Add pdf pages to zip

cozy.example.yaml

@@ -119,6 +119,9 @@ jobs:
  # path to the imagemagick convert binary
  # imagemagick_convert_cmd: convert
 
+ # path to the ghostscript binary
+ # ghostscript_cmd: gs
+
  # Specify whether the given list of jobs is an allowlist or blocklist. In case
  # of an allowlist, all jobs are deactivated by default and only the listed one
  # are activated.

docs/files.md

@@ -1403,6 +1403,10 @@ sub-directories in the archive.
 It's possible to give a file by its id (in the `ids` array) or by its path (in
 the `files` array).
 
+For PDF files, it's possible to put in the archive a single page, with the
+`pages` argument: it's an array of objects, with `id` the file identifier of
+the PDF file, and `page` the page number (1 is the first page).
+
 The generated archive is temporary and is not persisted.
 
 #### Request
@@ -1423,6 +1427,9 @@ Content-Type: application/vnd.api+json
  "/Documents/bills",
  "/Documents/images/sunset.jpg",
  "/Documents/images/eiffel-tower.jpg"
+ ],
+ "pages": [
+ { "id": "3780caf0-104f-013d-3619-18c04daba326", "page": 1 }
  ]
  }
  }

docs/workers.md

@@ -144,6 +144,9 @@ creates a zip archive from files in the VFS. The options are:
 - `dir_id`: the directory identifier where the zip archive will be put
 - `filename`: the name of the zip archive.
 
+**Note:** it is possible to include only a page for a PDF file, by using an
+object with `id` and `page` instead of just the file identifier.
+
 ### Example
 
 ```json
@@ -153,7 +156,9 @@ creates a zip archive from files in the VFS. The options are:
  "selection/two.pdf": "36eb54c8-90fe-11e9-aeca-03ddc3acf91c",
  "selection/three.pdf": "37284586-90fe-11e9-be6d-179f72076e43",
  "selection/four.pdf": "37655462-90fe-11e9-9059-8739e3746720",
- "selection/five.pdf": "379fedfc-90fe-11e9-849f-0bbe172eba5f"
+ "selection/five.pdf": "379fedfc-90fe-11e9-849f-0bbe172eba5f",
+ "selection/front.pdf": { "id": "49ca9e50-1074-013d-361a-18c04daba326", "page": 1 },
+ "selection/back.pdf": { "id": "49ca9e50-1074-013d-361a-18c04daba326", "page": 2 }
  },
  "dir_id": "3657ce9c-90fe-11e9-b40b-33baf841bcb8",
  "filename": "selection.zip"

model/vfs/archive.go

@@ -9,6 +9,7 @@ import (
  "path/filepath"
  "strings"
 
+ "github.com/cozy/cozy-stack/pkg/config/config"
  "github.com/cozy/cozy-stack/pkg/consts"
  "github.com/cozy/cozy-stack/pkg/couchdb"
  "github.com/labstack/echo/v4"
@@ -23,17 +24,24 @@ type Archive struct {
  Secret string `json:"-"`
  IDs []string `json:"ids"`
  Files []string `json:"files"`
+ Pages []Page `json:"pages"`
 
  // archiveEntries cache
  entries []ArchiveEntry
 }
 
+type Page struct {
+ ID string `json:"id"`
+ Page int `json:"page"`
+}
+
 // ArchiveEntry is an utility struct to store a file or doc to be placed
 // in the archive.
 type ArchiveEntry struct {
  root string
  Dir *DirDoc
  File *FileDoc
+ Page int
 }
 
 var plusEscaper = strings.NewReplacer("+", "%20")
@@ -66,9 +74,8 @@ func ContentDisposition(disposition, filename string) string {
 // GetEntries returns all files and folders in the archive as ArchiveEntry.
 func (a *Archive) GetEntries(fs VFS) ([]ArchiveEntry, error) {
  if a.entries == nil {
- n := len(a.IDs)
- entries := make([]ArchiveEntry, n+len(a.Files))
- for i, id := range a.IDs {
+ entries := make([]ArchiveEntry, 0, len(a.IDs)+len(a.Files)+len(a.Pages))
+ for _, id := range a.IDs {
  d, f, err := fs.DirOrFileByID(id)
  if err != nil {
  return nil, err
@@ -82,22 +89,37 @@ func (a *Archive) GetEntries(fs VFS) ([]ArchiveEntry, error) {
  return nil, err
  }
  }
- entries[i] = ArchiveEntry{
+ entries = append(entries, ArchiveEntry{
  root: root,
  Dir: d,
  File: f,
- }
+ })
  }
- for i, root := range a.Files {
+ for _, root := range a.Files {
  d, f, err := fs.DirOrFileByPath(root)
  if err != nil {
  return nil, err
  }
- entries[n+i] = ArchiveEntry{
+ entries = append(entries, ArchiveEntry{
  root: root,
  Dir: d,
  File: f,
+ })
+ }
+ for _, page := range a.Pages {
+ f, err := fs.FileByID(page.ID)
+ if err != nil {
+ return nil, err
  }
+ root, err := f.Path(fs)
+ if err != nil {
+ return nil, err
+ }
+ entries = append(entries, ArchiveEntry{
+ root: root,
+ File: f,
+ Page: page.Page,
+ })
  }
 
  a.entries = entries
@@ -140,6 +162,9 @@ func (a *Archive) Serve(fs VFS, w http.ResponseWriter) error {
  Method: zip.Deflate,
  Modified: file.UpdatedAt,
  }
+ if entry.Page >= 0 {
+ header.Name = addPageToName(header.Name, entry.Page)
+ }
  ze, err := zw.CreateHeader(header)
  if err != nil {
  return fmt.Errorf("Can't create zip entry <%s>: %s", name, err)
@@ -149,7 +174,16 @@ func (a *Archive) Serve(fs VFS, w http.ResponseWriter) error {
  return fmt.Errorf("Can't open file <%s>: %s", name, err)
  }
  defer f.Close()
- _, err = io.Copy(ze, f)
+
+ if entry.Page <= 0 {
+ _, err = io.Copy(ze, f)
+ return err
+ }
+ extracted, err := config.PDF().ExtractPage(f, entry.Page)
+ if err != nil {
+ return err
+ }
+ _, err = io.Copy(ze, extracted)
  return err
  }, 0)
  if err != nil {
@@ -189,3 +223,9 @@ func (a *Archive) SetID(_ string) {}
 
 // SetRev makes Archive a jsonapi.Object
 func (a *Archive) SetRev(_ string) {}
+
+func addPageToName(name string, page int) string {
+ ext := filepath.Ext(name)
+ basename := strings.TrimSuffix(name, ext)
+ return fmt.Sprintf("%s (%d)%s", basename, page, ext)
+}

model/vfs/archive_test.go

@@ -0,0 +1,12 @@
+package vfs
+
+import (
+ "testing"
+
+ "github.com/stretchr/testify/require"
+)
+
+func TestAddPageToName(t *testing.T) {
+ name := addPageToName("driving licence.pdf", 2)
+ require.Equal(t, "driving licence (2).pdf", name)
+}

pkg/config/config/config.go

@@ -29,6 +29,7 @@ import (
  "github.com/cozy/cozy-stack/pkg/limits"
  "github.com/cozy/cozy-stack/pkg/lock"
  "github.com/cozy/cozy-stack/pkg/logger"
+ "github.com/cozy/cozy-stack/pkg/pdf"
  "github.com/cozy/cozy-stack/pkg/tlsclient"
  "github.com/cozy/cozy-stack/pkg/utils"
  "github.com/cozy/gomail"
@@ -116,6 +117,7 @@ type Config struct {
  AuthorizedForConfirm []string
 
  Avatars *avatar.Service
+ PDF *pdf.Service
  Fs Fs
  Keyring keyring.Keyring
  CouchDB CouchDB
@@ -376,6 +378,11 @@ func Avatars() *avatar.Service {
  return config.Avatars
 }
 
+// PDF return the configured PDF service.
+func PDF() *pdf.Service {
+ return config.PDF
+}
+
 // GetKeyring returns the configured instance of [keyring.Keyring]
 func GetKeyring() keyring.Keyring {
  return config.Keyring
@@ -485,6 +492,7 @@ func Setup(cfgFile string) (err error) {
 
 func applyDefaults(v *viper.Viper) {
  v.SetDefault("password_reset_interval", defaultPasswordResetInterval)
+ v.SetDefault("jobs.ghostscript_cmd", "gs")
  v.SetDefault("jobs.imagemagick_convert_cmd", "convert")
  v.SetDefault("jobs.defaultDurationToKeep", "2W")
  v.SetDefault("assets_polling_disabled", false)
@@ -749,6 +757,7 @@ func UseViper(v *viper.Viper) error {
 
  cacheStorage := cache.New(cacheRedis)
  avatars := avatar.NewService(cacheStorage, v.GetString("jobs.imagemagick_convert_cmd"))
+ pdfService := pdf.NewService(v.GetString("jobs.ghostscript_cmd"))
 
  // Setup keyring
  var keyringCfg keyring.Config
@@ -823,6 +832,7 @@ func UseViper(v *viper.Viper) error {
  RemoteAssets: v.GetStringMapString("remote_assets"),
 
  Avatars: avatars,
+ PDF: pdfService,
  Keyring: keyring,
  Fs: Fs{
  URL: fsURL,

pkg/pdf/pdf.go

@@ -0,0 +1,50 @@
+// Package pdf is for manipulating PDF files.
+package pdf
+
+import (
+ "bytes"
+ "fmt"
+ "io"
+ "os/exec"
+
+ "github.com/cozy/cozy-stack/pkg/logger"
+)
+
+// Service provides methods for manipulating PDF files.
+type Service struct {
+ ghostscriptCmd string
+}
+
+// NewService instantiate a new [Service].
+func NewService(ghostscriptCmd string) *Service {
+ return &Service{ghostscriptCmd}
+}
+
+// ExtractPage extract a page from a PDF.
+func (s *Service) ExtractPage(stdin io.Reader, page int) (*bytes.Buffer, error) {
+ args := []string{
+ "-q",
+ "-sDEVICE=pdfwrite",
+ "-dNOPAUSE",
+ "-dBATCH",
+ "-dSAFER",
+ fmt.Sprintf("-dFirstPage=%d", page),
+ fmt.Sprintf("-dLastPage=%d", page),
+ "-sOutputFile=-",
+ "-", // Use stdin for input
+ }
+
+ var stdout, stderr bytes.Buffer
+ cmd := exec.Command(s.ghostscriptCmd, args...)
+ cmd.Stdin = stdin
+ cmd.Stdout = &stdout
+ cmd.Stderr = &stderr
+
+ if err := cmd.Run(); err != nil {
+ logger.WithNamespace("pdf").
+ WithField("stderr", stderr.String()).
+ Errorf("ghostscript failed: %s", err)
+ return nil, fmt.Errorf("failed to run the cmd %q: %w", s.ghostscriptCmd, err)
+ }
+ return &stdout, nil
+}

pkg/pdf/pdf_test.go

@@ -0,0 +1,38 @@
+package pdf
+
+import (
+ "os"
+ "os/exec"
+ "testing"
+
+ "github.com/stretchr/testify/require"
+)
+
+func Test_Extract_Page(t *testing.T) {
+ if testing.Short() {
+ t.Skipf("this test require the \"gs\" binary, skip it due to the \"--short\" flag")
+ }
+
+ service := NewService("gs")
+ input, err := os.Open("../../tests/fixtures/dev-desktop.pdf")
+ require.NoError(t, err)
+ defer input.Close()
+
+ extracted, err := service.ExtractPage(input, 1)
+ require.NoError(t, err)
+
+ // We cannot compare the output to an expected PDF file, as there many
+ // things that change from one run to another: CreationDate, uuid, etc.
+ // So, we are checking that it's a PDF, and it has the expected signature
+ // from ImageMagick.
+ content := extracted.Bytes()
+ start := []byte("%PDF-1.7")
+ require.Equal(t, start, content[:len(start)])
+
+ expected := "5b49b84d59866b2f6d825957c55c2c2681656e5de52e30c1439fbaf197fe1d14"
+ cmd := exec.Command("identify", "-quiet", "-format", "%#", "-")
+ cmd.Stdin = extracted
+ signature, err := cmd.Output()
+ require.NoError(t, err)
+ require.Equal(t, []byte(expected), signature)
+}

scripts/docker/cozy-app-dev/Dockerfile

@@ -45,6 +45,7 @@ RUN set -eux; apt-get update \
  openssl \
  fonts-lato \
  imagemagick \
+ ghostscript \
  git \
  && curl https://couchdb.apache.org/repo/keys.asc | gpg --dearmor > /usr/share/keyrings/couchdb-archive-keyring.gpg \
  && echo "deb [signed-by=/usr/share/keyrings/couchdb-archive-keyring.gpg] https://apache.jfrog.io/artifactory/couchdb-deb/ bullseye main" > /etc/apt/sources.list.d/couchdb.list \

worker/archive/unzip_test.go

@@ -87,7 +87,7 @@ func Test_archive(t *testing.T) {
  assert.NoError(t, err)
  assert.NoError(t, file.Close())
 
- files := map[string]string{
+ files := map[string]interface{}{
  "wet-cozy.jpg": one.ID(),
  "hello.txt": two.ID(),
  }