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         username, password, ok := r.BasicAuth()
 332         if !ok {
 333                 log.Println(r.RemoteAddr, "unauthenticated", username)
 334                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 335                 return
 336         }
 337         auther, ok := passwords[username]
 338         if !ok || !auther.Auth(password) {
 339                 log.Println(r.RemoteAddr, "unauthenticated", username)
 340                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 341                 return
 342         }
 343         var err error
 344         if err = r.ParseMultipartForm(1 << 20); err != nil {
 345                 http.Error(w, err.Error(), http.StatusBadRequest)
 346                 return
 347         }
 348         pkgNames, exists := r.MultipartForm.Value["name"]
 349         if !exists || len(pkgNames) != 1 {
 350                 http.Error(w, "name is expected in request", http.StatusBadRequest)
 351                 return
 352         }
 353         dir := normalizationRe.ReplaceAllString(pkgNames[0], "-")
 354         dirPath := filepath.Join(*root, dir)
 355         var digestExpected []byte
 356         if digestExpectedHex, exists := r.MultipartForm.Value["sha256_digest"]; exists {
 357                 digestExpected, err = hex.DecodeString(digestExpectedHex[0])
 358                 if err != nil {
 359                         http.Error(w, "bad sha256_digest: "+err.Error(), http.StatusBadRequest)
 360                         return
 361                 }
 362         }
 363         gpgSigsExpected := make(map[string]struct{})
 364         for _, file := range r.MultipartForm.File["content"] {
 365                 filename := file.Filename
 366                 gpgSigsExpected[filename+GPGSigExt] = struct{}{}
 367                 log.Println(r.RemoteAddr, "put", filename, "by", username)
 368                 path := filepath.Join(dirPath, filename)
 369                 if _, err = os.Stat(path); err == nil {
 370                         log.Println(r.RemoteAddr, "already exists", filename)
 371                         http.Error(w, "Already exists", http.StatusBadRequest)
 372                         return
 373                 }
 374                 if !mkdirForPkg(w, r, dir) {
 375                         return
 376                 }
 377                 internalPath := filepath.Join(dirPath, InternalFlag)
 378                 var dst *os.File
 379                 if _, err = os.Stat(internalPath); os.IsNotExist(err) {
 380                         if dst, err = os.Create(internalPath); err != nil {
 381                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 382                                 return
 383                         }
 384                         dst.Close()
 385                 }
 386                 src, err := file.Open()
 387                 defer src.Close()
 388                 if err != nil {
 389                         http.Error(w, err.Error(), http.StatusInternalServerError)
 390                         return
 391                 }
 392                 dst, err = TempFile(dirPath)
 393                 if err != nil {
 394                         http.Error(w, err.Error(), http.StatusInternalServerError)
 395                         return
 396                 }
 397                 hasher := sha256.New()
 398                 wr := io.MultiWriter(hasher, dst)
 399                 if _, err = io.Copy(wr, src); err != nil {
 400                         os.Remove(dst.Name())
 401                         dst.Close()
 402                         http.Error(w, err.Error(), http.StatusInternalServerError)
 403                         return
 404                 }
 405                 if err = dst.Sync(); err != nil {
 406                         os.Remove(dst.Name())
 407                         dst.Close()
 408                         http.Error(w, err.Error(), http.StatusInternalServerError)
 409                         return
 410                 }
 411                 dst.Close()
 412                 digest := hasher.Sum(nil)
 413                 if digestExpected != nil {
 414                         if bytes.Compare(digestExpected, digest) == 0 {
 415                                 log.Println(r.RemoteAddr, filename, "good checksum received")
 416                         } else {
 417                                 log.Println(r.RemoteAddr, filename, "bad checksum received")
 418                                 http.Error(w, "bad checksum", http.StatusBadRequest)
 419                                 os.Remove(dst.Name())
 420                                 return
 421                         }
 422                 }
 423                 if err = os.Rename(dst.Name(), path); err != nil {
 424                         http.Error(w, err.Error(), http.StatusInternalServerError)
 425                         return
 426                 }
 427                 if err = WriteFileSync(dirPath, path+SHA256Ext, digest); err != nil {
 428                         http.Error(w, err.Error(), http.StatusInternalServerError)
 429                         return
 430                 }
 431         }
 432         for _, file := range r.MultipartForm.File["gpg_signature"] {
 433                 filename := file.Filename
 434                 if _, exists := gpgSigsExpected[filename]; !exists {
 435                         http.Error(w, "unexpected GPG signature filename", http.StatusBadRequest)
 436                         return
 437                 }
 438                 delete(gpgSigsExpected, filename)
 439                 log.Println(r.RemoteAddr, "put", filename, "by", username)
 440                 path := filepath.Join(dirPath, filename)
 441                 if _, err = os.Stat(path); err == nil {
 442                         log.Println(r.RemoteAddr, "already exists", filename)
 443                         http.Error(w, "Already exists", http.StatusBadRequest)
 444                         return
 445                 }
 446                 src, err := file.Open()
 447                 if err != nil {
 448                         http.Error(w, err.Error(), http.StatusInternalServerError)
 449                         return
 450                 }
 451                 sig, err := ioutil.ReadAll(src)
 452                 src.Close()
 453                 if err != nil {
 454                         http.Error(w, err.Error(), http.StatusInternalServerError)
 455                         return
 456                 }
 457                 if err = WriteFileSync(dirPath, path, sig); err != nil {
 458                         http.Error(w, err.Error(), http.StatusInternalServerError)
 459                         return
 460                 }
 461         }
 462 }
 463
 464 func handler(w http.ResponseWriter, r *http.Request) {
 465         switch r.Method {
 466         case "GET":
 467                 var path string
 468                 var autorefresh bool
 469                 var gpgUpdate bool
 470                 if strings.HasPrefix(r.URL.Path, *norefreshURLPath) {
 471                         path = strings.TrimPrefix(r.URL.Path, *norefreshURLPath)
 472                 } else if strings.HasPrefix(r.URL.Path, *refreshURLPath) {
 473                         path = strings.TrimPrefix(r.URL.Path, *refreshURLPath)
 474                         autorefresh = true
 475                 } else if strings.HasPrefix(r.URL.Path, *gpgUpdateURLPath) {
 476                         path = strings.TrimPrefix(r.URL.Path, *gpgUpdateURLPath)
 477                         autorefresh = true
 478                         gpgUpdate = true
 479                 } else {
 480                         http.Error(w, "unknown action", http.StatusBadRequest)
 481                         return
 482                 }
 483                 parts := strings.Split(strings.TrimSuffix(path, "/"), "/")
 484                 if len(parts) > 2 {
 485                         http.Error(w, "invalid path", http.StatusBadRequest)
 486                         return
 487                 }
 488                 if len(parts) == 1 {
 489                         if parts[0] == "" {
 490                                 listRoot(w, r)
 491                         } else {
 492                                 listDir(w, r, parts[0], autorefresh, gpgUpdate)
 493                         }
 494                 } else {
 495                         servePkg(w, r, parts[0], parts[1])
 496                 }
 497         case "POST":
 498                 serveUpload(w, r)
 499         default:
 500                 http.Error(w, "unknown action", http.StatusBadRequest)
 501         }
 502 }
 503
 504 func goodIntegrity() bool {
 505         dirs, err := ioutil.ReadDir(*root)
 506         if err != nil {
 507                 log.Fatal(err)
 508         }
 509         hasher := sha256.New()
 510         digest := make([]byte, sha256.Size)
 511         isGood := true
 512         var data []byte
 513         var pkgName string
 514         for _, dir := range dirs {
 515                 files, err := ioutil.ReadDir(filepath.Join(*root, dir.Name()))
 516                 if err != nil {
 517                         log.Fatal(err)
 518                 }
 519                 for _, file := range files {
 520                         if !strings.HasSuffix(file.Name(), SHA256Ext) {
 521                                 continue
 522                         }
 523                         pkgName = strings.TrimSuffix(file.Name(), SHA256Ext)
 524                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), pkgName))
 525                         if err != nil {
 526                                 if os.IsNotExist(err) {
 527                                         continue
 528                                 }
 529                                 log.Fatal(err)
 530                         }
 531                         hasher.Write(data)
 532                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), file.Name()))
 533                         if err != nil {
 534                                 log.Fatal(err)
 535                         }
 536                         if bytes.Compare(hasher.Sum(digest[:0]), data) == 0 {
 537                                 log.Println(pkgName, "GOOD")
 538                         } else {
 539                                 isGood = false
 540                                 log.Println(pkgName, "BAD")
 541                         }
 542                         hasher.Reset()
 543                 }
 544         }
 545         return isGood
 546 }
 547
 548 func main() {
 549         flag.Parse()
 550         if *warranty {
 551                 fmt.Println(Warranty)
 552                 return
 553         }
 554         if *version {
 555                 fmt.Println("GoCheese version " + Version + " built with " + runtime.Version())
 556                 return
 557         }
 558         if *fsck {
 559                 if !goodIntegrity() {
 560                         os.Exit(1)
 561                 }
 562                 return
 563         }
 564         if *passwdCheck {
 565                 refreshPasswd()
 566                 return
 567         }
 568         if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
 569                 log.Fatalln("Both -tls-cert and -tls-key are required")
 570         }
 571         refreshPasswd()
 572         log.Println("root:", *root, "bind:", *bind)
 573
 574         ln, err := net.Listen("tcp", *bind)
 575         if err != nil {
 576                 log.Fatal(err)
 577         }
 578         ln = netutil.LimitListener(ln, *maxClients)
 579         server := &http.Server{
 580                 ReadTimeout:  time.Minute,
 581                 WriteTimeout: time.Minute,
 582         }
 583         http.HandleFunc(*norefreshURLPath, handler)
 584         http.HandleFunc(*refreshURLPath, handler)
 585         http.HandleFunc(*gpgUpdateURLPath, handler)
 586
 587         needsRefreshPasswd := make(chan os.Signal, 0)
 588         needsShutdown := make(chan os.Signal, 0)
 589         exitErr := make(chan error, 0)
 590         signal.Notify(needsRefreshPasswd, syscall.SIGHUP)
 591         signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
 592         go func() {
 593                 for range needsRefreshPasswd {
 594                         log.Println("Refreshing passwords")
 595                         refreshPasswd()
 596                 }
 597         }()
 598         go func(s *http.Server) {
 599                 <-needsShutdown
 600                 killed = true
 601                 log.Println("Shutting down")
 602                 ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
 603                 exitErr <- s.Shutdown(ctx)
 604                 cancel()
 605         }(server)
 606
 607         if *tlsCert == "" {
 608                 err = server.Serve(ln)
 609         } else {
 610                 err = server.ServeTLS(ln, *tlsCert, *tlsKey)
 611         }
 612         if err != http.ErrServerClosed {
 613                 log.Fatal(err)
 614         }
 615         if err := <-exitErr; err != nil {
 616                 log.Fatal(err)
 617         }
 618 }