/*
GoCheese -- Python private package repository and caching proxy
-Copyright (C) 2019 Sergey Matveev <stargrave@stargrave.org>
- 2019 Elena Balakhonova <balakhonova_e@riseup.net>
+Copyright (C) 2019-2021 Sergey Matveev <stargrave@stargrave.org>
+ 2019-2021 Elena Balakhonova <balakhonova_e@riseup.net>
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
"bytes"
"context"
"crypto/sha256"
+ "crypto/tls"
"encoding/hex"
+ "errors"
"flag"
"fmt"
- "io"
"io/ioutil"
"log"
"net"
)
const (
+ Version = "2.6.0"
HTMLBegin = `<!DOCTYPE html>
<html>
<head>
`
HTMLEnd = " </body>\n</html>\n"
HTMLElement = " <a href=\"%s\"%s>%s</a><br/>\n"
- SHA256Prefix = "sha256="
- SHA256Ext = ".sha256"
InternalFlag = ".internal"
GPGSigExt = ".asc"
- GPGSigAttr = " data-gpg-sig=true"
Warranty = `This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
along with this program. If not, see <http://www.gnu.org/licenses/>.`
)
+const (
+ HashAlgoSHA256 = "sha256"
+ HashAlgoBLAKE2b256 = "blake2_256"
+ HashAlgoSHA512 = "sha512"
+ HashAlgoMD5 = "md5"
+)
+
var (
- pkgPyPI = regexp.MustCompile(`^.*<a href="([^"]+)"[^>]*>(.+)</a><br/>.*$`)
- Version string = "UNKNOWN"
+ pkgPyPI = regexp.MustCompile(`^.*<a href="([^"]+)"[^>]*>(.+)</a><br/>.*$`)
+ normalizationRe = regexp.MustCompilePOSIX("[-_.]+")
+
+ knownHashAlgos []string = []string{
+ HashAlgoSHA256,
+ HashAlgoBLAKE2b256,
+ HashAlgoSHA512,
+ HashAlgoMD5,
+ }
root = flag.String("root", "./packages", "Path to packages directory")
bind = flag.String("bind", "[::]:8080", "Address to bind to")
refreshURLPath = flag.String("refresh", "/simple/", "Auto-refreshing URL path")
gpgUpdateURLPath = flag.String("gpgupdate", "/gpgupdate/", "GPG forceful refreshing URL path")
pypiURL = flag.String("pypi", "https://pypi.org/simple/", "Upstream PyPI URL")
+ pypiCertHash = flag.String("pypi-cert-hash", "", "Authenticate PyPI by its X.509 certificate's SHA256 hash")
passwdPath = flag.String("passwd", "passwd", "Path to file with authenticators")
+ logTimestamped = flag.Bool("log-timestamped", false, "Prepend timestmap to log messages")
passwdCheck = flag.Bool("passwd-check", false, "Test the -passwd file for syntax errors and exit")
- fsck = flag.Bool("fsck", false, "Check integrity of all packages")
+ fsck = flag.Bool("fsck", false, "Check integrity of all packages (errors are in stderr)")
maxClients = flag.Int("maxclients", 128, "Maximal amount of simultaneous clients")
version = flag.Bool("version", false, "Print version information")
warranty = flag.Bool("warranty", false, "Print warranty information")
- killed bool
-
- normalizationRe *regexp.Regexp = regexp.MustCompilePOSIX("[-_.]+")
+ killed bool
+ pypiURLParsed *url.URL
)
-func mkdirForPkg(w http.ResponseWriter, r *http.Request, dir string) bool {
- path := filepath.Join(*root, dir)
+func mkdirForPkg(w http.ResponseWriter, r *http.Request, pkgName string) bool {
+ path := filepath.Join(*root, pkgName)
if _, err := os.Stat(path); os.IsNotExist(err) {
if err = os.Mkdir(path, os.FileMode(0777)); err != nil {
+ log.Println("error", r.RemoteAddr, "mkdir", pkgName, err)
http.Error(w, err.Error(), http.StatusInternalServerError)
return false
}
- log.Println(r.RemoteAddr, "mkdir", dir)
- }
- return true
-}
-
-func refreshDir(
- w http.ResponseWriter,
- r *http.Request,
- dir,
- filenameGet string,
- gpgUpdate bool,
-) bool {
- if _, err := os.Stat(filepath.Join(*root, dir, InternalFlag)); err == nil {
- return true
- }
- resp, err := http.Get(*pypiURL + dir + "/")
- if err != nil {
- http.Error(w, err.Error(), http.StatusBadGateway)
- return false
- }
- body, err := ioutil.ReadAll(resp.Body)
- resp.Body.Close()
- if err != nil {
- http.Error(w, err.Error(), http.StatusBadGateway)
- return false
- }
- if !mkdirForPkg(w, r, dir) {
- return false
- }
- dirPath := filepath.Join(*root, dir)
- var submatches []string
- var uri string
- var filename string
- var path string
- var pkgURL *url.URL
- var digest []byte
- for _, lineRaw := range bytes.Split(body, []byte("\n")) {
- submatches = pkgPyPI.FindStringSubmatch(string(lineRaw))
- if len(submatches) == 0 {
- continue
- }
- uri = submatches[1]
- filename = submatches[2]
- if pkgURL, err = url.Parse(uri); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- if !strings.HasPrefix(pkgURL.Fragment, SHA256Prefix) {
- log.Println(r.RemoteAddr, "pypi", filename, "no SHA256 digest provided")
- http.Error(w, "no SHA256 digest provided", http.StatusBadGateway)
- return false
- }
- digest, err = hex.DecodeString(strings.TrimPrefix(pkgURL.Fragment, SHA256Prefix))
- if err != nil {
- http.Error(w, err.Error(), http.StatusBadGateway)
- return false
- }
- pkgURL.Fragment = ""
- uri = pkgURL.String()
- if pkgURL.Host == "" {
- uri = *pypiURL + strings.TrimPrefix(uri, "/")
- }
- path = filepath.Join(dirPath, filename)
- if filename == filenameGet {
- if killed {
- // Skip heavy remote call, when shutting down
- http.Error(w, "shutting down", http.StatusInternalServerError)
- return false
- }
- log.Println(r.RemoteAddr, "pypi download", filename)
- resp, err = http.Get(uri)
- if err != nil {
- log.Println(r.RemoteAddr, "pypi download error:", err.Error())
- http.Error(w, err.Error(), http.StatusBadGateway)
- return false
- }
- defer resp.Body.Close()
- hasher := sha256.New()
- dst, err := TempFile(dirPath)
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- wr := io.MultiWriter(hasher, dst)
- if _, err = io.Copy(wr, resp.Body); err != nil {
- os.Remove(dst.Name())
- dst.Close()
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- if bytes.Compare(hasher.Sum(nil), digest) != 0 {
- log.Println(r.RemoteAddr, "pypi", filename, "digest mismatch")
- os.Remove(dst.Name())
- dst.Close()
- http.Error(w, "digest mismatch", http.StatusBadGateway)
- return false
- }
- if err = dst.Sync(); err != nil {
- os.Remove(dst.Name())
- dst.Close()
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- dst.Close()
- if err = os.Rename(dst.Name(), path); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- if err = DirSync(dirPath); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- }
- if filename == filenameGet || gpgUpdate {
- if _, err = os.Stat(path); err != nil {
- goto GPGSigSkip
- }
- resp, err := http.Get(uri + GPGSigExt)
- if err != nil {
- goto GPGSigSkip
- }
- if resp.StatusCode != http.StatusOK {
- resp.Body.Close()
- goto GPGSigSkip
- }
- sig, err := ioutil.ReadAll(resp.Body)
- resp.Body.Close()
- if err != nil {
- goto GPGSigSkip
- }
- if err = WriteFileSync(dirPath, path+GPGSigExt, sig); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- log.Println(r.RemoteAddr, "pypi downloaded signature", filename)
- }
- GPGSigSkip:
- path = path + SHA256Ext
- _, err = os.Stat(path)
- if err == nil {
- continue
- }
- if !os.IsNotExist(err) {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
- log.Println(r.RemoteAddr, "pypi touch", filename)
- if err = WriteFileSync(dirPath, path, digest); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return false
- }
+ log.Println(r.RemoteAddr, "mkdir", pkgName)
}
return true
}
func listRoot(w http.ResponseWriter, r *http.Request) {
files, err := ioutil.ReadDir(*root)
if err != nil {
+ log.Println("error", r.RemoteAddr, "root", err)
http.Error(w, err.Error(), http.StatusInternalServerError)
return
}
result.WriteString(fmt.Sprintf(
HTMLElement,
*refreshURLPath+file.Name()+"/",
- file.Name(),
+ "", file.Name(),
))
}
}
func listDir(
w http.ResponseWriter,
r *http.Request,
- dir string,
- autorefresh,
- gpgUpdate bool,
+ pkgName string,
+ autorefresh, gpgUpdate bool,
) {
- dirPath := filepath.Join(*root, dir)
+ dirPath := filepath.Join(*root, pkgName)
if autorefresh {
- if !refreshDir(w, r, dir, "", gpgUpdate) {
+ if !refreshDir(w, r, pkgName, "", gpgUpdate) {
return
}
- } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, dir, "", false) {
+ } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, pkgName, "", false) {
return
}
- files, err := ioutil.ReadDir(dirPath)
+ fis, err := ioutil.ReadDir(dirPath)
if err != nil {
+ log.Println("error", r.RemoteAddr, "list", pkgName, err)
http.Error(w, err.Error(), http.StatusInternalServerError)
return
}
+ files := make(map[string]struct{}, len(fis)/2)
+ for _, fi := range fis {
+ files[fi.Name()] = struct{}{}
+ }
var result bytes.Buffer
- result.WriteString(fmt.Sprintf(HTMLBegin, dir))
- var data []byte
- var gpgSigAttr string
- var filenameClean string
- for _, file := range files {
- if !strings.HasSuffix(file.Name(), SHA256Ext) {
- continue
- }
- if killed {
- // Skip expensive I/O when shutting down
- http.Error(w, "shutting down", http.StatusInternalServerError)
- return
- }
- data, err = ioutil.ReadFile(filepath.Join(dirPath, file.Name()))
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- filenameClean = strings.TrimSuffix(file.Name(), SHA256Ext)
- if _, err = os.Stat(filepath.Join(dirPath, filenameClean+GPGSigExt)); os.IsNotExist(err) {
- gpgSigAttr = ""
- } else {
- gpgSigAttr = GPGSigAttr
+ result.WriteString(fmt.Sprintf(HTMLBegin, pkgName))
+ for _, algo := range knownHashAlgos {
+ for fn := range files {
+ if killed {
+ // Skip expensive I/O when shutting down
+ http.Error(w, "shutting down", http.StatusInternalServerError)
+ return
+ }
+ if !strings.HasSuffix(fn, "."+algo) {
+ continue
+ }
+ delete(files, fn)
+ digest, err := ioutil.ReadFile(filepath.Join(dirPath, fn))
+ if err != nil {
+ log.Println("error", r.RemoteAddr, "list", fn, err)
+ http.Error(w, err.Error(), http.StatusInternalServerError)
+ return
+ }
+ fnClean := strings.TrimSuffix(fn, "."+algo)
+ delete(files, fnClean)
+ gpgSigAttr := ""
+ if _, err = os.Stat(filepath.Join(dirPath, fnClean+GPGSigExt)); err == nil {
+ gpgSigAttr = " data-gpg-sig=true"
+ delete(files, fnClean+GPGSigExt)
+ }
+ result.WriteString(fmt.Sprintf(
+ HTMLElement,
+ strings.Join([]string{
+ *refreshURLPath, pkgName, "/", fnClean,
+ "#", algo, "=", hex.EncodeToString(digest),
+ }, ""),
+ gpgSigAttr,
+ fnClean,
+ ))
}
- result.WriteString(fmt.Sprintf(
- HTMLElement,
- strings.Join([]string{
- *refreshURLPath, dir, "/",
- filenameClean, "#", SHA256Prefix, hex.EncodeToString(data),
- }, ""),
- gpgSigAttr,
- filenameClean,
- ))
}
result.WriteString(HTMLEnd)
w.Write(result.Bytes())
}
-func servePkg(w http.ResponseWriter, r *http.Request, dir, filename string) {
+func servePkg(w http.ResponseWriter, r *http.Request, pkgName, filename string) {
log.Println(r.RemoteAddr, "get", filename)
- path := filepath.Join(*root, dir, filename)
+ path := filepath.Join(*root, pkgName, filename)
if _, err := os.Stat(path); os.IsNotExist(err) {
- if !refreshDir(w, r, dir, filename, false) {
+ if !refreshDir(w, r, pkgName, filename, false) {
return
}
}
http.ServeFile(w, r, path)
}
-func serveUpload(w http.ResponseWriter, r *http.Request) {
- // Authentication
- username, password, ok := r.BasicAuth()
- if !ok {
- log.Println(r.RemoteAddr, "unauthenticated", username)
- http.Error(w, "unauthenticated", http.StatusUnauthorized)
- return
- }
- auther, ok := passwords[username]
- if !ok || !auther.Auth(password) {
- log.Println(r.RemoteAddr, "unauthenticated", username)
- http.Error(w, "unauthenticated", http.StatusUnauthorized)
- return
- }
-
- // Form parsing
- var err error
- if err = r.ParseMultipartForm(1 << 20); err != nil {
- http.Error(w, err.Error(), http.StatusBadRequest)
- return
- }
- pkgNames, exists := r.MultipartForm.Value["name"]
- if !exists || len(pkgNames) != 1 {
- http.Error(w, "single name is expected in request", http.StatusBadRequest)
- return
- }
- dir := normalizationRe.ReplaceAllString(pkgNames[0], "-")
- dirPath := filepath.Join(*root, dir)
- var digestExpected []byte
- if digestExpectedHex, exists := r.MultipartForm.Value["sha256_digest"]; exists {
- digestExpected, err = hex.DecodeString(digestExpectedHex[0])
- if err != nil {
- http.Error(w, "bad sha256_digest: "+err.Error(), http.StatusBadRequest)
- return
- }
- }
- gpgSigsExpected := make(map[string]struct{})
-
- // Checking is it internal package
- if _, err = os.Stat(filepath.Join(dirPath, InternalFlag)); err != nil {
- log.Println(r.RemoteAddr, "non-internal package", dir)
- http.Error(w, "unknown internal package", http.StatusUnauthorized)
- return
- }
-
- for _, file := range r.MultipartForm.File["content"] {
- filename := file.Filename
- gpgSigsExpected[filename+GPGSigExt] = struct{}{}
- log.Println(r.RemoteAddr, "put", filename, "by", username)
- path := filepath.Join(dirPath, filename)
- if _, err = os.Stat(path); err == nil {
- log.Println(r.RemoteAddr, "already exists", filename)
- http.Error(w, "already exists", http.StatusBadRequest)
- return
- }
- if !mkdirForPkg(w, r, dir) {
- return
- }
- src, err := file.Open()
- defer src.Close()
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- dst, err := TempFile(dirPath)
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- hasher := sha256.New()
- wr := io.MultiWriter(hasher, dst)
- if _, err = io.Copy(wr, src); err != nil {
- os.Remove(dst.Name())
- dst.Close()
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- if err = dst.Sync(); err != nil {
- os.Remove(dst.Name())
- dst.Close()
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- dst.Close()
- digest := hasher.Sum(nil)
- if digestExpected != nil {
- if bytes.Compare(digestExpected, digest) == 0 {
- log.Println(r.RemoteAddr, filename, "good checksum received")
- } else {
- log.Println(r.RemoteAddr, filename, "bad checksum received")
- http.Error(w, "bad checksum", http.StatusBadRequest)
- os.Remove(dst.Name())
- return
- }
- }
- if err = os.Rename(dst.Name(), path); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- if err = DirSync(dirPath); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- if err = WriteFileSync(dirPath, path+SHA256Ext, digest); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- }
- for _, file := range r.MultipartForm.File["gpg_signature"] {
- filename := file.Filename
- if _, exists := gpgSigsExpected[filename]; !exists {
- http.Error(w, "unexpected GPG signature filename", http.StatusBadRequest)
- return
- }
- delete(gpgSigsExpected, filename)
- log.Println(r.RemoteAddr, "put", filename, "by", username)
- path := filepath.Join(dirPath, filename)
- if _, err = os.Stat(path); err == nil {
- log.Println(r.RemoteAddr, "already exists", filename)
- http.Error(w, "already exists", http.StatusBadRequest)
- return
- }
- src, err := file.Open()
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- sig, err := ioutil.ReadAll(src)
- src.Close()
- if err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- if err = WriteFileSync(dirPath, path, sig); err != nil {
- http.Error(w, err.Error(), http.StatusInternalServerError)
- return
- }
- }
-}
-
func handler(w http.ResponseWriter, r *http.Request) {
switch r.Method {
case "GET":
}
}
-func goodIntegrity() bool {
- dirs, err := ioutil.ReadDir(*root)
- if err != nil {
- log.Fatal(err)
- }
- hasher := sha256.New()
- digest := make([]byte, sha256.Size)
- isGood := true
- var data []byte
- var pkgName string
- for _, dir := range dirs {
- files, err := ioutil.ReadDir(filepath.Join(*root, dir.Name()))
- if err != nil {
- log.Fatal(err)
- }
- for _, file := range files {
- if !strings.HasSuffix(file.Name(), SHA256Ext) {
- continue
- }
- pkgName = strings.TrimSuffix(file.Name(), SHA256Ext)
- data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), pkgName))
- if err != nil {
- if os.IsNotExist(err) {
- continue
- }
- log.Fatal(err)
- }
- hasher.Write(data)
- data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), file.Name()))
- if err != nil {
- log.Fatal(err)
- }
- if bytes.Compare(hasher.Sum(digest[:0]), data) == 0 {
- fmt.Println(pkgName, "GOOD")
- } else {
- isGood = false
- fmt.Println(pkgName, "BAD")
- }
- hasher.Reset()
- }
- }
- return isGood
-}
-
func main() {
flag.Parse()
if *warranty {
return
}
if *version {
- fmt.Println("GoCheese version " + Version + " built with " + runtime.Version())
+ fmt.Println("GoCheese", Version, "built with", runtime.Version())
return
}
+
+ if *logTimestamped {
+ log.SetFlags(log.Ldate | log.Lmicroseconds | log.Lshortfile)
+ } else {
+ log.SetFlags(log.Lshortfile)
+ }
+ log.SetOutput(os.Stdout)
+
if *fsck {
if !goodIntegrity() {
os.Exit(1)
}
return
}
+
if *passwdCheck {
refreshPasswd()
return
}
+
if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
log.Fatalln("Both -tls-cert and -tls-key are required")
}
+
+ var err error
+ pypiURLParsed, err = url.Parse(*pypiURL)
+ if err != nil {
+ log.Fatalln(err)
+ }
refreshPasswd()
- log.Println("root:", *root, "bind:", *bind)
+ if *pypiCertHash == "" {
+ pypiHTTPTransport = http.Transport{}
+ } else {
+ ourDgst, err := hex.DecodeString(*pypiCertHash)
+ if err != nil {
+ log.Fatalln(err)
+ }
+ pypiHTTPTransport = http.Transport{
+ TLSClientConfig: &tls.Config{
+ VerifyConnection: func(s tls.ConnectionState) error {
+ spki := s.VerifiedChains[0][0].RawSubjectPublicKeyInfo
+ theirDgst := sha256.Sum256(spki)
+ if bytes.Compare(ourDgst, theirDgst[:]) != 0 {
+ return errors.New("certificate's digest mismatch")
+ }
+ return nil
+ }},
+ }
+ }
ln, err := net.Listen("tcp", *bind)
if err != nil {
}
http.HandleFunc(*norefreshURLPath, handler)
http.HandleFunc(*refreshURLPath, handler)
- http.HandleFunc(*gpgUpdateURLPath, handler)
+ if *gpgUpdateURLPath != "" {
+ http.HandleFunc(*gpgUpdateURLPath, handler)
+ }
needsRefreshPasswd := make(chan os.Signal, 0)
needsShutdown := make(chan os.Signal, 0)
signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
go func() {
for range needsRefreshPasswd {
- log.Println("Refreshing passwords")
+ log.Println("refreshing passwords")
refreshPasswd()
}
}()
go func(s *http.Server) {
<-needsShutdown
killed = true
- log.Println("Shutting down")
+ log.Println("shutting down")
ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
exitErr <- s.Shutdown(ctx)
cancel()
}(server)
+ log.Println(
+ "GoCheese", Version, "listens:",
+ "root:", *root,
+ "bind:", *bind,
+ "pypi:", *pypiURL,
+ )
if *tlsCert == "" {
err = server.Serve(ln)
} else {