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         body, err := ioutil.ReadAll(resp.Body)
 111         resp.Body.Close()
 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                 }
 187                 if !os.IsNotExist(err) {
 188                         http.Error(w, err.Error(), http.StatusInternalServerError)
 189                         return false
 190                 }
 191                 log.Println(r.RemoteAddr, "pypi touch", filename)
 192                 if err = ioutil.WriteFile(path, digest, os.FileMode(0666)); err != nil {
 193                         http.Error(w, err.Error(), http.StatusInternalServerError)
 194                         return false
 195                 }
 196         }
 197         return true
 198 }
 199
 200 func listRoot(w http.ResponseWriter, r *http.Request) {
 201         log.Println(r.RemoteAddr, "root")
 202         files, err := ioutil.ReadDir(*root)
 203         if err != nil {
 204                 http.Error(w, err.Error(), http.StatusInternalServerError)
 205                 return
 206         }
 207         w.Write([]byte(fmt.Sprintf(HTMLBegin, "root", "root")))
 208         for _, file := range files {
 209                 if file.Mode().IsDir() {
 210                         w.Write([]byte(fmt.Sprintf(
 211                                 HTMLElement,
 212                                 *refreshURLPath+file.Name()+"/",
 213                                 file.Name(),
 214                         )))
 215                 }
 216         }
 217         w.Write([]byte(HTMLEnd))
 218 }
 219
 220 func listDir(w http.ResponseWriter, r *http.Request, dir string, autorefresh bool) {
 221         log.Println(r.RemoteAddr, "dir", dir)
 222         dirPath := filepath.Join(*root, dir)
 223         if autorefresh {
 224                 if !refreshDir(w, r, dir, "") {
 225                         return
 226                 }
 227         } else if _, err := os.Stat(dirPath); os.IsNotExist(err) && !refreshDir(w, r, dir, "") {
 228                 return
 229         }
 230         files, err := ioutil.ReadDir(dirPath)
 231         if err != nil {
 232                 http.Error(w, err.Error(), http.StatusInternalServerError)
 233                 return
 234         }
 235         w.Write([]byte(fmt.Sprintf(HTMLBegin, dir, dir)))
 236         var data []byte
 237         var filenameClean string
 238         for _, file := range files {
 239                 if !strings.HasSuffix(file.Name(), SHA256Ext) {
 240                         continue
 241                 }
 242                 data, err = ioutil.ReadFile(filepath.Join(dirPath, file.Name()))
 243                 if err != nil {
 244                         http.Error(w, err.Error(), http.StatusInternalServerError)
 245                         return
 246                 }
 247                 filenameClean = strings.TrimSuffix(file.Name(), SHA256Ext)
 248                 w.Write([]byte(fmt.Sprintf(
 249                         HTMLElement,
 250                         strings.Join([]string{
 251                                 *refreshURLPath, dir, "/",
 252                                 filenameClean, "#", SHA256Prefix, string(data),
 253                         }, ""),
 254                         filenameClean,
 255                 )))
 256         }
 257         w.Write([]byte(HTMLEnd))
 258 }
 259
 260 func servePkg(w http.ResponseWriter, r *http.Request, dir, filename string) {
 261         log.Println(r.RemoteAddr, "pkg", filename)
 262         path := filepath.Join(*root, dir, filename)
 263         if _, err := os.Stat(path); os.IsNotExist(err) {
 264                 if !refreshDir(w, r, dir, filename) {
 265                         return
 266                 }
 267         }
 268         http.ServeFile(w, r, path)
 269 }
 270
 271 func serveUpload(w http.ResponseWriter, r *http.Request) {
 272         username, password, ok := r.BasicAuth()
 273         if !ok {
 274                 log.Println(r.RemoteAddr, "unauthenticated", username)
 275                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 276                 return
 277         }
 278         auther, ok := passwords[username]
 279         if !ok || !auther.Auth(password) {
 280                 log.Println(r.RemoteAddr, "unauthenticated", username)
 281                 http.Error(w, "unauthenticated", http.StatusUnauthorized)
 282                 return
 283         }
 284         var err error
 285         if err = r.ParseMultipartForm(1 << 20); err != nil {
 286                 http.Error(w, err.Error(), http.StatusBadRequest)
 287                 return
 288         }
 289         for _, file := range r.MultipartForm.File["content"] {
 290                 filename := file.Filename
 291                 log.Println(r.RemoteAddr, "upload", filename, "by", username)
 292                 dir := filename[:strings.LastIndex(filename, "-")]
 293                 dirPath := filepath.Join(*root, dir)
 294                 path := filepath.Join(dirPath, filename)
 295                 if _, err = os.Stat(path); err == nil {
 296                         log.Println(r.RemoteAddr, "already exists", filename)
 297                         http.Error(w, "Already exists", http.StatusBadRequest)
 298                         return
 299                 }
 300                 if !mkdirForPkg(w, r, dir) {
 301                         return
 302                 }
 303                 internalPath := filepath.Join(dirPath, InternalFlag)
 304                 var dst *os.File
 305                 if _, err = os.Stat(internalPath); os.IsNotExist(err) {
 306                         if dst, err = os.Create(internalPath); err != nil {
 307                                 http.Error(w, err.Error(), http.StatusInternalServerError)
 308                                 return
 309                         }
 310                         dst.Close()
 311                 }
 312                 src, err := file.Open()
 313                 defer src.Close()
 314                 if err != nil {
 315                         http.Error(w, err.Error(), http.StatusInternalServerError)
 316                         return
 317                 }
 318                 dst, err = TempFile(dirPath)
 319                 if err != nil {
 320                         http.Error(w, err.Error(), http.StatusInternalServerError)
 321                         return
 322                 }
 323                 hasher := sha256.New()
 324                 wr := io.MultiWriter(hasher, dst)
 325                 if _, err = io.Copy(wr, src); err != nil {
 326                         os.Remove(dst.Name())
 327                         dst.Close()
 328                         http.Error(w, err.Error(), http.StatusInternalServerError)
 329                         return
 330                 }
 331                 if err = dst.Sync(); err != nil {
 332                         os.Remove(dst.Name())
 333                         dst.Close()
 334                         http.Error(w, err.Error(), http.StatusInternalServerError)
 335                         return
 336                 }
 337                 dst.Close()
 338                 if err = os.Rename(dst.Name(), path); err != nil {
 339                         http.Error(w, err.Error(), http.StatusInternalServerError)
 340                         return
 341                 }
 342                 if err = ioutil.WriteFile(path+SHA256Ext, hasher.Sum(nil), os.FileMode(0666)); err != nil {
 343                         http.Error(w, err.Error(), http.StatusInternalServerError)
 344                         return
 345                 }
 346         }
 347 }
 348
 349 func handler(w http.ResponseWriter, r *http.Request) {
 350         if r.Method == "GET" {
 351                 var path string
 352                 var autorefresh bool
 353                 if strings.HasPrefix(r.URL.Path, *norefreshURLPath) {
 354                         path = strings.TrimPrefix(r.URL.Path, *norefreshURLPath)
 355                         autorefresh = false
 356                 } else {
 357                         path = strings.TrimPrefix(r.URL.Path, *refreshURLPath)
 358                         autorefresh = true
 359                 }
 360                 parts := strings.Split(strings.TrimSuffix(path, "/"), "/")
 361                 if len(parts) > 2 {
 362                         http.Error(w, "invalid path", http.StatusBadRequest)
 363                         return
 364                 }
 365                 if len(parts) == 1 {
 366                         if parts[0] == "" {
 367                                 listRoot(w, r)
 368                         } else {
 369                                 listDir(w, r, parts[0], autorefresh)
 370                         }
 371                 } else {
 372                         servePkg(w, r, parts[0], parts[1])
 373                 }
 374         } else if r.Method == "POST" {
 375                 serveUpload(w, r)
 376         }
 377 }
 378
 379 func goodIntegrity() bool {
 380         dirs, err := ioutil.ReadDir(*root)
 381         if err != nil {
 382                 log.Fatal(err)
 383         }
 384         hasher := sha256.New()
 385         digest := make([]byte, sha256.Size)
 386         isGood := true
 387         var data []byte
 388         var pkgName string
 389         for _, dir := range dirs {
 390                 files, err := ioutil.ReadDir(filepath.Join(*root, dir.Name()))
 391                 if err != nil {
 392                         log.Fatal(err)
 393                 }
 394                 for _, file := range files {
 395                         if !strings.HasSuffix(file.Name(), SHA256Ext) {
 396                                 continue
 397                         }
 398                         pkgName = strings.TrimSuffix(file.Name(), SHA256Ext)
 399                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), pkgName))
 400                         if err != nil {
 401                                 if os.IsNotExist(err) {
 402                                         continue
 403                                 }
 404                                 log.Fatal(err)
 405                         }
 406                         hasher.Write(data)
 407                         data, err = ioutil.ReadFile(filepath.Join(*root, dir.Name(), file.Name()))
 408                         if err != nil {
 409                                 log.Fatal(err)
 410                         }
 411                         if bytes.Compare(hasher.Sum(digest[:0]), data) == 0 {
 412                                 log.Println(pkgName, "GOOD")
 413                         } else {
 414                                 isGood = false
 415                                 log.Println(pkgName, "BAD")
 416                         }
 417                         hasher.Reset()
 418                 }
 419         }
 420         return isGood
 421 }
 422
 423 func main() {
 424         flag.Parse()
 425         if *warranty {
 426                 fmt.Println(Warranty)
 427                 return
 428         }
 429         if *version {
 430                 fmt.Println("GoCheese version " + Version + " built with " + runtime.Version())
 431                 return
 432         }
 433         if *fsck {
 434                 if !goodIntegrity() {
 435                         os.Exit(1)
 436                 }
 437                 return
 438         }
 439         if *passwdCheck {
 440                 refreshPasswd()
 441                 return
 442         }
 443         if (*tlsCert != "" && *tlsKey == "") || (*tlsCert == "" && *tlsKey != "") {
 444                 log.Fatalln("Both -tls-cert and -tls-key are required")
 445         }
 446         refreshPasswd()
 447         log.Println("root:", *root, "bind:", *bind)
 448
 449         ln, err := net.Listen("tcp", *bind)
 450         if err != nil {
 451                 log.Fatal(err)
 452         }
 453         ln = netutil.LimitListener(ln, *maxClients)
 454         server := &http.Server{
 455                 ReadTimeout:  time.Minute,
 456                 WriteTimeout: time.Minute,
 457         }
 458         http.HandleFunc(*norefreshURLPath, handler)
 459         http.HandleFunc(*refreshURLPath, handler)
 460
 461         needsRefreshPasswd := make(chan os.Signal, 0)
 462         needsShutdown := make(chan os.Signal, 0)
 463         killed := make(chan error, 0)
 464         signal.Notify(needsRefreshPasswd, syscall.SIGHUP)
 465         signal.Notify(needsShutdown, syscall.SIGTERM, syscall.SIGINT)
 466         go func() {
 467                 for range needsRefreshPasswd {
 468                         log.Println("Refreshing passwords")
 469                         refreshPasswd()
 470                 }
 471         }()
 472         go func(s *http.Server) {
 473                 <-needsShutdown
 474                 log.Println("Shutting down")
 475                 ctx, cancel := context.WithTimeout(context.TODO(), time.Minute)
 476                 killed <- s.Shutdown(ctx)
 477                 cancel()
 478         }(server)
 479
 480         if *tlsCert == "" {
 481                 err = server.Serve(ln)
 482         } else {
 483                 err = server.ServeTLS(ln, *tlsCert, *tlsKey)
 484         }
 485         if err != http.ErrServerClosed {
 486                 log.Fatal(err)
 487         }
 488         if err := <-killed; err != nil {
 489                 log.Fatal(err)
 490         }
 491 }