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><html><head><title>Links for %s</title></head><body><h1>Links for %s</h1>\n"
  49         HTMLEnd      = "</body></html>"
  50         HTMLElement  = "<a href='%s'>%s</a><br/>\n"
  51         SHA256Prefix = "sha256="
  52         SHA256Ext    = ".sha256"
  53         InternalFlag = ".internal"
  54
  55         Warranty = `This program is free software: you can redistribute it and/or modify
  56 it under the terms of the GNU General Public License as published by
  57 the Free Software Foundation, version 3 of the License.
  58
  59 This program is distributed in the hope that it will be useful,
  60 but WITHOUT ANY WARRANTY; without even the implied warranty of
  61 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  62 GNU General Public License for more details.
  63
  64 You should have received a copy of the GNU General Public License
  65 along with this program.  If not, see <http://www.gnu.org/licenses/>.`
  66 )
  67
  68 var (
  69         pkgPyPI        = regexp.MustCompile(`^.*<a href="([^"]+)"[^>]*>(.+)</a><br/>.*$`)
  70         Version string = "UNKNOWN"
  71
  72         root             = flag.String("root", "./packages", "Path to packages directory")
  73         bind             = flag.String("bind", "[::]:8080", "Address to bind to")
  74         tlsCert          = flag.String("tls-cert", "", "Path to TLS X.509 certificate")
  75         tlsKey           = flag.String("tls-key", "", "Path to TLS X.509 private key")
  76         norefreshURLPath = flag.String("norefresh", "/norefresh/", "Non-refreshing URL path")
  77         refreshURLPath   = flag.String("refresh", "/simple/", "Auto-refreshing URL path")
  78         pypiURL          = flag.String("pypi", "https://pypi.org/simple/", "Upstream PyPI URL")
  79         passwdPath       = flag.String("passwd", "passwd", "Path to file with authenticators")
  80         passwdCheck      = flag.Bool("passwd-check", false, "Test the -passwd file for syntax errors and exit")
  81         fsck             = flag.Bool("fsck", false, "Check integrity of all packages")
  82         maxClients       = flag.Int("maxclients", 128, "Maximal amount of simultaneous clients")
  83         version          = flag.Bool("version", false, "Print version information")
  84         warranty         = flag.Bool("warranty", false, "Print warranty information")
  85 )
  86
  87 func mkdirForPkg(w http.ResponseWriter, r *http.Request, dir string) bool {
  88         path := filepath.Join(*root, dir)
  89         if _, err := os.Stat(path); os.IsNotExist(err) {
  90                 if err = os.Mkdir(path, os.FileMode(0777)); err != nil {
  91                         http.Error(w, err.Error(), http.StatusInternalServerError)
  92                         return false
  93                 }
  94                 log.Println(r.RemoteAddr, "mkdir", dir)
  95         }
  96         return true
  97 }
  98
  99 func refreshDir(w http.ResponseWriter, r *http.Request, dir, filenameGet string) bool {
 100         if _, err := os.Stat(filepath.Join(*root, dir, InternalFlag)); err == nil {
 101                 log.Println(r.RemoteAddr, "pypi refresh skip, internal package", dir)
 102                 return true
 103         }
 104         log.Println(r.RemoteAddr, "pypi refresh", dir)
 105         resp, err := http.Get(*pypiURL + dir + "/")
 106         if err != nil {
 107                 http.Error(w, err.Error(), http.StatusBadGateway)
 108                 return false
 109         }
 110         defer resp.Body.Close()
 111         body, err := ioutil.ReadAll(resp.Body)
 112         if err != nil {
 113                 http.Error(w, err.Error(), http.StatusBadGateway)
 114                 return false
 115         }
 116         if !mkdirForPkg(w, r, dir) {
 117                 return false
 118         }
 119         var submatches []string
 120         var uri string
 121         var filename string
 122         var path string
 123         var pkgURL *url.URL
 124         var digest []byte
 125         for _, lineRaw := range bytes.Split(body, []byte("\n")) {
 126                 submatches = pkgPyPI.FindStringSubmatch(string(lineRaw))
 127                 if len(submatches) == 0 {
 128                         continue
 129                 }
 130                 uri = submatches[1]
 131                 filename = submatches[2]
 132                 if pkgURL, err = url.Parse(uri); err != nil {
 133                         http.Error(w, err.Error(), http.StatusInternalServerError)
 134                         return false
 135                 }
 136                 digest, err = hex.DecodeString(strings.TrimPrefix(pkgURL.Fragment, SHA256Prefix))
 137                 if err != nil {
 138                         http.Error(w, err.Error(), http.StatusBadGateway)
 139                         return false
 140                 }
 141                 if filename == filenameGet {
 142                         log.Println(r.RemoteAddr, "pypi download", filename)
 143                         path = filepath.Join(*root, dir, filename)
 144                         resp, err = http.Get(uri)
 145                         if err != nil {
 146                                 http.Error(w, err.Error(), http.StatusBadGateway)
 147                                 return false
 148                         }
 149                         defer resp.Body.Close()
 150                         hasher := sha256.New()
 151                         dst, err := TempFile(filepath.Join(*root, dir))
 152                         if err != nil {
 153                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 154                                 return false
 155                         }
 156                         wr := io.MultiWriter(hasher, dst)
 157                         if _, err = io.Copy(wr, resp.Body); err != nil {
 158                                 os.Remove(dst.Name())
 159                                 dst.Close()
 160                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 161                                 return false
 162                         }
 163                         if bytes.Compare(hasher.Sum(nil), digest) != 0 {
 164                                 log.Println(r.RemoteAddr, "pypi", filename, "digest mismatch")
 165                                 os.Remove(dst.Name())
 166                                 dst.Close()
 167                                 http.Error(w, err.Error(), http.StatusBadGateway)
 168                                 return false
 169                         }
 170                         if err = dst.Sync(); err != nil {
 171                                 os.Remove(dst.Name())
 172                                 dst.Close()
 173                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 174                                 return false
 175                         }
 176                         dst.Close()
 177                         if err = os.Rename(dst.Name(), path); err != nil {
 178                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 179                                 return false
 180                         }
 181                 }
 182                 path = filepath.Join(*root, dir, filename+SHA256Ext)
 183                 _, err = os.Stat(path)
 184                 if err == nil {
 185                         continue
 186                 } else {
 187                         if !os.IsNotExist(err) {
 188                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 189                                 return false
 190                         }
 191                 }
 192                 log.Println(r.RemoteAddr, "pypi touch", filename)
 193                 if err = ioutil.WriteFile(path, digest, os.FileMode(0666)); err != nil {
 194                         http.Error(w, err.Error(), http.StatusInternalServerError)
 195                         return false
 196                 }
 197         }
 198         return true
 199 }
 200
 201 func listRoot(w http.ResponseWriter, r *http.Request) {
 202         log.Println(r.RemoteAddr, "root")
 203         files, err := ioutil.ReadDir(*root)
 204         if err != nil {
 205                 http.Error(w, err.Error(), http.StatusInternalServerError)
 206                 return
 207         }
 208         w.Write([]byte(fmt.Sprintf(HTMLBegin, "root", "root")))
 209         for _, file := range files {
 210                 if file.Mode().IsDir() {
 211                         w.Write([]byte(fmt.Sprintf(
 212                                 HTMLElement,
 213                                 *refreshURLPath+file.Name()+"/",
 214                                 file.Name(),
 215                         )))
 216                 }
 217         }
 218         w.Write([]byte(HTMLEnd))
 219 }
 220
 221 func listDir(w http.ResponseWriter, r *http.Request, dir string, autorefresh bool) {
 222         log.Println(r.RemoteAddr, "dir", dir)
 223         dirPath := filepath.Join(*root, dir)
 224         if autorefresh {
 225                 if !refreshDir(w, r, dir, "") {
 226                         return
 227                 }
 228         } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, dir, "") {
 229                 return
 230         }
 231         files, err := ioutil.ReadDir(dirPath)
 232         if err != nil {
 233                 http.Error(w, err.Error(), http.StatusInternalServerError)
 234                 return
 235         }
 236         w.Write([]byte(fmt.Sprintf(HTMLBegin, dir, dir)))
 237         var data []byte
 238         var filenameClean string
 239         for _, file := range files {
 240                 if !strings.HasSuffix(file.Name(), SHA256Ext) {
 241                         continue
 242                 }
 243                 data, err = ioutil.ReadFile(filepath.Join(dirPath, file.Name()))
 244                 if err != nil {
 245                         http.Error(w, err.Error(), http.StatusInternalServerError)
 246                         return
 247                 }
 248                 filenameClean = strings.TrimSuffix(file.Name(), SHA256Ext)
 249                 w.Write([]byte(fmt.Sprintf(
 250                         HTMLElement,
 251                         strings.Join([]string{
 252                                 *refreshURLPath, dir, "/",
 253                                 filenameClean, "#", SHA256Prefix, string(data),
 254                         }, ""),
 255                         filenameClean,
 256                 )))
 257         }
 258         w.Write([]byte(HTMLEnd))
 259 }
 260
 261 func servePkg(w http.ResponseWriter, r *http.Request, dir, filename string) {
 262         log.Println(r.RemoteAddr, "pkg", filename)
 263         path := filepath.Join(*root, dir, filename)
 264         if _, err := os.Stat(path); os.IsNotExist(err) {
 265                 if !refreshDir(w, r, dir, filename) {
 266                         return
 267                 }
 268         }
 269         http.ServeFile(w, r, path)
 270 }
 271
 272 func serveUpload(w http.ResponseWriter, r *http.Request) {
 273         username, password, ok := r.BasicAuth()
 274         if !ok {
 275                 log.Println(r.RemoteAddr, "unauthenticated", username)
 276                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 277                 return
 278         }
 279         auther, ok := passwords[username]
 280         if !ok || !auther.Auth(password) {
 281                 log.Println(r.RemoteAddr, "unauthenticated", username)
 282                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 283                 return
 284         }
 285         var err error
 286         if err = r.ParseMultipartForm(1 << 20); err != nil {
 287                 http.Error(w, err.Error(), http.StatusBadRequest)
 288                 return
 289         }
 290         for _, file := range r.MultipartForm.File["content"] {
 291                 filename := file.Filename
 292                 log.Println(r.RemoteAddr, "upload", filename, "by", username)
 293                 dir := filename[:strings.LastIndex(filename, "-")]
 294                 dirPath := filepath.Join(*root, dir)
 295                 path := filepath.Join(dirPath, filename)
 296                 if _, err = os.Stat(path); err == nil {
 297                         log.Println(r.RemoteAddr, "already exists", filename)
 298                         http.Error(w, "Already exists", http.StatusBadRequest)
 299                         return
 300                 }
 301                 if !mkdirForPkg(w, r, dir) {
 302                         return
 303                 }
 304                 internalPath := filepath.Join(dirPath, InternalFlag)
 305                 var dst *os.File
 306                 if _, err = os.Stat(internalPath); os.IsNotExist(err) {
 307                         if dst, err = os.Create(internalPath); err != nil {
 308                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 309                                 return
 310                         }
 311                         dst.Close()
 312                 }
 313                 src, err := file.Open()
 314                 defer src.Close()
 315                 if err != nil {
 316                         http.Error(w, err.Error(), http.StatusInternalServerError)
 317                         return
 318                 }
 319                 dst, err = TempFile(dirPath)
 320                 if err != nil {
 321                         http.Error(w, err.Error(), http.StatusInternalServerError)
 322                         return
 323                 }
 324                 hasher := sha256.New()
 325                 wr := io.MultiWriter(hasher, dst)
 326                 if _, err = io.Copy(wr, src); err != nil {
 327                         os.Remove(dst.Name())
 328                         dst.Close()
 329                         http.Error(w, err.Error(), http.StatusInternalServerError)
 330                         return
 331                 }
 332                 if err = dst.Sync(); err != nil {
 333                         os.Remove(dst.Name())
 334                         dst.Close()
 335                         http.Error(w, err.Error(), http.StatusInternalServerError)
 336                         return
 337                 }
 338                 dst.Close()
 339                 if err = os.Rename(dst.Name(), path); err != nil {
 340                         http.Error(w, err.Error(), http.StatusInternalServerError)
 341                         return
 342                 }
 343                 if err = ioutil.WriteFile(path+SHA256Ext, hasher.Sum(nil), os.FileMode(0666)); err != nil {
 344                         http.Error(w, err.Error(), http.StatusInternalServerError)
 345                         return
 346                 }
 347         }
 348 }
 349
 350 func handler(w http.ResponseWriter, r *http.Request) {
 351         if r.Method == "GET" {
 352                 var path string
 353                 var autorefresh bool
 354                 if strings.HasPrefix(r.URL.Path, *norefreshURLPath) {
 355                         path = strings.TrimPrefix(r.URL.Path, *norefreshURLPath)
 356                         autorefresh = false
 357                 } else {
 358                         path = strings.TrimPrefix(r.URL.Path, *refreshURLPath)
 359                         autorefresh = true
 360                 }
 361                 parts := strings.Split(strings.TrimSuffix(path, "/"), "/")
 362                 if len(parts) > 2 {
 363                         http.Error(w, "invalid path", http.StatusBadRequest)
 364                         return
 365                 }
 366                 if len(parts) == 1 {
 367                         if parts[0] == "" {
 368                                 listRoot(w, r)
 369                         } else {
 370                                 listDir(w, r, parts[0], autorefresh)
 371                         }
 372                 } else {
 373                         servePkg(w, r, parts[0], parts[1])
 374                 }
 375         } else if r.Method == "POST" {
 376                 serveUpload(w, r)
 377         }
 378 }
 379
 380 func goodIntegrity() bool {
 381         dirs, err := ioutil.ReadDir(*root)
 382         if err != nil {
 383                 log.Fatal(err)
 384         }
 385         hasher := sha256.New()
 386         digest := make([]byte, sha256.Size)
 387         isGood := true
 388         var data []byte
 389         var pkgName string
 390         for _, dir := range dirs {
 391                 files, err := ioutil.ReadDir(filepath.Join(*root, dir.Name()))
 392                 if err != nil {
 393                         log.Fatal(err)
 394                 }
 395                 for _, file := range files {
 396                         if !strings.HasSuffix(file.Name(), SHA256Ext) {
 397                                 continue
 398                         }
 399                         pkgName = strings.TrimSuffix(file.Name(), SHA256Ext)
 400                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), pkgName))
 401                         if err != nil {
 402                                 if os.IsNotExist(err) {
 403                                         continue
 404                                 }
 405                                 log.Fatal(err)
 406                         }
 407                         hasher.Write(data)
 408                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), file.Name()))
 409                         if err != nil {
 410                                 log.Fatal(err)
 411                         }
 412                         if bytes.Compare(hasher.Sum(digest[:0]), data) == 0 {
 413                                 log.Println(pkgName, "GOOD")
 414                         } else {
 415                                 isGood = false
 416                                 log.Println(pkgName, "BAD")
 417                         }
 418                         hasher.Reset()
 419                 }
 420         }
 421         return isGood
 422 }
 423
 424 func main() {
 425         flag.Parse()
 426         if *warranty {
 427                 fmt.Println(Warranty)
 428                 return
 429         }
 430         if *version {
 431                 fmt.Println("GoCheese version " + Version + " built with " + runtime.Version())
 432                 return
 433         }
 434         if *fsck {
 435                 if !goodIntegrity() {
 436                         os.Exit(1)
 437                 }
 438                 return
 439         }
 440         if *passwdCheck {
 441                 refreshPasswd()
 442                 return
 443         }
 444         if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
 445                 log.Fatalln("Both -tls-cert and -tls-key are required")
 446         }
 447         refreshPasswd()
 448         log.Println("root:", *root, "bind:", *bind)
 449
 450         ln, err := net.Listen("tcp", *bind)
 451         if err != nil {
 452                 log.Fatal(err)
 453         }
 454         ln = netutil.LimitListener(ln, *maxClients)
 455         server := &http.Server{
 456                 ReadTimeout:  time.Minute,
 457                 WriteTimeout: time.Minute,
 458         }
 459         http.HandleFunc(*norefreshURLPath, handler)
 460         http.HandleFunc(*refreshURLPath, handler)
 461
 462         needsRefreshPasswd := make(chan os.Signal, 0)
 463         needsShutdown := make(chan os.Signal, 0)
 464         killed := make(chan error, 0)
 465         signal.Notify(needsRefreshPasswd, syscall.SIGHUP)
 466         signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
 467         go func() {
 468                 for range needsRefreshPasswd {
 469                         log.Println("Refreshing passwords")
 470                         refreshPasswd()
 471                 }
 472         }()
 473         go func(s *http.Server) {
 474                 <-needsShutdown
 475                 log.Println("Shutting down")
 476                 ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
 477                 killed <- s.Shutdown(ctx)
 478                 cancel()
 479         }(server)
 480
 481         if *tlsCert == "" {
 482                 err = server.Serve(ln)
 483         } else {
 484                 err = server.ServeTLS(ln, *tlsCert, *tlsKey)
 485         }
 486         if err != http.ErrServerClosed {
 487                 log.Fatal(err)
 488         }
 489         if err := <-killed; err != nil {
 490                 log.Fatal(err)
 491         }
 492 }