gocheese.go

   1 /*
   2 GoCheese -- Python private package repository and caching proxy
   3 Copyright (C) 2019 Sergey Matveev <stargrave@stargrave.org>
   4               2019 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         "encoding/hex"
  27         "flag"
  28         "fmt"
  29         "io"
  30         "io/ioutil"
  31         "log"
  32         "net"
  33         "net/http"
  34         "net/url"
  35         "os"
  36         "os/signal"
  37         "path/filepath"
  38         "regexp"
  39         "runtime"
  40         "strings"
  41         "syscall"
  42         "time"
  43
  44         "golang.org/x/net/netutil"
  45 )
  46
  47 const (
  48         HTMLBegin = `<!DOCTYPE html>
  49 <html>
  50   <head>
  51     <title>Links for %s</title>
  52   </head>
  53   <body>
  54 `
  55         HTMLEnd      = "  </body>\n</html>\n"
  56         HTMLElement  = "    <a href=\"%s\"%s>%s</a><br/>\n"
  57         SHA256Prefix = "sha256="
  58         SHA256Ext    = ".sha256"
  59         InternalFlag = ".internal"
  60         GPGSigExt    = ".asc"
  61         GPGSigAttr   = " data-gpg-sig=true"
  62
  63         Warranty = `This program is free software: you can redistribute it and/or modify
  64 it under the terms of the GNU General Public License as published by
  65 the Free Software Foundation, version 3 of the License.
  66
  67 This program is distributed in the hope that it will be useful,
  68 but WITHOUT ANY WARRANTY; without even the implied warranty of
  69 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  70 GNU General Public License for more details.
  71
  72 You should have received a copy of the GNU General Public License
  73 along with this program.  If not, see <http://www.gnu.org/licenses/>.`
  74 )
  75
  76 var (
  77         pkgPyPI        = regexp.MustCompile(`^.*<a href="([^"]+)"[^>]*>(.+)</a><br/>.*$`)
  78         Version string = "UNKNOWN"
  79
  80         root             = flag.String("root", "./packages", "Path to packages directory")
  81         bind             = flag.String("bind", "[::]:8080", "Address to bind to")
  82         tlsCert          = flag.String("tls-cert", "", "Path to TLS X.509 certificate")
  83         tlsKey           = flag.String("tls-key", "", "Path to TLS X.509 private key")
  84         norefreshURLPath = flag.String("norefresh", "/norefresh/", "Non-refreshing URL path")
  85         refreshURLPath   = flag.String("refresh", "/simple/", "Auto-refreshing URL path")
  86         gpgUpdateURLPath = flag.String("gpgupdate", "/gpgupdate/", "GPG forceful refreshing URL path")
  87         pypiURL          = flag.String("pypi", "https://pypi.org/simple/", "Upstream PyPI URL")
  88         passwdPath       = flag.String("passwd", "passwd", "Path to file with authenticators")
  89         passwdCheck      = flag.Bool("passwd-check", false, "Test the -passwd file for syntax errors and exit")
  90         fsck             = flag.Bool("fsck", false, "Check integrity of all packages")
  91         maxClients       = flag.Int("maxclients", 128, "Maximal amount of simultaneous clients")
  92         version          = flag.Bool("version", false, "Print version information")
  93         warranty         = flag.Bool("warranty", false, "Print warranty information")
  94
  95         killed bool
  96
  97         normalizationRe *regexp.Regexp = regexp.MustCompilePOSIX("[-_.]+")
  98 )
  99
 100 func mkdirForPkg(w http.ResponseWriter, r *http.Request, dir string) bool {
 101         path := filepath.Join(*root, dir)
 102         if _, err := os.Stat(path); os.IsNotExist(err) {
 103                 if err = os.Mkdir(path, os.FileMode(0777)); err != nil {
 104                         http.Error(w, err.Error(), http.StatusInternalServerError)
 105                         return false
 106                 }
 107                 log.Println(r.RemoteAddr, "mkdir", dir)
 108         }
 109         return true
 110 }
 111
 112 func refreshDir(
 113         w http.ResponseWriter,
 114         r *http.Request,
 115         dir,
 116         filenameGet string,
 117         gpgUpdate bool,
 118 ) bool {
 119         if _, err := os.Stat(filepath.Join(*root, dir, InternalFlag)); err == nil {
 120                 return true
 121         }
 122         resp, err := http.Get(*pypiURL + dir + "/")
 123         if err != nil {
 124                 http.Error(w, err.Error(), http.StatusBadGateway)
 125                 return false
 126         }
 127         body, err := ioutil.ReadAll(resp.Body)
 128         resp.Body.Close()
 129         if err != nil {
 130                 http.Error(w, err.Error(), http.StatusBadGateway)
 131                 return false
 132         }
 133         if !mkdirForPkg(w, r, dir) {
 134                 return false
 135         }
 136         dirPath := filepath.Join(*root, dir)
 137         var submatches []string
 138         var uri string
 139         var filename string
 140         var path string
 141         var pkgURL *url.URL
 142         var digest []byte
 143         for _, lineRaw := range bytes.Split(body, []byte("\n")) {
 144                 submatches = pkgPyPI.FindStringSubmatch(string(lineRaw))
 145                 if len(submatches) == 0 {
 146                         continue
 147                 }
 148                 uri = submatches[1]
 149                 filename = submatches[2]
 150                 if pkgURL, err = url.Parse(uri); err != nil {
 151                         http.Error(w, err.Error(), http.StatusInternalServerError)
 152                         return false
 153                 }
 154                 digest, err = hex.DecodeString(strings.TrimPrefix(pkgURL.Fragment, SHA256Prefix))
 155                 if err != nil {
 156                         http.Error(w, err.Error(), http.StatusBadGateway)
 157                         return false
 158                 }
 159                 pkgURL.Fragment = ""
 160                 path = filepath.Join(dirPath, filename)
 161                 if filename == filenameGet {
 162                         if killed {
 163                                 // Skip heavy remote call, when shutting down
 164                                 http.Error(w, "shutting down", http.StatusInternalServerError)
 165                                 return false
 166                         }
 167                         log.Println(r.RemoteAddr, "pypi download", filename)
 168                         resp, err = http.Get(pkgURL.String())
 169                         if err != nil {
 170                                 http.Error(w, err.Error(), http.StatusBadGateway)
 171                                 return false
 172                         }
 173                         defer resp.Body.Close()
 174                         hasher := sha256.New()
 175                         dst, err := TempFile(dirPath)
 176                         if err != nil {
 177                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 178                                 return false
 179                         }
 180                         wr := io.MultiWriter(hasher, dst)
 181                         if _, err = io.Copy(wr, resp.Body); err != nil {
 182                                 os.Remove(dst.Name())
 183                                 dst.Close()
 184                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 185                                 return false
 186                         }
 187                         if bytes.Compare(hasher.Sum(nil), digest) != 0 {
 188                                 log.Println(r.RemoteAddr, "pypi", filename, "digest mismatch")
 189                                 os.Remove(dst.Name())
 190                                 dst.Close()
 191                                 http.Error(w, err.Error(), http.StatusBadGateway)
 192                                 return false
 193                         }
 194                         if err = dst.Sync(); err != nil {
 195                                 os.Remove(dst.Name())
 196                                 dst.Close()
 197                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 198                                 return false
 199                         }
 200                         dst.Close()
 201                         if err = os.Rename(dst.Name(), path); err != nil {
 202                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 203                                 return false
 204                         }
 205                 }
 206                 if filename == filenameGet || gpgUpdate {
 207                         if _, err = os.Stat(path); err == nil {
 208                                 if resp, err := http.Get(pkgURL.String() + GPGSigExt); err == nil {
 209                                         sig, err := ioutil.ReadAll(resp.Body)
 210                                         resp.Body.Close()
 211                                         if err == nil {
 212                                                 if err = WriteFileSync(dirPath, path+GPGSigExt, sig); err != nil {
 213                                                         http.Error(w, err.Error(), http.StatusInternalServerError)
 214                                                         return false
 215                                                 }
 216                                                 log.Println(r.RemoteAddr, "pypi downloaded signature", filename)
 217                                         }
 218                                 }
 219                         }
 220                 }
 221                 path = path + SHA256Ext
 222                 _, err = os.Stat(path)
 223                 if err == nil {
 224                         continue
 225                 }
 226                 if !os.IsNotExist(err) {
 227                         http.Error(w, err.Error(), http.StatusInternalServerError)
 228                         return false
 229                 }
 230                 log.Println(r.RemoteAddr, "pypi touch", filename)
 231                 if err = WriteFileSync(dirPath, path, digest); err != nil {
 232                         http.Error(w, err.Error(), http.StatusInternalServerError)
 233                         return false
 234                 }
 235         }
 236         return true
 237 }
 238
 239 func listRoot(w http.ResponseWriter, r *http.Request) {
 240         files, err := ioutil.ReadDir(*root)
 241         if err != nil {
 242                 http.Error(w, err.Error(), http.StatusInternalServerError)
 243                 return
 244         }
 245         var result bytes.Buffer
 246         result.WriteString(fmt.Sprintf(HTMLBegin, "root"))
 247         for _, file := range files {
 248                 if file.Mode().IsDir() {
 249                         result.WriteString(fmt.Sprintf(
 250                                 HTMLElement,
 251                                 *refreshURLPath+file.Name()+"/",
 252                                 file.Name(),
 253                         ))
 254                 }
 255         }
 256         result.WriteString(HTMLEnd)
 257         w.Write(result.Bytes())
 258 }
 259
 260 func listDir(
 261         w http.ResponseWriter,
 262         r *http.Request,
 263         dir string,
 264         autorefresh,
 265         gpgUpdate bool,
 266 ) {
 267         dirPath := filepath.Join(*root, dir)
 268         if autorefresh {
 269                 if !refreshDir(w, r, dir, "", gpgUpdate) {
 270                         return
 271                 }
 272         } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, dir, "", false) {
 273                 return
 274         }
 275         files, err := ioutil.ReadDir(dirPath)
 276         if err != nil {
 277                 http.Error(w, err.Error(), http.StatusInternalServerError)
 278                 return
 279         }
 280         var result bytes.Buffer
 281         result.WriteString(fmt.Sprintf(HTMLBegin, dir))
 282         var data []byte
 283         var gpgSigAttr string
 284         var filenameClean string
 285         for _, file := range files {
 286                 if !strings.HasSuffix(file.Name(), SHA256Ext) {
 287                         continue
 288                 }
 289                 if killed {
 290                         // Skip expensive I/O when shutting down
 291                         http.Error(w, "shutting down", http.StatusInternalServerError)
 292                         return
 293                 }
 294                 data, err = ioutil.ReadFile(filepath.Join(dirPath, file.Name()))
 295                 if err != nil {
 296                         http.Error(w, err.Error(), http.StatusInternalServerError)
 297                         return
 298                 }
 299                 filenameClean = strings.TrimSuffix(file.Name(), SHA256Ext)
 300                 if _, err = os.Stat(filepath.Join(dirPath, filenameClean+GPGSigExt)); os.IsNotExist(err) {
 301                         gpgSigAttr = ""
 302                 } else {
 303                         gpgSigAttr = GPGSigAttr
 304                 }
 305                 result.WriteString(fmt.Sprintf(
 306                         HTMLElement,
 307                         strings.Join([]string{
 308                                 *refreshURLPath, dir, "/",
 309                                 filenameClean, "#", SHA256Prefix, hex.EncodeToString(data),
 310                         }, ""),
 311                         gpgSigAttr,
 312                         filenameClean,
 313                 ))
 314         }
 315         result.WriteString(HTMLEnd)
 316         w.Write(result.Bytes())
 317 }
 318
 319 func servePkg(w http.ResponseWriter, r *http.Request, dir, filename string) {
 320         log.Println(r.RemoteAddr, "get", filename)
 321         path := filepath.Join(*root, dir, filename)
 322         if _, err := os.Stat(path); os.IsNotExist(err) {
 323                 if !refreshDir(w, r, dir, filename, false) {
 324                         return
 325                 }
 326         }
 327         http.ServeFile(w, r, path)
 328 }
 329
 330 func serveUpload(w http.ResponseWriter, r *http.Request) {
 331         // Authentication
 332         username, password, ok := r.BasicAuth()
 333         if !ok {
 334                 log.Println(r.RemoteAddr, "unauthenticated", username)
 335                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 336                 return
 337         }
 338         auther, ok := passwords[username]
 339         if !ok || !auther.Auth(password) {
 340                 log.Println(r.RemoteAddr, "unauthenticated", username)
 341                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 342                 return
 343         }
 344
 345         // Form parsing
 346         var err error
 347         if err = r.ParseMultipartForm(1 << 20); err != nil {
 348                 http.Error(w, err.Error(), http.StatusBadRequest)
 349                 return
 350         }
 351         pkgNames, exists := r.MultipartForm.Value["name"]
 352         if !exists || len(pkgNames) != 1 {
 353                 http.Error(w, "single name is expected in request", http.StatusBadRequest)
 354                 return
 355         }
 356         dir := normalizationRe.ReplaceAllString(pkgNames[0], "-")
 357         dirPath := filepath.Join(*root, dir)
 358         var digestExpected []byte
 359         if digestExpectedHex, exists := r.MultipartForm.Value["sha256_digest"]; exists {
 360                 digestExpected, err = hex.DecodeString(digestExpectedHex[0])
 361                 if err != nil {
 362                         http.Error(w, "bad sha256_digest: "+err.Error(), http.StatusBadRequest)
 363                         return
 364                 }
 365         }
 366         gpgSigsExpected := make(map[string]struct{})
 367
 368         // Checking is it internal package
 369         if _, err = os.Stat(filepath.Join(dirPath, InternalFlag)); err != nil {
 370                 log.Println(r.RemoteAddr, "non-internal package", dir)
 371                 http.Error(w, "unknown internal package", http.StatusUnauthorized)
 372                 return
 373         }
 374
 375         for _, file := range r.MultipartForm.File["content"] {
 376                 filename := file.Filename
 377                 gpgSigsExpected[filename+GPGSigExt] = struct{}{}
 378                 log.Println(r.RemoteAddr, "put", filename, "by", username)
 379                 path := filepath.Join(dirPath, filename)
 380                 if _, err = os.Stat(path); err == nil {
 381                         log.Println(r.RemoteAddr, "already exists", filename)
 382                         http.Error(w, "already exists", http.StatusBadRequest)
 383                         return
 384                 }
 385                 if !mkdirForPkg(w, r, dir) {
 386                         return
 387                 }
 388                 src, err := file.Open()
 389                 defer src.Close()
 390                 if err != nil {
 391                         http.Error(w, err.Error(), http.StatusInternalServerError)
 392                         return
 393                 }
 394                 dst, err := TempFile(dirPath)
 395                 if err != nil {
 396                         http.Error(w, err.Error(), http.StatusInternalServerError)
 397                         return
 398                 }
 399                 hasher := sha256.New()
 400                 wr := io.MultiWriter(hasher, dst)
 401                 if _, err = io.Copy(wr, src); err != nil {
 402                         os.Remove(dst.Name())
 403                         dst.Close()
 404                         http.Error(w, err.Error(), http.StatusInternalServerError)
 405                         return
 406                 }
 407                 if err = dst.Sync(); err != nil {
 408                         os.Remove(dst.Name())
 409                         dst.Close()
 410                         http.Error(w, err.Error(), http.StatusInternalServerError)
 411                         return
 412                 }
 413                 dst.Close()
 414                 digest := hasher.Sum(nil)
 415                 if digestExpected != nil {
 416                         if bytes.Compare(digestExpected, digest) == 0 {
 417                                 log.Println(r.RemoteAddr, filename, "good checksum received")
 418                         } else {
 419                                 log.Println(r.RemoteAddr, filename, "bad checksum received")
 420                                 http.Error(w, "bad checksum", http.StatusBadRequest)
 421                                 os.Remove(dst.Name())
 422                                 return
 423                         }
 424                 }
 425                 if err = os.Rename(dst.Name(), path); err != nil {
 426                         http.Error(w, err.Error(), http.StatusInternalServerError)
 427                         return
 428                 }
 429                 if err = WriteFileSync(dirPath, path+SHA256Ext, digest); err != nil {
 430                         http.Error(w, err.Error(), http.StatusInternalServerError)
 431                         return
 432                 }
 433         }
 434         for _, file := range r.MultipartForm.File["gpg_signature"] {
 435                 filename := file.Filename
 436                 if _, exists := gpgSigsExpected[filename]; !exists {
 437                         http.Error(w, "unexpected GPG signature filename", http.StatusBadRequest)
 438                         return
 439                 }
 440                 delete(gpgSigsExpected, filename)
 441                 log.Println(r.RemoteAddr, "put", filename, "by", username)
 442                 path := filepath.Join(dirPath, filename)
 443                 if _, err = os.Stat(path); err == nil {
 444                         log.Println(r.RemoteAddr, "already exists", filename)
 445                         http.Error(w, "already exists", http.StatusBadRequest)
 446                         return
 447                 }
 448                 src, err := file.Open()
 449                 if err != nil {
 450                         http.Error(w, err.Error(), http.StatusInternalServerError)
 451                         return
 452                 }
 453                 sig, err := ioutil.ReadAll(src)
 454                 src.Close()
 455                 if err != nil {
 456                         http.Error(w, err.Error(), http.StatusInternalServerError)
 457                         return
 458                 }
 459                 if err = WriteFileSync(dirPath, path, sig); err != nil {
 460                         http.Error(w, err.Error(), http.StatusInternalServerError)
 461                         return
 462                 }
 463         }
 464 }
 465
 466 func handler(w http.ResponseWriter, r *http.Request) {
 467         switch r.Method {
 468         case "GET":
 469                 var path string
 470                 var autorefresh bool
 471                 var gpgUpdate bool
 472                 if strings.HasPrefix(r.URL.Path, *norefreshURLPath) {
 473                         path = strings.TrimPrefix(r.URL.Path, *norefreshURLPath)
 474                 } else if strings.HasPrefix(r.URL.Path, *refreshURLPath) {
 475                         path = strings.TrimPrefix(r.URL.Path, *refreshURLPath)
 476                         autorefresh = true
 477                 } else if strings.HasPrefix(r.URL.Path, *gpgUpdateURLPath) {
 478                         path = strings.TrimPrefix(r.URL.Path, *gpgUpdateURLPath)
 479                         autorefresh = true
 480                         gpgUpdate = true
 481                 } else {
 482                         http.Error(w, "unknown action", http.StatusBadRequest)
 483                         return
 484                 }
 485                 parts := strings.Split(strings.TrimSuffix(path, "/"), "/")
 486                 if len(parts) > 2 {
 487                         http.Error(w, "invalid path", http.StatusBadRequest)
 488                         return
 489                 }
 490                 if len(parts) == 1 {
 491                         if parts[0] == "" {
 492                                 listRoot(w, r)
 493                         } else {
 494                                 listDir(w, r, parts[0], autorefresh, gpgUpdate)
 495                         }
 496                 } else {
 497                         servePkg(w, r, parts[0], parts[1])
 498                 }
 499         case "POST":
 500                 serveUpload(w, r)
 501         default:
 502                 http.Error(w, "unknown action", http.StatusBadRequest)
 503         }
 504 }
 505
 506 func goodIntegrity() bool {
 507         dirs, err := ioutil.ReadDir(*root)
 508         if err != nil {
 509                 log.Fatal(err)
 510         }
 511         hasher := sha256.New()
 512         digest := make([]byte, sha256.Size)
 513         isGood := true
 514         var data []byte
 515         var pkgName string
 516         for _, dir := range dirs {
 517                 files, err := ioutil.ReadDir(filepath.Join(*root, dir.Name()))
 518                 if err != nil {
 519                         log.Fatal(err)
 520                 }
 521                 for _, file := range files {
 522                         if !strings.HasSuffix(file.Name(), SHA256Ext) {
 523                                 continue
 524                         }
 525                         pkgName = strings.TrimSuffix(file.Name(), SHA256Ext)
 526                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), pkgName))
 527                         if err != nil {
 528                                 if os.IsNotExist(err) {
 529                                         continue
 530                                 }
 531                                 log.Fatal(err)
 532                         }
 533                         hasher.Write(data)
 534                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), file.Name()))
 535                         if err != nil {
 536                                 log.Fatal(err)
 537                         }
 538                         if bytes.Compare(hasher.Sum(digest[:0]), data) == 0 {
 539                                 log.Println(pkgName, "GOOD")
 540                         } else {
 541                                 isGood = false
 542                                 log.Println(pkgName, "BAD")
 543                         }
 544                         hasher.Reset()
 545                 }
 546         }
 547         return isGood
 548 }
 549
 550 func main() {
 551         flag.Parse()
 552         if *warranty {
 553                 fmt.Println(Warranty)
 554                 return
 555         }
 556         if *version {
 557                 fmt.Println("GoCheese version " + Version + " built with " + runtime.Version())
 558                 return
 559         }
 560         if *fsck {
 561                 if !goodIntegrity() {
 562                         os.Exit(1)
 563                 }
 564                 return
 565         }
 566         if *passwdCheck {
 567                 refreshPasswd()
 568                 return
 569         }
 570         if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
 571                 log.Fatalln("Both -tls-cert and -tls-key are required")
 572         }
 573         refreshPasswd()
 574         log.Println("root:", *root, "bind:", *bind)
 575
 576         ln, err := net.Listen("tcp", *bind)
 577         if err != nil {
 578                 log.Fatal(err)
 579         }
 580         ln = netutil.LimitListener(ln, *maxClients)
 581         server := &http.Server{
 582                 ReadTimeout:  time.Minute,
 583                 WriteTimeout: time.Minute,
 584         }
 585         http.HandleFunc(*norefreshURLPath, handler)
 586         http.HandleFunc(*refreshURLPath, handler)
 587         http.HandleFunc(*gpgUpdateURLPath, handler)
 588
 589         needsRefreshPasswd := make(chan os.Signal, 0)
 590         needsShutdown := make(chan os.Signal, 0)
 591         exitErr := make(chan error, 0)
 592         signal.Notify(needsRefreshPasswd, syscall.SIGHUP)
 593         signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
 594         go func() {
 595                 for range needsRefreshPasswd {
 596                         log.Println("Refreshing passwords")
 597                         refreshPasswd()
 598                 }
 599         }()
 600         go func(s *http.Server) {
 601                 <-needsShutdown
 602                 killed = true
 603                 log.Println("Shutting down")
 604                 ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
 605                 exitErr <- s.Shutdown(ctx)
 606                 cancel()
 607         }(server)
 608
 609         if *tlsCert == "" {
 610                 err = server.Serve(ln)
 611         } else {
 612                 err = server.ServeTLS(ln, *tlsCert, *tlsKey)
 613         }
 614         if err != http.ErrServerClosed {
 615                 log.Fatal(err)
 616         }
 617         if err := <-exitErr; err != nil {
 618                 log.Fatal(err)
 619         }
 620 }