Allow to download a single page from a PDF (#4434)

cozy · Jul 2, 2024 · b950976 · b950976
2 parents 8a145e0 + 48c1485
commit b950976
Show file tree

Hide file tree

Showing 3 changed files with 51 additions and 2 deletions.
diff --git a/docs/files.md b/docs/files.md
@@ -905,6 +905,9 @@ Download the file content.
 By default the `content-disposition` will be `inline`, but it will be
 `attachment` if the query string contains the parameter `Dl=1`
 
+For a PDF file, it's possible to get only a single page by using the `Page`
+parameter in the query-string (1 is the first page).
+
 #### Request
 
 ```http
@@ -929,6 +932,9 @@ Download the file content from its path.
 By default the `content-disposition` will be `inline`, but it will be
 `attachment` if the query string contains the parameter `Dl=1`
 
+For a PDF file, it's possible to get only a single page by using the `Page`
+parameter in the query-string (1 is the first page).
+
 #### Request
 
 ```http

diff --git a/model/vfs/file.go b/model/vfs/file.go
@@ -1,15 +1,18 @@
 package vfs
 
 import (
+	"bytes"
 	"encoding/base64"
 	"fmt"
 	"mime"
 	"net/http"
 	"os"
 	"path"
+	"path/filepath"
 	"strings"
 	"time"
 
+	"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"
@@ -252,6 +255,28 @@ func ServeFileContent(fs VFS, doc *FileDoc, version *Version, filename, disposit
 	return nil
 }
 
+// ServePDFPage replies to an http request with a single page from a PDF file.
+func ServePDFPage(fs VFS, doc *FileDoc, disposition string, page int, req *http.Request, w http.ResponseWriter) error {
+	ext := filepath.Ext(doc.DocName)
+	basename := strings.TrimSuffix(doc.DocName, ext)
+	filename := fmt.Sprintf("%s (%d)%s", basename, page, ext)
+
+	f, err := fs.OpenFile(doc)
+	if err != nil {
+		return err
+	}
+	defer f.Close()
+
+	extracted, err := config.PDF().ExtractPage(f, page)
+	if err != nil {
+		return err
+	}
+	content := bytes.NewReader(extracted.Bytes())
+
+	http.ServeContent(w, req, filename, doc.UpdatedAt, content)
+	return nil
+}
+
 // ModifyFileMetadata modify the metadata associated to a file. It can
 // be used to rename or move the file in the VFS.
 func ModifyFileMetadata(fs VFS, olddoc *FileDoc, patch *DocPatch) (*FileDoc, error) {

diff --git a/web/files/files.go b/web/files/files.go
@@ -905,7 +905,16 @@ func ReadFileContentFromIDHandler(c echo.Context) error {
 	if c.QueryParam("Dl") == "1" {
 		disposition = "attachment"
 	}
-	err = vfs.ServeFileContent(instance.VFS(), doc, nil, "", disposition, c.Request(), c.Response())
+
+	if page := c.QueryParam("Page"); page != "" {
+		p, errp := strconv.Atoi(page)
+		if errp != nil {
+			return jsonapi.InvalidParameter("Page", errp)
+		}
+		err = vfs.ServePDFPage(instance.VFS(), doc, disposition, p, c.Request(), c.Response())
+	} else {
+		err = vfs.ServeFileContent(instance.VFS(), doc, nil, "", disposition, c.Request(), c.Response())
+	}
 	if err != nil {
 		return WrapVfsError(err)
 	}
@@ -1115,7 +1124,16 @@ func sendFileFromPath(c echo.Context, path string, checkPermission bool) error {
 	} else if !checkPermission {
 		addCSPRuleForDirectLink(c, doc.Class, doc.Mime)
 	}
-	err = vfs.ServeFileContent(instance.VFS(), doc, nil, "", disposition, c.Request(), c.Response())
+
+	if page := c.QueryParam("Page"); page != "" {
+		p, errp := strconv.Atoi(page)
+		if errp != nil {
+			return jsonapi.InvalidParameter("Page", errp)
+		}
+		err = vfs.ServePDFPage(instance.VFS(), doc, disposition, p, c.Request(), c.Response())
+	} else {
+		err = vfs.ServeFileContent(instance.VFS(), doc, nil, "", disposition, c.Request(), c.Response())
+	}
 	if err != nil {
 		return WrapVfsError(err)
 	}