more caching in-client

cmd/main.go

@@ -76,13 +76,10 @@ func Serve(ctx *cli.Context) error {
 	canonicalDomainCache := cache.NewKeyValueCache()
 	// dnsLookupCache stores DNS lookups for custom domains
 	dnsLookupCache := cache.NewKeyValueCache()
-	// branchTimestampCache stores branch timestamps for faster cache checking
+	// clientResponseCache stores responses from the Gitea server
-	branchTimestampCache := cache.NewKeyValueCache()
+	clientResponseCache := cache.NewKeyValueCache()
-	// fileResponseCache stores responses from the Gitea server
-	// TODO: make this an MRU cache with a size limit
-	fileResponseCache := cache.NewKeyValueCache()
 
-	giteaClient, err := gitea.NewClient(giteaRoot, giteaAPIToken, fileResponseCache, ctx.Bool("enable-symlink-support"), ctx.Bool("enable-lfs-support"))
+	giteaClient, err := gitea.NewClient(giteaRoot, giteaAPIToken, clientResponseCache, ctx.Bool("enable-symlink-support"), ctx.Bool("enable-lfs-support"))
 	if err != nil {
 		return fmt.Errorf("could not create new gitea client: %v", err)
 	}
@@ -92,7 +89,7 @@ func Serve(ctx *cli.Context) error {
 		giteaClient,
 		giteaRoot, rawInfoPage,
 		BlacklistedPaths, allowedCorsDomains,
-		dnsLookupCache, canonicalDomainCache, branchTimestampCache, fileResponseCache)
+		dnsLookupCache, canonicalDomainCache)
 
 	fastServer := server.SetupServer(handler)
 	httpServer := server.SetupHTTPACMEChallengeServer(challengeCache)

server/gitea/cache.go

@@ -1,5 +1,7 @@
 package gitea
 
+import "time"
+
 type FileResponse struct {
 	Exists   bool
 	ETag     []byte
@@ -10,3 +12,18 @@ type FileResponse struct {
 func (f FileResponse) IsEmpty() bool {
 	return len(f.Body) != 0
 }
+
+type BranchTimestamp struct {
+	Branch    string
+	Timestamp time.Time
+}
+
+var (
+	// defaultBranchCacheTimeout specifies the timeout for the default branch cache. It can be quite long.
+	defaultBranchCacheTimeout = 15 * time.Minute
+
+	// branchExistenceCacheTimeout specifies the timeout for the branch timestamp & existence cache. It should be shorter
+	// than fileCacheTimeout, as that gets invalidated if the branch timestamp has changed. That way, repo changes will be
+	// picked up faster, while still allowing the content to be cached longer if nothing changes.
+	branchExistenceCacheTimeout = 5 * time.Minute
+)

server/gitea/client.go

@@ -5,3 +5,8 @@ import (
 )
 
 var ErrorNotFound = errors.New("not found")
+
+const (
+	branchTimestampCacheKeyPrefix = "branchTime"
+	defaultBranchCacheKeyPrefix   = "defaultBranch"
+)

server/gitea/client_fasthttp.go

@@ -26,13 +26,13 @@ type Client struct {
 	infoTimeout    time.Duration
 	contentTimeout time.Duration
 	fastClient     *fasthttp.Client
-	fileResponseCache cache.SetGetKey
+	responseCache  cache.SetGetKey
 
 	followSymlinks bool
 	supportLFS     bool
 }
 
-func NewClient(giteaRoot, giteaAPIToken string, fileResponseCache cache.SetGetKey, followSymlinks, supportLFS bool) (*Client, error) {
+func NewClient(giteaRoot, giteaAPIToken string, respCache cache.SetGetKey, followSymlinks, supportLFS bool) (*Client, error) {
 	rootURL, err := url.Parse(giteaRoot)
 	giteaRoot = strings.Trim(rootURL.String(), "/")
 
@@ -42,7 +42,7 @@ func NewClient(giteaRoot, giteaAPIToken string, fileResponseCache cache.SetGetKe
 		infoTimeout:    5 * time.Second,
 		contentTimeout: 10 * time.Second,
 		fastClient:     getFastHTTPClient(),
-		fileResponseCache: fileResponseCache,
+		responseCache:  respCache,
 
 		followSymlinks: followSymlinks,
 		supportLFS:     supportLFS,

server/gitea/client_std.go

@@ -17,13 +17,13 @@ import (
 
 type Client struct {
 	sdkClient     *gitea.Client
-	fileResponseCache cache.SetGetKey
+	responseCache cache.SetGetKey
 
 	followSymlinks bool
 	supportLFS     bool
 }
 
-func NewClient(giteaRoot, giteaAPIToken string, fileResponseCache cache.SetGetKey, followSymlinks, supportLFS bool) (*Client, error) {
+func NewClient(giteaRoot, giteaAPIToken string, respCache cache.SetGetKey, followSymlinks, supportLFS bool) (*Client, error) {
 	rootURL, err := url.Parse(giteaRoot)
 	giteaRoot = strings.Trim(rootURL.String(), "/")
 
@@ -32,7 +32,7 @@ func NewClient(giteaRoot, giteaAPIToken string, fileResponseCache cache.SetGetKe
 	sdk, err := gitea.NewClient(giteaRoot, gitea.SetHTTPClient(&stdClient), gitea.SetToken(giteaAPIToken))
 	return &Client{
 		sdkClient:     sdk,
-		fileResponseCache: fileResponseCache,
+		responseCache: respCache,
 	}, err
 }
 
@@ -68,18 +68,40 @@ func (client *Client) ServeRawContent(targetOwner, targetRepo, ref, resource str
 	}
 }
 
-func (client *Client) GiteaGetRepoBranchTimestamp(repoOwner, repoName, branchName string) (time.Time, error) {
+func (client *Client) GiteaGetRepoBranchTimestamp(repoOwner, repoName, branchName string) (*BranchTimestamp, error) {
+	cacheKey := fmt.Sprintf("%s/%s/%s/%s", branchTimestampCacheKeyPrefix, repoOwner, repoName, branchName)
+
+	if stamp, ok := client.responseCache.Get(cacheKey); ok && stamp != nil {
+		return stamp.(*BranchTimestamp), nil
+	}
+
 	branch, resp, err := client.sdkClient.GetRepoBranch(repoOwner, repoName, branchName)
 	if err != nil {
-		return time.Time{}, err
+		if resp != nil && resp.StatusCode == http.StatusNotFound {
+			return &BranchTimestamp{}, ErrorNotFound
+		}
+		return &BranchTimestamp{}, err
 	}
 	if resp.StatusCode != http.StatusOK {
-		return time.Time{}, fmt.Errorf("unexpected status code '%d'", resp.StatusCode)
+		return &BranchTimestamp{}, fmt.Errorf("unexpected status code '%d'", resp.StatusCode)
 	}
-	return branch.Commit.Timestamp, nil
+
+	stamp := &BranchTimestamp{
+		Branch:    branch.Name,
+		Timestamp: branch.Commit.Timestamp,
+	}
+
+	client.responseCache.Set(cacheKey, stamp, branchExistenceCacheTimeout)
+	return stamp, nil
 }
 
 func (client *Client) GiteaGetRepoDefaultBranch(repoOwner, repoName string) (string, error) {
+	cacheKey := fmt.Sprintf("%s/%s/%s", defaultBranchCacheKeyPrefix, repoOwner, repoName)
+
+	if branch, ok := client.responseCache.Get(cacheKey); ok && branch != nil {
+		return branch.(string), nil
+	}
+
 	repo, resp, err := client.sdkClient.GetRepo(repoOwner, repoName)
 	if err != nil {
 		return "", err
@@ -87,5 +109,8 @@ func (client *Client) GiteaGetRepoDefaultBranch(repoOwner, repoName string) (str
 	if resp.StatusCode != http.StatusOK {
 		return "", fmt.Errorf("unexpected status code '%d'", resp.StatusCode)
 	}
-	return repo.DefaultBranch, nil
+
+	branch := repo.DefaultBranch
+	client.responseCache.Set(cacheKey, branch, defaultBranchCacheTimeout)
+	return branch, nil
 }

server/handler.go

@@ -24,7 +24,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 	giteaClient *gitea.Client,
 	giteaRoot, rawInfoPage string,
 	blacklistedPaths, allowedCorsDomains [][]byte,
-	dnsLookupCache, canonicalDomainCache, branchTimestampCache, fileResponseCache cache.SetGetKey,
+	dnsLookupCache, canonicalDomainCache cache.SetGetKey,
 ) func(ctx *fasthttp.RequestCtx) {
 	return func(ctx *fasthttp.RequestCtx) {
 		log := log.With().Str("Handler", string(ctx.Request.Header.RequestURI())).Logger()
@@ -96,7 +96,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 			branch = strings.ReplaceAll(branch, "~", "/")
 
 			// Check if the branch exists, otherwise treat it as a file path
-			branchTimestampResult := upstream.GetBranchTimestamp(giteaClient, targetOwner, repo, branch, branchTimestampCache)
+			branchTimestampResult := upstream.GetBranchTimestamp(giteaClient, targetOwner, repo, branch)
 			if branchTimestampResult == nil {
 				log.Debug().Msg("tryBranch: branch doesn't exist")
 				return false
@@ -153,7 +153,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 					log.Debug().Msg("tryBranch, now trying upstream 1")
 					tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 						targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-						canonicalDomainCache, branchTimestampCache, fileResponseCache)
+						canonicalDomainCache)
 					return
 				}
 				log.Debug().Msg("missing branch")
@@ -169,7 +169,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 			log.Debug().Msg("tryBranch, now trying upstream 2")
 			tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 				targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-				canonicalDomainCache, branchTimestampCache, fileResponseCache)
+				canonicalDomainCache)
 			return
 
 		} else if bytes.HasSuffix(trimmedHost, mainDomainSuffix) {
@@ -204,7 +204,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 					log.Debug().Msg("tryBranch, now trying upstream 3")
 					tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 						targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-						canonicalDomainCache, branchTimestampCache, fileResponseCache)
+						canonicalDomainCache)
 				} else {
 					html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
 				}
@@ -220,7 +220,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 					log.Debug().Msg("tryBranch, now trying upstream 4")
 					tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 						targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-						canonicalDomainCache, branchTimestampCache, fileResponseCache)
+						canonicalDomainCache)
 				} else {
 					html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
 				}
@@ -236,7 +236,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 				log.Debug().Msg("tryBranch, now trying upstream 5")
 				tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 					targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-					canonicalDomainCache, branchTimestampCache, fileResponseCache)
+					canonicalDomainCache)
 				return
 			}
 
@@ -248,7 +248,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 				log.Debug().Msg("tryBranch, now trying upstream 6")
 				tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 					targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-					canonicalDomainCache, branchTimestampCache, fileResponseCache)
+					canonicalDomainCache)
 				return
 			}
 
@@ -296,7 +296,7 @@ func Handler(mainDomainSuffix, rawDomain []byte,
 				log.Debug().Msg("tryBranch, now trying upstream 7")
 				tryUpstream(ctx, giteaClient, mainDomainSuffix, trimmedHost,
 					targetOptions, targetOwner, targetRepo, targetBranch, targetPath,
-					canonicalDomainCache, branchTimestampCache, fileResponseCache)
+					canonicalDomainCache)
 				return
 			}
 

server/handler_test.go

@@ -24,8 +24,6 @@ func TestHandlerPerformance(t *testing.T) {
 		[][]byte{[]byte("raw.codeberg.org"), []byte("fonts.codeberg.org"), []byte("design.codeberg.org")},
 		cache.NewKeyValueCache(),
 		cache.NewKeyValueCache(),
-		cache.NewKeyValueCache(),
-		cache.NewKeyValueCache(),
 	)
 
 	testCase := func(uri string, status int) {

server/try.go

@@ -21,7 +21,7 @@ func tryUpstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client,
 	targetOptions *upstream.Options,
 	targetOwner, targetRepo, targetBranch, targetPath string,
 
-	canonicalDomainCache, branchTimestampCache, fileResponseCache cache.SetGetKey,
+	canonicalDomainCache cache.SetGetKey,
 ) {
 	// check if a canonical domain exists on a request on MainDomain
 	if bytes.HasSuffix(trimmedHost, mainDomainSuffix) {
@@ -45,7 +45,7 @@ func tryUpstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client,
 	targetOptions.TargetPath = targetPath
 
 	// Try to request the file from the Gitea API
-	if !targetOptions.Upstream(ctx, giteaClient, branchTimestampCache, fileResponseCache) {
+	if !targetOptions.Upstream(ctx, giteaClient, branchTimestampCache) {
 		html.ReturnErrorPage(ctx, ctx.Response.StatusCode())
 	}
 }

server/upstream/const.go

@@ -2,14 +2,6 @@ package upstream
 
 import "time"
 
-// defaultBranchCacheTimeout specifies the timeout for the default branch cache. It can be quite long.
-var defaultBranchCacheTimeout = 15 * time.Minute
-
-// branchExistenceCacheTimeout specifies the timeout for the branch timestamp & existence cache. It should be shorter
-// than fileCacheTimeout, as that gets invalidated if the branch timestamp has changed. That way, repo changes will be
-// picked up faster, while still allowing the content to be cached longer if nothing changes.
-var branchExistenceCacheTimeout = 5 * time.Minute
-
 // fileCacheTimeout specifies the timeout for the file content cache - you might want to make this quite long, depending
 // on your available memory.
 // TODO: move as option into cache interface

server/upstream/helper.go

@@ -5,46 +5,27 @@ import (
 	"path"
 	"strconv"
 	"strings"
-	"time"
 
-	"codeberg.org/codeberg/pages/server/cache"
 	"codeberg.org/codeberg/pages/server/gitea"
 )
 
-type branchTimestamp struct {
-	Branch    string
-	Timestamp time.Time
-}
-
 // GetBranchTimestamp finds the default branch (if branch is "") and returns the last modification time of the branch
 // (or nil if the branch doesn't exist)
-func GetBranchTimestamp(giteaClient *gitea.Client, owner, repo, branch string, branchTimestampCache cache.SetGetKey) *branchTimestamp {
+func GetBranchTimestamp(giteaClient *gitea.Client, owner, repo, branch string) *gitea.BranchTimestamp {
-	if result, ok := branchTimestampCache.Get(owner + "/" + repo + "/" + branch); ok {
-		if result == nil {
-			return nil
-		}
-		return result.(*branchTimestamp)
-	}
-	result := &branchTimestamp{
-		Branch: branch,
-	}
 	if len(branch) == 0 {
 		// Get default branch
 		defaultBranch, err := giteaClient.GiteaGetRepoDefaultBranch(owner, repo)
 		if err != nil {
-			_ = branchTimestampCache.Set(owner+"/"+repo+"/", nil, defaultBranchCacheTimeout)
 			return nil
 		}
-		result.Branch = defaultBranch
+		branch = defaultBranch
 	}
 
-	timestamp, err := giteaClient.GiteaGetRepoBranchTimestamp(owner, repo, result.Branch)
+	timestamp, err := giteaClient.GiteaGetRepoBranchTimestamp(owner, repo, branch)
 	if err != nil {
 		return nil
 	}
-	result.Timestamp = timestamp
+	return timestamp
-	_ = branchTimestampCache.Set(owner+"/"+repo+"/"+branch, result, branchExistenceCacheTimeout)
-	return result
 }
 
 func (o *Options) getMimeTypeByExtension() string {

server/upstream/upstream.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"net/http"
 	"strings"
 	"time"
 
@@ -12,7 +13,6 @@ import (
 	"github.com/valyala/fasthttp"
 
 	"codeberg.org/codeberg/pages/html"
-	"codeberg.org/codeberg/pages/server/cache"
 	"codeberg.org/codeberg/pages/server/gitea"
 )
 
@@ -43,12 +43,12 @@ type Options struct {
 }
 
 // Upstream requests a file from the Gitea API at GiteaRoot and writes it to the request context.
-func (o *Options) Upstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client, branchTimestampCache, fileResponseCache cache.SetGetKey) (final bool) {
+func (o *Options) Upstream(ctx *http.Response, giteaClient *gitea.Client) (final bool) {
 	log := log.With().Strs("upstream", []string{o.TargetOwner, o.TargetRepo, o.TargetBranch, o.TargetPath}).Logger()
 
 	// Check if the branch exists and when it was modified
 	if o.BranchTimestamp.IsZero() {
-		branch := GetBranchTimestamp(giteaClient, o.TargetOwner, o.TargetRepo, o.TargetBranch, branchTimestampCache)
+		branch := GetBranchTimestamp(giteaClient, o.TargetOwner, o.TargetRepo, o.TargetBranch)
 
 		if branch == nil {
 			html.ReturnErrorPage(ctx, fasthttp.StatusFailedDependency)
@@ -93,7 +93,7 @@ func (o *Options) Upstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client,
 			optionsForIndexPages.appendTrailingSlash = true
 			for _, indexPage := range upstreamIndexPages {
 				optionsForIndexPages.TargetPath = strings.TrimSuffix(o.TargetPath, "/") + "/" + indexPage
-				if optionsForIndexPages.Upstream(ctx, giteaClient, branchTimestampCache, fileResponseCache) {
+				if optionsForIndexPages.Upstream(ctx, giteaClient, branchTimestampCache) {
 					_ = fileResponseCache.Set(uri+"?timestamp="+o.timestamp(), gitea.FileResponse{
 						Exists: false,
 					}, fileCacheTimeout)
@@ -104,7 +104,7 @@ func (o *Options) Upstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client,
 			optionsForIndexPages.appendTrailingSlash = false
 			optionsForIndexPages.redirectIfExists = strings.TrimSuffix(string(ctx.Request.URI().Path()), "/") + ".html"
 			optionsForIndexPages.TargetPath = o.TargetPath + ".html"
-			if optionsForIndexPages.Upstream(ctx, giteaClient, branchTimestampCache, fileResponseCache) {
+			if optionsForIndexPages.Upstream(ctx, giteaClient, branchTimestampCache) {
 				_ = fileResponseCache.Set(uri+"?timestamp="+o.timestamp(), gitea.FileResponse{
 					Exists: false,
 				}, fileCacheTimeout)
@@ -119,7 +119,7 @@ func (o *Options) Upstream(ctx *fasthttp.RequestCtx, giteaClient *gitea.Client,
 			optionsForNotFoundPages.appendTrailingSlash = false
 			for _, notFoundPage := range upstreamNotFoundPages {
 				optionsForNotFoundPages.TargetPath = "/" + notFoundPage
-				if optionsForNotFoundPages.Upstream(ctx, giteaClient, branchTimestampCache, fileResponseCache) {
+				if optionsForNotFoundPages.Upstream(ctx, giteaClient, branchTimestampCache) {
 					_ = fileResponseCache.Set(uri+"?timestamp="+o.timestamp(), gitea.FileResponse{
 						Exists: false,
 					}, fileCacheTimeout)