helpers/convert.go:452:14: undefined: boilergql.NullDotInt16ToPointerInt
graph/helpers/convert.go:500:19: undefined: boilergql.NullDotFloat32ToPointerFloat64
graph/helpers/convert.go:501:19: undefined: boilergql.NullDotInt16ToPointerInt
graph/helpers/convert.go:502:19: undefined: boilergql.NullDotFloat32ToPointerFloat64
graph/helpers/convert.go:503:19: undefined: boilergql.NullDotInt16ToPointerInt
graph/helpers/convert_input.go:186:3: cannot use boilergql.IDToNullBoiler(boilergql.PointerStringToString(m.MeetingID)) (type null.Uint) as type null.String in field value
graph/helpers/convert_input.go:254:3: cannot use boilergql.IDToNullBoiler(boilergql.PointerStringToString(m.MeetingID)) (type null.Uint) as type null.String in field value
graph/helpers/convert_input.go:256:14: undefined: boilergql.PointerIntToTimeTime

Wrong field name & empty type

I got a wrong schema which I think it is caused due to bug while parsing the boiler struct. As you see below I have a field UserID which is not parsed correctly and eventually results to wrong generated field and empty type. Some other irrelevant part of the structs are omitted.

sqlboiler:

type Member struct {
	UserID   int    `boil:"user_id" json:"user_id" toml:"user_id" yaml:"user_id"`
	Username string `boil:"username" json:"username" toml:"username" yaml:"username"`
	RealName string `boil:"real_name" json:"real_name" toml:"real_name" yaml:"real_name"`
}

generated

type Member {
	user: !
	username: String!
	realName: String!
}

Variable transformations called by convert_input.go not defined in utils-go/boilergql/[email protected]/convert.go

Following transformation functions not declared by package boilergql ("github.com/web-ridge/utils-go/boilergql/v3/convert.go")

Float32ToFloat64
Float64ToFloat32
PointerFloat64ToFloat32

PointerIntToInt8
PointerIntToUint
UintToString
StringToUint
PointerStringToUint

PointerStringToNullDotJSON
PointerStringToTypesJSON
StringToTypesJSON
TypesJSONToString

Special case this one, In order to get the SQLBoiler type, null.JSON to pass the schema step, I added a condition to output the json type as a string field.

Not sure how to fix this problem, I was thinking the library could just send out the value of the json object as a string, or undefined/empty if !Valid or null (github.com/volatiletech/null/v8/json.go)

ambiguous use of UserIDFromContext

UserIDFromContext seems to be required with 2 different returns types:

string
null.String

Is issue coming from our schema? or it something generated inappropriately?

Helpers package output breaks if in sub directory

If I attempt to put the outputted helpers dir as a sub directory, for example graph/helpers then the helpers package name will be package graph/helpers which obviously doesn't work

func main() {
	cfg, err := config.LoadConfigFromDefaultLocations()
	if err != nil {
		fmt.Fprintln(os.Stderr, "failed to load config", err.Error())
		os.Exit(2)
	}

	convertHelpersDir := "graph/helpers"
	sqlboilerDir := "models"
	gqlgenModelDir := "graphql_models"

	err = api.Generate(cfg,
		api.AddPlugin(gbgen.NewConvertPlugin(
			convertHelpersDir, // directory where convert.go, convert_input.go and preload.go should live
			sqlboilerDir,      // directory where sqlboiler files are put
			gqlgenModelDir,    // directory where gqlgen models live
		)),
      
       // ...

Make preloads more memory efficient with less things generated

Preload generating generates too much preload code. We need something else at runtime to prevent too much generated clutter: https://github.com/web-ridge/gqlgen-sqlboiler-examples/blob/master/social-network/helpers/preload.go

IDToRemove should be send as pointer

{"level":"error","error":"obj type should be *Type, *[]Type, or *[]*Type but was \"[]boilergql.RemovedStringID\"","time":"2020-05-22T20:19:54Z","message":"could not delete dealSearchStates"}

Generated code is

	var IDsToRemove []boilergql.RemovedStringID
	if err := dm.UserSocials(mods...).Bind(ctx, r.db, IDsToRemove); err != nil {
		log.Error().Err(err).Msg(publicUserSocialBatchDeleteError)
		return nil, errors.New(publicUserSocialBatchDeleteError)
	}

expected code

	var IDsToRemove []boilergql.RemovedStringID
	if err := dm.UserSocials(mods...).Bind(ctx, r.db, &IDsToRemove); err != nil {
		log.Error().Err(err).Msg(publicUserSocialBatchDeleteError)
		return nil, errors.New(publicUserSocialBatchDeleteError)
	}

Batch create generation in resolvers

Support all gqlgen scalars

Issue while running `go run convert_plugin.go`

Thanks for this amazing library!

Currently when executing go run convert_plugin.go I'm getting the following error

panic: recursive or concurrent call to RenderToFile detected

my gqlgen.yml looks like this

schema:
  - schema.graphql
exec:
  filename: graphql/generated.go
  package: graphql
model:
  filename: graphql/models/generated_models.go
  package: gqlmodels
resolver:
  filename: graphql/resolver.go
  type: Resolver

any idea why this would be an issue?

Use npx instead of npm install for prettier

Support Global Object Identification

Implement https://graphql.org/learn/global-object-identification/ to fully support Relay.dev and other specs

MIT License?

Saw your issue at 99designs/gqlgen#646, i can't use LGPL. Can you do MIT (same license as gqlgen)?

v3 error: package [PRJ NAME]/helpers is not in GOROOT

I'm getting this error now:

9:43PM DBG write GraphQL schema to disk bytes=7869 file=schema.graphql
9:43PM DBG [convert] get boiler models
9:43PM DBG [convert] get extra's from schema
9:43PM DBG [convert] get model with information
9:43PM WRN no models found in graphql so skipping generation
9:43PM DBG [resolver] get boiler models
9:43PM DBG [resolver] get models with information
9:43PM DBG [resolver] generate file
error while trying generate resolver and converts
validation failed: packages.Load: resolver.go:6:2: package api_gqlgen_sqlboiler/helpers is not in GOROOT (c:\go\src\api_gqlgen_sqlboiler\helpers)
C:\api-gqlgen-sqlboiler\resolver.go:6:4: could not import api_gqlgen_sqlboiler/helpers (invalid package name: "")
C:\api-gqlgen-sqlboiler\resolver.go:19:10: undeclared name: queryResolver
C:\api-gqlgen-sqlboiler\resolver.go:19:77: Node not declared by package graphql_models
C:\api-gqlgen-sqlboiler\resolver.go:6:2: "api_gqlgen_sqlboiler/helpers" imported but not used

exit status 3

I'm using this go.mod file:

module api_gqlgen_sqlboiler

go 1.15

require(...)

Helpers funcs always assumes graphql_models as the package

It would be nice to be able to customize the package the helper package / funcs return. From what I can tell it will always assume the package name is graphql_models even if I change the directory to be app as shown below.

func main() {
	cfg, err := config.LoadConfigFromDefaultLocations()
	if err != nil {
		fmt.Fprintln(os.Stderr, "failed to load config", err.Error())
		os.Exit(2)
	}

	convertHelpersDir := "helpers"
	sqlboilerDir := "models"
	gqlgenModelDir := "app"

	err = api.Generate(cfg,
		api.AddPlugin(gbgen.NewConvertPlugin(
			convertHelpersDir, // directory where convert.go, convert_input.go and preload.go should live
			sqlboilerDir,      // directory where sqlboiler files are put
			gqlgenModelDir,    // directory where gqlgen models live
		)),

If this package name could be customized or at least respect the gqlgenModelDir variable that would be nice
https://github.com/web-ridge/gqlgen-sqlboiler/blob/master/convert.gotpl#L31

Order of createdAt, deletedAt and updatedAt have different order if schema.graphql is re-generated

I see some fields in the schema.graphql are written everytime in different order.

Example

First time:

input UserWhere {
  id: IDFilter
  email: StringFilter
  password: StringFilter
  updatedAt: IntFilter // this is here now
  createdAt: IntFilter
}

Second time:

input UserWhere {
  id: IDFilter
  email: StringFilter
  password: StringFilter
  createdAt: IntFilter
  updatedAt: IntFilter // this is here now
}

And so on...

This is a problem as you can imagine.

empty ambient import error

when i run convert_plugin.go after setting up everything i encounter this error
[covert] render filter.gotp; renderError template.gotpl: template: template.gotpl:27:3: executing "template.gotpl" at <reserveImport $.BackendModelsPath>: error calling reserveImport: empty ambient import

Record composite primary key incorrectly referred to by non existent single field "ID" and input translation not generated

Model with composite primary key incorrectly referred to by non existent single field "ID"

Take a model with two primary keys called "ThreadVote"
ThreadID
UserID
CreatedAt
Vote

Parts of the generated code incorrectly assumes that it has a single primary key called ID,

For example instead of

dm.ThreadVoteWhere.ThreadID AND dm.ThreadVoteWhere.UserID

It refers to the table by

dm.ThreadVoteWhere.ID.EQ(dbID)

Key translations in mutations are also not generated

For example;

dbID := ThreadVoteID(id)

Does not exist.

Config file from 3.0?

Following the new Readme I'm getting this error:

failed to load config file does not exist.

I think we need a config file but there is no mention in the Readme.

NewResolverPlugin issue

Hey, finally got time to come back to using gqlgen-sqlboiler.

I'm running into another issue.

Here is my convert_plugin.go

package main

import (
	"fmt"
	"os"

	"github.com/99designs/gqlgen/api"
	"github.com/99designs/gqlgen/codegen/config"
	gbgen "github.com/web-ridge/gqlgen-sqlboiler/v3"
)

func main() {
	cfg, err := config.LoadConfigFromDefaultLocations()
	if err != nil {
		fmt.Fprintln(os.Stderr, "failed to load config", err.Error())
		os.Exit(2)
	}

	output := gbgen.Config{
		Directory: "generated/graphql/helpers",
		// Directory:   "helpers", // supports root or sub directories
		PackageName: "helpers",
	}

	backend := gbgen.Config{
		Directory: "generated/db/models",
		// Directory:   "models",
		PackageName: "models",
	}

	frontend := gbgen.Config{
		Directory: "generated/graphql/models",
		// Directory:   "graphql_models",
		PackageName: "gqlmodels",
	}

	if err = gbgen.SchemaWrite(
		gbgen.SchemaConfig{
			BoilerModelDirectory: backend,
			// Directives:           []string{"IsAuthenticated"},
			// GenerateBatchCreate:  false, // Not implemented yet
			GenerateMutations:   true,
			GenerateBatchDelete: false,
			GenerateBatchUpdate: false,
		},
		"schema/schema.graphqls",
		gbgen.SchemaGenerateConfig{
			MergeSchema: false, // uses three way merge to keep your customization
		},
	); err != nil {
		fmt.Println("error while trying to generate schema.graphql")
		fmt.Fprintln(os.Stderr, err.Error())
		os.Exit(3)
	}

	if err = api.Generate(cfg,
		api.AddPlugin(gbgen.NewConvertPlugin(
			output,   // directory where convert.go, convert_input.go and preload.go should live
			backend,  // directory where sqlboiler files are put
			frontend, // directory where gqlgen models live
			gbgen.ConvertPluginConfig{
				//UseReflectWorkaroundForSubModelFilteringInPostgresIssue25: true, // see issue #25 on GitHub
			},
		)),
		api.AddPlugin(gbgen.NewResolverPlugin(
			output,
			backend,
			frontend,
			"", // leave empty if you don't have auth
		)),
	); err != nil {
		fmt.Println("error while trying generate resolver and converts")
		fmt.Fprintln(os.Stderr, err.Error())
		os.Exit(3)
	}
}

When I run it I get the following output/error

% go run convert_plugin.go
...
9:53PM DBG write GraphQL schema to disk bytes=36923 file=schema/schema.graphqls
9:53PM DBG [convert] get boiler models
9:53PM DBG [convert] get extra's from schema
9:53PM DBG [convert] get model with information
9:53PM DBG [convert] render preload.gotpl
9:53PM DBG [convert] render convert.gotpl
9:53PM DBG [convert] render convert_input.gotpl
9:53PM DBG [convert] render filter.gotpl
9:53PM DBG [convert] render sort.gotpl
9:54PM DBG [resolver] get boiler models
9:54PM DBG [resolver] get models with information
9:54PM DBG [resolver] generate file
error while trying generate resolver and converts
resolvergen: convert.gotpl: template: convert.gotpl:30:18: executing "convert.gotpl" at <.Enums>: can't evaluate field Enums in type *gbgen.ResolverBuild
exit status 3

It seems strange to me that the resolver is using the convert.gotpl so I dug in a little further.

This Render statement doesn't include a Template option is this ok?

gqlgen-sqlboiler/resolver.go

Line 191 in 049ffb3

err := templates.Render(templates.Options{

Locally I modified that file to

		err := templates.Render(templates.Options{
			Template:    getTemplate("resolver.gotpl"),
			PackageName: data.Config.Resolver.Package,
			PackageDoc: `
				// This file will be automatically regenerated based on the schema, any resolver implementations
				// will be copied through when generating and any unknown code will be moved to the end.`,
			Filename: filename,
			Data:     resolverBuild,
			Packages: data.Config.Packages,
		})

When I run convert_plugin.go using the modified file I get the following output

% go run convert_plugin.go
...
9:57PM DBG write GraphQL schema to disk bytes=36923 file=schema/schema.graphqls
9:58PM DBG [convert] get boiler models
9:58PM DBG [convert] get extra's from schema
9:58PM DBG [convert] get model with information
9:58PM DBG [convert] render preload.gotpl
9:58PM DBG [convert] render convert.gotpl
9:58PM DBG [convert] render convert_input.gotpl
9:58PM DBG [convert] render filter.gotpl
9:58PM DBG [convert] render sort.gotpl
gofmt failed on schema.resolvers.go: /Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:475:8: expected 'IDENT', found '=' (and 607 more errors)
9:58PM DBG [resolver] get boiler models
9:58PM DBG [resolver] get models with information
9:58PM DBG [resolver] generate file
gofmt failed on schema.resolvers.go: /Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:64:8: expected 'IDENT', found '=' (and 911 more errors)
error while trying generate resolver and converts
validation failed: packages.Load: /Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:64:8: expected 'IDENT', found '='
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:64:8: missing constant value
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:64:10: expected ';', found ""
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:69:8: expected 'IDENT', found '='
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:69:8: missing constant value
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:69:10: expected ';', found ""
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:74:8: expected 'IDENT', found '='
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:74:8: missing constant value
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:74:10: expected ';', found ""
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:79:8: expected 'IDENT', found '='
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:79:8: missing constant value
/Users/steve/src/github.com/sjhitchner/soapcalc/backend/graphql/resolver/schema.resolvers.go:79:10: expected ';', found ""

It looks like the template executes correctly but the resolver code generated are not valid Go.

const inputKey = "input"

const  = ""

	func (r *mutationResolver) CreateAdditive (ctx context.Context, input gmodels.AdditiveCreateInput) (*gmodels.AdditivePayload, error) {
	}

const  = ""

	func (r *mutationResolver) UpdateAdditive (ctx context.Context, id string, input gmodels.AdditiveUpdateInput) (*gmodels.AdditivePayload, error) {
	}

const  = ""

	func (r *mutationResolver) DeleteAdditive (ctx context.Context, id string) (*gmodels.AdditiveDeletePayload, error) {
	}

const  = ""

	func (r *mutationResolver) CreateAdditiveInventory (ctx context.Context, input gmodels.AdditiveInventoryCreateInput) (*gmodels.AdditiveInventoryPayload, error) {
	}

const  = ""

	func (r *mutationResolver) UpdateAdditiveInventory (ctx context.Context, id string, input gmodels.AdditiveInventoryUpdateInput) (*gmodels.AdditiveInventoryPayload, error) {
	}

Any ideas?

Example social-network issue of `gitlab.com/webridge-git/api` missing.

clone this repo
run cd example/social-network && go run convert_plugin.go
step 2 would throw message followed

go: gitlab.com/webridge-git/[email protected]: reading https://goproxy.cn/gitlab.com/webridge-git/api/@v/v0.0.0-20181123203207-501884b46b27.mod: 404 Not Found

Generation of pagination resolver

Does gqlgen-sqlboiler currently support the generation of offset pagination resolvers?

I see that sqlboiler-graphql-schema can generate a schema with offset pagination support but the resolvers are not getting created when I run convert.

I didn't see the generation of pagination resolvers on the roadmap or done, so I figured I would ask to see if I'm just missing something or if this feature is not available.

Thanks.

Apply a filter even if it is nil if it is explicitly specified in the GraphQL request

Not sure if this is the right place for the issue. So direct me if I'm wrong. But as the resolver is generated by gqlgen-sqlboiler it might be the right spot.

Example below returns all fields from the database, but should return empty list

query {
    deals(filter: {
        where: {
            creator: {
                id: {
                    #equalTo: ""
                }
            }
        }
        }){
            id
        }
}

Sort preloads/other maps to prevent commit clutter when re-generating

When we re-run the go run convert_plugin.go there is everytime changes since maps are random in golang. That means that there is a lot of diff change while in reality nothing has changed only the order of preloads.

v3 error: RemovedIDsToBoilerInt64 not declared by package boilergql

From v3:

9:52PM DBG write GraphQL schema to disk bytes=6293 file=schema.graphql
9:52PM DBG [convert] get boiler models
9:52PM DBG [convert] get extra's from schema
9:52PM DBG [convert] get model with information
9:52PM DBG [convert] render preload.gotpl
9:52PM DBG [convert] render convert.gotpl
9:52PM DBG [convert] render convert_input.gotpl
9:52PM DBG [convert] render filter.gotpl
9:52PM DBG [convert] render sort.gotpl
9:52PM DBG [resolver] get boiler models
9:52PM DBG [resolver] get models with information
9:52PM DBG [resolver] generate file
error while trying generate resolver and converts
validation failed: packages.Load: C:\api-gqlgen-sqlboiler\resolver.go:129:25: RemovedIDsToBoilerInt64 not declared by package boilergql
C:\api-gqlgen-sqlboiler\resolver.go:136:18: Int64IDsToGraphQL not declared by package boilergql
C:\api-gqlgen-sqlboiler\resolver.go:245:25: RemovedIDsToBoilerInt64 not declared by package boilergql
C:\api-gqlgen-sqlboiler\resolver.go:252:18: Int64IDsToGraphQL not declared by package boilergql
C:\api-gqlgen-sqlboiler\resolver.go:361:25: RemovedIDsToBoilerInt64 not declared by package boilergql
C:\api-gqlgen-sqlboiler\resolver.go:368:18: Int64IDsToGraphQL not declared by package boilergql

exit status 3

Blacklist/Whitelist for sqlboiler models

I think it should be possible to have a blacklist/whitelist for the models I want to convert from sqlboiler ones.

Is it already possible today?