phares3757/gatus
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

#205: Work on supporting OpenID Connect for auth

Browse Source
This commit is contained in:
TwiN
2021-12-31 00:10:54 -05:00
parent 4ab5724fc1
commit be9087bee3
123 changed files with 27693 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
vendor/github.com/TwiN/g8/.gitattributes generated vendored Normal file
View File

@ -0,0 +1 @@
* text=lf

2
vendor/github.com/TwiN/g8/.gitignore generated vendored Normal file
View File

@ -0,0 +1,2 @@
.idea
*.iml

9
vendor/github.com/TwiN/g8/LICENSE.md generated vendored Normal file
View File

@ -0,0 +1,9 @@
MIT License
Copyright (c) 2021 TwiN
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

238
vendor/github.com/TwiN/g8/README.md generated vendored Normal file
View File

@ -0,0 +1,238 @@
# g8
![build](https://github.com/TwiN/g8/workflows/build/badge.svg?branch=master)
[![Go Report Card](https://goreportcard.com/badge/github.com/TwiN/g8)](https://goreportcard.com/report/github.com/TwiN/g8)
[![codecov](https://codecov.io/gh/TwiN/g8/branch/master/graph/badge.svg)](https://codecov.io/gh/TwiN/g8)
[![Go version](https://img.shields.io/github/go-mod/go-version/TwiN/g8.svg)](https://github.com/TwiN/g8)
[![Go Reference](https://pkg.go.dev/badge/github.com/TwiN/g8.svg)](https://pkg.go.dev/github.com/TwiN/g8)
[![Follow TwiN](https://img.shields.io/github/followers/TwiN?label=Follow&style=social)](https://github.com/TwiN)
g8, pronounced gate, is a simple Go library for protecting HTTP handlers.
Tired of constantly re-implementing a security layer for each application? Me too, that's why I made g8.
## Installation
```console
go get -u github.com/TwiN/g8
```
## Usage
Because the entire purpose of g8 is to NOT waste time configuring the layer of security, the primary emphasis is to
keep it as simple as possible.
### Simple
Just want a simple layer of security without the need for advanced permissions? This configuration is what you're
looking for.
```go
authorizationService := g8.NewAuthorizationService().WithToken("mytoken")
gate := g8.New().WithAuthorizationService(authorizationService)
router := http.NewServeMux()
router.Handle("/unprotected", yourHandler)
router.Handle("/protected", gate.Protect(yourHandler))
http.ListenAndServe(":8080", router)
```
The endpoint `/protected` is now only accessible if you pass the header `Authorization: Bearer mytoken`.
If you use `http.HandleFunc` instead of `http.Handle`, you may use `gate.ProtectFunc(yourHandler)` instead.
If you're not using the `Authorization` header, you can specify a custom token extractor.
This enables use cases like [Protecting a handler using session cookie](#protecting-a-handler-using-session-cookie)
### Advanced permissions
If you have tokens with more permissions than others, g8's permission system will make managing authorization a breeze.
Rather than registering tokens, think of it as registering clients, the only difference being that clients may be
configured with permissions while tokens cannot.
```go
authorizationService := g8.NewAuthorizationService().WithClient(g8.NewClient("mytoken").WithPermission("admin"))
gate := g8.New().WithAuthorizationService(authorizationService)
router := http.NewServeMux()
router.Handle("/unprotected", yourHandler)
router.Handle("/protected-with-admin", gate.ProtectWithPermissions(yourHandler, []string{"admin"}))
http.ListenAndServe(":8080", router)
```
The endpoint `/protected-with-admin` is now only accessible if you pass the header `Authorization: Bearer mytoken`,
because the client with the token `mytoken` has the permission `admin`. Note that the following handler would also be
accessible with that token:
```go
router.Handle("/protected", gate.Protect(yourHandler))
```
To clarify, both clients and tokens have access to handlers that aren't protected with extra permissions, and
essentially, tokens are registered as clients with no extra permissions in the background.
Creating a token like so:
```go
authorizationService := g8.NewAuthorizationService().WithToken("mytoken")
```
is the equivalent of creating the following client:
```go
authorizationService := g8.NewAuthorizationService().WithClient(g8.NewClient("mytoken"))
```
### With client provider
A client provider's task is to retrieve a Client from an external source (e.g. a database) when provided with a token.
You should use a client provider when you have a lot of tokens and it wouldn't make sense to register all of them using
`AuthorizationService`'s `WithToken`/`WithTokens`/`WithClient`/`WithClients`.
Note that the provider is used as a fallback source. As such, if a token is explicitly registered using one of the 4
aforementioned functions, the client provider will not be used.
```go
clientProvider := g8.NewClientProvider(func(token string) *g8.Client {
// We'll assume that the following function calls your database and returns a struct "User" that
// has the user's token as well as the permissions granted to said user
user := database.GetUserByToken(token)
if user != nil {
return g8.NewClient(user.Token).WithPermissions(user.Permissions)
}
return nil
})
authorizationService := g8.NewAuthorizationService().WithClientProvider(clientProvider)
gate := g8.New().WithAuthorizationService(authorizationService)
```
You can also configure the client provider to cache the output of the function you provide to retrieve clients by token:
```go
clientProvider := g8.NewClientProvider(...).WithCache(ttl, maxSize)
```
Since g8 leverages [TwiN/gocache](https://github.com/TwiN/gocache), you can also use gocache's
constants for configuring the TTL and the maximum size:
- Setting the TTL to `gocache.NoExpiration` (-1) will disable the TTL.
- Setting the maximum size to `gocache.NoMaxSize` (0) will disable the maximum cache size
If you're using a TTL and have a lot of tokens (100k+), you may want to use `clientProvider.StartJanitor()` to allow
the cache to passively delete expired entries. If you have to re-initialize the client provider after the janitor has
been started, make sure to stop the janitor first (`clientProvider.StopJanitor()`). This is because the janitor runs on
a separate goroutine, thus, if you were to re-create a client provider and re-assign it, the old client provider would
still exist in memory with the old cache. I'm only specifying this for completeness, because for the overwhelming
majority of people, the gate will be created on application start and never modified again until the application shuts
down, in which case, you don't even need to worry about stopping the janitor.
To avoid any misunderstandings, using a client provider is not mandatory. If you only have a few tokens and you can load
them on application start, you can just leverage `AuthorizationService`'s `WithToken`/`WithTokens`/`WithClient`/`WithClients`.
## AuthorizationService
As the previous examples may have hinted, there are several ways to create clients. The one thing they have
in common is that they all go through AuthorizationService, which is in charge of both managing clients and determining
whether a request should be blocked or allowed through.
| Function | Description |
|:-------------------|:---------------------------------------------------------------------------------------------------------------------------------|
| WithToken | Creates a single static client with no extra permissions |
| WithTokens | Creates a slice of static clients with no extra permissions |
| WithClient | Creates a single static client |
| WithClients | Creates a slice of static clients |
| WithClientProvider | Creates a client provider which will allow a fallback to a dynamic source (e.g. to a database) when a static client is not found |
Except for `WithClientProvider`, every functions listed above can be called more than once.
As a result, you may safely perform actions like this:
```go
authorizationService := g8.NewAuthorizationService().
WithToken("123").
WithToken("456").
WithClient(g8.NewClient("789").WithPermission("admin"))
gate := g8.New().WithAuthorizationService(authorizationService)
```
Be aware that g8.Client supports a list of permissions as well. You may call `WithPermission` several times, or call
`WithPermissions` with a slice of permissions instead.
### Permissions
Unlike client permissions, handler permissions are requirements.
A client may have as many permissions as you want, but for said client to have access to a handler protected by
permissions, the client must have all permissions defined by said handler in order to have access to it.
In other words, a client with the permissions `create`, `read`, `update` and `delete` would have access to all of these handlers:
```go
gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClient(g8.NewClient("mytoken").WithPermissions([]string{"create", "read", "update", "delete"})))
router := http.NewServeMux()
router.Handle("/", gate.Protect(homeHandler)) // equivalent of gate.ProtectWithPermissions(homeHandler, []string{})
router.Handle("/create", gate.ProtectWithPermissions(createHandler, []string{"create"}))
router.Handle("/read", gate.ProtectWithPermissions(readHandler, []string{"read"}))
router.Handle("/update", gate.ProtectWithPermissions(updateHandler, []string{"update"}))
router.Handle("/delete", gate.ProtectWithPermissions(deleteHandler, []string{"delete"}))
router.Handle("/crud", gate.ProtectWithPermissions(crudHandler, []string{"create", "read", "update", "delete"}))
```
But it would not have access to the following handler, because while `mytoken` has the `read` permission, it does not
have the `backup` permission:
```go
router.Handle("/backup", gate.ProtectWithPermissions(&testHandler{}, []string{"read", "backup"}))
```
## Rate limiting
To add a rate limit of 100 requests per second:
```
gate := g8.New().WithRateLimit(100)
```
## Special use cases
### Protecting a handler using session cookie
If you want to only allow authenticated users to access a handler, you can use a custom token extractor function
combined with a client provider.
First, we'll create a function to extract the session ID from the session cookie. While a session ID does not
theoretically refer to a token, g8 uses the term `token` as a blanket term to refer to any string that can be used to
identify a client.
```go
customTokenExtractorFunc := func(request *http.Request) string {
sessionCookie, err := request.Cookie("session")
if err != nil {
return ""
}
return sessionCookie.Value
}
```
Next, we need to create a client provider that will validate our token, which refers to the session ID in this case.
```go
clientProvider := g8.NewClientProvider(func(token string) *g8.Client {
// We'll assume that the following function calls your database and validates whether the session is valid.
isSessionValid := database.CheckIfSessionIsValid(token)
if !isSessionValid {
return nil // Returning nil will cause the gate to return a 401 Unauthorized.
}
// You could also retrieve the user and their permissions if you wanted instead, but for this example,
// all we care about is confirming whether the session is valid or not.
return g8.NewClient(token)
})
```
Keep in mind that you can get really creative with the client provider above.
For instance, you could refresh the session's expiration time, which will allow the user to stay logged in for
as long as they're active.
You're also not limited to using something stateful like the example above. You could use a JWT and have your client
provider validate said JWT.
Finally, we can create the authorization service and the gate:
```go
authorizationService := g8.NewAuthorizationService().WithClientProvider(clientProvider)
gate := g8.New().WithAuthorizationService(authorizationService).WithCustomTokenExtractor(customTokenExtractorFunc)
```
If you need to access the token (session ID in this case) from the protected handlers, you can retrieve it from the
request context by using the key `g8.TokenContextKey`:
```go
http.Handle("/handle", gate.ProtectFunc(func(w http.ResponseWriter, r *http.Request) {
sessionID, _ := r.Context().Value(g8.TokenContextKey).(string)
// ...
}))
```

122
vendor/github.com/TwiN/g8/authorization.go generated vendored Normal file
View File

@ -0,0 +1,122 @@
package g8
import (
"sync"
)
// AuthorizationService is the service that manages client/token registry and client fallback as well as the service
// that determines whether a token meets the specific requirements to be authorized by a Gate or not.
type AuthorizationService struct {
clients map[string]*Client
clientProvider *ClientProvider
mutex sync.RWMutex
}
// NewAuthorizationService creates a new AuthorizationService
func NewAuthorizationService() *AuthorizationService {
return &AuthorizationService{
clients: make(map[string]*Client),
}
}
// WithToken is used to specify a single token for which authorization will be granted
//
// The client that will be created from this token will have access to all handlers that are not protected with a
// specific permission.
//
// In other words, if you were to do the following:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithToken("12345"))
//
// The following handler would be accessible with the token 12345:
// router.Handle("/1st-handler", gate.Protect(yourHandler))
//
// But not this one would not be accessible with the token 12345:
// router.Handle("/2nd-handler", gate.ProtectWithPermissions(yourOtherHandler, []string{"admin"}))
//
// Calling this function multiple times will add multiple clients, though you may want to use WithTokens instead
// if you plan to add multiple clients
//
// If you wish to configure advanced permissions, consider using WithClient instead.
//
func (authorizationService *AuthorizationService) WithToken(token string) *AuthorizationService {
authorizationService.mutex.Lock()
authorizationService.clients[token] = NewClient(token)
authorizationService.mutex.Unlock()
return authorizationService
}
// WithTokens is used to specify a slice of tokens for which authorization will be granted
func (authorizationService *AuthorizationService) WithTokens(tokens []string) *AuthorizationService {
authorizationService.mutex.Lock()
for _, token := range tokens {
authorizationService.clients[token] = NewClient(token)
}
authorizationService.mutex.Unlock()
return authorizationService
}
// WithClient is used to specify a single client for which authorization will be granted
//
// When compared to WithToken, the advantage of using this function is that you may specify the client's
// permissions and thus, be a lot more granular with what endpoint a token has access to.
//
// In other words, if you were to do the following:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClient(g8.NewClient("12345").WithPermission("mod")))
//
// The following handlers would be accessible with the token 12345:
// router.Handle("/1st-handler", gate.ProtectWithPermissions(yourHandler, []string{"mod"}))
// router.Handle("/2nd-handler", gate.Protect(yourOtherHandler))
//
// But not this one, because the user does not have the permission "admin":
// router.Handle("/3rd-handler", gate.ProtectWithPermissions(yetAnotherHandler, []string{"admin"}))
//
// Calling this function multiple times will add multiple clients, though you may want to use WithClients instead
// if you plan to add multiple clients
func (authorizationService *AuthorizationService) WithClient(client *Client) *AuthorizationService {
authorizationService.mutex.Lock()
authorizationService.clients[client.Token] = client
authorizationService.mutex.Unlock()
return authorizationService
}
// WithClients is used to specify a slice of clients for which authorization will be granted
func (authorizationService *AuthorizationService) WithClients(clients []*Client) *AuthorizationService {
authorizationService.mutex.Lock()
for _, client := range clients {
authorizationService.clients[client.Token] = client
}
authorizationService.mutex.Unlock()
return authorizationService
}
// WithClientProvider allows specifying a custom provider to fetch clients by token.
//
// For example, you can use it to fallback to making a call in your database when a request is made with a token that
// hasn't been specified via WithToken, WithTokens, WithClient or WithClients.
func (authorizationService *AuthorizationService) WithClientProvider(provider *ClientProvider) *AuthorizationService {
authorizationService.clientProvider = provider
return authorizationService
}
// IsAuthorized checks whether a client with a given token exists and has the permissions required.
//
// If permissionsRequired is nil or empty and a client with the given token exists, said client will have access to all
// handlers that are not protected by a given permission.
func (authorizationService *AuthorizationService) IsAuthorized(token string, permissionsRequired []string) bool {
if len(token) == 0 {
return false
}
authorizationService.mutex.RLock()
client, _ := authorizationService.clients[token]
authorizationService.mutex.RUnlock()
// If there's no clients with the given token directly stored in the AuthorizationService, fall back to the
// client provider, if there's one configured.
if client == nil && authorizationService.clientProvider != nil {
client = authorizationService.clientProvider.GetClientByToken(token)
}
if client != nil {
return client.HasPermissions(permissionsRequired)
}
return false
}

58
vendor/github.com/TwiN/g8/client.go generated vendored Normal file
View File

@ -0,0 +1,58 @@
package g8
// Client is a struct containing both a Token and a slice of extra Permissions that said token has.
type Client struct {
// Token is the value used to authenticate with the API.
Token string
// Permissions is a slice of extra permissions that may be used for more granular access control.
//
// If you only wish to use Gate.Protect and Gate.ProtectFunc, you do not have to worry about this,
// since they're only used by Gate.ProtectWithPermissions and Gate.ProtectFuncWithPermissions
Permissions []string
}
// NewClient creates a Client with a given token
func NewClient(token string) *Client {
return &Client{
Token: token,
}
}
// NewClientWithPermissions creates a Client with a slice of permissions
// Equivalent to using NewClient and WithPermissions
func NewClientWithPermissions(token string, permissions []string) *Client {
return NewClient(token).WithPermissions(permissions)
}
// WithPermissions adds a slice of permissions to a client
func (client *Client) WithPermissions(permissions []string) *Client {
client.Permissions = append(client.Permissions, permissions...)
return client
}
// WithPermission adds a permission to a client
func (client *Client) WithPermission(permission string) *Client {
client.Permissions = append(client.Permissions, permission)
return client
}
// HasPermission checks whether a client has a given permission
func (client Client) HasPermission(permissionRequired string) bool {
for _, permission := range client.Permissions {
if permissionRequired == permission {
return true
}
}
return false
}
// HasPermissions checks whether a client has the all permissions passed
func (client Client) HasPermissions(permissionsRequired []string) bool {
for _, permissionRequired := range permissionsRequired {
if !client.HasPermission(permissionRequired) {
return false
}
}
return true
}

140
vendor/github.com/TwiN/g8/clientprovider.go generated vendored Normal file
View File

@ -0,0 +1,140 @@
package g8
import (
"errors"
"time"
"github.com/TwiN/gocache/v2"
)
var (
// ErrNoExpiration is the error returned by ClientProvider.StartCacheJanitor if there was an attempt to start the
// janitor despite no expiration being configured.
// To clarify, this is because the cache janitor is only useful when an expiration is set.
ErrNoExpiration = errors.New("no point starting the janitor if the TTL is set to not expire")
// ErrCacheNotInitialized is the error returned by ClientProvider.StartCacheJanitor if there was an attempt to start
// the janitor despite the cache not having been initialized using ClientProvider.WithCache
ErrCacheNotInitialized = errors.New("cannot start janitor because cache is not configured")
)
// ClientProvider has the task of retrieving a Client from an external source (e.g. a database) when provided with a
// token. It should be used when you have a lot of tokens, and it wouldn't make sense to register all of them using
// AuthorizationService's WithToken, WithTokens, WithClient or WithClients.
//
// Note that the provider is used as a fallback source. As such, if a token is explicitly registered using one of the 4
// aforementioned functions, the client provider will not be used by the AuthorizationService when a request is made
// with said token. It will, however, be called upon if a token that is not explicitly registered in
// AuthorizationService is sent alongside a request going through the Gate.
//
// clientProvider := g8.NewClientProvider(func(token string) *g8.Client {
// // We'll assume that the following function calls your database and returns a struct "User" that
// // has the user's token as well as the permissions granted to said user
// user := database.GetUserByToken(token)
// if user != nil {
// return g8.NewClient(user.Token).WithPermissions(user.Permissions)
// }
// return nil
// })
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClientProvider(clientProvider))
//
type ClientProvider struct {
getClientByTokenFunc func(token string) *Client
cache *gocache.Cache
ttl time.Duration
}
// NewClientProvider creates a ClientProvider
// The parameter that must be passed is a function that the provider will use to retrieve a client by a given token
//
// Example:
// clientProvider := g8.NewClientProvider(func(token string) *g8.Client {
// // We'll assume that the following function calls your database and returns a struct "User" that
// // has the user's token as well as the permissions granted to said user
// user := database.GetUserByToken(token)
// if user == nil {
// return nil
// }
// return g8.NewClient(user.Token).WithPermissions(user.Permissions)
// })
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClientProvider(clientProvider))
//
func NewClientProvider(getClientByTokenFunc func(token string) *Client) *ClientProvider {
return &ClientProvider{
getClientByTokenFunc: getClientByTokenFunc,
}
}
// WithCache adds cache options to the ClientProvider.
//
// ttl is the time until the cache entry will expire. A TTL of gocache.NoExpiration (-1) means no expiration
// maxSize is the maximum amount of entries that can be in the cache at any given time.
// If a value of gocache.NoMaxSize (0) or less is provided for maxSize, there will be no maximum size.
//
// Example:
// clientProvider := g8.NewClientProvider(func(token string) *g8.Client {
// // We'll assume that the following function calls your database and returns a struct "User" that
// // has the user's token as well as the permissions granted to said user
// user := database.GetUserByToken(token)
// if user != nil {
// return g8.NewClient(user.Token).WithPermissions(user.Permissions)
// }
// return nil
// })
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClientProvider(clientProvider.WithCache(time.Hour, 70000)))
//
func (provider *ClientProvider) WithCache(ttl time.Duration, maxSize int) *ClientProvider {
provider.cache = gocache.NewCache().WithEvictionPolicy(gocache.LeastRecentlyUsed).WithMaxSize(maxSize)
provider.ttl = ttl
return provider
}
// StartCacheJanitor starts the cache janitor, which passively deletes expired cache entries in the background.
//
// Not really necessary unless you have a lot of clients (100000+).
//
// Even without the janitor, active eviction will still happen (i.e. when GetClientByToken is called, but the cache
// entry for the given token has expired, the cache entry will be automatically deleted and re-fetched from the
// user-defined getClientByTokenFunc)
func (provider *ClientProvider) StartCacheJanitor() error {
if provider.cache == nil {
// Can't start the cache janitor if there's no cache
return ErrCacheNotInitialized
}
if provider.ttl != gocache.NoExpiration {
return provider.cache.StartJanitor()
}
return ErrNoExpiration
}
// StopCacheJanitor stops the cache janitor
//
// Not required unless your application initializes multiple providers over the course of its lifecycle.
// In English, that means if you initialize a ClientProvider only once on application start and it stays up
// until your application shuts down, you don't need to call this function.
func (provider *ClientProvider) StopCacheJanitor() {
if provider.cache != nil {
provider.cache.StopJanitor()
}
}
// GetClientByToken retrieves a client by its token through the provided getClientByTokenFunc.
func (provider *ClientProvider) GetClientByToken(token string) *Client {
if provider.cache == nil {
return provider.getClientByTokenFunc(token)
}
if cachedClient, exists := provider.cache.Get(token); exists {
if cachedClient == nil {
return nil
}
// Safely typecast the client.
// Regardless of whether the typecast is successful or not, we return client since it'll be either client or
// nil. Technically, it should never be nil, but it's better to be safe than sorry.
client, _ := cachedClient.(*Client)
return client
}
client := provider.getClientByTokenFunc(token)
provider.cache.SetWithTTL(token, client, provider.ttl)
return client
}

212
vendor/github.com/TwiN/g8/gate.go generated vendored Normal file
View File

@ -0,0 +1,212 @@
package g8
import (
"context"
"net/http"
"strings"
)
const (
// AuthorizationHeader is the header in which g8 looks for the authorization bearer token
AuthorizationHeader = "Authorization"
// DefaultUnauthorizedResponseBody is the default response body returned if a request was sent with a missing or invalid token
DefaultUnauthorizedResponseBody = "token is missing or invalid"
// DefaultTooManyRequestsResponseBody is the default response body returned if a request exceeded the allowed rate limit
DefaultTooManyRequestsResponseBody = "too many requests"
// TokenContextKey is the key used to store the token in the context.
TokenContextKey = "g8.token"
)
// Gate is lock to the front door of your API, letting only those you allow through.
type Gate struct {
authorizationService *AuthorizationService
unauthorizedResponseBody []byte
customTokenExtractorFunc func(request *http.Request) string
rateLimiter *RateLimiter
tooManyRequestsResponseBody []byte
}
// Deprecated: use New instead.
func NewGate(authorizationService *AuthorizationService) *Gate {
return &Gate{
authorizationService: authorizationService,
unauthorizedResponseBody: []byte(DefaultUnauthorizedResponseBody),
tooManyRequestsResponseBody: []byte(DefaultTooManyRequestsResponseBody),
}
}
// New creates a new Gate.
func New() *Gate {
return &Gate{
unauthorizedResponseBody: []byte(DefaultUnauthorizedResponseBody),
tooManyRequestsResponseBody: []byte(DefaultTooManyRequestsResponseBody),
}
}
// WithAuthorizationService sets the authorization service to use.
//
// If there is no authorization service, Gate will not enforce authorization.
func (gate *Gate) WithAuthorizationService(authorizationService *AuthorizationService) *Gate {
gate.authorizationService = authorizationService
return gate
}
// WithCustomUnauthorizedResponseBody sets a custom response body when Gate determines that a request must be blocked
func (gate *Gate) WithCustomUnauthorizedResponseBody(unauthorizedResponseBody []byte) *Gate {
gate.unauthorizedResponseBody = unauthorizedResponseBody
return gate
}
// WithCustomTokenExtractor allows the specification of a custom function to extract a token from a request.
// If a custom token extractor is not specified, the token will be extracted from the Authorization header.
//
// For instance, if you're using a session cookie, you can extract the token from the cookie like so:
// authorizationService := g8.NewAuthorizationService()
// customTokenExtractorFunc := func(request *http.Request) string {
// sessionCookie, err := request.Cookie("session")
// if err != nil {
// return ""
// }
// return sessionCookie.Value
// }
// gate := g8.New().WithAuthorizationService(authorizationService).WithCustomTokenExtractor(customTokenExtractorFunc)
//
// You would normally use this with a client provider that matches whatever need you have.
// For example, if you're using a session cookie, your client provider would retrieve the user from the session ID
// extracted by this custom token extractor.
//
// Note that for the sake of convenience, the token extracted from the request is passed the protected handlers request
// context under the key TokenContextKey. This is especially useful if the token is in fact a session ID.
func (gate *Gate) WithCustomTokenExtractor(customTokenExtractorFunc func(request *http.Request) string) *Gate {
gate.customTokenExtractorFunc = customTokenExtractorFunc
return gate
}
// WithRateLimit adds rate limiting to the Gate
//
// If you just want to use a gate for rate limiting purposes:
// gate := g8.New().WithRateLimit(50)
//
func (gate *Gate) WithRateLimit(maximumRequestsPerSecond int) *Gate {
gate.rateLimiter = NewRateLimiter(maximumRequestsPerSecond)
return gate
}
// Protect secures a handler, requiring requests going through to have a valid Authorization Bearer token.
// Unlike ProtectWithPermissions, Protect will allow access to any registered tokens, regardless of their permissions
// or lack thereof.
//
// Example:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithToken("token"))
// router := http.NewServeMux()
// // Without protection
// router.Handle("/handle", yourHandler)
// // With protection
// router.Handle("/handle", gate.Protect(yourHandler))
//
// The token extracted from the request is passed to the handlerFunc request context under the key TokenContextKey
func (gate *Gate) Protect(handler http.Handler) http.Handler {
return gate.ProtectWithPermissions(handler, nil)
}
// ProtectWithPermissions secures a handler, requiring requests going through to have a valid Authorization Bearer token
// as well as a slice of permissions that must be met.
//
// Example:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClient(g8.NewClient("token").WithPermission("admin")))
// router := http.NewServeMux()
// // Without protection
// router.Handle("/handle", yourHandler)
// // With protection
// router.Handle("/handle", gate.ProtectWithPermissions(yourHandler, []string{"admin"}))
//
// The token extracted from the request is passed to the handlerFunc request context under the key TokenContextKey
func (gate *Gate) ProtectWithPermissions(handler http.Handler, permissions []string) http.Handler {
return gate.ProtectFuncWithPermissions(func(writer http.ResponseWriter, request *http.Request) {
handler.ServeHTTP(writer, request)
}, permissions)
}
// ProtectWithPermission does the same thing as ProtectWithPermissions, but for a single permission instead of a
// slice of permissions
//
// See ProtectWithPermissions for further documentation
func (gate *Gate) ProtectWithPermission(handler http.Handler, permission string) http.Handler {
return gate.ProtectFuncWithPermissions(func(writer http.ResponseWriter, request *http.Request) {
handler.ServeHTTP(writer, request)
}, []string{permission})
}
// ProtectFunc secures a handlerFunc, requiring requests going through to have a valid Authorization Bearer token.
// Unlike ProtectFuncWithPermissions, ProtectFunc will allow access to any registered tokens, regardless of their
// permissions or lack thereof.
//
// Example:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithToken("token"))
// router := http.NewServeMux()
// // Without protection
// router.HandleFunc("/handle", yourHandlerFunc)
// // With protection
// router.HandleFunc("/handle", gate.ProtectFunc(yourHandlerFunc))
//
// The token extracted from the request is passed to the handlerFunc request context under the key TokenContextKey
func (gate *Gate) ProtectFunc(handlerFunc http.HandlerFunc) http.HandlerFunc {
return gate.ProtectFuncWithPermissions(handlerFunc, nil)
}
// ProtectFuncWithPermissions secures a handler, requiring requests going through to have a valid Authorization Bearer
// token as well as a slice of permissions that must be met.
//
// Example:
// gate := g8.New().WithAuthorizationService(g8.NewAuthorizationService().WithClient(g8.NewClient("token").WithPermission("admin")))
// router := http.NewServeMux()
// // Without protection
// router.HandleFunc("/handle", yourHandlerFunc)
// // With protection
// router.HandleFunc("/handle", gate.ProtectFuncWithPermissions(yourHandlerFunc, []string{"admin"}))
//
// The token extracted from the request is passed to the handlerFunc request context under the key TokenContextKey
func (gate *Gate) ProtectFuncWithPermissions(handlerFunc http.HandlerFunc, permissions []string) http.HandlerFunc {
return func(writer http.ResponseWriter, request *http.Request) {
if gate.rateLimiter != nil {
if !gate.rateLimiter.Try() {
writer.WriteHeader(http.StatusTooManyRequests)
_, _ = writer.Write(gate.tooManyRequestsResponseBody)
return
}
}
if gate.authorizationService != nil {
var token string
if gate.customTokenExtractorFunc != nil {
token = gate.customTokenExtractorFunc(request)
} else {
token = extractTokenFromRequest(request)
}
if !gate.authorizationService.IsAuthorized(token, permissions) {
writer.WriteHeader(http.StatusUnauthorized)
_, _ = writer.Write(gate.unauthorizedResponseBody)
return
}
request = request.WithContext(context.WithValue(request.Context(), TokenContextKey, token))
}
handlerFunc(writer, request)
}
}
// ProtectFuncWithPermission does the same thing as ProtectFuncWithPermissions, but for a single permission instead of a
// slice of permissions
//
// See ProtectFuncWithPermissions for further documentation
func (gate *Gate) ProtectFuncWithPermission(handlerFunc http.HandlerFunc, permission string) http.HandlerFunc {
return gate.ProtectFuncWithPermissions(handlerFunc, []string{permission})
}
// extractTokenFromRequest extracts the bearer token from the AuthorizationHeader
func extractTokenFromRequest(request *http.Request) string {
return strings.TrimPrefix(request.Header.Get(AuthorizationHeader), "Bearer ")
}

42
vendor/github.com/TwiN/g8/ratelimiter.go generated vendored Normal file
View File

@ -0,0 +1,42 @@
package g8
import (
"sync"
"time"
)
// RateLimiter is a fixed rate limiter
type RateLimiter struct {
maximumExecutionsPerSecond int
executionsLeftInWindow int
windowStartTime time.Time
mutex sync.Mutex
}
// NewRateLimiter creates a RateLimiter
func NewRateLimiter(maximumExecutionsPerSecond int) *RateLimiter {
return &RateLimiter{
windowStartTime: time.Now(),
executionsLeftInWindow: maximumExecutionsPerSecond,
maximumExecutionsPerSecond: maximumExecutionsPerSecond,
}
}
// Try updates the number of executions if the rate limit quota hasn't been reached and returns whether the
// attempt was successful or not.
//
// Returns false if the execution was not successful (rate limit quota has been reached)
// Returns true if the execution was successful (rate limit quota has not been reached)
func (r *RateLimiter) Try() bool {
r.mutex.Lock()
defer r.mutex.Unlock()
if time.Now().Add(-time.Second).After(r.windowStartTime) {
r.windowStartTime = time.Now()
r.executionsLeftInWindow = r.maximumExecutionsPerSecond
}
if r.executionsLeftInWindow == 0 {
return false
}
r.executionsLeftInWindow--
return true
}

1
vendor/github.com/TwiN/gocache/v2/.gitattributes generated vendored Normal file
View File

@ -0,0 +1 @@
* text=lf

1
vendor/github.com/TwiN/gocache/v2/.gitignore generated vendored Normal file
View File

@ -0,0 +1 @@
.idea

9
vendor/github.com/TwiN/gocache/v2/LICENSE.md generated vendored Normal file
View File

@ -0,0 +1,9 @@
MIT License
Copyright (c) 2021 TwiN
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

370
vendor/github.com/TwiN/gocache/v2/README.md generated vendored Normal file
View File

@ -0,0 +1,370 @@
# gocache
![build](https://github.com/TwiN/gocache/workflows/build/badge.svg?branch=master)
[![Go Report Card](https://goreportcard.com/badge/github.com/TwiN/gocache)](https://goreportcard.com/report/github.com/TwiN/gocache)
[![codecov](https://codecov.io/gh/TwiN/gocache/branch/master/graph/badge.svg)](https://codecov.io/gh/TwiN/gocache)
[![Go version](https://img.shields.io/github/go-mod/go-version/TwiN/gocache.svg)](https://github.com/TwiN/gocache)
[![Go Reference](https://pkg.go.dev/badge/github.com/TwiN/gocache.svg)](https://pkg.go.dev/github.com/TwiN/gocache)
[![Follow TwiN](https://img.shields.io/github/followers/TwiN?label=Follow&style=social)](https://github.com/TwiN)
gocache is an easy-to-use, high-performance, lightweight and thread-safe (goroutine-safe) in-memory key-value cache
with support for LRU and FIFO eviction policies as well as expiration, bulk operations and even retrieval of keys by pattern.
## Table of Contents
- [Features](#features)
- [Usage](#usage)
- [Initializing the cache](#initializing-the-cache)
- [Functions](#functions)
- [Examples](#examples)
- [Creating or updating an entry](#creating-or-updating-an-entry)
- [Getting an entry](#getting-an-entry)
- [Deleting an entry](#deleting-an-entry)
- [Complex example](#complex-example)
- [Persistence](#persistence)
- [Eviction](#eviction)
- [MaxSize](#maxsize)
- [MaxMemoryUsage](#maxmemoryusage)
- [Expiration](#expiration)
- [Performance](#performance)
- [Summary](#summary)
- [Results](#results)
- [FAQ](#faq)
- [How can I persist the data on application termination?](#how-can-i-persist-the-data-on-application-termination)
## Features
gocache supports the following cache eviction policies:
- First in first out (FIFO)
- Least recently used (LRU)
It also supports cache entry TTL, which is both active and passive. Active expiration means that if you attempt
to retrieve a cache key that has already expired, it will delete it on the spot and the behavior will be as if
the cache key didn't exist. As for passive expiration, there's a background task that will take care of deleting
expired keys.
It also includes what you'd expect from a cache, like GET/SET, bulk operations and get by pattern.
## Usage
```
go get -u github.com/TwiN/gocache/v2
```
### Initializing the cache
```go
cache := gocache.NewCache().WithMaxSize(1000).WithEvictionPolicy(gocache.LeastRecentlyUsed)
```
If you're planning on using expiration (`SetWithTTL` or `Expire`) and you want expired entries to be automatically deleted
in the background, make sure to start the janitor when you instantiate the cache:
```go
cache.StartJanitor()
```
### Functions
| Function | Description |
| --------------------------------- | ----------- |
| WithMaxSize | Sets the max size of the cache. `gocache.NoMaxSize` means there is no limit. If not set, the default max size is `gocache.DefaultMaxSize`.
| WithMaxMemoryUsage | Sets the max memory usage of the cache. `gocache.NoMaxMemoryUsage` means there is no limit. The default behavior is to not evict based on memory usage.
| WithEvictionPolicy | Sets the eviction algorithm to be used when the cache reaches the max size. If not set, the default eviction policy is `gocache.FirstInFirstOut` (FIFO).
| WithForceNilInterfaceOnNilPointer | Configures whether values with a nil pointer passed to write functions should be forcefully set to nil. Defaults to true.
| StartJanitor | Starts the janitor, which is in charge of deleting expired cache entries in the background.
| StopJanitor | Stops the janitor.
| Set | Same as `SetWithTTL`, but with no expiration (`gocache.NoExpiration`)
| SetAll | Same as `Set`, but in bulk
| SetWithTTL | Creates or updates a cache entry with the given key, value and expiration time. If the max size after the aforementioned operation is above the configured max size, the tail will be evicted. Depending on the eviction policy, the tail is defined as the oldest
| Get | Gets a cache entry by its key.
| GetByKeys | Gets a map of entries by their keys. The resulting map will contain all keys, even if some of the keys in the slice passed as parameter were not present in the cache.
| GetAll | Gets all cache entries.
| GetKeysByPattern | Retrieves a slice of keys that matches a given pattern.
| Delete | Removes a key from the cache.
| DeleteAll | Removes multiple keys from the cache.
| Count | Gets the size of the cache. This includes cache keys which may have already expired, but have not been removed yet.
| Clear | Wipes the cache.
| TTL | Gets the time until a cache key expires.
| Expire | Sets the expiration time of an existing cache key.
For further documentation, please refer to [Go Reference](https://pkg.go.dev/github.com/TwiN/gocache)
### Examples
#### Creating or updating an entry
```go
cache.Set("key", "value")
cache.Set("key", 1)
cache.Set("key", struct{ Text string }{Test: "value"})
cache.SetWithTTL("key", []byte("value"), 24*time.Hour)
```
#### Getting an entry
```go
value, exists := cache.Get("key")
```
You can also get multiple entries by using `cache.GetByKeys([]string{"key1", "key2"})`
#### Deleting an entry
```go
cache.Delete("key")
```
You can also delete multiple entries by using `cache.DeleteAll([]string{"key1", "key2"})`
#### Complex example
```go
package main
import (
"fmt"
"time"
"github.com/TwiN/gocache/v2"
)
func main() {
cache := gocache.NewCache().WithEvictionPolicy(gocache.LeastRecentlyUsed).WithMaxSize(10000)
cache.StartJanitor() // Passively manages expired entries
defer cache.StopJanitor()
cache.Set("key", "value")
cache.SetWithTTL("key-with-ttl", "value", 60*time.Minute)
cache.SetAll(map[string]interface{}{"k1": "v1", "k2": "v2", "k3": "v3"})
fmt.Println("[Count] Cache size:", cache.Count())
value, exists := cache.Get("key")
fmt.Printf("[Get] key=key; value=%s; exists=%v\n", value, exists)
for key, value := range cache.GetByKeys([]string{"k1", "k2", "k3"}) {
fmt.Printf("[GetByKeys] key=%s; value=%s\n", key, value)
}
for _, key := range cache.GetKeysByPattern("key*", 0) {
fmt.Printf("[GetKeysByPattern] pattern=key*; key=%s\n", key)
}
cache.Expire("key", time.Hour)
time.Sleep(500*time.Millisecond)
timeUntilExpiration, _ := cache.TTL("key")
fmt.Println("[TTL] Number of minutes before 'key' expires:", int(timeUntilExpiration.Seconds()))
cache.Delete("key")
cache.DeleteAll([]string{"k1", "k2", "k3"})
cache.Clear()
fmt.Println("[Count] Cache size after clearing the cache:", cache.Count())
}
```
<details>
<summary>Output</summary>
```
[Count] Cache size: 5
[Get] key=key; value=value; exists=true
[GetByKeys] key=k1; value=v1
[GetByKeys] key=k2; value=v2
[GetByKeys] key=k3; value=v3
[GetKeysByPattern] pattern=key*; key=key-with-ttl
[GetKeysByPattern] pattern=key*; key=key
[TTL] Number of minutes before 'key' expires: 3599
[Count] Cache size after clearing the cache: 0
```
</details>
## Persistence
Prior to v2, gocache supported persistence out of the box.
After some thinking, I decided that persistence added too many dependencies, and given than this is a cache library
and most people wouldn't be interested in persistence, I decided to get rid of it.
That being said, you can use the `GetAll` and `SetAll` methods of `gocache.Cache` to implement persistence yourself.
## Eviction
### MaxSize
Eviction by MaxSize is the default behavior, and is also the most efficient.
The code below will create a cache that has a maximum size of 1000:
```go
cache := gocache.NewCache().WithMaxSize(1000)
```
This means that whenever an operation causes the total size of the cache to go above 1000, the tail will be evicted.
### MaxMemoryUsage
Eviction by MaxMemoryUsage is **disabled by default**, and is in alpha.
The code below will create a cache that has a maximum memory usage of 50MB:
```go
cache := gocache.NewCache().WithMaxSize(0).WithMaxMemoryUsage(50*gocache.Megabyte)
```
This means that whenever an operation causes the total memory usage of the cache to go above 50MB, one or more tails
will be evicted.
Unlike evictions caused by reaching the MaxSize, evictions triggered by MaxMemoryUsage may lead to multiple entries
being evicted in a row. The reason for this is that if, for instance, you had 100 entries of 0.1MB each and you suddenly added
a single entry of 10MB, 100 entries would need to be evicted to make enough space for that new big entry.
It's very important to keep in mind that eviction by MaxMemoryUsage is approximate.
**The only memory taken into consideration is the size of the cache, not the size of the entire application.**
If you pass along 100MB worth of data in a matter of seconds, even though the cache's memory usage will remain
under 50MB (or whatever you configure the MaxMemoryUsage to), the memory footprint generated by that 100MB will
still exist until the next GC cycle.
As previously mentioned, this is a work in progress, and here's a list of the things you should keep in mind:
- The memory usage of structs are a gross estimation and may not reflect the actual memory usage.
- Native types (string, int, bool, []byte, etc.) are the most accurate for calculating the memory usage.
- Adding an entry bigger than the configured MaxMemoryUsage will work, but it will evict all other entries.
## Expiration
There are two ways that the deletion of expired keys can take place:
- Active
- Passive
**Active deletion of expired keys** happens when an attempt is made to access the value of a cache entry that expired.
`Get`, `GetByKeys` and `GetAll` are the only functions that can trigger active deletion of expired keys.
**Passive deletion of expired keys** runs in the background and is managed by the janitor.
If you do not start the janitor, there will be no passive deletion of expired keys.
## Performance
### Summary
- **Set**: Both map and gocache have the same performance.
- **Get**: Map is faster than gocache.
This is because gocache keeps track of the head and the tail for eviction and expiration/TTL.
Ultimately, the difference is negligible.
We could add a way to disable eviction or disable expiration altogether just to match the map's performance,
but if you're looking into using a library like gocache, odds are, you want more than just a map.
### Results
| key | value |
|:------ |:-------- |
| goos | windows |
| goarch | amd64 |
| cpu | i7-9700K |
| mem | 32G DDR4 |
```
// Normal map
BenchmarkMap_Get
BenchmarkMap_Get-8 46087372 26.7 ns/op
BenchmarkMap_Set
BenchmarkMap_Set/small_value-8 3841911 389 ns/op
BenchmarkMap_Set/medium_value-8 3887074 391 ns/op
BenchmarkMap_Set/large_value-8 3921956 393 ns/op
// Gocache
BenchmarkCache_Get
BenchmarkCache_Get/FirstInFirstOut-8 27273036 46.4 ns/op
BenchmarkCache_Get/LeastRecentlyUsed-8 26648248 46.3 ns/op
BenchmarkCache_Set
BenchmarkCache_Set/FirstInFirstOut_small_value-8 2919584 405 ns/op
BenchmarkCache_Set/FirstInFirstOut_medium_value-8 2990841 391 ns/op
BenchmarkCache_Set/FirstInFirstOut_large_value-8 2970513 391 ns/op
BenchmarkCache_Set/LeastRecentlyUsed_small_value-8 2962939 402 ns/op
BenchmarkCache_Set/LeastRecentlyUsed_medium_value-8 2962963 390 ns/op
BenchmarkCache_Set/LeastRecentlyUsed_large_value-8 2962928 394 ns/op
BenchmarkCache_SetUsingMaxMemoryUsage
BenchmarkCache_SetUsingMaxMemoryUsage/small_value-8 2683356 447 ns/op
BenchmarkCache_SetUsingMaxMemoryUsage/medium_value-8 2637578 441 ns/op
BenchmarkCache_SetUsingMaxMemoryUsage/large_value-8 2672434 443 ns/op
BenchmarkCache_SetWithMaxSize
BenchmarkCache_SetWithMaxSize/100_small_value-8 4782966 252 ns/op
BenchmarkCache_SetWithMaxSize/10000_small_value-8 4067967 296 ns/op
BenchmarkCache_SetWithMaxSize/100000_small_value-8 3762055 328 ns/op
BenchmarkCache_SetWithMaxSize/100_medium_value-8 4760479 252 ns/op
BenchmarkCache_SetWithMaxSize/10000_medium_value-8 4081050 295 ns/op
BenchmarkCache_SetWithMaxSize/100000_medium_value-8 3785050 330 ns/op
BenchmarkCache_SetWithMaxSize/100_large_value-8 4732909 254 ns/op
BenchmarkCache_SetWithMaxSize/10000_large_value-8 4079533 297 ns/op
BenchmarkCache_SetWithMaxSize/100000_large_value-8 3712820 331 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU
BenchmarkCache_SetWithMaxSizeAndLRU/100_small_value-8 4761732 254 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/10000_small_value-8 4084474 296 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/100000_small_value-8 3761402 329 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/100_medium_value-8 4783075 254 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/10000_medium_value-8 4103980 296 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/100000_medium_value-8 3646023 331 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/100_large_value-8 4779025 254 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/10000_large_value-8 4096192 296 ns/op
BenchmarkCache_SetWithMaxSizeAndLRU/100000_large_value-8 3726823 331 ns/op
BenchmarkCache_GetSetMultipleConcurrent
BenchmarkCache_GetSetMultipleConcurrent-8 707142 1698 ns/op
BenchmarkCache_GetSetConcurrentWithFrequentEviction
BenchmarkCache_GetSetConcurrentWithFrequentEviction/FirstInFirstOut-8 3616256 334 ns/op
BenchmarkCache_GetSetConcurrentWithFrequentEviction/LeastRecentlyUsed-8 3636367 331 ns/op
BenchmarkCache_GetConcurrentWithLRU
BenchmarkCache_GetConcurrentWithLRU/FirstInFirstOut-8 4405557 268 ns/op
BenchmarkCache_GetConcurrentWithLRU/LeastRecentlyUsed-8 4445475 269 ns/op
BenchmarkCache_WithForceNilInterfaceOnNilPointer
BenchmarkCache_WithForceNilInterfaceOnNilPointer/true_with_nil_struct_pointer-8 6184591 191 ns/op
BenchmarkCache_WithForceNilInterfaceOnNilPointer/true-8 6090482 191 ns/op
BenchmarkCache_WithForceNilInterfaceOnNilPointer/false_with_nil_struct_pointer-8 6184629 187 ns/op
BenchmarkCache_WithForceNilInterfaceOnNilPointer/false-8 6281781 186 ns/op
(Trimmed "BenchmarkCache_" for readability)
WithForceNilInterfaceOnNilPointerWithConcurrency
WithForceNilInterfaceOnNilPointerWithConcurrency/true_with_nil_struct_pointer-8 4379564 268 ns/op
WithForceNilInterfaceOnNilPointerWithConcurrency/true-8 4379558 265 ns/op
WithForceNilInterfaceOnNilPointerWithConcurrency/false_with_nil_struct_pointer-8 4444456 261 ns/op
WithForceNilInterfaceOnNilPointerWithConcurrency/false-8 4493896 262 ns/op
```
## FAQ
### How can I persist the data on application termination?
While creating your own auto save feature might come in handy, it may still lead to loss of data if the application
automatically saves every 10 minutes and your application crashes 9 minutes after the previous save.
To increase your odds of not losing any data, you can use Go's `signal` package, more specifically its `Notify` function
which allows listening for termination signals like SIGTERM and SIGINT. Once a termination signal is caught, you can
add the necessary logic for a graceful shutdown.
In the following example, the code that would usually be present in the `main` function is moved to a different function
named `Start` which is launched on a different goroutine so that listening for a termination signals is what blocks the
main goroutine instead:
```go
package main
import (
"log"
"os"
"os/signal"
"syscall"
"github.com/TwiN/gocache/v2"
)
var cache = gocache.NewCache()
func main() {
data := retrieveCacheEntriesUsingWhateverMeanYouUsedToPersistIt()
cache.SetAll(data)
// Start everything else on another goroutine to prevent blocking the main goroutine
go Start()
// Wait for termination signal
sig := make(chan os.Signal, 1)
done := make(chan bool, 1)
signal.Notify(sig, os.Interrupt, syscall.SIGTERM)
go func() {
<-sig
log.Println("Received termination signal, attempting to gracefully shut down")
// Persist the cache entries
cacheEntries := cache.GetAll()
persistCacheEntriesHoweverYouWant(cacheEntries)
// Tell the main goroutine that we're done
done <- true
}()
<-done
log.Println("Shutting down")
}
```
Note that this won't protect you from a SIGKILL, as this signal cannot be caught.

108
vendor/github.com/TwiN/gocache/v2/entry.go generated vendored Normal file
View File

@ -0,0 +1,108 @@
package gocache
import (
"fmt"
"time"
"unsafe"
)
// Entry is a cache entry
type Entry struct {
// Key is the name of the cache entry
Key string
// Value is the value of the cache entry
Value interface{}
// RelevantTimestamp is the variable used to store either:
// - creation timestamp, if the Cache's EvictionPolicy is FirstInFirstOut
// - last access timestamp, if the Cache's EvictionPolicy is LeastRecentlyUsed
//
// Note that updating an existing entry will also update this value
RelevantTimestamp time.Time
// Expiration is the unix time in nanoseconds at which the entry will expire (-1 means no expiration)
Expiration int64
next *Entry
previous *Entry
}
// Accessed updates the Entry's RelevantTimestamp to now
func (entry *Entry) Accessed() {
entry.RelevantTimestamp = time.Now()
}
// Expired returns whether the Entry has expired
func (entry Entry) Expired() bool {
if entry.Expiration > 0 {
if time.Now().UnixNano() > entry.Expiration {
return true
}
}
return false
}
// SizeInBytes returns the size of an entry in bytes, approximately.
func (entry *Entry) SizeInBytes() int {
return toBytes(entry.Key) + toBytes(entry.Value) + 32
}
func toBytes(value interface{}) int {
switch value.(type) {
case string:
return int(unsafe.Sizeof(value)) + len(value.(string))
case int8, uint8, bool:
return int(unsafe.Sizeof(value)) + 1
case int16, uint16:
return int(unsafe.Sizeof(value)) + 2
case int32, uint32, float32, complex64:
return int(unsafe.Sizeof(value)) + 4
case int64, uint64, int, uint, float64, complex128:
return int(unsafe.Sizeof(value)) + 8
case []interface{}:
size := 0
for _, v := range value.([]interface{}) {
size += toBytes(v)
}
return int(unsafe.Sizeof(value)) + size
case []string:
size := 0
for _, v := range value.([]string) {
size += toBytes(v)
}
return int(unsafe.Sizeof(value)) + size
case []int8:
return int(unsafe.Sizeof(value)) + len(value.([]int8))
case []uint8:
return int(unsafe.Sizeof(value)) + len(value.([]uint8))
case []bool:
return int(unsafe.Sizeof(value)) + len(value.([]bool))
case []int16:
return int(unsafe.Sizeof(value)) + (len(value.([]int16)) * 2)
case []uint16:
return int(unsafe.Sizeof(value)) + (len(value.([]uint16)) * 2)
case []int32:
return int(unsafe.Sizeof(value)) + (len(value.([]int32)) * 4)
case []uint32:
return int(unsafe.Sizeof(value)) + (len(value.([]uint32)) * 4)
case []float32:
return int(unsafe.Sizeof(value)) + (len(value.([]float32)) * 4)
case []complex64:
return int(unsafe.Sizeof(value)) + (len(value.([]complex64)) * 4)
case []int64:
return int(unsafe.Sizeof(value)) + (len(value.([]int64)) * 8)
case []uint64:
return int(unsafe.Sizeof(value)) + (len(value.([]uint64)) * 8)
case []int:
return int(unsafe.Sizeof(value)) + (len(value.([]int)) * 8)
case []uint:
return int(unsafe.Sizeof(value)) + (len(value.([]uint)) * 8)
case []float64:
return int(unsafe.Sizeof(value)) + (len(value.([]float64)) * 8)
case []complex128:
return int(unsafe.Sizeof(value)) + (len(value.([]complex128)) * 8)
default:
return int(unsafe.Sizeof(value)) + len(fmt.Sprintf("%v", value))
}
}

567
vendor/github.com/TwiN/gocache/v2/gocache.go generated vendored Normal file
View File

@ -0,0 +1,567 @@
package gocache
import (
"errors"
"reflect"
"sync"
"time"
)
var (
Debug = false
)
const (
// NoMaxSize means that the cache has no maximum number of entries in the cache
// Setting Cache.maxSize to this value also means there will be no eviction
NoMaxSize = 0
// NoMaxMemoryUsage means that the cache has no maximum number of entries in the cache
NoMaxMemoryUsage = 0
// DefaultMaxSize is the max size set if no max size is specified
DefaultMaxSize = 100000
// NoExpiration is the value that must be used as TTL to specify that the given key should never expire
NoExpiration = -1
Kilobyte = 1024
Megabyte = 1024 * Kilobyte
Gigabyte = 1024 * Megabyte
)
var (
ErrKeyDoesNotExist = errors.New("key does not exist") // Returned when a cache key does not exist
ErrKeyHasNoExpiration = errors.New("key has no expiration") // Returned when a cache key has no expiration
ErrJanitorAlreadyRunning = errors.New("janitor is already running") // Returned when the janitor has already been started
)
// Cache is the core struct of gocache which contains the data as well as all relevant configuration fields
type Cache struct {
// maxSize is the maximum amount of entries that can be in the cache at any given time
// By default, this is set to DefaultMaxSize
maxSize int
// maxMemoryUsage is the maximum amount of memory that can be taken up by the cache at any time
// By default, this is set to NoMaxMemoryUsage, meaning that the default behavior is to not evict
// based on maximum memory usage
maxMemoryUsage int
// evictionPolicy is the eviction policy
evictionPolicy EvictionPolicy
// stats is the object that contains cache statistics/metrics
stats *Statistics
// entries is the content of the cache
entries map[string]*Entry
// mutex is the lock for making concurrent operations on the cache
mutex sync.RWMutex
// head is the cache entry at the head of the cache
head *Entry
// tail is the last cache node and also the next entry that will be evicted
tail *Entry
// stopJanitor is the channel used to stop the janitor
stopJanitor chan bool
// memoryUsage is the approximate memory usage of the cache (dataset only) in bytes
memoryUsage int
// forceNilInterfaceOnNilPointer determines whether all Set-like functions should set a value as nil if the
// interface passed has a nil value but not a nil type.
//
// By default, interfaces are only nil when both their type and value is nil.
// This means that when you pass a pointer to a nil value, the type of the interface
// will still show as nil, which means that if you don't cast the interface after
// retrieving it, a nil check will return that the value is not false.
forceNilInterfaceOnNilPointer bool
}
// MaxSize returns the maximum amount of keys that can be present in the cache before
// new entries trigger the eviction of the tail
func (cache *Cache) MaxSize() int {
return cache.maxSize
}
// MaxMemoryUsage returns the configured maxMemoryUsage of the cache
func (cache *Cache) MaxMemoryUsage() int {
return cache.maxMemoryUsage
}
// EvictionPolicy returns the EvictionPolicy of the Cache
func (cache *Cache) EvictionPolicy() EvictionPolicy {
return cache.evictionPolicy
}
// Stats returns statistics from the cache
func (cache *Cache) Stats() Statistics {
cache.mutex.RLock()
stats := Statistics{
EvictedKeys: cache.stats.EvictedKeys,
ExpiredKeys: cache.stats.ExpiredKeys,
Hits: cache.stats.Hits,
Misses: cache.stats.Misses,
}
cache.mutex.RUnlock()
return stats
}
// MemoryUsage returns the current memory usage of the cache's dataset in bytes
// If MaxMemoryUsage is set to NoMaxMemoryUsage, this will return 0
func (cache *Cache) MemoryUsage() int {
return cache.memoryUsage
}
// WithMaxSize sets the maximum amount of entries that can be in the cache at any given time
// A maxSize of 0 or less means infinite
func (cache *Cache) WithMaxSize(maxSize int) *Cache {
if maxSize < 0 {
maxSize = NoMaxSize
}
if maxSize != NoMaxSize && cache.Count() == 0 {
cache.entries = make(map[string]*Entry, maxSize)
}
cache.maxSize = maxSize
return cache
}
// WithMaxMemoryUsage sets the maximum amount of memory that can be used by the cache at any given time
//
// NOTE: This is approximate.
//
// Setting this to NoMaxMemoryUsage will disable eviction by memory usage
func (cache *Cache) WithMaxMemoryUsage(maxMemoryUsageInBytes int) *Cache {
if maxMemoryUsageInBytes < 0 {
maxMemoryUsageInBytes = NoMaxMemoryUsage
}
cache.maxMemoryUsage = maxMemoryUsageInBytes
return cache
}
// WithEvictionPolicy sets eviction algorithm.
// Defaults to FirstInFirstOut (FIFO)
func (cache *Cache) WithEvictionPolicy(policy EvictionPolicy) *Cache {
cache.evictionPolicy = policy
return cache
}
// WithForceNilInterfaceOnNilPointer sets whether all Set-like functions should set a value as nil if the
// interface passed has a nil value but not a nil type.
//
// In Go, an interface is only nil if both its type and value are nil, which means that a nil pointer
// (e.g. (*Struct)(nil)) will retain its attribution to the type, and the unmodified value returned from
// Cache.Get, for instance, would return false when compared with nil if this option is set to false.
//
// We can bypass this by detecting if the interface's value is nil and setting it to nil rather than
// a nil pointer, which will make the value returned from Cache.Get return true when compared with nil.
// This is exactly what passing true to WithForceNilInterfaceOnNilPointer does, and it's also the default behavior.
//
// Alternatively, you may pass false to WithForceNilInterfaceOnNilPointer, which will mean that you'll have
// to cast the value returned from Cache.Get to its original type to check for whether the pointer returned
// is nil or not.
//
// If set to true (default):
// cache := gocache.NewCache().WithForceNilInterfaceOnNilPointer(true)
// cache.Set("key", (*Struct)(nil))
// value, _ := cache.Get("key")
// // the following returns true, because the interface{} was forcefully set to nil
// if value == nil {}
// // the following will panic, because the value has been casted to its type (which is nil)
// if value.(*Struct) == nil {}
//
// If set to false:
// cache := gocache.NewCache().WithForceNilInterfaceOnNilPointer(false)
// cache.Set("key", (*Struct)(nil))
// value, _ := cache.Get("key")
// // the following returns false, because the interface{} returned has a non-nil type (*Struct)
// if value == nil {}
// // the following returns true, because the value has been casted to its type
// if value.(*Struct) == nil {}
//
// In other words, if set to true, you do not need to cast the value returned from the cache to
// to check if the value is nil.
//
// Defaults to true
func (cache *Cache) WithForceNilInterfaceOnNilPointer(forceNilInterfaceOnNilPointer bool) *Cache {
cache.forceNilInterfaceOnNilPointer = forceNilInterfaceOnNilPointer
return cache
}
// NewCache creates a new Cache
//
// Should be used in conjunction with Cache.WithMaxSize, Cache.WithMaxMemoryUsage and/or Cache.WithEvictionPolicy
// gocache.NewCache().WithMaxSize(10000).WithEvictionPolicy(gocache.LeastRecentlyUsed)
//
func NewCache() *Cache {
return &Cache{
maxSize: DefaultMaxSize,
evictionPolicy: FirstInFirstOut,
stats: &Statistics{},
entries: make(map[string]*Entry),
mutex: sync.RWMutex{},
stopJanitor: nil,
forceNilInterfaceOnNilPointer: true,
}
}
// Set creates or updates a key with a given value
func (cache *Cache) Set(key string, value interface{}) {
cache.SetWithTTL(key, value, NoExpiration)
}
// SetWithTTL creates or updates a key with a given value and sets an expiration time (-1 is NoExpiration)
//
// The TTL provided must be greater than 0, or NoExpiration (-1). If a negative value that isn't -1 (NoExpiration) is
// provided, the entry will not be created if the key doesn't exist
func (cache *Cache) SetWithTTL(key string, value interface{}, ttl time.Duration) {
// An interface is only nil if both its value and its type are nil, however, passing a nil pointer as an interface{}
// means that the interface itself is not nil, because the interface value is nil but not the type.
if cache.forceNilInterfaceOnNilPointer {
if value != nil && (reflect.ValueOf(value).Kind() == reflect.Ptr && reflect.ValueOf(value).IsNil()) {
value = nil
}
}
cache.mutex.Lock()
entry, ok := cache.get(key)
if !ok {
// A negative TTL that isn't -1 (NoExpiration) or 0 is an entry that will expire instantly,
// so might as well just not create it in the first place
if ttl != NoExpiration && ttl < 1 {
cache.mutex.Unlock()
return
}
// Cache entry doesn't exist, so we have to create a new one
entry = &Entry{
Key: key,
Value: value,
RelevantTimestamp: time.Now(),
next: cache.head,
}
if cache.head == nil {
cache.tail = entry
} else {
cache.head.previous = entry
}
cache.head = entry
cache.entries[key] = entry
if cache.maxMemoryUsage != NoMaxMemoryUsage {
cache.memoryUsage += entry.SizeInBytes()
}
} else {
// A negative TTL that isn't -1 (NoExpiration) or 0 is an entry that will expire instantly,
// so might as well just delete it immediately instead of updating it
if ttl != NoExpiration && ttl < 1 {
cache.delete(key)
cache.mutex.Unlock()
return
}
if cache.maxMemoryUsage != NoMaxMemoryUsage {
// Subtract the old entry from the cache's memoryUsage
cache.memoryUsage -= entry.SizeInBytes()
}
// Update existing entry's value
entry.Value = value
entry.RelevantTimestamp = time.Now()
if cache.maxMemoryUsage != NoMaxMemoryUsage {
// Add the memory usage of the new entry to the cache's memoryUsage
cache.memoryUsage += entry.SizeInBytes()
}
// Because we just updated the entry, we need to move it back to HEAD
cache.moveExistingEntryToHead(entry)
}
if ttl != NoExpiration {
entry.Expiration = time.Now().Add(ttl).UnixNano()
} else {
entry.Expiration = NoExpiration
}
// If the cache doesn't have a maxSize/maxMemoryUsage, then there's no point
// checking if we need to evict an entry, so we'll just return now
if cache.maxSize == NoMaxSize && cache.maxMemoryUsage == NoMaxMemoryUsage {
cache.mutex.Unlock()
return
}
// If there's a maxSize and the cache has more entries than the maxSize, evict
if cache.maxSize != NoMaxSize && len(cache.entries) > cache.maxSize {
cache.evict()
}
// If there's a maxMemoryUsage and the memoryUsage is above the maxMemoryUsage, evict
if cache.maxMemoryUsage != NoMaxMemoryUsage && cache.memoryUsage > cache.maxMemoryUsage {
for cache.memoryUsage > cache.maxMemoryUsage && len(cache.entries) > 0 {
cache.evict()
}
}
cache.mutex.Unlock()
}
// SetAll creates or updates multiple values
func (cache *Cache) SetAll(entries map[string]interface{}) {
for key, value := range entries {
cache.SetWithTTL(key, value, NoExpiration)
}
}
// Get retrieves an entry using the key passed as parameter
// If there is no such entry, the value returned will be nil and the boolean will be false
// If there is an entry, the value returned will be the value cached and the boolean will be true
func (cache *Cache) Get(key string) (interface{}, bool) {
cache.mutex.Lock()
entry, ok := cache.get(key)
if !ok {
cache.mutex.Unlock()
cache.stats.Misses++
return nil, false
}
if entry.Expired() {
cache.stats.ExpiredKeys++
cache.delete(key)
cache.mutex.Unlock()
return nil, false
}
cache.stats.Hits++
if cache.evictionPolicy == LeastRecentlyUsed {
entry.Accessed()
if cache.head == entry {
cache.mutex.Unlock()
return entry.Value, true
}
// Because the eviction policy is LRU, we need to move the entry back to HEAD
cache.moveExistingEntryToHead(entry)
}
cache.mutex.Unlock()
return entry.Value, true
}
// GetValue retrieves an entry using the key passed as parameter
// Unlike Get, this function only returns the value
func (cache *Cache) GetValue(key string) interface{} {
value, _ := cache.Get(key)
return value
}
// GetByKeys retrieves multiple entries using the keys passed as parameter
// All keys are returned in the map, regardless of whether they exist or not, however, entries that do not exist in the
// cache will return nil, meaning that there is no way of determining whether a key genuinely has the value nil, or
// whether it doesn't exist in the cache using only this function.
func (cache *Cache) GetByKeys(keys []string) map[string]interface{} {
entries := make(map[string]interface{})
for _, key := range keys {
entries[key], _ = cache.Get(key)
}
return entries
}
// GetAll retrieves all cache entries
//
// If the eviction policy is LeastRecentlyUsed, note that unlike Get and GetByKeys, this does not update the last access
// timestamp. The reason for this is that since all cache entries will be accessed, updating the last access timestamp
// would provide very little benefit while harming the ability to accurately determine the next key that will be evicted
//
// You should probably avoid using this if you have a lot of entries.
//
// GetKeysByPattern is a good alternative if you want to retrieve entries that you do not have the key for, as it only
// retrieves the keys and does not trigger active eviction and has a parameter for setting a limit to the number of keys
// you wish to retrieve.
func (cache *Cache) GetAll() map[string]interface{} {
entries := make(map[string]interface{})
cache.mutex.Lock()
for key, entry := range cache.entries {
if entry.Expired() {
cache.delete(key)
continue
}
entries[key] = entry.Value
}
cache.stats.Hits += uint64(len(entries))
cache.mutex.Unlock()
return entries
}
// GetKeysByPattern retrieves a slice of keys that match a given pattern
// If the limit is set to 0, the entire cache will be searched for matching keys.
// If the limit is above 0, the search will stop once the specified number of matching keys have been found.
//
// e.g.
// cache.GetKeysByPattern("*some*", 0) will return all keys containing "some" in them
// cache.GetKeysByPattern("*some*", 5) will return 5 keys (or less) containing "some" in them
//
// Note that GetKeysByPattern does not trigger active evictions, nor does it count as accessing the entry, the latter
// only applying if the cache uses the LeastRecentlyUsed eviction policy.
// The reason for that behavior is that these two (active eviction and access) only applies when you access the value
// of the cache entry, and this function only returns the keys.
func (cache *Cache) GetKeysByPattern(pattern string, limit int) []string {
var matchingKeys []string
cache.mutex.Lock()
for key, value := range cache.entries {
if value.Expired() {
continue
}
if MatchPattern(pattern, key) {
matchingKeys = append(matchingKeys, key)
if limit > 0 && len(matchingKeys) >= limit {
break
}
}
}
cache.mutex.Unlock()
return matchingKeys
}
// Delete removes a key from the cache
//
// Returns false if the key did not exist.
func (cache *Cache) Delete(key string) bool {
cache.mutex.Lock()
ok := cache.delete(key)
cache.mutex.Unlock()
return ok
}
// DeleteAll deletes multiple entries based on the keys passed as parameter
//
// Returns the number of keys deleted
func (cache *Cache) DeleteAll(keys []string) int {
numberOfKeysDeleted := 0
cache.mutex.Lock()
for _, key := range keys {
if cache.delete(key) {
numberOfKeysDeleted++
}
}
cache.mutex.Unlock()
return numberOfKeysDeleted
}
// Count returns the total amount of entries in the cache, regardless of whether they're expired or not
func (cache *Cache) Count() int {
cache.mutex.RLock()
count := len(cache.entries)
cache.mutex.RUnlock()
return count
}
// Clear deletes all entries from the cache
func (cache *Cache) Clear() {
cache.mutex.Lock()
cache.entries = make(map[string]*Entry)
cache.memoryUsage = 0
cache.head = nil
cache.tail = nil
cache.mutex.Unlock()
}
// TTL returns the time until the cache entry specified by the key passed as parameter
// will be deleted.
func (cache *Cache) TTL(key string) (time.Duration, error) {
cache.mutex.RLock()
entry, ok := cache.get(key)
cache.mutex.RUnlock()
if !ok {
return 0, ErrKeyDoesNotExist
}
if entry.Expiration == NoExpiration {
return 0, ErrKeyHasNoExpiration
}
timeUntilExpiration := time.Until(time.Unix(0, entry.Expiration))
if timeUntilExpiration < 0 {
// The key has already expired but hasn't been deleted yet.
// From the client's perspective, this means that the cache entry doesn't exist
return 0, ErrKeyDoesNotExist
}
return timeUntilExpiration, nil
}
// Expire sets a key's expiration time
//
// A TTL of -1 means that the key will never expire
// A TTL of 0 means that the key will expire immediately
// If using LRU, note that this does not reset the position of the key
//
// Returns true if the cache key exists and has had its expiration time altered
func (cache *Cache) Expire(key string, ttl time.Duration) bool {
entry, ok := cache.get(key)
if !ok || entry.Expired() {
return false
}
if ttl != NoExpiration {
entry.Expiration = time.Now().Add(ttl).UnixNano()
} else {
entry.Expiration = NoExpiration
}
return true
}
// get retrieves an entry using the key passed as parameter, but unlike Get, it doesn't update the access time or
// move the position of the entry to the head
func (cache *Cache) get(key string) (*Entry, bool) {
entry, ok := cache.entries[key]
return entry, ok
}
func (cache *Cache) delete(key string) bool {
entry, ok := cache.entries[key]
if ok {
if cache.maxMemoryUsage != NoMaxMemoryUsage {
cache.memoryUsage -= entry.SizeInBytes()
}
cache.removeExistingEntryReferences(entry)
delete(cache.entries, key)
}
return ok
}
// moveExistingEntryToHead replaces the current cache head for an existing entry
func (cache *Cache) moveExistingEntryToHead(entry *Entry) {
if !(entry == cache.head && entry == cache.tail) {
cache.removeExistingEntryReferences(entry)
}
if entry != cache.head {
entry.next = cache.head
entry.previous = nil
if cache.head != nil {
cache.head.previous = entry
}
cache.head = entry
}
}
// removeExistingEntryReferences modifies the next and previous reference of an existing entry and re-links
// the next and previous entry accordingly, as well as the cache head or/and the cache tail if necessary.
// Note that it does not remove the entry from the cache, only the references.
func (cache *Cache) removeExistingEntryReferences(entry *Entry) {
if cache.tail == entry && cache.head == entry {
cache.tail = nil
cache.head = nil
} else if cache.tail == entry {
cache.tail = cache.tail.previous
} else if cache.head == entry {
cache.head = cache.head.next
}
if entry.previous != nil {
entry.previous.next = entry.next
}
if entry.next != nil {
entry.next.previous = entry.previous
}
entry.next = nil
entry.previous = nil
}
// evict removes the tail from the cache
func (cache *Cache) evict() {
if cache.tail == nil || len(cache.entries) == 0 {
return
}
if cache.tail != nil {
oldTail := cache.tail
cache.removeExistingEntryReferences(oldTail)
delete(cache.entries, oldTail.Key)
if cache.maxMemoryUsage != NoMaxMemoryUsage {
cache.memoryUsage -= oldTail.SizeInBytes()
}
cache.stats.EvictedKeys++
}
}

146
vendor/github.com/TwiN/gocache/v2/janitor.go generated vendored Normal file
View File

@ -0,0 +1,146 @@
package gocache
import (
"log"
"time"
)
const (
// JanitorShiftTarget is the target number of expired keys to find during passive clean up duty
// before pausing the passive expired keys eviction process
JanitorShiftTarget = 25
// JanitorMaxIterationsPerShift is the maximum number of nodes to traverse before pausing
//
// This is to prevent the janitor from traversing the entire cache, which could take a long time
// to complete depending on the size of the cache.
//
// By limiting it to a small number, we are effectively reducing the impact of passive eviction.
JanitorMaxIterationsPerShift = 1000
// JanitorMinShiftBackOff is the minimum interval between each iteration of steps
// defined by JanitorMaxIterationsPerShift
JanitorMinShiftBackOff = 50 * time.Millisecond
// JanitorMaxShiftBackOff is the maximum interval between each iteration of steps
// defined by JanitorMaxIterationsPerShift
JanitorMaxShiftBackOff = 500 * time.Millisecond
)
// StartJanitor starts the janitor on a different goroutine
// The janitor's job is to delete expired keys in the background, in other words, it takes care of passive eviction.
// It can be stopped by calling Cache.StopJanitor.
// If you do not start the janitor, expired keys will only be deleted when they are accessed through Get, GetByKeys, or
// GetAll.
func (cache *Cache) StartJanitor() error {
if cache.stopJanitor != nil {
return ErrJanitorAlreadyRunning
}
cache.stopJanitor = make(chan bool)
go func() {
// rather than starting from the tail on every run, we can try to start from the last traversed entry
var lastTraversedNode *Entry
totalNumberOfExpiredKeysInPreviousRunFromTailToHead := 0
backOff := JanitorMinShiftBackOff
for {
select {
case <-time.After(backOff):
// Passive clean up duty
cache.mutex.Lock()
if cache.tail != nil {
start := time.Now()
steps := 0
expiredEntriesFound := 0
current := cache.tail
if lastTraversedNode != nil {
// Make sure the lastTraversedNode is still in the cache, otherwise we might be traversing nodes that were already deleted.
// Furthermore, we need to make sure that the entry from the cache has the same pointer as the lastTraversedNode
// to verify that there isn't just a new cache entry with the same key (i.e. in case lastTraversedNode got evicted)
if entryFromCache, isInCache := cache.get(lastTraversedNode.Key); isInCache && entryFromCache == lastTraversedNode {
current = lastTraversedNode
}
}
if current == cache.tail {
if Debug {
log.Printf("There are currently %d entries in the cache. The last walk resulted in finding %d expired keys", len(cache.entries), totalNumberOfExpiredKeysInPreviousRunFromTailToHead)
}
totalNumberOfExpiredKeysInPreviousRunFromTailToHead = 0
}
for current != nil {
// since we're walking from the tail to the head, we get the previous reference
var previous *Entry
steps++
if current.Expired() {
expiredEntriesFound++
// Because delete will remove the previous reference from the entry, we need to store the
// previous reference before we delete it
previous = current.previous
cache.delete(current.Key)
cache.stats.ExpiredKeys++
}
if current == cache.head {
lastTraversedNode = nil
break
}
// Travel to the current node's previous node only if no specific previous node has been specified
if previous != nil {
current = previous
} else {
current = current.previous
}
lastTraversedNode = current
if steps == JanitorMaxIterationsPerShift || expiredEntriesFound >= JanitorShiftTarget {
if expiredEntriesFound > 0 {
backOff = JanitorMinShiftBackOff
} else {
if backOff*2 <= JanitorMaxShiftBackOff {
backOff *= 2
} else {
backOff = JanitorMaxShiftBackOff
}
}
break
}
}
if Debug {
log.Printf("traversed %d nodes and found %d expired entries in %s before stopping\n", steps, expiredEntriesFound, time.Since(start))
}
totalNumberOfExpiredKeysInPreviousRunFromTailToHead += expiredEntriesFound
} else {
if backOff*2 < JanitorMaxShiftBackOff {
backOff *= 2
} else {
backOff = JanitorMaxShiftBackOff
}
}
cache.mutex.Unlock()
case <-cache.stopJanitor:
cache.stopJanitor <- true
return
}
}
}()
//if Debug {
// go func() {
// var m runtime.MemStats
// for {
// runtime.ReadMemStats(&m)
// log.Printf("Alloc=%vMB; HeapReleased=%vMB; Sys=%vMB; HeapInUse=%vMB; HeapObjects=%v; HeapObjectsFreed=%v; GC=%v; cache.memoryUsage=%vMB; cacheSize=%d\n", m.Alloc/1024/1024, m.HeapReleased/1024/1024, m.Sys/1024/1024, m.HeapInuse/1024/1024, m.HeapObjects, m.Frees, m.NumGC, cache.memoryUsage/1024/1024, cache.Count())
// time.Sleep(3 * time.Second)
// }
// }()
//}
return nil
}
// StopJanitor stops the janitor
func (cache *Cache) StopJanitor() {
if cache.stopJanitor != nil {
// Tell the janitor to stop, and then wait for the janitor to reply on the same channel that it's stopping
// This may seem a bit odd, but this allows us to avoid a data race condition when trying to set
// cache.stopJanitor to nil
cache.stopJanitor <- true
<-cache.stopJanitor
cache.stopJanitor = nil
}
}

12
vendor/github.com/TwiN/gocache/v2/pattern.go generated vendored Normal file
View File

@ -0,0 +1,12 @@
package gocache
import "path/filepath"
// MatchPattern checks whether a string matches a pattern
func MatchPattern(pattern, s string) bool {
if pattern == "*" {
return true
}
matched, _ := filepath.Match(pattern, s)
return matched
}

33
vendor/github.com/TwiN/gocache/v2/policy.go generated vendored Normal file
View File

@ -0,0 +1,33 @@
package gocache
// EvictionPolicy is what dictates how evictions are handled
type EvictionPolicy string
var (
// LeastRecentlyUsed is an eviction policy that causes the most recently accessed cache entry to be moved to the
// head of the cache. Effectively, this causes the cache entries that have not been accessed for some time to
// gradually move closer and closer to the tail, and since the tail is the entry that gets deleted when an eviction
// is required, it allows less used cache entries to be evicted while keeping recently accessed entries at or close
// to the head.
//
// For instance, creating a Cache with a Cache.MaxSize of 3 and creating the entries 1, 2 and 3 in that order would
// put 3 at the head and 1 at the tail:
// 3 (head) -> 2 -> 1 (tail)
// If the cache entry 1 was then accessed, 1 would become the head and 2 the tail:
// 1 (head) -> 3 -> 2 (tail)
// If a cache entry 4 was then created, because the Cache.MaxSize is 3, the tail (2) would then be evicted:
// 4 (head) -> 1 -> 3 (tail)
LeastRecentlyUsed EvictionPolicy = "LeastRecentlyUsed"
// FirstInFirstOut is an eviction policy that causes cache entries to be evicted in the same order that they are
// created.
//
// For instance, creating a Cache with a Cache.MaxSize of 3 and creating the entries 1, 2 and 3 in that order would
// put 3 at the head and 1 at the tail:
// 3 (head) -> 2 -> 1 (tail)
// If the cache entry 1 was then accessed, unlike with LeastRecentlyUsed, nothing would change:
// 3 (head) -> 2 -> 1 (tail)
// If a cache entry 4 was then created, because the Cache.MaxSize is 3, the tail (1) would then be evicted:
// 4 (head) -> 3 -> 2 (tail)
FirstInFirstOut EvictionPolicy = "FirstInFirstOut"
)

15
vendor/github.com/TwiN/gocache/v2/statistics.go generated vendored Normal file
View File

@ -0,0 +1,15 @@
package gocache
type Statistics struct {
// EvictedKeys is the number of keys that were evicted
EvictedKeys uint64
// ExpiredKeys is the number of keys that were automatically deleted as a result of expiring
ExpiredKeys uint64
// Hits is the number of cache hits
Hits uint64
// Misses is the number of cache misses
Misses uint64
}