gocheese.go

   1 /*
   2 GoCheese -- Python private package repository and caching proxy
   3 Copyright (C) 2019-2021 Sergey Matveev <stargrave@stargrave.org>
   4               2019-2021 Elena Balakhonova <balakhonova_e@riseup.net>
   5
   6 This program is free software: you can redistribute it and/or modify
   7 it under the terms of the GNU General Public License as published by
   8 the Free Software Foundation, version 3 of the License.
   9
  10 This program is distributed in the hope that it will be useful,
  11 but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 GNU General Public License for more details.
  14
  15 You should have received a copy of the GNU General Public License
  16 along with this program.  If not, see <http://www.gnu.org/licenses/>.
  17 */
  18
  19 // Python private package repository and caching proxy
  20 package main
  21
  22 import (
  23         "bytes"
  24         "context"
  25         "crypto/sha256"
  26         "crypto/tls"
  27         "encoding/hex"
  28         "errors"
  29         "flag"
  30         "fmt"
  31         "io/ioutil"
  32         "log"
  33         "net"
  34         "net/http"
  35         "net/url"
  36         "os"
  37         "os/signal"
  38         "path/filepath"
  39         "regexp"
  40         "runtime"
  41         "strings"
  42         "syscall"
  43         "time"
  44
  45         "golang.org/x/net/netutil"
  46 )
  47
  48 const (
  49         Version   = "3.0.0"
  50         HTMLBegin = `<!DOCTYPE html>
  51 <html>
  52   <head>
  53     <title>Links for %s</title>
  54   </head>
  55   <body>
  56 `
  57         HTMLEnd      = "  </body>\n</html>\n"
  58         HTMLElement  = "    <a href=\"%s\"%s>%s</a><br/>\n"
  59         InternalFlag = ".internal"
  60         GPGSigExt    = ".asc"
  61
  62         Warranty = `This program is free software: you can redistribute it and/or modify
  63 it under the terms of the GNU General Public License as published by
  64 the Free Software Foundation, version 3 of the License.
  65
  66 This program is distributed in the hope that it will be useful,
  67 but WITHOUT ANY WARRANTY; without even the implied warranty of
  68 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  69 GNU General Public License for more details.
  70
  71 You should have received a copy of the GNU General Public License
  72 along with this program.  If not, see <http://www.gnu.org/licenses/>.`
  73 )
  74
  75 const (
  76         HashAlgoSHA256     = "sha256"
  77         HashAlgoBLAKE2b256 = "blake2_256"
  78         HashAlgoSHA512     = "sha512"
  79         HashAlgoMD5        = "md5"
  80 )
  81
  82 var (
  83         pkgPyPI         = regexp.MustCompile(`^.*<a href="([^"]+)"[^>]*>(.+)</a><br/>.*$`)
  84         normalizationRe = regexp.MustCompilePOSIX("[-_.]+")
  85
  86         knownHashAlgos []string = []string{
  87                 HashAlgoSHA256,
  88                 HashAlgoBLAKE2b256,
  89                 HashAlgoSHA512,
  90                 HashAlgoMD5,
  91         }
  92
  93         root             = flag.String("root", "./packages", "Path to packages directory")
  94         bind             = flag.String("bind", "[::]:8080", "Address to bind to")
  95         tlsCert          = flag.String("tls-cert", "", "Path to TLS X.509 certificate")
  96         tlsKey           = flag.String("tls-key", "", "Path to TLS X.509 private key")
  97         norefreshURLPath = flag.String("norefresh", "/norefresh/", "Non-refreshing URL path")
  98         refreshURLPath   = flag.String("refresh", "/simple/", "Auto-refreshing URL path")
  99         gpgUpdateURLPath = flag.String("gpgupdate", "/gpgupdate/", "GPG forceful refreshing URL path")
 100         pypiURL          = flag.String("pypi", "https://pypi.org/simple/", "Upstream (PyPI) URL")
 101         pypiCertHash     = flag.String("pypi-cert-hash", "", "Authenticate upstream by its X.509 certificate's SPKI SHA256 hash")
 102         logTimestamped   = flag.Bool("log-timestamped", false, "Prepend timestmap to log messages")
 103         passwdPath       = flag.String("passwd", "", "Path to FIFO for upload authentication")
 104         passwdCheck      = flag.Bool("passwd-check", false, "Run password checker")
 105         fsck             = flag.Bool("fsck", false, "Check integrity of all packages (errors are in stderr)")
 106         maxClients       = flag.Int("maxclients", 128, "Maximal amount of simultaneous clients")
 107         version          = flag.Bool("version", false, "Print version information")
 108         warranty         = flag.Bool("warranty", false, "Print warranty information")
 109
 110         killed        bool
 111         pypiURLParsed *url.URL
 112 )
 113
 114 func mkdirForPkg(w http.ResponseWriter, r *http.Request, pkgName string) bool {
 115         path := filepath.Join(*root, pkgName)
 116         if _, err := os.Stat(path); os.IsNotExist(err) {
 117                 if err = os.Mkdir(path, os.FileMode(0777)); err != nil {
 118                         log.Println("error", r.RemoteAddr, "mkdir", pkgName, err)
 119                         http.Error(w, err.Error(), http.StatusInternalServerError)
 120                         return false
 121                 }
 122                 log.Println(r.RemoteAddr, "mkdir", pkgName)
 123         }
 124         return true
 125 }
 126
 127 func listRoot(w http.ResponseWriter, r *http.Request) {
 128         files, err := ioutil.ReadDir(*root)
 129         if err != nil {
 130                 log.Println("error", r.RemoteAddr, "root", err)
 131                 http.Error(w, err.Error(), http.StatusInternalServerError)
 132                 return
 133         }
 134         var result bytes.Buffer
 135         result.WriteString(fmt.Sprintf(HTMLBegin, "root"))
 136         for _, file := range files {
 137                 if file.Mode().IsDir() {
 138                         result.WriteString(fmt.Sprintf(
 139                                 HTMLElement,
 140                                 *refreshURLPath+file.Name()+"/",
 141                                 "", file.Name(),
 142                         ))
 143                 }
 144         }
 145         result.WriteString(HTMLEnd)
 146         w.Write(result.Bytes())
 147 }
 148
 149 func listDir(
 150         w http.ResponseWriter,
 151         r *http.Request,
 152         pkgName string,
 153         autorefresh, gpgUpdate bool,
 154 ) {
 155         dirPath := filepath.Join(*root, pkgName)
 156         if autorefresh {
 157                 if !refreshDir(w, r, pkgName, "", gpgUpdate) {
 158                         return
 159                 }
 160         } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, pkgName, "", false) {
 161                 return
 162         }
 163         fis, err := ioutil.ReadDir(dirPath)
 164         if err != nil {
 165                 log.Println("error", r.RemoteAddr, "list", pkgName, err)
 166                 http.Error(w, err.Error(), http.StatusInternalServerError)
 167                 return
 168         }
 169         files := make(map[string]struct{}, len(fis)/2)
 170         for _, fi := range fis {
 171                 files[fi.Name()] = struct{}{}
 172         }
 173         var result bytes.Buffer
 174         result.WriteString(fmt.Sprintf(HTMLBegin, pkgName))
 175         for _, algo := range knownHashAlgos {
 176                 for fn := range files {
 177                         if killed {
 178                                 // Skip expensive I/O when shutting down
 179                                 http.Error(w, "shutting down", http.StatusInternalServerError)
 180                                 return
 181                         }
 182                         if !strings.HasSuffix(fn, "."+algo) {
 183                                 continue
 184                         }
 185                         delete(files, fn)
 186                         digest, err := ioutil.ReadFile(filepath.Join(dirPath, fn))
 187                         if err != nil {
 188                                 log.Println("error", r.RemoteAddr, "list", fn, err)
 189                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 190                                 return
 191                         }
 192                         fnClean := strings.TrimSuffix(fn, "."+algo)
 193                         delete(files, fnClean)
 194                         gpgSigAttr := ""
 195                         if _, err = os.Stat(filepath.Join(dirPath, fnClean+GPGSigExt)); err == nil {
 196                                 gpgSigAttr = " data-gpg-sig=true"
 197                                 delete(files, fnClean+GPGSigExt)
 198                         }
 199                         result.WriteString(fmt.Sprintf(
 200                                 HTMLElement,
 201                                 strings.Join([]string{
 202                                         *refreshURLPath, pkgName, "/", fnClean,
 203                                         "#", algo, "=", hex.EncodeToString(digest),
 204                                 }, ""),
 205                                 gpgSigAttr,
 206                                 fnClean,
 207                         ))
 208                 }
 209         }
 210         result.WriteString(HTMLEnd)
 211         w.Write(result.Bytes())
 212 }
 213
 214 func servePkg(w http.ResponseWriter, r *http.Request, pkgName, filename string) {
 215         log.Println(r.RemoteAddr, "get", filename)
 216         path := filepath.Join(*root, pkgName, filename)
 217         if _, err := os.Stat(path); os.IsNotExist(err) {
 218                 if !refreshDir(w, r, pkgName, filename, false) {
 219                         return
 220                 }
 221         }
 222         http.ServeFile(w, r, path)
 223 }
 224
 225 func handler(w http.ResponseWriter, r *http.Request) {
 226         switch r.Method {
 227         case "GET":
 228                 var path string
 229                 var autorefresh bool
 230                 var gpgUpdate bool
 231                 if strings.HasPrefix(r.URL.Path, *norefreshURLPath) {
 232                         path = strings.TrimPrefix(r.URL.Path, *norefreshURLPath)
 233                 } else if strings.HasPrefix(r.URL.Path, *refreshURLPath) {
 234                         path = strings.TrimPrefix(r.URL.Path, *refreshURLPath)
 235                         autorefresh = true
 236                 } else if strings.HasPrefix(r.URL.Path, *gpgUpdateURLPath) {
 237                         path = strings.TrimPrefix(r.URL.Path, *gpgUpdateURLPath)
 238                         autorefresh = true
 239                         gpgUpdate = true
 240                 } else {
 241                         http.Error(w, "unknown action", http.StatusBadRequest)
 242                         return
 243                 }
 244                 parts := strings.Split(strings.TrimSuffix(path, "/"), "/")
 245                 if len(parts) > 2 {
 246                         http.Error(w, "invalid path", http.StatusBadRequest)
 247                         return
 248                 }
 249                 if len(parts) == 1 {
 250                         if parts[0] == "" {
 251                                 listRoot(w, r)
 252                         } else {
 253                                 listDir(w, r, parts[0], autorefresh, gpgUpdate)
 254                         }
 255                 } else {
 256                         servePkg(w, r, parts[0], parts[1])
 257                 }
 258         case "POST":
 259                 serveUpload(w, r)
 260         default:
 261                 http.Error(w, "unknown action", http.StatusBadRequest)
 262         }
 263 }
 264
 265 func main() {
 266         flag.Parse()
 267         if *warranty {
 268                 fmt.Println(Warranty)
 269                 return
 270         }
 271         if *version {
 272                 fmt.Println("GoCheese", Version, "built with", runtime.Version())
 273                 return
 274         }
 275
 276         if *logTimestamped {
 277                 log.SetFlags(log.Ldate | log.Lmicroseconds | log.Lshortfile)
 278         } else {
 279                 log.SetFlags(log.Lshortfile)
 280         }
 281         log.SetOutput(os.Stdout)
 282
 283         if *fsck {
 284                 if !goodIntegrity() {
 285                         os.Exit(1)
 286                 }
 287                 return
 288         }
 289
 290         if *passwdCheck {
 291                 if passwdReader(os.Stdin) {
 292                         os.Exit(0)
 293                 } else {
 294                         os.Exit(1)
 295                 }
 296         }
 297
 298         if *passwdPath != "" {
 299                 go func() {
 300                         for {
 301                                 fd, err := os.OpenFile(*passwdPath, os.O_RDONLY, os.FileMode(0666))
 302                                 if err != nil {
 303                                         log.Fatalln(err)
 304                                 }
 305                                 passwdReader(fd)
 306                                 fd.Close()
 307                         }
 308                 }()
 309         }
 310
 311         if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
 312                 log.Fatalln("Both -tls-cert and -tls-key are required")
 313         }
 314
 315         var err error
 316         pypiURLParsed, err = url.Parse(*pypiURL)
 317         if err != nil {
 318                 log.Fatalln(err)
 319         }
 320         tlsConfig := tls.Config{
 321                 ClientSessionCache: tls.NewLRUClientSessionCache(16),
 322                 NextProtos:         []string{"h2", "http/1.1"},
 323         }
 324         pypiHTTPTransport = http.Transport{
 325                 ForceAttemptHTTP2: true,
 326                 TLSClientConfig:   &tlsConfig,
 327         }
 328         if *pypiCertHash != "" {
 329                 ourDgst, err := hex.DecodeString(*pypiCertHash)
 330                 if err != nil {
 331                         log.Fatalln(err)
 332                 }
 333                 tlsConfig.VerifyConnection = func(s tls.ConnectionState) error {
 334                         spki := s.VerifiedChains[0][0].RawSubjectPublicKeyInfo
 335                         theirDgst := sha256.Sum256(spki)
 336                         if bytes.Compare(ourDgst, theirDgst[:]) != 0 {
 337                                 return errors.New("certificate's SPKI digest mismatch")
 338                         }
 339                         return nil
 340                 }
 341         }
 342
 343         ln, err := net.Listen("tcp", *bind)
 344         if err != nil {
 345                 log.Fatal(err)
 346         }
 347         ln = netutil.LimitListener(ln, *maxClients)
 348         server := &http.Server{
 349                 ReadTimeout:  time.Minute,
 350                 WriteTimeout: time.Minute,
 351         }
 352         http.HandleFunc(*norefreshURLPath, handler)
 353         http.HandleFunc(*refreshURLPath, handler)
 354         if *gpgUpdateURLPath != "" {
 355                 http.HandleFunc(*gpgUpdateURLPath, handler)
 356         }
 357
 358         needsShutdown := make(chan os.Signal, 0)
 359         exitErr := make(chan error, 0)
 360         signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
 361         go func(s *http.Server) {
 362                 <-needsShutdown
 363                 killed = true
 364                 log.Println("shutting down")
 365                 ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
 366                 exitErr <- s.Shutdown(ctx)
 367                 cancel()
 368         }(server)
 369
 370         log.Println(
 371                 "GoCheese", Version, "listens:",
 372                 "root:", *root,
 373                 "bind:", *bind,
 374                 "pypi:", *pypiURL,
 375         )
 376         if *tlsCert == "" {
 377                 err = server.Serve(ln)
 378         } else {
 379                 err = server.ServeTLS(ln, *tlsCert, *tlsKey)
 380         }
 381         if err != http.ErrServerClosed {
 382                 log.Fatal(err)
 383         }
 384         if err := <-exitErr; err != nil {
 385                 log.Fatal(err)
 386         }
 387 }