1
0
mirror of https://github.com/IBM/fp-go.git synced 2026-03-04 13:21:02 +02:00

Compare commits

...

9 Commits

Author SHA1 Message Date
Dr. Carsten Leue
94b1ea30d1 fix: improved doc
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-03-02 13:23:59 +01:00
Dr. Carsten Leue
a77d61f632 fix: add Bind for Codec
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-03-02 13:01:21 +01:00
renovate[bot]
66b2f57d73 fix(deps): update module github.com/urfave/cli/v3 to v3.7.0 (#157)
Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>
2026-03-02 08:40:21 +00:00
Dr. Carsten Leue
92eb2a50a2 fix: support MarshalJSON and MarshalText types
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-03-01 22:38:20 +01:00
Dr. Carsten Leue
3df1dca146 fix: better result type for Pipe
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-02-27 17:03:12 +01:00
Dr. Carsten Leue
a0132e2e92 fix: parameter order
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-02-27 15:08:11 +01:00
Dr. Carsten Leue
c6b342908d doc: better explanation for logger
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-02-27 13:52:08 +01:00
Dr. Carsten Leue
962237492f doc: explain Effect
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-02-27 13:27:20 +01:00
Dr. Carsten Leue
168a6e1072 fix: add Eitherize to Effect
Signed-off-by: Dr. Carsten Leue <carsten.leue@de.ibm.com>
2026-02-27 12:55:03 +01:00
16 changed files with 3191 additions and 880 deletions

View File

@@ -1,6 +1,6 @@
# Functional I/O in Go: Context, Errors, and the Reader Pattern
This document explores how functional programming principles apply to I/O operations in Go, comparing traditional imperative approaches with functional patterns using the `context/readerioresult` and `idiomatic/context/readerresult` packages.
This document explores how functional programming principles apply to I/O operations in Go, comparing traditional imperative approaches with functional patterns using the `context/readerioresult`, `idiomatic/context/readerresult`, and `effect` packages.
## Table of Contents
@@ -10,6 +10,7 @@ This document explores how functional programming principles apply to I/O operat
- [Benefits of the Functional Approach](#benefits-of-the-functional-approach)
- [Side-by-Side Comparison](#side-by-side-comparison)
- [Advanced Patterns](#advanced-patterns)
- [The Effect Package: Higher-Level Abstraction](#the-effect-package-higher-level-abstraction)
- [When to Use Each Approach](#when-to-use-each-approach)
## Why Context in I/O Operations
@@ -775,6 +776,191 @@ func FetchWithRetry(url string, maxRetries int) RIO.ReaderIOResult[[]byte] {
}
```
## The Effect Package: Higher-Level Abstraction
### What is Effect?
The `effect` package provides a higher-level abstraction over `ReaderReaderIOResult`, offering a complete effect system for managing dependencies, errors, and side effects in a composable way. It's inspired by [effect-ts](https://effect.website/) and provides a cleaner API for complex workflows.
### Core Type
```go
// Effect represents an effectful computation that:
// - Requires a context of type C (dependency injection)
// - Can perform I/O operations
// - Can fail with an error
// - Produces a value of type A on success
type Effect[C, A any] = readerreaderioresult.ReaderReaderIOResult[C, A]
```
**Key difference from ReaderIOResult**: Effect adds an additional layer of dependency injection (the `C` type parameter) on top of the runtime `context.Context`, enabling type-safe dependency management.
### When to Use Effect
Use the Effect package when you need:
1. **Type-Safe Dependency Injection**: Your application has typed dependencies (config, services, repositories) that need to be threaded through operations
2. **Complex Workflows**: Multiple services and dependencies need to be composed
3. **Testability**: You want to easily mock dependencies by providing different contexts
4. **Separation of Concerns**: Clear separation between business logic, dependencies, and I/O
### Effect vs ReaderIOResult
```go
// ReaderIOResult - depends only on runtime context
type ReaderIOResult[A any] = func(context.Context) (A, error)
// Effect - depends on typed context C AND runtime context
type Effect[C, A any] = func(C) func(context.Context) (A, error)
```
**ReaderIOResult** is simpler and suitable when you only need runtime context (cancellation, deadlines, request-scoped values).
**Effect** adds typed dependency injection, making it ideal for applications with complex service dependencies.
### Basic Usage
#### Creating Effects
```go
type AppConfig struct {
DatabaseURL string
APIKey string
}
// Create a successful effect
successEffect := effect.Succeed[AppConfig, string]("hello")
// Create a failed effect
failEffect := effect.Fail[AppConfig, string](errors.New("failed"))
// Lift a pure value
pureEffect := effect.Of[AppConfig, int](42)
```
#### Integrating Standard Go Functions
The `Eitherize` function makes it easy to integrate standard Go functions that return `(value, error)`:
```go
type Database struct {
conn *sql.DB
}
// Convert a standard Go function to an Effect using Eitherize
func fetchUser(id int) effect.Effect[Database, User] {
return effect.Eitherize(func(db Database, ctx context.Context) (User, error) {
var user User
err := db.conn.QueryRowContext(ctx, "SELECT * FROM users WHERE id = ?", id).Scan(&user)
return user, err
})
}
// Use Eitherize1 for Kleisli arrows (functions with an additional parameter)
fetchUserKleisli := effect.Eitherize1(func(db Database, ctx context.Context, id int) (User, error) {
var user User
err := db.conn.QueryRowContext(ctx, "SELECT * FROM users WHERE id = ?", id).Scan(&user)
return user, err
})
// fetchUserKleisli has type: func(int) Effect[Database, User]
```
#### Composing Effects
```go
type Services struct {
UserRepo UserRepository
EmailSvc EmailService
}
// Compose multiple effects with typed dependencies
func processUser(id int, newEmail string) effect.Effect[Services, User] {
return F.Pipe3(
// Fetch user from repository
effect.Eitherize(func(svc Services, ctx context.Context) (User, error) {
return svc.UserRepo.GetUser(ctx, id)
}),
// Validate user (pure function lifted into Effect)
effect.ChainEitherK[Services](validateUser),
// Update email
effect.Chain[Services](func(user User) effect.Effect[Services, User] {
return effect.Eitherize(func(svc Services, ctx context.Context) (User, error) {
if err := svc.EmailSvc.SendVerification(ctx, newEmail); err != nil {
return User{}, err
}
return svc.UserRepo.UpdateEmail(ctx, user.ID, newEmail)
})
}),
)
}
```
#### Running Effects
```go
func main() {
// Set up typed dependencies once
services := Services{
UserRepo: &PostgresUserRepo{db: db},
EmailSvc: &SMTPEmailService{host: "smtp.example.com"},
}
// Build the effect pipeline (no execution yet)
userEffect := processUser(42, "new@email.com")
// Provide dependencies - returns a Thunk (ReaderIOResult)
thunk := effect.Provide(services)(userEffect)
// Run synchronously - returns a func(context.Context) (User, error)
readerResult := effect.RunSync(thunk)
// Execute with runtime context
user, err := readerResult(context.Background())
if err != nil {
log.Fatal(err)
}
fmt.Printf("Updated user: %+v\n", user)
}
```
### Comparison: Traditional vs ReaderIOResult vs Effect
| Aspect | Traditional | ReaderIOResult | Effect |
|---|---|---|---|
| Error propagation | Manual | Automatic | Automatic |
| Dependency injection | Function parameters | Closure / `context.Context` | Typed `C` parameter |
| Testability | Requires mocking | Mock `ReaderIOResult` | Provide mock `C` |
| Composability | Low | High | High |
| Type-safe dependencies | No | No | Yes |
| Complexity | Low | Medium | Medium-High |
### Testing with Effect
One of the key benefits of Effect is easy testing through dependency substitution:
```go
func TestProcessUser(t *testing.T) {
// Create mock services
mockServices := Services{
UserRepo: &MockUserRepo{
users: map[int]User{42: {ID: 42, Age: 25, Email: "old@email.com"}},
},
EmailSvc: &MockEmailService{},
}
// Run the effect with mock dependencies
user, err := effect.RunSync(
effect.Provide(mockServices)(processUser(42, "new@email.com")),
)(context.Background())
assert.NoError(t, err)
assert.Equal(t, "new@email.com", user.Email)
}
```
No database or SMTP server needed — just provide mock implementations of the `Services` struct.
## When to Use Each Approach
### Use Traditional Go Style When:
@@ -794,6 +980,17 @@ func FetchWithRetry(url string, maxRetries int) RIO.ReaderIOResult[[]byte] {
5. **Resource management**: Need guaranteed cleanup (Bracket)
6. **Parallel execution**: Need to parallelize operations easily
7. **Type safety**: Want the type system to track I/O effects
8. **Simple dependencies**: Only need runtime `context.Context`, no typed dependencies
### Use Effect When:
1. **Type-safe dependency injection**: Application has typed dependencies (config, services, repositories)
2. **Complex service architectures**: Multiple services need to be composed with clear dependency management
3. **Testability with mocks**: Want to easily substitute dependencies for testing
4. **Separation of concerns**: Need clear separation between business logic, dependencies, and I/O
5. **Large applications**: Building applications where dependency management is critical
6. **Team experience**: Team is comfortable with functional programming and effect systems
7. **Integration with standard Go**: Need to integrate many standard `(value, error)` functions using `Eitherize`
### Use Idiomatic Functional Style (idiomatic/context/readerresult) When:
@@ -803,6 +1000,7 @@ func FetchWithRetry(url string, maxRetries int) RIO.ReaderIOResult[[]byte] {
4. **Go integration**: Want seamless integration with existing Go code
5. **Production services**: Building high-throughput services
6. **Best of both worlds**: Want functional composition with Go's native patterns
7. **Simple dependencies**: Only need runtime `context.Context`, no typed dependencies
## Summary
@@ -814,11 +1012,18 @@ The functional approach to I/O in Go offers significant advantages:
4. **Type Safety**: I/O effects visible in the type system
5. **Lazy Evaluation**: Build pipelines, execute when ready
6. **Context Propagation**: Automatic threading of context
7. **Performance**: Idiomatic version offers 2-10x speedup
7. **Dependency Injection**: Type-safe dependency management with Effect
8. **Performance**: Idiomatic version offers 2-10x speedup
The key insight is that **I/O operations return descriptions of effects** (ReaderIOResult) rather than performing effects immediately. This enables powerful composition patterns while maintaining Go's idiomatic error handling through the `(value, error)` tuple pattern.
The key insight is that **I/O operations return descriptions of effects** rather than performing effects immediately. This enables powerful composition patterns while maintaining Go's idiomatic error handling through the `(value, error)` tuple pattern.
For production Go services, the **idiomatic/context/readerresult** package provides the best balance: full functional programming capabilities with native Go performance and familiar error handling patterns.
### Choosing the Right Abstraction
- **ReaderIOResult**: Best for simple I/O pipelines that only need runtime `context.Context`
- **Effect**: Best for complex applications with typed dependencies and service architectures
- **idiomatic/context/readerresult**: Best for production services needing high performance with functional patterns
For production Go services, the **idiomatic/context/readerresult** package provides the best balance of performance and functional capabilities. For applications with complex dependency management, the **effect** package provides type-safe dependency injection with a clean, composable API.
## Further Reading
@@ -826,4 +1031,6 @@ For production Go services, the **idiomatic/context/readerresult** package provi
- [IDIOMATIC_COMPARISON.md](./IDIOMATIC_COMPARISON.md) - Performance comparison
- [idiomatic/doc.go](./idiomatic/doc.go) - Idiomatic package overview
- [context/readerioresult](./context/readerioresult/) - ReaderIOResult package
- [idiomatic/context/readerresult](./idiomatic/context/readerresult/) - Idiomatic ReaderResult package
- [idiomatic/context/readerresult](./idiomatic/context/readerresult/) - Idiomatic ReaderResult package
- [effect](./effect/) - Effect package for type-safe dependency injection
- [effect-ts](https://effect.website/) - TypeScript effect system that inspired this package

View File

@@ -461,7 +461,8 @@ func process() IOResult[string] {
- **Result** - Simplified Either with error as left type (recommended for error handling)
- **IO** - Lazy evaluation and side effect management
- **IOOption** - Combine IO with Option for optional values with side effects
- **IOResult** - Combine IO with Result for error handling (recommended over IOEither)
- **IOResult** - Combine IO with Result for error handling (recommended over IOEither when using standard `error` type)
- **Effect** - Composable effects with dependency injection and error handling
- **Reader** - Dependency injection pattern
- **ReaderOption** - Combine Reader with Option for optional values with dependency injection
- **ReaderIOOption** - Combine Reader, IO, and Option for optional values with dependency injection and side effects

View File

@@ -369,7 +369,7 @@ func onExitAny(
// )
func LogEntryExitWithCallback[A any](
logLevel slog.Level,
cb func(context.Context) *slog.Logger,
cb Reader[context.Context, *slog.Logger],
name string) Operator[A, A] {
nameAttr := slog.String("name", name)
@@ -499,12 +499,12 @@ func LogEntryExit[A any](name string) Operator[A, A] {
func curriedLog(
logLevel slog.Level,
cb func(context.Context) *slog.Logger,
message string) func(slog.Attr) func(context.Context) func() struct{} {
return F.Curry2(func(a slog.Attr, ctx context.Context) func() struct{} {
message string) func(slog.Attr) ReaderIO[Void] {
return F.Curry2(func(a slog.Attr, ctx context.Context) IO[Void] {
logger := cb(ctx)
return func() struct{} {
return func() Void {
logger.LogAttrs(ctx, logLevel, message, a)
return struct{}{}
return F.VOID
}
})
}
@@ -571,7 +571,7 @@ func curriedLog(
// - Conditional logging: Enable/disable logging based on logger configuration
func SLogWithCallback[A any](
logLevel slog.Level,
cb func(context.Context) *slog.Logger,
cb Reader[context.Context, *slog.Logger],
message string) Kleisli[Result[A], A] {
return F.Pipe1(
@@ -582,18 +582,23 @@ func SLogWithCallback[A any](
curriedLog(logLevel, cb, message),
),
// preserve the original context
reader.Chain(reader.Sequence(readerio.MapTo[struct{}, Result[A]])),
reader.Chain(reader.Sequence(readerio.MapTo[Void, Result[A]])),
)
}
// SLog creates a Kleisli arrow that logs a Result value (success or error) with a message.
//
// This function logs both successful values and errors at Info level using the logger from the context.
// This function logs both successful values and errors at Info level. It retrieves the logger
// using logging.GetLoggerFromContext, which returns either:
// - The logger stored in the context via logging.WithLogger, or
// - The global logger (set via logging.SetLogger or slog.Default())
//
// It's a convenience wrapper around SLogWithCallback with standard settings.
//
// The logged output includes:
// - For success: The message with the value as a structured "value" attribute
// - For error: The message with the error as a structured "error" attribute
// The message parameter becomes the main log message text, and the Result value or error
// is attached as a structured logging attribute:
// - For success: Logs message with attribute value=<the actual value>
// - For error: Logs message with attribute error=<the error>
//
// The Result is passed through unchanged after logging, making this function transparent in the
// computation pipeline.
@@ -647,25 +652,47 @@ func SLog[A any](message string) Kleisli[Result[A], A] {
return SLogWithCallback[A](slog.LevelInfo, logging.GetLoggerFromContext, message)
}
// TapSLog creates an operator that logs only successful values with a message and passes them through unchanged.
// TapSLog creates an operator that logs both successful values and errors with a message,
// and passes the ReaderIOResult through unchanged.
//
// This function is useful for debugging and monitoring values as they flow through a ReaderIOResult
// computation chain. Unlike SLog which logs both successes and errors, TapSLog only logs when the
// computation is successful. If the computation contains an error, no logging occurs and the error
// is propagated unchanged.
// computation chain. It logs both successful values and errors at Info level. It retrieves the logger
// using logging.GetLoggerFromContext, which returns either:
// - The logger stored in the context via logging.WithLogger, or
// - The global logger (set via logging.SetLogger or slog.Default())
//
// The logged output includes:
// - The provided message
// - The value being passed through (as a structured "value" attribute)
// The ReaderIOResult is returned unchanged after logging.
//
// The difference between TapSLog and SLog is their position in the pipeline:
// - SLog is a Kleisli[Result[A], A] used with Chain to intercept the raw Result
// - TapSLog is an Operator[A, A] used directly in a pipe on a ReaderIOResult[A]
//
// Both log the same information (success value or error), but TapSLog is more ergonomic
// when composing ReaderIOResult pipelines with F.Pipe.
//
// The message parameter becomes the main log message text, and the Result value or error
// is attached as a structured logging attribute:
// - For success: Logs message with attribute value=<the actual value>
// - For error: Logs message with attribute error=<the error>
//
// For example, TapSLog[User]("Fetched user") with a successful result produces:
//
// Log message: "Fetched user"
// Structured attribute: value={ID:123 Name:"Alice"}
//
// With an error result, it produces:
//
// Log message: "Fetched user"
// Structured attribute: error="user not found"
//
// Type Parameters:
// - A: The type of the value to log and pass through
// - A: The success type of the ReaderIOResult to log and pass through
//
// Parameters:
// - message: A descriptive message to include in the log entry
//
// Returns:
// - An Operator that logs successful values and returns them unchanged
// - An Operator that logs the Result (value or error) and returns the ReaderIOResult unchanged
//
// Example with simple value logging:
//
@@ -680,7 +707,7 @@ func SLog[A any](message string) Kleisli[Result[A], A] {
// )
//
// result := pipeline(t.Context())()
// // Logs: "Fetched user" value={ID:123 Name:"Alice"}
// // If successful, logs: "Fetched user" value={ID:123 Name:"Alice"}
// // Returns: result.Of("Alice")
//
// Example in a processing pipeline:
@@ -695,36 +722,36 @@ func SLog[A any](message string) Kleisli[Result[A], A] {
// )
//
// result := processOrder(t.Context())()
// // Logs each successful step with the intermediate values
// // If any step fails, subsequent TapSLog calls don't log
// // Logs each step with its value or error
//
// Example with error handling:
//
// pipeline := F.Pipe3(
// fetchData(id),
// TapSLog[Data]("Data fetched"),
// Chain(func(d Data) ReaderIOResult[Result] {
// Chain(func(d Data) ReaderIOResult[Data] {
// if d.IsValid() {
// return Of(processData(d))
// }
// return Left[Result](errors.New("invalid data"))
// return Left[Data](errors.New("invalid data"))
// }),
// TapSLog[Result]("Data processed"),
// TapSLog[Data]("Data processed"),
// )
//
// // If fetchData succeeds: logs "Data fetched" with the data
// // If processing succeeds: logs "Data processed" with the result
// // If processing fails: "Data processed" is NOT logged (error propagates)
// // If fetchData succeeds: logs "Data fetched" value={...}
// // If fetchData fails: logs "Data fetched" error="..."
// // If processing succeeds: logs "Data processed" value={...}
// // If processing fails: logs "Data processed" error="invalid data"
//
// Use Cases:
// - Debugging: Inspect intermediate successful values in a computation pipeline
// - Monitoring: Track successful data flow through complex operations
// - Troubleshooting: Identify where successful computations stop (last logged value before error)
// - Auditing: Log important successful values for compliance or security
// - Development: Understand data transformations during development
// - Debugging: Inspect intermediate values and errors in a computation pipeline
// - Monitoring: Track both successful and failed data flow through complex operations
// - Troubleshooting: Identify where errors are introduced or propagated
// - Auditing: Log important values and failures for compliance or security
// - Development: Understand data transformations and error paths during development
//
// Note: This function only logs successful values. Errors are silently propagated without logging.
// For logging both successes and errors, use SLog instead.
// Note: This function logs both successful values and errors. It is equivalent to SLog
// but expressed as an Operator for direct use in F.Pipe pipelines on ReaderIOResult values.
//
//go:inline
func TapSLog[A any](message string) Operator[A, A] {

View File

@@ -0,0 +1,213 @@
// Copyright (c) 2023 - 2025 IBM Corp.
// All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package readerreaderioresult
import (
"context"
F "github.com/IBM/fp-go/v2/function"
"github.com/IBM/fp-go/v2/ioresult"
)
// Eitherize converts a function that returns a value and error into a ReaderReaderIOResult.
//
// This function takes a function that accepts an outer context R and context.Context,
// returning a value T and an error, and converts it into a ReaderReaderIOResult[R, T].
// The error is automatically converted into the Left case of the Result, while successful
// values become the Right case.
//
// This is particularly useful for integrating standard Go error-handling patterns into
// the functional programming style of ReaderReaderIOResult. It is especially helpful
// for adapting interface member functions that accept a context. When you have an
// interface method with signature (receiver, context.Context) (T, error), you can
// use Eitherize to convert it into a ReaderReaderIOResult where the receiver becomes
// the outer reader context R.
//
// # Type Parameters
//
// - R: The outer reader context type (e.g., application configuration)
// - T: The success value type
//
// # Parameters
//
// - f: A function that takes R and context.Context and returns (T, error)
//
// # Returns
//
// - ReaderReaderIOResult[R, T]: A computation that depends on R and context.Context,
// performs IO, and produces a Result[T]
//
// # Example Usage
//
// type AppConfig struct {
// DatabaseURL string
// }
//
// // A function using standard Go error handling
// func fetchUser(cfg AppConfig, ctx context.Context) (*User, error) {
// // Implementation that may return an error
// return &User{ID: 1, Name: "Alice"}, nil
// }
//
// // Convert to ReaderReaderIOResult
// fetchUserRR := Eitherize(fetchUser)
//
// // Use in functional composition
// result := F.Pipe1(
// fetchUserRR,
// Map[AppConfig](func(u *User) string { return u.Name }),
// )
//
// // Execute with config and context
// cfg := AppConfig{DatabaseURL: "postgres://localhost"}
// outcome := result(cfg)(context.Background())()
//
// # Adapting Interface Methods
//
// Eitherize is particularly useful for adapting interface member functions:
//
// type UserRepository interface {
// GetUser(ctx context.Context, id int) (*User, error)
// }
//
// type UserRepo struct {
// db *sql.DB
// }
//
// func (r *UserRepo) GetUser(ctx context.Context, id int) (*User, error) {
// // Implementation
// return &User{ID: id}, nil
// }
//
// // Adapt the method by binding the first parameter (receiver)
// repo := &UserRepo{db: db}
// getUserRR := Eitherize(func(id int, ctx context.Context) (*User, error) {
// return repo.GetUser(ctx, id)
// })
//
// // Now getUserRR has type: ReaderReaderIOResult[int, *User]
// // The receiver (repo) is captured in the closure
// // The id becomes the outer reader context R
//
// # See Also
//
// - Eitherize1: For functions that take an additional parameter
// - ioresult.Eitherize2: The underlying conversion function
func Eitherize[R, T any](f func(R, context.Context) (T, error)) ReaderReaderIOResult[R, T] {
return F.Pipe1(
ioresult.Eitherize2(f),
F.Curry2,
)
}
// Eitherize1 converts a function that takes an additional parameter and returns a value
// and error into a Kleisli arrow.
//
// This function takes a function that accepts an outer context R, context.Context, and
// an additional parameter A, returning a value T and an error, and converts it into a
// Kleisli arrow (A -> ReaderReaderIOResult[R, T]). The error is automatically converted
// into the Left case of the Result, while successful values become the Right case.
//
// This is useful for creating composable operations that depend on both contexts and
// an input value, following standard Go error-handling patterns. It is especially helpful
// for adapting interface member functions that accept a context and additional parameters.
// When you have an interface method with signature (receiver, context.Context, A) (T, error),
// you can use Eitherize1 to convert it into a Kleisli arrow where the receiver becomes
// the outer reader context R and A becomes the input parameter.
//
// # Type Parameters
//
// - R: The outer reader context type (e.g., application configuration)
// - A: The input parameter type
// - T: The success value type
//
// # Parameters
//
// - f: A function that takes R, context.Context, and A, returning (T, error)
//
// # Returns
//
// - Kleisli[R, A, T]: A function from A to ReaderReaderIOResult[R, T]
//
// # Example Usage
//
// type AppConfig struct {
// DatabaseURL string
// }
//
// // A function using standard Go error handling
// func fetchUserByID(cfg AppConfig, ctx context.Context, id int) (*User, error) {
// // Implementation that may return an error
// return &User{ID: id, Name: "Alice"}, nil
// }
//
// // Convert to Kleisli arrow
// fetchUserKleisli := Eitherize1(fetchUserByID)
//
// // Use in functional composition with Chain
// pipeline := F.Pipe1(
// Of[AppConfig](123),
// Chain[AppConfig](fetchUserKleisli),
// )
//
// // Execute with config and context
// cfg := AppConfig{DatabaseURL: "postgres://localhost"}
// outcome := pipeline(cfg)(context.Background())()
//
// # Adapting Interface Methods
//
// Eitherize1 is particularly useful for adapting interface member functions with parameters:
//
// type UserRepository interface {
// GetUserByID(ctx context.Context, id int) (*User, error)
// UpdateUser(ctx context.Context, user *User) error
// }
//
// type UserRepo struct {
// db *sql.DB
// }
//
// func (r *UserRepo) GetUserByID(ctx context.Context, id int) (*User, error) {
// // Implementation
// return &User{ID: id}, nil
// }
//
// // Adapt the method - receiver becomes R, id becomes A
// repo := &UserRepo{db: db}
// getUserKleisli := Eitherize1(func(r *UserRepo, ctx context.Context, id int) (*User, error) {
// return r.GetUserByID(ctx, id)
// })
//
// // Now getUserKleisli has type: Kleisli[*UserRepo, int, *User]
// // Which is: func(int) ReaderReaderIOResult[*UserRepo, *User]
// // Use it in composition:
// pipeline := F.Pipe1(
// Of[*UserRepo](123),
// Chain[*UserRepo](getUserKleisli),
// )
// result := pipeline(repo)(context.Background())()
//
// # See Also
//
// - Eitherize: For functions without an additional parameter
// - Chain: For composing Kleisli arrows
// - ioresult.Eitherize3: The underlying conversion function
func Eitherize1[R, A, T any](f func(R, context.Context, A) (T, error)) Kleisli[R, A, T] {
return F.Flow2(
F.Bind3of3(ioresult.Eitherize3(f)),
F.Curry2,
)
}

View File

@@ -0,0 +1,507 @@
// Copyright (c) 2023 - 2025 IBM Corp.
// All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package readerreaderioresult
import (
"context"
"errors"
"fmt"
"strconv"
"testing"
F "github.com/IBM/fp-go/v2/function"
"github.com/IBM/fp-go/v2/result"
"github.com/stretchr/testify/assert"
)
type TestConfig struct {
Prefix string
MaxLen int
}
var testConfig = TestConfig{
Prefix: "test",
MaxLen: 100,
}
// TestEitherize_Success tests successful conversion with Eitherize
func TestEitherize_Success(t *testing.T) {
t.Run("converts successful function to ReaderReaderIOResult", func(t *testing.T) {
// Arrange
successFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return cfg.Prefix + "-success", nil
}
rr := Eitherize(successFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of("test-success"), outcome)
})
t.Run("preserves context values", func(t *testing.T) {
// Arrange
type ctxKey string
key := ctxKey("testKey")
expectedValue := "contextValue"
contextFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
value := ctx.Value(key)
if value == nil {
return "", errors.New("context value not found")
}
return value.(string), nil
}
rr := Eitherize(contextFunc)
ctx := context.WithValue(context.Background(), key, expectedValue)
// Act
outcome := rr(testConfig)(ctx)()
// Assert
assert.Equal(t, result.Of(expectedValue), outcome)
})
t.Run("works with different types", func(t *testing.T) {
// Arrange
intFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return cfg.MaxLen, nil
}
rr := Eitherize(intFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(100), outcome)
})
}
// TestEitherize_Failure tests error handling with Eitherize
func TestEitherize_Failure(t *testing.T) {
t.Run("converts error to Left", func(t *testing.T) {
// Arrange
expectedErr := errors.New("operation failed")
failFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return "", expectedErr
}
rr := Eitherize(failFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
assert.Equal(t, result.Left[string](expectedErr), outcome)
})
t.Run("preserves error message", func(t *testing.T) {
// Arrange
expectedErr := fmt.Errorf("validation error: field is required")
failFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return 0, expectedErr
}
rr := Eitherize(failFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
leftValue := result.MonadFold(outcome,
F.Identity[error],
func(int) error { return nil },
)
assert.Equal(t, expectedErr, leftValue)
})
}
// TestEitherize_EdgeCases tests edge cases for Eitherize
func TestEitherize_EdgeCases(t *testing.T) {
t.Run("handles nil context", func(t *testing.T) {
// Arrange
nilCtxFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
if ctx == nil {
return "nil-context", nil
}
return "non-nil-context", nil
}
rr := Eitherize(nilCtxFunc)
// Act
outcome := rr(testConfig)(nil)()
// Assert
assert.Equal(t, result.Of("nil-context"), outcome)
})
t.Run("handles zero value config", func(t *testing.T) {
// Arrange
zeroFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return cfg.Prefix, nil
}
rr := Eitherize(zeroFunc)
// Act
outcome := rr(TestConfig{})(context.Background())()
// Assert
assert.Equal(t, result.Of(""), outcome)
})
t.Run("handles pointer types", func(t *testing.T) {
// Arrange
type User struct {
Name string
}
ptrFunc := func(cfg TestConfig, ctx context.Context) (*User, error) {
return &User{Name: "Alice"}, nil
}
rr := Eitherize(ptrFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsRight(outcome))
user := result.MonadFold(outcome,
func(error) *User { return nil },
F.Identity[*User],
)
assert.NotNil(t, user)
assert.Equal(t, "Alice", user.Name)
})
}
// TestEitherize_Integration tests integration with other operations
func TestEitherize_Integration(t *testing.T) {
t.Run("composes with Map", func(t *testing.T) {
// Arrange
baseFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return 42, nil
}
rr := Eitherize(baseFunc)
// Act
pipeline := F.Pipe1(
rr,
Map[TestConfig](func(n int) string { return strconv.Itoa(n) }),
)
outcome := pipeline(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of("42"), outcome)
})
t.Run("composes with Chain", func(t *testing.T) {
// Arrange
firstFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return 10, nil
}
secondFunc := func(n int) ReaderReaderIOResult[TestConfig, string] {
return Of[TestConfig](fmt.Sprintf("value: %d", n))
}
// Act
pipeline := F.Pipe1(
Eitherize(firstFunc),
Chain[TestConfig](secondFunc),
)
outcome := pipeline(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of("value: 10"), outcome)
})
}
// TestEitherize1_Success tests successful conversion with Eitherize1
func TestEitherize1_Success(t *testing.T) {
t.Run("converts successful function to Kleisli", func(t *testing.T) {
// Arrange
addFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
return n + cfg.MaxLen, nil
}
kleisli := Eitherize1(addFunc)
// Act
outcome := kleisli(10)(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(110), outcome)
})
t.Run("works with string input", func(t *testing.T) {
// Arrange
concatFunc := func(cfg TestConfig, ctx context.Context, s string) (string, error) {
return cfg.Prefix + "-" + s, nil
}
kleisli := Eitherize1(concatFunc)
// Act
outcome := kleisli("input")(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of("test-input"), outcome)
})
t.Run("preserves context in Kleisli", func(t *testing.T) {
// Arrange
type ctxKey string
key := ctxKey("multiplier")
multiplyFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
multiplier := ctx.Value(key)
if multiplier == nil {
return n, nil
}
return n * multiplier.(int), nil
}
kleisli := Eitherize1(multiplyFunc)
ctx := context.WithValue(context.Background(), key, 3)
// Act
outcome := kleisli(5)(testConfig)(ctx)()
// Assert
assert.Equal(t, result.Of(15), outcome)
})
}
// TestEitherize1_Failure tests error handling with Eitherize1
func TestEitherize1_Failure(t *testing.T) {
t.Run("converts error to Left in Kleisli", func(t *testing.T) {
// Arrange
expectedErr := errors.New("division by zero")
divideFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
if n == 0 {
return 0, expectedErr
}
return 100 / n, nil
}
kleisli := Eitherize1(divideFunc)
// Act
outcome := kleisli(0)(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
assert.Equal(t, result.Left[int](expectedErr), outcome)
})
t.Run("preserves error context", func(t *testing.T) {
// Arrange
validateFunc := func(cfg TestConfig, ctx context.Context, s string) (string, error) {
if len(s) > cfg.MaxLen {
return "", fmt.Errorf("string too long: %d > %d", len(s), cfg.MaxLen)
}
return s, nil
}
kleisli := Eitherize1(validateFunc)
longString := string(make([]byte, 200))
// Act
outcome := kleisli(longString)(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
leftValue := result.MonadFold(outcome,
F.Identity[error],
func(string) error { return nil },
)
assert.Contains(t, leftValue.Error(), "string too long")
})
}
// TestEitherize1_EdgeCases tests edge cases for Eitherize1
func TestEitherize1_EdgeCases(t *testing.T) {
t.Run("handles zero value input", func(t *testing.T) {
// Arrange
zeroFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
return n, nil
}
kleisli := Eitherize1(zeroFunc)
// Act
outcome := kleisli(0)(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(0), outcome)
})
t.Run("handles pointer input", func(t *testing.T) {
// Arrange
type Input struct {
Value int
}
ptrFunc := func(cfg TestConfig, ctx context.Context, in *Input) (int, error) {
if in == nil {
return 0, errors.New("nil input")
}
return in.Value, nil
}
kleisli := Eitherize1(ptrFunc)
// Act
outcome := kleisli(&Input{Value: 42})(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(42), outcome)
})
t.Run("handles nil pointer input", func(t *testing.T) {
// Arrange
type Input struct {
Value int
}
ptrFunc := func(cfg TestConfig, ctx context.Context, in *Input) (int, error) {
if in == nil {
return 0, errors.New("nil input")
}
return in.Value, nil
}
kleisli := Eitherize1(ptrFunc)
// Act
outcome := kleisli((*Input)(nil))(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
})
}
// TestEitherize1_Integration tests integration with other operations
func TestEitherize1_Integration(t *testing.T) {
t.Run("composes with Chain", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
doubleFunc := func(n int) ReaderReaderIOResult[TestConfig, int] {
return Of[TestConfig](n * 2)
}
parseKleisli := Eitherize1(parseFunc)
// Act
pipeline := F.Pipe2(
Of[TestConfig]("42"),
Chain[TestConfig](parseKleisli),
Chain[TestConfig](doubleFunc),
)
outcome := pipeline(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(84), outcome)
})
t.Run("handles error in chain", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
parseKleisli := Eitherize1(parseFunc)
// Act
pipeline := F.Pipe1(
Of[TestConfig]("not-a-number"),
Chain(parseKleisli),
)
outcome := pipeline(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsLeft(outcome))
})
t.Run("composes multiple Kleisli arrows", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
formatFunc := func(cfg TestConfig, ctx context.Context, n int) (string, error) {
return fmt.Sprintf("%s-%d", cfg.Prefix, n), nil
}
parseKleisli := Eitherize1(parseFunc)
formatKleisli := Eitherize1(formatFunc)
// Act
pipeline := F.Pipe2(
Of[TestConfig]("123"),
Chain[TestConfig](parseKleisli),
Chain[TestConfig](formatKleisli),
)
outcome := pipeline(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of("test-123"), outcome)
})
}
// TestEitherize_TypeSafety tests type safety across different scenarios
func TestEitherize_TypeSafety(t *testing.T) {
t.Run("Eitherize with complex types", func(t *testing.T) {
// Arrange
type ComplexResult struct {
Data map[string]int
Count int
}
complexFunc := func(cfg TestConfig, ctx context.Context) (ComplexResult, error) {
return ComplexResult{
Data: map[string]int{"key": 42},
Count: 1,
}, nil
}
rr := Eitherize(complexFunc)
// Act
outcome := rr(testConfig)(context.Background())()
// Assert
assert.True(t, result.IsRight(outcome))
value := result.MonadFold(outcome,
func(error) ComplexResult { return ComplexResult{} },
F.Identity[ComplexResult],
)
assert.Equal(t, 42, value.Data["key"])
assert.Equal(t, 1, value.Count)
})
t.Run("Eitherize1 with different input and output types", func(t *testing.T) {
// Arrange
type Input struct {
ID int
}
type Output struct {
Name string
}
convertFunc := func(cfg TestConfig, ctx context.Context, in Input) (Output, error) {
return Output{Name: fmt.Sprintf("%s-%d", cfg.Prefix, in.ID)}, nil
}
kleisli := Eitherize1(convertFunc)
// Act
outcome := kleisli(Input{ID: 99})(testConfig)(context.Background())()
// Assert
assert.Equal(t, result.Of(Output{Name: "test-99"}), outcome)
})
}

208
v2/effect/eitherize.go Normal file
View File

@@ -0,0 +1,208 @@
// Copyright (c) 2023 - 2025 IBM Corp.
// All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package effect
import (
"context"
"github.com/IBM/fp-go/v2/context/readerreaderioresult"
)
// Eitherize converts a function that returns a value and error into an Effect.
//
// This function takes a function that accepts a context C and context.Context,
// returning a value T and an error, and converts it into an Effect[C, T].
// The error is automatically converted into a failure, while successful
// values become successes.
//
// This is particularly useful for integrating standard Go error-handling patterns into
// the effect system. It is especially helpful for adapting interface member functions
// that accept a context. When you have an interface method with signature
// (receiver, context.Context) (T, error), you can use Eitherize to convert it into
// an Effect where the receiver becomes the context C.
//
// # Type Parameters
//
// - C: The context type required by the effect
// - T: The success value type
//
// # Parameters
//
// - f: A function that takes C and context.Context and returns (T, error)
//
// # Returns
//
// - Effect[C, T]: An effect that depends on C, performs IO, and produces T
//
// # Example Usage
//
// type AppConfig struct {
// DatabaseURL string
// }
//
// // A function using standard Go error handling
// func fetchUser(cfg AppConfig, ctx context.Context) (*User, error) {
// // Implementation that may return an error
// return &User{ID: 1, Name: "Alice"}, nil
// }
//
// // Convert to Effect
// fetchUserEffect := effect.Eitherize(fetchUser)
//
// // Use in functional composition
// pipeline := F.Pipe1(
// fetchUserEffect,
// effect.Map[AppConfig](func(u *User) string { return u.Name }),
// )
//
// // Execute with config
// cfg := AppConfig{DatabaseURL: "postgres://localhost"}
// result, err := effect.RunSync(effect.Provide[*User](cfg)(pipeline))(context.Background())
//
// # Adapting Interface Methods
//
// Eitherize is particularly useful for adapting interface member functions:
//
// type UserRepository interface {
// GetUser(ctx context.Context, id int) (*User, error)
// }
//
// type UserRepo struct {
// db *sql.DB
// }
//
// func (r *UserRepo) GetUser(ctx context.Context, id int) (*User, error) {
// // Implementation
// return &User{ID: id}, nil
// }
//
// // Adapt the method by binding the first parameter (receiver)
// repo := &UserRepo{db: db}
// getUserEffect := effect.Eitherize(func(id int, ctx context.Context) (*User, error) {
// return repo.GetUser(ctx, id)
// })
//
// // Now getUserEffect has type: Effect[int, *User]
// // The receiver (repo) is captured in the closure
// // The id becomes the context C
//
// # See Also
//
// - Eitherize1: For functions that take an additional parameter
// - readerreaderioresult.Eitherize: The underlying implementation
//
//go:inline
func Eitherize[C, T any](f func(C, context.Context) (T, error)) Effect[C, T] {
return readerreaderioresult.Eitherize(f)
}
// Eitherize1 converts a function that takes an additional parameter and returns a value
// and error into a Kleisli arrow.
//
// This function takes a function that accepts a context C, context.Context, and
// an additional parameter A, returning a value T and an error, and converts it into a
// Kleisli arrow (A -> Effect[C, T]). The error is automatically converted into a failure,
// while successful values become successes.
//
// This is useful for creating composable operations that depend on context and
// an input value, following standard Go error-handling patterns. It is especially helpful
// for adapting interface member functions that accept a context and additional parameters.
// When you have an interface method with signature (receiver, context.Context, A) (T, error),
// you can use Eitherize1 to convert it into a Kleisli arrow where the receiver becomes
// the context C and A becomes the input parameter.
//
// # Type Parameters
//
// - C: The context type required by the effect
// - A: The input parameter type
// - T: The success value type
//
// # Parameters
//
// - f: A function that takes C, context.Context, and A, returning (T, error)
//
// # Returns
//
// - Kleisli[C, A, T]: A function from A to Effect[C, T]
//
// # Example Usage
//
// type AppConfig struct {
// DatabaseURL string
// }
//
// // A function using standard Go error handling
// func fetchUserByID(cfg AppConfig, ctx context.Context, id int) (*User, error) {
// // Implementation that may return an error
// return &User{ID: id, Name: "Alice"}, nil
// }
//
// // Convert to Kleisli arrow
// fetchUserKleisli := effect.Eitherize1(fetchUserByID)
//
// // Use in functional composition with Chain
// pipeline := F.Pipe1(
// effect.Succeed[AppConfig](123),
// effect.Chain[AppConfig](fetchUserKleisli),
// )
//
// // Execute with config
// cfg := AppConfig{DatabaseURL: "postgres://localhost"}
// result, err := effect.RunSync(effect.Provide[*User](cfg)(pipeline))(context.Background())
//
// # Adapting Interface Methods
//
// Eitherize1 is particularly useful for adapting interface member functions with parameters:
//
// type UserRepository interface {
// GetUserByID(ctx context.Context, id int) (*User, error)
// UpdateUser(ctx context.Context, user *User) error
// }
//
// type UserRepo struct {
// db *sql.DB
// }
//
// func (r *UserRepo) GetUserByID(ctx context.Context, id int) (*User, error) {
// // Implementation
// return &User{ID: id}, nil
// }
//
// // Adapt the method - receiver becomes C, id becomes A
// repo := &UserRepo{db: db}
// getUserKleisli := effect.Eitherize1(func(r *UserRepo, ctx context.Context, id int) (*User, error) {
// return r.GetUserByID(ctx, id)
// })
//
// // Now getUserKleisli has type: Kleisli[*UserRepo, int, *User]
// // Which is: func(int) Effect[*UserRepo, *User]
// // Use it in composition:
// pipeline := F.Pipe1(
// effect.Succeed[*UserRepo](123),
// effect.Chain[*UserRepo](getUserKleisli),
// )
// result, err := effect.RunSync(effect.Provide[*User](repo)(pipeline))(context.Background())
//
// # See Also
//
// - Eitherize: For functions without an additional parameter
// - Chain: For composing Kleisli arrows
// - readerreaderioresult.Eitherize1: The underlying implementation
//
//go:inline
func Eitherize1[C, A, T any](f func(C, context.Context, A) (T, error)) Kleisli[C, A, T] {
return readerreaderioresult.Eitherize1(f)
}

507
v2/effect/eitherize_test.go Normal file
View File

@@ -0,0 +1,507 @@
// Copyright (c) 2023 - 2025 IBM Corp.
// All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package effect
import (
"context"
"errors"
"fmt"
"strconv"
"testing"
F "github.com/IBM/fp-go/v2/function"
"github.com/stretchr/testify/assert"
)
// TestEitherize_Success tests successful conversion with Eitherize
func TestEitherize_Success(t *testing.T) {
t.Run("converts successful function to Effect", func(t *testing.T) {
// Arrange
successFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return cfg.Prefix + "-success", nil
}
eff := Eitherize(successFunc)
// Act
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "LOG-success", result)
})
t.Run("preserves context values", func(t *testing.T) {
// Arrange
type ctxKey string
key := ctxKey("testKey")
expectedValue := "contextValue"
contextFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
value := ctx.Value(key)
if value == nil {
return "", errors.New("context value not found")
}
return value.(string), nil
}
eff := Eitherize(contextFunc)
// Act
ioResult := Provide[string](testConfig)(eff)
readerResult := RunSync(ioResult)
ctx := context.WithValue(context.Background(), key, expectedValue)
result, err := readerResult(ctx)
// Assert
assert.NoError(t, err)
assert.Equal(t, expectedValue, result)
})
t.Run("works with different types", func(t *testing.T) {
// Arrange
intFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return cfg.Multiplier, nil
}
eff := Eitherize(intFunc)
// Act
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 3, result)
})
}
// TestEitherize_Failure tests error handling with Eitherize
func TestEitherize_Failure(t *testing.T) {
t.Run("converts error to failure", func(t *testing.T) {
// Arrange
expectedErr := errors.New("operation failed")
failFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return "", expectedErr
}
eff := Eitherize(failFunc)
// Act
_, err := runEffect(eff, testConfig)
// Assert
assert.Error(t, err)
assert.Equal(t, expectedErr, err)
})
t.Run("preserves error message", func(t *testing.T) {
// Arrange
expectedErr := fmt.Errorf("validation error: field is required")
failFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return 0, expectedErr
}
eff := Eitherize(failFunc)
// Act
_, err := runEffect(eff, testConfig)
// Assert
assert.Error(t, err)
assert.Equal(t, expectedErr, err)
})
}
// TestEitherize_EdgeCases tests edge cases for Eitherize
func TestEitherize_EdgeCases(t *testing.T) {
t.Run("handles nil context", func(t *testing.T) {
// Arrange
nilCtxFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
if ctx == nil {
return "nil-context", nil
}
return "non-nil-context", nil
}
eff := Eitherize(nilCtxFunc)
// Act
ioResult := Provide[string](testConfig)(eff)
readerResult := RunSync(ioResult)
result, err := readerResult(nil)
// Assert
assert.NoError(t, err)
assert.Equal(t, "nil-context", result)
})
t.Run("handles zero value config", func(t *testing.T) {
// Arrange
zeroFunc := func(cfg TestConfig, ctx context.Context) (string, error) {
return cfg.Prefix, nil
}
eff := Eitherize(zeroFunc)
// Act
result, err := runEffect(eff, TestConfig{})
// Assert
assert.NoError(t, err)
assert.Equal(t, "", result)
})
t.Run("handles pointer types", func(t *testing.T) {
// Arrange
type User struct {
Name string
}
ptrFunc := func(cfg TestConfig, ctx context.Context) (*User, error) {
return &User{Name: cfg.Prefix}, nil
}
eff := Eitherize(ptrFunc)
// Act
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.NotNil(t, result)
assert.Equal(t, "LOG", result.Name)
})
}
// TestEitherize_Integration tests integration with other operations
func TestEitherize_Integration(t *testing.T) {
t.Run("composes with Map", func(t *testing.T) {
// Arrange
baseFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return cfg.Multiplier, nil
}
eff := Eitherize(baseFunc)
// Act
pipeline := F.Pipe1(
eff,
Map[TestConfig](func(n int) string { return strconv.Itoa(n) }),
)
result, err := runEffect(pipeline, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "3", result)
})
t.Run("composes with Chain", func(t *testing.T) {
// Arrange
firstFunc := func(cfg TestConfig, ctx context.Context) (int, error) {
return cfg.Multiplier, nil
}
secondFunc := func(n int) Effect[TestConfig, string] {
return Succeed[TestConfig](fmt.Sprintf("value: %d", n))
}
// Act
pipeline := F.Pipe1(
Eitherize(firstFunc),
Chain[TestConfig](secondFunc),
)
result, err := runEffect(pipeline, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "value: 3", result)
})
}
// TestEitherize1_Success tests successful conversion with Eitherize1
func TestEitherize1_Success(t *testing.T) {
t.Run("converts successful function to Kleisli", func(t *testing.T) {
// Arrange
multiplyFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
return n * cfg.Multiplier, nil
}
kleisli := Eitherize1(multiplyFunc)
// Act
eff := kleisli(10)
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 30, result)
})
t.Run("works with string input", func(t *testing.T) {
// Arrange
concatFunc := func(cfg TestConfig, ctx context.Context, s string) (string, error) {
return cfg.Prefix + "-" + s, nil
}
kleisli := Eitherize1(concatFunc)
// Act
eff := kleisli("input")
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "LOG-input", result)
})
t.Run("preserves context in Kleisli", func(t *testing.T) {
// Arrange
type ctxKey string
key := ctxKey("factor")
scaleFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
factor := ctx.Value(key)
if factor == nil {
return n * cfg.Multiplier, nil
}
return n * factor.(int), nil
}
kleisli := Eitherize1(scaleFunc)
// Act
eff := kleisli(5)
ioResult := Provide[int](testConfig)(eff)
readerResult := RunSync(ioResult)
ctx := context.WithValue(context.Background(), key, 7)
result, err := readerResult(ctx)
// Assert
assert.NoError(t, err)
assert.Equal(t, 35, result)
})
}
// TestEitherize1_Failure tests error handling with Eitherize1
func TestEitherize1_Failure(t *testing.T) {
t.Run("converts error to failure in Kleisli", func(t *testing.T) {
// Arrange
expectedErr := errors.New("division by zero")
divideFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
if n == 0 {
return 0, expectedErr
}
return 100 / n, nil
}
kleisli := Eitherize1(divideFunc)
// Act
eff := kleisli(0)
_, err := runEffect(eff, testConfig)
// Assert
assert.Error(t, err)
assert.Equal(t, expectedErr, err)
})
t.Run("preserves error context", func(t *testing.T) {
// Arrange
validateFunc := func(cfg TestConfig, ctx context.Context, s string) (string, error) {
if len(s) > 10 {
return "", fmt.Errorf("string too long: %d > 10", len(s))
}
return s, nil
}
kleisli := Eitherize1(validateFunc)
// Act
eff := kleisli("this-string-is-too-long")
_, err := runEffect(eff, testConfig)
// Assert
assert.Error(t, err)
assert.Contains(t, err.Error(), "string too long")
})
}
// TestEitherize1_EdgeCases tests edge cases for Eitherize1
func TestEitherize1_EdgeCases(t *testing.T) {
t.Run("handles zero value input", func(t *testing.T) {
// Arrange
zeroFunc := func(cfg TestConfig, ctx context.Context, n int) (int, error) {
return n, nil
}
kleisli := Eitherize1(zeroFunc)
// Act
eff := kleisli(0)
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 0, result)
})
t.Run("handles pointer input", func(t *testing.T) {
// Arrange
type Input struct {
Value int
}
ptrFunc := func(cfg TestConfig, ctx context.Context, in *Input) (int, error) {
if in == nil {
return 0, errors.New("nil input")
}
return in.Value * cfg.Multiplier, nil
}
kleisli := Eitherize1(ptrFunc)
// Act
eff := kleisli(&Input{Value: 7})
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 21, result)
})
t.Run("handles nil pointer input", func(t *testing.T) {
// Arrange
type Input struct {
Value int
}
ptrFunc := func(cfg TestConfig, ctx context.Context, in *Input) (int, error) {
if in == nil {
return 0, errors.New("nil input")
}
return in.Value, nil
}
kleisli := Eitherize1(ptrFunc)
// Act
eff := kleisli((*Input)(nil))
_, err := runEffect(eff, testConfig)
// Assert
assert.Error(t, err)
assert.Contains(t, err.Error(), "nil input")
})
}
// TestEitherize1_Integration tests integration with other operations
func TestEitherize1_Integration(t *testing.T) {
t.Run("composes with Chain", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
doubleFunc := func(n int) Effect[TestConfig, int] {
return Succeed[TestConfig](n * 2)
}
parseKleisli := Eitherize1(parseFunc)
// Act
pipeline := F.Pipe2(
Succeed[TestConfig]("42"),
Chain[TestConfig](parseKleisli),
Chain[TestConfig](doubleFunc),
)
result, err := runEffect(pipeline, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 84, result)
})
t.Run("handles error in chain", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
parseKleisli := Eitherize1(parseFunc)
// Act
pipeline := F.Pipe1(
Succeed[TestConfig]("not-a-number"),
Chain[TestConfig](parseKleisli),
)
_, err := runEffect(pipeline, testConfig)
// Assert
assert.Error(t, err)
})
t.Run("composes multiple Kleisli arrows", func(t *testing.T) {
// Arrange
parseFunc := func(cfg TestConfig, ctx context.Context, s string) (int, error) {
return strconv.Atoi(s)
}
formatFunc := func(cfg TestConfig, ctx context.Context, n int) (string, error) {
return fmt.Sprintf("%s-%d", cfg.Prefix, n), nil
}
parseKleisli := Eitherize1(parseFunc)
formatKleisli := Eitherize1(formatFunc)
// Act
pipeline := F.Pipe2(
Succeed[TestConfig]("123"),
Chain[TestConfig](parseKleisli),
Chain[TestConfig](formatKleisli),
)
result, err := runEffect(pipeline, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "LOG-123", result)
})
}
// TestEitherize_TypeSafety tests type safety across different scenarios
func TestEitherize_TypeSafety(t *testing.T) {
t.Run("Eitherize with complex types", func(t *testing.T) {
// Arrange
type ComplexResult struct {
Data map[string]int
Count int
}
complexFunc := func(cfg TestConfig, ctx context.Context) (ComplexResult, error) {
return ComplexResult{
Data: map[string]int{cfg.Prefix: cfg.Multiplier},
Count: cfg.Multiplier,
}, nil
}
eff := Eitherize(complexFunc)
// Act
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, 3, result.Data["LOG"])
assert.Equal(t, 3, result.Count)
})
t.Run("Eitherize1 with different input and output types", func(t *testing.T) {
// Arrange
type Input struct {
ID int
}
type Output struct {
Name string
}
convertFunc := func(cfg TestConfig, ctx context.Context, in Input) (Output, error) {
return Output{Name: fmt.Sprintf("%s-%d", cfg.Prefix, in.ID)}, nil
}
kleisli := Eitherize1(convertFunc)
// Act
eff := kleisli(Input{ID: 99})
result, err := runEffect(eff, testConfig)
// Assert
assert.NoError(t, err)
assert.Equal(t, "LOG-99", result.Name)
})
}

View File

@@ -4,7 +4,7 @@ go 1.24
require (
github.com/stretchr/testify v1.11.1
github.com/urfave/cli/v3 v3.6.2
github.com/urfave/cli/v3 v3.7.0
)
require (

View File

@@ -6,6 +6,8 @@ github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu
github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
github.com/urfave/cli/v3 v3.6.2 h1:lQuqiPrZ1cIz8hz+HcrG0TNZFxU70dPZ3Yl+pSrH9A8=
github.com/urfave/cli/v3 v3.6.2/go.mod h1:ysVLtOEmg2tOy6PknnYVhDoouyC/6N42TMeoMzskhso=
github.com/urfave/cli/v3 v3.7.0 h1:AGSnbUyjtLiM+WJUb4dzXKldl/gL+F8OwmRDtVr6g2U=
github.com/urfave/cli/v3 v3.7.0/go.mod h1:ysVLtOEmg2tOy6PknnYVhDoouyC/6N42TMeoMzskhso=
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=

View File

@@ -288,5 +288,3 @@ func BenchmarkVoidMonoid_Empty(b *testing.B) {
_ = m.Empty()
}
}
// Made with Bob

View File

@@ -26,6 +26,83 @@ import (
"github.com/IBM/fp-go/v2/semigroup"
)
// Do creates the initial empty codec to be used as the starting point for
// do-notation style codec construction.
//
// This is the entry point for building up a struct codec field-by-field using
// the applicative and monadic sequencing operators ApSL, ApSO, and Bind.
// It wraps Empty and lifts a lazily-evaluated default Pair[O, A] into a
// Type[A, O, I] that ignores its input and always succeeds with the default value.
//
// # Type Parameters
//
// - I: The input type for decoding (what the codec reads from)
// - A: The target struct type being built up (what the codec decodes to)
// - O: The output type for encoding (what the codec writes to)
//
// # Parameters
//
// - e: A Lazy[Pair[O, A]] providing the initial default values:
// - pair.Head(e()): The default encoded output O (e.g. an empty monoid value)
// - pair.Tail(e()): The initial zero value of the struct A (e.g. MyStruct{})
//
// # Returns
//
// - A Type[A, O, I] that always decodes to the default A and encodes to the
// default O, regardless of input. This is then transformed by chaining
// ApSL, ApSO, or Bind operators to add fields one by one.
//
// # Example Usage
//
// Building a struct codec using do-notation style:
//
// import (
// "github.com/IBM/fp-go/v2/function"
// "github.com/IBM/fp-go/v2/lazy"
// "github.com/IBM/fp-go/v2/optics/codec"
// "github.com/IBM/fp-go/v2/optics/lens"
// "github.com/IBM/fp-go/v2/pair"
// S "github.com/IBM/fp-go/v2/string"
// )
//
// type Person struct {
// Name string
// Age int
// }
//
// nameLens := lens.MakeLens(
// func(p Person) string { return p.Name },
// func(p Person, name string) Person { p.Name = name; return p },
// )
// ageLens := lens.MakeLens(
// func(p Person) int { return p.Age },
// func(p Person, age int) Person { p.Age = age; return p },
// )
//
// personCodec := F.Pipe2(
// codec.Do[any, Person, string](lazy.Of(pair.MakePair("", Person{}))),
// codec.ApSL(S.Monoid, nameLens, codec.String()),
// codec.ApSL(S.Monoid, ageLens, codec.Int()),
// )
//
// # Notes
//
// - Do is typically the first call in a codec pipeline, followed by ApSL, ApSO, or Bind
// - The lazy pair should use the monoid's empty value for O and the zero value for A
// - For convenience, use Struct to create the initial codec for named struct types
//
// # See Also
//
// - Empty: The underlying codec constructor that Do delegates to
// - ApSL: Applicative sequencing for required struct fields via Lens
// - ApSO: Applicative sequencing for optional struct fields via Optional
// - Bind: Monadic sequencing for context-dependent field codecs
//
//go:inline
func Do[I, A, O any](e Lazy[Pair[O, A]]) Type[A, O, I] {
return Empty[I](e)
}
// ApSL creates an applicative sequencing operator for codecs using a lens.
//
// This function implements the "ApS" (Applicative Sequencing) pattern for codecs,
@@ -297,3 +374,132 @@ func ApSO[S, T, O, I any](
)
}
}
// Bind creates a monadic sequencing operator for codecs using a lens and a Kleisli arrow.
//
// This function implements the "Bind" (monadic bind / chain) pattern for codecs,
// allowing you to build up complex codecs where the codec for a field depends on
// the current decoded value of the struct. Unlike ApSL which uses a fixed field
// codec, Bind accepts a Kleisli arrow — a function from the current struct value S
// to a Type[T, O, I] — enabling context-sensitive codec construction.
//
// The function combines:
// - Encoding: Evaluates the Kleisli arrow f on the current struct value s to obtain
// the field codec, extracts the field T using the lens, encodes it with that codec,
// and combines it with the base encoding using the monoid.
// - Validation: Validates the base struct first (monadic sequencing), then uses the
// Kleisli arrow to obtain the field codec for the decoded struct value, and validates
// the field through the lens. Errors are propagated but NOT accumulated (fail-fast
// semantics, unlike ApSL which accumulates errors).
//
// # Type Parameters
//
// - S: The source struct type (what we're building a codec for)
// - T: The field type accessed by the lens
// - O: The output type for encoding (must have a monoid)
// - I: The input type for decoding
//
// # Parameters
//
// - m: A Monoid[O] for combining encoded outputs
// - l: A Lens[S, T] that focuses on a specific field in S
// - f: A Kleisli[S, T, O, I] — a function from S to Type[T, O, I] — that produces
// the field codec based on the current struct value
//
// # Returns
//
// An Operator[S, S, O, I] that transforms a base codec by adding the field
// specified by the lens, where the field codec is determined by the Kleisli arrow.
//
// # How It Works
//
// 1. **Encoding**: When encoding a value of type S:
// - Evaluate f(s) to obtain the field codec fa
// - Extract the field T using l.Get
// - Encode T to O using fa.Encode
// - Combine with the base encoding using the monoid
//
// 2. **Validation**: When validating input I:
// - Run the base validation to obtain a decoded S (fail-fast: stop on base failure)
// - For the decoded S, evaluate f(s) to obtain the field codec fa
// - Validate the input I using fa.Validate
// - Set the validated T into S using l.Set
//
// 3. **Type Checking**: Preserves the base type checker
//
// # Difference from ApSL
//
// Unlike ApSL which uses a fixed field codec:
// - ApSL: Field codec is fixed at construction time; errors are accumulated
// - Bind: Field codec depends on the current struct value (Kleisli arrow); validation
// uses monadic sequencing (fail-fast on base failure)
// - Bind is more powerful but less parallel than ApSL
//
// # Example
//
// import (
// "github.com/IBM/fp-go/v2/optics/codec"
// "github.com/IBM/fp-go/v2/optics/lens"
// S "github.com/IBM/fp-go/v2/string"
// )
//
// type Config struct {
// Mode string
// Value int
// }
//
// modeLens := lens.MakeLens(
// func(c Config) string { return c.Mode },
// func(c Config, mode string) Config { c.Mode = mode; return c },
// )
//
// // Build a Config codec where the Value codec depends on the Mode
// configCodec := F.Pipe1(
// codec.Struct[Config]("Config"),
// codec.Bind(S.Monoid, modeLens, func(c Config) codec.Type[string, string, any] {
// return codec.String()
// }),
// )
//
// # Use Cases
//
// - Building codecs where a field's codec depends on another field's value
// - Implementing discriminated unions or tagged variants
// - Context-sensitive validation (e.g., validate field B differently based on field A)
// - Dependent type-like patterns in codec construction
//
// # Notes
//
// - The monoid determines how encoded outputs are combined
// - The lens must be total (handle all cases safely)
// - Validation uses monadic (fail-fast) sequencing: if the base codec fails,
// the Kleisli arrow is never evaluated
// - The name is automatically generated for debugging purposes
//
// See also:
// - ApSL: Applicative sequencing with a fixed lens codec (error accumulation)
// - Kleisli: The function type from S to Type[T, O, I]
// - decode.Bind: The underlying decode-level bind combinator
func Bind[S, T, O, I any](
m Monoid[O],
l Lens[S, T],
f Kleisli[S, T, O, I],
) Operator[S, S, O, I] {
name := fmt.Sprintf("Bind[%s]", l)
val := F.Curry2(Type[T, O, I].Validate)
return func(t Type[S, O, I]) Type[S, O, I] {
return MakeType(
name,
t.Is,
F.Pipe1(
t.Validate,
validate.Bind(l.Set, F.Flow2(f, val)),
),
func(s S) O {
return m.Concat(t.Encode(s), f(s).Encode(l.Get(s)))
},
)
}
}

View File

@@ -814,3 +814,588 @@ func TestApSO_ErrorAccumulation(t *testing.T) {
assert.NotEmpty(t, errors, "Should have validation errors")
})
}
// TestBind_EncodingCombination verifies that Bind combines the base encoding with
// the field encoding produced by the Kleisli arrow using the monoid.
func TestBind_EncodingCombination(t *testing.T) {
t.Run("combines base and field encodings using monoid", func(t *testing.T) {
// Lens for Person.Name
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
// Base codec encodes to "Person:"
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "Person:" },
)
// Kleisli arrow: always returns a string identity codec regardless of struct value
kleisli := func(p Person) Type[string, string, any] {
return MakeType(
"Name",
func(i any) validation.Result[string] {
if s, ok := i.(string); ok {
return validation.ToResult(validation.Success(s))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected string"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if s, ok := i.(string); ok {
return validation.Success(s)
}
return validation.FailureWithMessage[string](i, "expected string")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
person := Person{Name: "Alice", Age: 30}
encoded := enhancedCodec.Encode(person)
// Encoding should include both the base prefix and the field value
assert.Contains(t, encoded, "Person:")
assert.Contains(t, encoded, "Alice")
})
}
// TestBind_KleisliArrowReceivesCurrentValue verifies that the Kleisli arrow f
// receives the current struct value when producing the field codec.
func TestBind_KleisliArrowReceivesCurrentValue(t *testing.T) {
t.Run("kleisli arrow receives current struct value during encoding", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
// Kleisli arrow that uses the struct value to produce a prefix in the encoding
var capturedPerson Person
kleisli := func(p Person) Type[string, string, any] {
capturedPerson = p
return MakeType(
"Name",
func(i any) validation.Result[string] {
if s, ok := i.(string); ok {
return validation.ToResult(validation.Success(s))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected string"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if s, ok := i.(string); ok {
return validation.Success(s)
}
return validation.FailureWithMessage[string](i, "expected string")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
person := Person{Name: "Bob", Age: 25}
enhancedCodec.Encode(person)
// The Kleisli arrow should have been called with the actual struct value
assert.Equal(t, person, capturedPerson)
})
}
// TestBind_ValidationSuccess verifies that Bind correctly validates and decodes
// a struct when both the base and field validations succeed.
func TestBind_ValidationSuccess(t *testing.T) {
t.Run("succeeds when base and field validations pass", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
// The field codec receives the same input I (any = Person struct).
// It must extract the Name field from the Person input.
kleisli := func(p Person) Type[string, string, any] {
return MakeType(
"Name",
func(i any) validation.Result[string] {
if person, ok := i.(Person); ok {
return validation.ToResult(validation.Success(person.Name))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if person, ok := i.(Person); ok {
return validation.Success(person.Name)
}
return validation.FailureWithMessage[string](i, "expected Person")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
person := Person{Name: "Carol", Age: 28}
result := enhancedCodec.Decode(person)
assert.True(t, either.IsRight(result), "Should succeed when both validations pass")
})
}
// TestBind_ValidationFailsOnBaseFailure verifies that Bind uses fail-fast (monadic)
// semantics: if the base codec fails, the Kleisli arrow is never evaluated.
func TestBind_ValidationFailsOnBaseFailure(t *testing.T) {
t.Run("fails fast when base validation fails", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
// Base codec always fails
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "base always fails"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
return validation.FailureWithMessage[Person](i, "base always fails")(ctx)
}
},
func(p Person) string { return "" },
)
kleisliCalled := false
kleisli := func(p Person) Type[string, string, any] {
kleisliCalled = true
return MakeType(
"Name",
func(i any) validation.Result[string] {
if s, ok := i.(string); ok {
return validation.ToResult(validation.Success(s))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected string"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if s, ok := i.(string); ok {
return validation.Success(s)
}
return validation.FailureWithMessage[string](i, "expected string")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
person := Person{Name: "Dave", Age: 40}
result := enhancedCodec.Decode(person)
assert.True(t, either.IsLeft(result), "Should fail when base validation fails")
assert.False(t, kleisliCalled, "Kleisli arrow should NOT be called when base fails")
})
}
// TestBind_ValidationFailsOnFieldFailure verifies that Bind propagates field
// validation errors when the Kleisli arrow's codec fails.
func TestBind_ValidationFailsOnFieldFailure(t *testing.T) {
t.Run("fails when field validation from kleisli codec fails", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
// Base codec succeeds
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
// Kleisli arrow returns a codec that always fails regardless of input
kleisli := func(p Person) Type[string, string, any] {
return MakeType(
"Name",
func(i any) validation.Result[string] {
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "field always fails"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
return validation.FailureWithMessage[string](i, "field always fails")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
// The field codec receives the same input (Person) and always fails
person := Person{Name: "Eve", Age: 22}
result := enhancedCodec.Decode(person)
assert.True(t, either.IsLeft(result), "Should fail when field validation fails")
errors := either.MonadFold(result,
F.Identity[validation.Errors],
func(Person) validation.Errors { return nil },
)
assert.NotEmpty(t, errors, "Should have validation errors from field codec")
})
}
// TestBind_TypeCheckingPreserved verifies that Bind preserves the base type checker.
func TestBind_TypeCheckingPreserved(t *testing.T) {
t.Run("preserves base type checker", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
kleisli := func(p Person) Type[string, string, any] {
return MakeType(
"Name",
func(i any) validation.Result[string] {
if s, ok := i.(string); ok {
return validation.ToResult(validation.Success(s))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected string"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if s, ok := i.(string); ok {
return validation.Success(s)
}
return validation.FailureWithMessage[string](i, "expected string")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
// Valid type
person := Person{Name: "Frank", Age: 35}
isResult := enhancedCodec.Is(person)
assert.True(t, either.IsRight(isResult), "Should accept Person type")
// Invalid type
invalidResult := enhancedCodec.Is("not a person")
assert.True(t, either.IsLeft(invalidResult), "Should reject non-Person type")
})
}
// TestBind_Naming verifies that Bind generates a descriptive name for the codec.
func TestBind_Naming(t *testing.T) {
t.Run("generates descriptive name containing Bind and lens info", func(t *testing.T) {
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
kleisli := func(p Person) Type[string, string, any] {
return MakeType(
"Name",
func(i any) validation.Result[string] {
if s, ok := i.(string); ok {
return validation.ToResult(validation.Success(s))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected string"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if s, ok := i.(string); ok {
return validation.Success(s)
}
return validation.FailureWithMessage[string](i, "expected string")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
name := enhancedCodec.Name()
assert.Contains(t, name, "Bind", "Name should contain 'Bind'")
})
}
// TestBind_DependentFieldCodec verifies that the Kleisli arrow can produce
// different codecs based on the current struct value (the key differentiator
// from ApSL).
//
// The field codec Type[T, O, I] receives the same input I as the base codec.
// It must extract the field value from that input. The Kleisli arrow f(s)
// produces a different codec depending on the already-decoded struct value s.
func TestBind_DependentFieldCodec(t *testing.T) {
t.Run("kleisli arrow produces different codecs based on struct value", func(t *testing.T) {
// Lens for Person.Name
nameLens := lens.MakeLens(
func(p Person) string { return p.Name },
func(p Person, name string) Person {
return Person{Name: name, Age: p.Age}
},
)
// Base codec succeeds for any Person
baseCodec := MakeType(
"Person",
func(i any) validation.Result[Person] {
if p, ok := i.(Person); ok {
return validation.ToResult(validation.Success(p))
}
return validation.ToResult(validation.Failures[Person](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, Person] {
return func(ctx Context) validation.Validation[Person] {
if p, ok := i.(Person); ok {
return validation.Success(p)
}
return validation.FailureWithMessage[Person](i, "expected Person")(ctx)
}
},
func(p Person) string { return "" },
)
// Kleisli arrow: the field codec receives the same input I (any = Person).
// It extracts the Name from the Person input.
// If the decoded struct's Age > 18, accept any name (including empty).
// If Age <= 18, reject empty names.
kleisli := func(p Person) Type[string, string, any] {
if p.Age > 18 {
// Adult: accept any name extracted from the Person input
return MakeType(
"AnyName",
func(i any) validation.Result[string] {
if person, ok := i.(Person); ok {
return validation.ToResult(validation.Success(person.Name))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if person, ok := i.(Person); ok {
return validation.Success(person.Name)
}
return validation.FailureWithMessage[string](i, "expected Person")(ctx)
}
},
F.Identity[string],
)
}
// Minor: reject empty names
return MakeType(
"NonEmptyName",
func(i any) validation.Result[string] {
if person, ok := i.(Person); ok {
if person.Name != "" {
return validation.ToResult(validation.Success(person.Name))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: person.Name, Messsage: "name must not be empty for minors"},
}))
}
return validation.ToResult(validation.Failures[string](validation.Errors{
&validation.ValidationError{Value: i, Messsage: "expected Person"},
}))
},
func(i any) Decode[Context, string] {
return func(ctx Context) validation.Validation[string] {
if person, ok := i.(Person); ok {
if person.Name != "" {
return validation.Success(person.Name)
}
return validation.FailureWithMessage[string](person.Name, "name must not be empty for minors")(ctx)
}
return validation.FailureWithMessage[string](i, "expected Person")(ctx)
}
},
F.Identity[string],
)
}
operator := Bind(S.Monoid, nameLens, kleisli)
enhancedCodec := operator(baseCodec)
// Adult (Age=30) with empty name: should succeed (adult codec accepts any name)
adultPerson := Person{Name: "", Age: 30}
adultResult := enhancedCodec.Decode(adultPerson)
assert.True(t, either.IsRight(adultResult), "Adult should accept empty name")
// Minor (Age=15) with empty name: should fail (minor codec rejects empty names)
minorPerson := Person{Name: "", Age: 15}
minorResult := enhancedCodec.Decode(minorPerson)
assert.True(t, either.IsLeft(minorResult), "Minor with empty name should fail")
// Minor (Age=15) with non-empty name: should succeed
minorWithName := Person{Name: "Junior", Age: 15}
minorWithNameResult := enhancedCodec.Decode(minorWithName)
assert.True(t, either.IsRight(minorWithNameResult), "Minor with non-empty name should succeed")
})
}

View File

@@ -101,7 +101,7 @@ func (t *typeImpl[A, O, I]) Is(i any) Result[A] {
// stringToInt := codec.MakeType(...) // Type[int, string, string]
// intToPositive := codec.MakeType(...) // Type[PositiveInt, int, int]
// composed := codec.Pipe(intToPositive)(stringToInt) // Type[PositiveInt, string, string]
func Pipe[O, I, A, B any](ab Type[B, A, A]) func(Type[A, O, I]) Type[B, O, I] {
func Pipe[O, I, A, B any](ab Type[B, A, A]) Operator[A, B, O, I] {
return func(this Type[A, O, I]) Type[B, O, I] {
return MakeType(
fmt.Sprintf("Pipe(%s, %s)", this.Name(), ab.Name()),
@@ -851,7 +851,7 @@ func FromRefinement[A, B any](refinement Refinement[A, B]) Type[B, A, A] {
// See also:
// - Id: For identity codecs that preserve values
// - MakeType: For creating custom codecs with validation logic
func Empty[A, O, I any](e Lazy[Pair[O, A]]) Type[A, O, I] {
func Empty[I, A, O any](e Lazy[Pair[O, A]]) Type[A, O, I] {
return MakeType(
"Empty",
Is[A](),

View File

@@ -1742,7 +1742,7 @@ func TestFromRefinementValidationContext(t *testing.T) {
// TestEmpty_Success tests that Empty always succeeds during decoding
func TestEmpty_Success(t *testing.T) {
t.Run("decodes any input to default value", func(t *testing.T) {
defaultCodec := Empty[int, string, any](lazy.Of(pair.MakePair("default", 42)))
defaultCodec := Empty[any, int, string](lazy.Of(pair.MakePair("default", 42)))
// Test with various input types
testCases := []struct {
@@ -1766,7 +1766,7 @@ func TestEmpty_Success(t *testing.T) {
})
t.Run("always returns same default value", func(t *testing.T) {
defaultCodec := Empty[string, string, any](lazy.Of(pair.MakePair("output", "default")))
defaultCodec := Empty[any, string, string](lazy.Of(pair.MakePair("output", "default")))
result1 := defaultCodec.Decode(123)
result2 := defaultCodec.Decode("different")
@@ -1789,7 +1789,7 @@ func TestEmpty_Success(t *testing.T) {
// TestEmpty_Encoding tests that Empty always uses default output during encoding
func TestEmpty_Encoding(t *testing.T) {
t.Run("encodes any value to default output", func(t *testing.T) {
defaultCodec := Empty[int, string, any](lazy.Of(pair.MakePair("default", 42)))
defaultCodec := Empty[any, int, string](lazy.Of(pair.MakePair("default", 42)))
// Test with various input values
testCases := []struct {
@@ -1811,7 +1811,7 @@ func TestEmpty_Encoding(t *testing.T) {
})
t.Run("always returns same default output", func(t *testing.T) {
defaultCodec := Empty[string, int, any](lazy.Of(pair.MakePair(999, "ignored")))
defaultCodec := Empty[any, string, int](lazy.Of(pair.MakePair(999, "ignored")))
encoded1 := defaultCodec.Encode("value1")
encoded2 := defaultCodec.Encode("value2")
@@ -1826,7 +1826,7 @@ func TestEmpty_Encoding(t *testing.T) {
// TestEmpty_Name tests that Empty has correct name
func TestEmpty_Name(t *testing.T) {
t.Run("has name 'Empty'", func(t *testing.T) {
defaultCodec := Empty[int, int, any](lazy.Of(pair.MakePair(0, 0)))
defaultCodec := Empty[any, int, int](lazy.Of(pair.MakePair(0, 0)))
assert.Equal(t, "Empty", defaultCodec.Name())
})
}
@@ -1834,7 +1834,7 @@ func TestEmpty_Name(t *testing.T) {
// TestEmpty_TypeChecking tests that Empty performs standard type checking
func TestEmpty_TypeChecking(t *testing.T) {
t.Run("Is checks for correct type", func(t *testing.T) {
defaultCodec := Empty[int, string, any](lazy.Of(pair.MakePair("default", 42)))
defaultCodec := Empty[any, int, string](lazy.Of(pair.MakePair("default", 42)))
// Should succeed for int
result := defaultCodec.Is(100)
@@ -1846,7 +1846,7 @@ func TestEmpty_TypeChecking(t *testing.T) {
})
t.Run("Is checks for string type", func(t *testing.T) {
defaultCodec := Empty[string, string, any](lazy.Of(pair.MakePair("out", "in")))
defaultCodec := Empty[any, string, string](lazy.Of(pair.MakePair("out", "in")))
// Should succeed for string
result := defaultCodec.Is("hello")
@@ -1867,7 +1867,7 @@ func TestEmpty_LazyEvaluation(t *testing.T) {
return pair.MakePair(counter, counter*10)
}
defaultCodec := Empty[int, int, any](lazyPair)
defaultCodec := Empty[any, int, int](lazyPair)
// Each decode can get a different value if the lazy function is dynamic
result1 := defaultCodec.Decode("input1")
@@ -1897,7 +1897,7 @@ func TestEmpty_WithStructs(t *testing.T) {
t.Run("provides default struct value", func(t *testing.T) {
defaultConfig := Config{Timeout: 30, Retries: 3}
defaultCodec := Empty[Config, Config, any](lazy.Of(pair.MakePair(defaultConfig, defaultConfig)))
defaultCodec := Empty[any, Config, Config](lazy.Of(pair.MakePair(defaultConfig, defaultConfig)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -1914,7 +1914,7 @@ func TestEmpty_WithStructs(t *testing.T) {
defaultConfig := Config{Timeout: 30, Retries: 3}
inputConfig := Config{Timeout: 60, Retries: 5}
defaultCodec := Empty[Config, Config, any](lazy.Of(pair.MakePair(defaultConfig, defaultConfig)))
defaultCodec := Empty[any, Config, Config](lazy.Of(pair.MakePair(defaultConfig, defaultConfig)))
encoded := defaultCodec.Encode(inputConfig)
assert.Equal(t, 30, encoded.Timeout)
@@ -1926,7 +1926,7 @@ func TestEmpty_WithStructs(t *testing.T) {
func TestEmpty_WithPointers(t *testing.T) {
t.Run("provides default pointer value", func(t *testing.T) {
defaultValue := 42
defaultCodec := Empty[*int, *int, any](lazy.Of(pair.MakePair(&defaultValue, &defaultValue)))
defaultCodec := Empty[any, *int, *int](lazy.Of(pair.MakePair(&defaultValue, &defaultValue)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -1941,7 +1941,7 @@ func TestEmpty_WithPointers(t *testing.T) {
t.Run("provides nil pointer as default", func(t *testing.T) {
var nilPtr *int
defaultCodec := Empty[*int, *int, any](lazy.Of(pair.MakePair(nilPtr, nilPtr)))
defaultCodec := Empty[any, *int, *int](lazy.Of(pair.MakePair(nilPtr, nilPtr)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -1958,7 +1958,7 @@ func TestEmpty_WithPointers(t *testing.T) {
func TestEmpty_WithSlices(t *testing.T) {
t.Run("provides default slice value", func(t *testing.T) {
defaultSlice := []int{1, 2, 3}
defaultCodec := Empty[[]int, []int, any](lazy.Of(pair.MakePair(defaultSlice, defaultSlice)))
defaultCodec := Empty[any, []int, []int](lazy.Of(pair.MakePair(defaultSlice, defaultSlice)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -1972,7 +1972,7 @@ func TestEmpty_WithSlices(t *testing.T) {
t.Run("provides empty slice as default", func(t *testing.T) {
emptySlice := []int{}
defaultCodec := Empty[[]int, []int, any](lazy.Of(pair.MakePair(emptySlice, emptySlice)))
defaultCodec := Empty[any, []int, []int](lazy.Of(pair.MakePair(emptySlice, emptySlice)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -1988,7 +1988,7 @@ func TestEmpty_WithSlices(t *testing.T) {
// TestEmpty_DifferentInputOutput tests Empty with different input and output types
func TestEmpty_DifferentInputOutput(t *testing.T) {
t.Run("decodes to int, encodes to string", func(t *testing.T) {
defaultCodec := Empty[int, string, any](lazy.Of(pair.MakePair("default-output", 42)))
defaultCodec := Empty[any, int, string](lazy.Of(pair.MakePair("default-output", 42)))
// Decode always returns 42
result := defaultCodec.Decode("any input")
@@ -2000,7 +2000,7 @@ func TestEmpty_DifferentInputOutput(t *testing.T) {
})
t.Run("decodes to string, encodes to int", func(t *testing.T) {
defaultCodec := Empty[string, int, any](lazy.Of(pair.MakePair(999, "default-value")))
defaultCodec := Empty[any, string, int](lazy.Of(pair.MakePair(999, "default-value")))
// Decode always returns "default-value"
result := defaultCodec.Decode(123)
@@ -2020,7 +2020,7 @@ func TestEmpty_DifferentInputOutput(t *testing.T) {
// TestEmpty_EdgeCases tests edge cases for Empty
func TestEmpty_EdgeCases(t *testing.T) {
t.Run("with zero values", func(t *testing.T) {
defaultCodec := Empty[int, int, any](lazy.Of(pair.MakePair(0, 0)))
defaultCodec := Empty[any, int, int](lazy.Of(pair.MakePair(0, 0)))
result := defaultCodec.Decode("anything")
assert.True(t, either.IsRight(result))
@@ -2035,7 +2035,7 @@ func TestEmpty_EdgeCases(t *testing.T) {
})
t.Run("with empty string", func(t *testing.T) {
defaultCodec := Empty[string, string, any](lazy.Of(pair.MakePair("", "")))
defaultCodec := Empty[any, string, string](lazy.Of(pair.MakePair("", "")))
result := defaultCodec.Decode("non-empty")
assert.True(t, either.IsRight(result))
@@ -2050,7 +2050,7 @@ func TestEmpty_EdgeCases(t *testing.T) {
})
t.Run("with false boolean", func(t *testing.T) {
defaultCodec := Empty[bool, bool, any](lazy.Of(pair.MakePair(false, false)))
defaultCodec := Empty[any, bool, bool](lazy.Of(pair.MakePair(false, false)))
result := defaultCodec.Decode(true)
assert.Equal(t, validation.Of(false), result)
@@ -2064,7 +2064,7 @@ func TestEmpty_EdgeCases(t *testing.T) {
func TestEmpty_Integration(t *testing.T) {
t.Run("composes with other codecs using Pipe", func(t *testing.T) {
// Create a codec that always provides a default int
defaultIntCodec := Empty[int, int, any](lazy.Of(pair.MakePair(42, 42)))
defaultIntCodec := Empty[any, int, int](lazy.Of(pair.MakePair(42, 42)))
// Create a refinement that only accepts positive integers
positiveIntPrism := prism.MakePrismWithName(
@@ -2090,7 +2090,7 @@ func TestEmpty_Integration(t *testing.T) {
t.Run("used as placeholder in generic contexts", func(t *testing.T) {
// Empty can be used where a codec is required but not actually used
unitCodec := Empty[Void, Void, any](
unitCodec := Empty[any, Void, Void](
lazy.Of(pair.MakePair(F.VOID, F.VOID)),
)
@@ -2105,7 +2105,7 @@ func TestEmpty_Integration(t *testing.T) {
// TestEmpty_RoundTrip tests that Empty maintains consistency
func TestEmpty_RoundTrip(t *testing.T) {
t.Run("decode then encode returns default output", func(t *testing.T) {
defaultCodec := Empty[int, string, any](lazy.Of(pair.MakePair("output", 42)))
defaultCodec := Empty[any, int, string](lazy.Of(pair.MakePair("output", 42)))
// Decode
result := defaultCodec.Decode("input")
@@ -2124,7 +2124,7 @@ func TestEmpty_RoundTrip(t *testing.T) {
})
t.Run("multiple round trips are consistent", func(t *testing.T) {
defaultCodec := Empty[int, int, any](lazy.Of(pair.MakePair(100, 50)))
defaultCodec := Empty[any, int, int](lazy.Of(pair.MakePair(100, 50)))
// First round trip
result1 := defaultCodec.Decode("input1")

View File

@@ -22,15 +22,19 @@
package codec
import (
"encoding"
"encoding/json"
"net/url"
"regexp"
"strconv"
"time"
F "github.com/IBM/fp-go/v2/function"
"github.com/IBM/fp-go/v2/optics/codec/validate"
"github.com/IBM/fp-go/v2/optics/codec/validation"
"github.com/IBM/fp-go/v2/optics/prism"
"github.com/IBM/fp-go/v2/reader"
"github.com/IBM/fp-go/v2/result"
)
// validateFromParser creates a validation function from a parser that may fail.
@@ -338,3 +342,128 @@ func Int64FromString() Type[int64, string, string] {
prism.ParseInt64().ReverseGet,
)
}
func decodeJSON[T any](dec json.Unmarshaler) ReaderResult[[]byte, T] {
return func(b []byte) Result[T] {
var t T
err := dec.UnmarshalJSON(b)
return result.TryCatchError(t, err)
}
}
func decodeText[T any](dec encoding.TextUnmarshaler) ReaderResult[[]byte, T] {
return func(b []byte) Result[T] {
var t T
err := dec.UnmarshalText(b)
return result.TryCatchError(t, err)
}
}
// MarshalText creates a bidirectional codec for types that implement encoding.TextMarshaler
// and encoding.TextUnmarshaler. This codec handles binary text serialization formats.
//
// The codec:
// - Decodes: Calls dec.UnmarshalText(b) to deserialize []byte into the target type T
// - Encodes: Calls enc.MarshalText() to serialize the value to []byte
// - Validates: Returns a failure if UnmarshalText returns an error
//
// Note: The enc and dec parameters are external marshaler/unmarshaler instances. The
// decoded value is the zero value of T after UnmarshalText has been called on dec
// (the caller is responsible for ensuring dec holds the decoded state).
//
// Type Parameters:
// - T: The Go type to encode/decode
//
// Parameters:
// - enc: An encoding.TextMarshaler used for encoding values to []byte
// - dec: An encoding.TextUnmarshaler used for decoding []byte to the target type
//
// Returns:
// - A Type[T, []byte, []byte] codec that handles text marshaling/unmarshaling
//
// Example:
//
// type MyType struct{ Value string }
//
// var instance MyType
// codec := MarshalText[MyType](instance, &instance)
//
// // Decode bytes to MyType
// result := codec.Decode([]byte(`some text`))
//
// // Encode MyType to bytes
// encoded := codec.Encode(instance)
func MarshalText[T any](
enc encoding.TextMarshaler,
dec encoding.TextUnmarshaler,
) Type[T, []byte, []byte] {
return MakeType(
"UnmarshalText",
Is[T](),
F.Pipe2(
dec,
decodeText[T],
validate.FromReaderResult,
),
func(t T) []byte {
b, _ := enc.MarshalText()
return b
},
)
}
// MarshalJSON creates a bidirectional codec for types that implement encoding/json's
// json.Marshaler and json.Unmarshaler interfaces. This codec handles JSON serialization.
//
// The codec:
// - Decodes: Calls dec.UnmarshalJSON(b) to deserialize []byte JSON into the target type T
// - Encodes: Calls enc.MarshalJSON() to serialize the value to []byte JSON
// - Validates: Returns a failure if UnmarshalJSON returns an error
//
// Note: The enc and dec parameters are external marshaler/unmarshaler instances. The
// decoded value is the zero value of T after UnmarshalJSON has been called on dec
// (the caller is responsible for ensuring dec holds the decoded state).
//
// Type Parameters:
// - T: The Go type to encode/decode
//
// Parameters:
// - enc: A json.Marshaler used for encoding values to JSON []byte
// - dec: A json.Unmarshaler used for decoding JSON []byte to the target type
//
// Returns:
// - A Type[T, []byte, []byte] codec that handles JSON marshaling/unmarshaling
//
// Example:
//
// type MyData struct {
// Name string `json:"name"`
// Value int `json:"value"`
// }
//
// var instance MyData
// codec := MarshalJSON[MyData](&instance, &instance)
//
// // Decode JSON bytes to MyData
// result := codec.Decode([]byte(`{"name":"test","value":42}`))
//
// // Encode MyData to JSON bytes
// encoded := codec.Encode(instance)
func MarshalJSON[T any](
enc json.Marshaler,
dec json.Unmarshaler,
) Type[T, []byte, []byte] {
return MakeType(
"UnmarshalJSON",
Is[T](),
F.Pipe2(
dec,
decodeJSON[T],
validate.FromReaderResult,
),
func(t T) []byte {
b, _ := enc.MarshalJSON()
return b
},
)
}

File diff suppressed because it is too large Load Diff