Merge PR #1542 into 2.0.0-beta.2

This commit is contained in:
Levi committed 2026-09-25 16:23:55 +08:00
commit f400137ead
17 files changed
+862 -28

No files matched your search

+106 -16
View File
@@ -2,8 +2,13 @@ package http
import (
"context"
"crypto/md5"
"crypto/sha1"
"crypto/sha256"
"encoding/hex"
"errors"
"fmt"
"hash"
"io"
"net/http"
"net/url"
@@ -280,6 +285,12 @@ type Fetcher struct {
primaryReadyOnce sync.Once
primaryReadyCh chan struct{}
// checksumFailed marks that the last completed download failed
// checksum verification, so the next retry must force a full
// re-download instead of skipping connections that already
// reported Completed.
checksumFailed atomic.Bool
// Start pending mechanism
startPending atomic.Bool
resolvedCh chan struct{} // Signal when resolve completes
@@ -735,12 +746,26 @@ func (f *Fetcher) doStart() error {
// If retrying after error, reset connection states for retry
if state == stateError {
forceFullReset := f.checksumFailed.Swap(false)
if forceFullReset {
// Checksum mismatch: the file content itself is wrong even
// though every connection reported Completed. Truncate the
// corrupt file so stale bytes cannot linger if the retry is
// interrupted before every connection re-writes its chunk.
if err := os.Truncate(f.meta.SingleFilepath(), 0); err != nil && !os.IsNotExist(err) {
return fmt.Errorf("failed to reset file after checksum mismatch: %w", err)
}
}
f.connMu.Lock()
for _, conn := range f.connections {
// Reset connections that can be retried
if !conn.Completed && conn.State != connCompleted {
// On a normal error, only reset connections that didn't
// finish. On a checksum mismatch, reset EVERY connection —
// "Completed" only means the download loop finished, not
// that the bytes were verified correct.
shouldReset := forceFullReset || (!conn.Completed && conn.State != connCompleted)
if shouldReset {
if !f.hasSequentialPrefixLocked(conn) {
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, forceFullReset)
}
conn.State = connNotStarted
conn.failed = false
@@ -970,8 +995,14 @@ func (f *Fetcher) expandConnections() {
}
f.fileMu.Unlock()
f.setState(stateDone)
f.doneCh <- nil
var err error
if chkErr := f.verifyChecksum(); chkErr != nil {
err = chkErr
f.setState(stateError)
} else {
f.setState(stateDone)
}
f.doneCh <- err
return
}
@@ -1107,7 +1138,7 @@ func (f *Fetcher) runConnection(conn *connection) {
// it only when no safe validator exists or If-Range returns a full 200.
f.connMu.Lock()
if !f.hasSequentialPrefixLocked(conn) {
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, false)
}
f.connMu.Unlock()
}
@@ -1214,7 +1245,7 @@ func (f *Fetcher) downloadChunkOnce(conn *connection, client *http.Client, buf [
resumeProbe := f.canProbeSequentialResumeLocked(conn)
intentionalRestart := sequentialSizeUnknown
if sequentialSizeUnknown || (hasSequentialPrefix && !resumeProbe) {
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, false)
f.resolveDataPos.Store(0)
f.rangeValidatorPinned = false
intentionalRestart = true
@@ -1659,7 +1690,7 @@ func (f *Fetcher) fallbackToSequentialDownload(conn *connection, ifRange string)
f.rangeReprobeEligible = true
f.rangeValidatorPinned = false
f.ifRange = ifRange
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, false)
f.resolveDataPos.Store(0)
return nil
}
@@ -1675,7 +1706,7 @@ func (f *Fetcher) restartSequentialDownload(conn *connection, ifRange string) {
f.rangeReprobeEligible = true
f.rangeValidatorPinned = false
f.ifRange = ifRange
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, false)
f.resolveDataPos.Store(0)
}
@@ -2058,16 +2089,25 @@ func (f *Fetcher) helpOtherConnection(helper *connection) bool {
return true
}
func (f *Fetcher) resetConnectionForRestart(conn *connection) {
if f.meta.Res.Range {
func (f *Fetcher) resetConnectionForRestart(conn *connection, force bool) {
if f.meta.Res.Range && !force {
return
}
if f.meta.Res.Range && force {
// Checksum mismatch: the on-disk content for this chunk is
// confirmed wrong, so its bytes must be re-fetched from the
// origin server even though range support would normally let
// us just resume from the existing Downloaded position.
if conn.Chunk != nil {
conn.Chunk.Downloaded = 0
}
}
// Without range support a new request always starts from byte 0,
// so pause/retry must restart instead of continuing from the old offset.
if conn.Chunk == nil {
conn.Chunk = newChunk(0, 0)
} else {
} else if !f.meta.Res.Range {
conn.Chunk.Begin = 0
conn.Chunk.setEnd(0)
conn.Chunk.Downloaded = 0
@@ -2131,7 +2171,7 @@ func (f *Fetcher) resumeConnections() {
}
}
if !f.hasSequentialPrefixLocked(conn) {
f.resetConnectionForRestart(conn)
f.resetConnectionForRestart(conn, false)
}
// Reset the connection state for resume
conn.ctx, conn.cancel = context.WithCancel(f.ctx)
@@ -2238,10 +2278,15 @@ func (f *Fetcher) onDownloadComplete() {
}
f.fileMu.Unlock()
if finalErr != nil {
f.setState(stateError)
if finalErr == nil {
if chkErr := f.verifyChecksum(); chkErr != nil {
finalErr = chkErr
f.setState(stateError)
} else {
f.setState(stateDone)
}
} else {
f.setState(stateDone)
f.setState(stateError)
}
select {
@@ -2250,6 +2295,51 @@ func (f *Fetcher) onDownloadComplete() {
}
}
func (f *Fetcher) verifyChecksum() error {
if f.meta == nil || f.meta.Opts == nil || f.meta.Opts.Extra == nil {
return nil
}
extra, ok := f.meta.Opts.Extra.(*fhttp.OptsExtra)
if !ok || extra.Checksum == nil {
return nil
}
chk := extra.Checksum
if chk.Algorithm == "" && chk.Expected == "" {
return nil
}
var h hash.Hash
switch strings.ToLower(strings.TrimSpace(chk.Algorithm)) {
case "md5":
h = md5.New()
case "sha1", "sha-1":
h = sha1.New()
case "sha256", "sha-256":
h = sha256.New()
default:
return fmt.Errorf("unsupported checksum algorithm: %s", chk.Algorithm)
}
filePath := f.meta.SingleFilepath()
file, err := os.Open(filePath)
if err != nil {
return fmt.Errorf("open file for checksum verification failed: %w", err)
}
defer file.Close()
buf := make([]byte, 64*1024)
if _, err := io.CopyBuffer(h, file, buf); err != nil {
return fmt.Errorf("calculate checksum failed: %w", err)
}
actual := hex.EncodeToString(h.Sum(nil))
expected := strings.ToLower(strings.TrimSpace(chk.Expected))
if !strings.EqualFold(actual, expected) {
f.checksumFailed.Store(true)
return fmt.Errorf("checksum mismatch: expected %s, got %s", expected, actual)
}
return nil
}
func (f *Fetcher) checkCompletion() bool {
// Check if all data has been downloaded
f.connMu.Lock()
@@ -0,0 +1,108 @@
package http
import (
"crypto/sha256"
"encoding/hex"
"fmt"
gohttp "net/http"
"net/http/httptest"
"os"
"strings"
"sync/atomic"
"testing"
"github.com/GopeedLab/gopeed/pkg/base"
"github.com/GopeedLab/gopeed/pkg/protocol/http"
)
func TestFetcher_ChecksumMismatch_Retry(t *testing.T) {
corruptContent := []byte("Corrupt Initial Download Data!")
correctContent := []byte("Correct Retried Download Data!")
correctSha256Hasher := sha256.New()
correctSha256Hasher.Write(correctContent)
correctSha256 := hex.EncodeToString(correctSha256Hasher.Sum(nil))
var reqCount atomic.Int32
server := httptest.NewServer(gohttp.HandlerFunc(func(w gohttp.ResponseWriter, r *gohttp.Request) {
count := reqCount.Add(1)
var served []byte
if count == 1 {
// First request: serve corrupted content
served = corruptContent
} else {
// Second / subsequent request: serve correct content
served = correctContent
}
w.Header().Set(base.HttpHeaderContentLength, fmt.Sprintf("%d", len(served)))
w.Header().Set(base.HttpHeaderAcceptRanges, base.HttpHeaderBytes)
w.WriteHeader(gohttp.StatusOK)
_, _ = w.Write(served)
}))
defer server.Close()
downloadDir := t.TempDir()
fileName := "test_checksum_retry.txt"
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: downloadDir,
Name: fileName,
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "sha256",
Expected: correctSha256,
},
},
}
// 1. Initial attempt -> will download corruptContent, failing checksum verification
if err := f.Resolve(&base.Request{URL: server.URL + "/retry_test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
err := f.Wait()
if err == nil {
t.Fatal("expected checksum mismatch error on initial attempt, got nil")
}
if !strings.Contains(err.Error(), "checksum mismatch") {
t.Fatalf("expected 'checksum mismatch' in error, got %v", err)
}
if f.getState() != stateError {
t.Fatalf("expected stateError after checksum mismatch, got %v", f.getState())
}
// 2. Retry download -> should trigger forced full reset, truncate file, and re-download correctContent
if err := f.Start(); err != nil {
t.Fatalf("Start error on retry: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error on retry (expected success): %v", err)
}
if f.getState() != stateDone {
t.Fatalf("expected stateDone after successful retry, got %v", f.getState())
}
// 3. Verify on-disk file content matches the re-downloaded bytes
filePath := f.meta.SingleFilepath()
fileBytes, err := os.ReadFile(filePath)
if err != nil {
t.Fatalf("failed to read completed file: %v", err)
}
if string(fileBytes) != string(correctContent) {
t.Fatalf("on-disk file content mismatch:\ngot: %q\nwant: %q", string(fileBytes), string(correctContent))
}
// 4. Compute hash independently and verify
actualHasher := sha256.New()
actualHasher.Write(fileBytes)
actualHash := hex.EncodeToString(actualHasher.Sum(nil))
if actualHash != correctSha256 {
t.Fatalf("re-computed on-disk hash mismatch: got %s, want %s", actualHash, correctSha256)
}
}
@@ -0,0 +1,307 @@
package http
import (
"crypto/md5"
"crypto/sha256"
"encoding/hex"
"fmt"
gohttp "net/http"
"net/http/httptest"
"strings"
"testing"
"github.com/GopeedLab/gopeed/internal/fetcher"
"github.com/GopeedLab/gopeed/internal/test"
"github.com/GopeedLab/gopeed/pkg/base"
"github.com/GopeedLab/gopeed/pkg/protocol/http"
)
func TestFetcher_ChecksumVerification(t *testing.T) {
content := []byte("Hello, Gopeed Checksum Verification Test Data!")
md5Hasher := md5.New()
md5Hasher.Write(content)
correctMd5 := hex.EncodeToString(md5Hasher.Sum(nil))
sha256Hasher := sha256.New()
sha256Hasher.Write(content)
correctSha256 := hex.EncodeToString(sha256Hasher.Sum(nil))
wrongHash := "0123456789abcdef0123456789abcdef"
server := httptest.NewServer(gohttp.HandlerFunc(func(w gohttp.ResponseWriter, r *gohttp.Request) {
w.Header().Set(base.HttpHeaderContentLength, fmt.Sprintf("%d", len(content)))
w.Header().Set(base.HttpHeaderAcceptRanges, base.HttpHeaderBytes)
w.WriteHeader(gohttp.StatusOK)
_, _ = w.Write(content)
}))
defer server.Close()
t.Run("Valid MD5", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_valid_md5.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "md5",
Expected: correctMd5,
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error (expected success): %v", err)
}
})
t.Run("Valid SHA256", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_valid_sha256.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "sha256",
Expected: correctSha256,
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error (expected success): %v", err)
}
})
t.Run("Wrong MD5 mismatch", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_wrong_md5.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "md5",
Expected: wrongHash,
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
err := f.Wait()
if err == nil {
t.Fatal("Wait expected checksum mismatch error, got nil")
}
if !strings.Contains(err.Error(), "checksum mismatch") {
t.Fatalf("expected 'checksum mismatch' in error, got %v", err)
}
})
t.Run("Unsupported Algorithm", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_unsupported.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "invalid-algo",
Expected: wrongHash,
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
err := f.Wait()
if err == nil {
t.Fatal("Wait expected error for unsupported algorithm, got nil")
}
if !strings.Contains(err.Error(), "unsupported checksum algorithm") {
t.Fatalf("expected 'unsupported checksum algorithm' in error, got %v", err)
}
if f.checksumFailed.Load() {
t.Fatal("checksumFailed flag should NOT be set on unsupported algorithm error")
}
})
t.Run("Case-Insensitive Hash Comparison", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_case_insensitive.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "sha256",
Expected: strings.ToUpper(correctSha256),
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error (expected success with upper case hash): %v", err)
}
})
t.Run("Empty Expected Checksum (Skipped)", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_empty_expected.txt",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "",
Expected: "",
},
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error (expected skip and success): %v", err)
}
})
t.Run("Large File Multi-Chunk Streaming (>64KB)", func(t *testing.T) {
largeData := make([]byte, 128*1024)
for i := range largeData {
largeData[i] = byte(i % 256)
}
largeSha256Hasher := sha256.New()
largeSha256Hasher.Write(largeData)
largeSha256 := hex.EncodeToString(largeSha256Hasher.Sum(nil))
largeServer := httptest.NewServer(gohttp.HandlerFunc(func(w gohttp.ResponseWriter, r *gohttp.Request) {
w.Header().Set(base.HttpHeaderContentLength, fmt.Sprintf("%d", len(largeData)))
w.Header().Set(base.HttpHeaderAcceptRanges, base.HttpHeaderBytes)
w.WriteHeader(gohttp.StatusOK)
_, _ = w.Write(largeData)
}))
defer largeServer.Close()
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: "test_large_file.bin",
Extra: &http.OptsExtra{
Connections: 1,
Checksum: &http.ChecksumOption{
Algorithm: "sha256",
Expected: largeSha256,
},
},
}
if err := f.Resolve(&base.Request{URL: largeServer.URL + "/large.bin"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error (expected large file checksum success): %v", err)
}
})
t.Run("No Checksum (Opt-in backward compatibility)", func(t *testing.T) {
f := buildFetcher()
defer f.Close()
opts := &base.Options{
Path: t.TempDir(),
Name: test.DownloadName,
Extra: &http.OptsExtra{
Connections: 1,
},
}
if err := f.Resolve(&base.Request{URL: server.URL + "/test.txt"}, opts); err != nil {
t.Fatalf("Resolve error: %v", err)
}
if err := f.Start(); err != nil {
t.Fatalf("Start error: %v", err)
}
if err := f.Wait(); err != nil {
t.Fatalf("Wait error: %v", err)
}
})
}
func TestResetConnectionForRestart_ForceOnRangeDownload(t *testing.T) {
f := buildFetcher()
f.meta = &fetcher.FetcherMeta{
Res: &base.Resource{
Range: true,
},
}
conn := &connection{
Chunk: newChunk(0, 1024),
Downloaded: 512,
Completed: true,
}
conn.Chunk.Downloaded = 512
// Non-forced reset on range-supported download does not reset Chunk.Downloaded
f.resetConnectionForRestart(conn, false)
if conn.Chunk.Downloaded != 512 {
t.Fatalf("expected Chunk.Downloaded to remain 512 when force is false, got %d", conn.Chunk.Downloaded)
}
// Forced reset on range-supported download resets Chunk.Downloaded to 0
f.resetConnectionForRestart(conn, true)
if conn.Chunk.Downloaded != 0 {
t.Fatalf("expected Chunk.Downloaded to be reset to 0 when force is true, got %d", conn.Chunk.Downloaded)
}
if conn.Completed {
t.Fatal("expected Completed to be false after reset")
}
}