add no-archive and no-thumbnail options to administration exporter and importer

docs/administration.md

@@ -227,12 +227,14 @@ is not a TTY" errors. For example:
 `docker-compose exec -T webserver document_exporter ../export`
 
 ```
-document_exporter target [-c] [-f] [-d]
+document_exporter target [-c] [-f] [-d] [-na] [-nt]
 
 optional arguments:
 -c, --compare-checksums
 -f, --use-filename-format
 -d, --delete
+-na, --no-archive
+-nt, --no-thumbnail
 -z  --zip
 ```
 
@@ -259,6 +261,14 @@ current export such as files from deleted documents, specify `--delete`.
 Be careful when pointing paperless to a directory that already contains
 other files.
 
+Paperless will not export archive files if you use `--no-archive`, or will
+not export thumbnails if you use `--no-thumbnail`. After importing, These
+files can be generated again by using `document_archiver` or
+`document_thumbnails`. It can make sense to omit these files from backup
+as their content and checksum can change (new archiver or thumbnail
+generator algorithm) and may then cause additional used space in
+a deduplicated backup.
+
 If `-z` or `--zip` is provided, the export will be a zipfile
 in the target directory, named according to the current date.
 

src/documents/management/commands/document_exporter.py

@@ -82,6 +82,21 @@ class Command(BaseCommand):
             "deleted documents.",
         )
 
+        parser.add_argument(
+            "-na",
+            "--no-archive",
+            default=False,
+            action="store_true",
+            help="Avoid exporting archive files",
+        )
+
+        parser.add_argument(
+            "-nt",
+            "--no-thumbnail",
+            default=False,
+            action="store_true",
+            help="Avoid exporting thumbnail files",
+        )
         parser.add_argument(
             "--no-progress-bar",
             default=False,
@@ -105,6 +120,8 @@ class Command(BaseCommand):
         self.compare_checksums = False
         self.use_filename_format = False
         self.delete = False
+        self.no_archive = False
+        self.no_thumbnail = False
 
     def handle(self, *args, **options):
 
@@ -112,6 +129,8 @@ class Command(BaseCommand):
         self.compare_checksums = options["compare_checksums"]
         self.use_filename_format = options["use_filename_format"]
         self.delete = options["delete"]
+        self.no_archive = options["no_archive"]
+        self.no_thumbnail = options["no_thumbnail"]
         zip_export: bool = options["zip"]
 
         # If zipping, save the original target for later and
@@ -246,11 +265,14 @@ class Command(BaseCommand):
             original_target = (self.target / Path(original_name)).resolve()
             document_dict[EXPORTER_FILE_NAME] = original_name
 
-            thumbnail_name = base_name + "-thumbnail.webp"
-            thumbnail_target = (self.target / Path(thumbnail_name)).resolve()
-            document_dict[EXPORTER_THUMBNAIL_NAME] = thumbnail_name
+            if not self.no_thumbnail:
+                thumbnail_name = base_name + "-thumbnail.webp"
+                thumbnail_target = (self.target / Path(thumbnail_name)).resolve()
+                document_dict[EXPORTER_THUMBNAIL_NAME] = thumbnail_name
+            else:
+                thumbnail_target = None
 
-            if document.has_archive_version:
+            if not self.no_archive and document.has_archive_version:
                 archive_name = base_name + "-archive.pdf"
                 archive_target = (self.target / Path(archive_name)).resolve()
                 document_dict[EXPORTER_ARCHIVE_NAME] = archive_name
@@ -266,10 +288,11 @@ class Command(BaseCommand):
                     original_target.write_bytes(GnuPG.decrypted(out_file))
                     os.utime(original_target, times=(t, t))
 
-                thumbnail_target.parent.mkdir(parents=True, exist_ok=True)
-                with document.thumbnail_file as out_file:
-                    thumbnail_target.write_bytes(GnuPG.decrypted(out_file))
-                    os.utime(thumbnail_target, times=(t, t))
+                if thumbnail_target:
+                    thumbnail_target.parent.mkdir(parents=True, exist_ok=True)
+                    with document.thumbnail_file as out_file:
+                        thumbnail_target.write_bytes(GnuPG.decrypted(out_file))
+                        os.utime(thumbnail_target, times=(t, t))
 
                 if archive_target:
                     archive_target.parent.mkdir(parents=True, exist_ok=True)
@@ -283,7 +306,8 @@ class Command(BaseCommand):
                     original_target,
                 )
 
-                self.check_and_copy(document.thumbnail_path, None, thumbnail_target)
+                if thumbnail_target:
+                    self.check_and_copy(document.thumbnail_path, None, thumbnail_target)
 
                 if archive_target:
                     self.check_and_copy(

src/documents/management/commands/document_importer.py

@@ -193,8 +193,11 @@ class Command(BaseCommand):
             doc_file = record[EXPORTER_FILE_NAME]
             document_path = os.path.join(self.source, doc_file)
 
-            thumb_file = record[EXPORTER_THUMBNAIL_NAME]
-            thumbnail_path = Path(os.path.join(self.source, thumb_file)).resolve()
+            if EXPORTER_THUMBNAIL_NAME in record:
+                thumb_file = record[EXPORTER_THUMBNAIL_NAME]
+                thumbnail_path = Path(os.path.join(self.source, thumb_file)).resolve()
+            else:
+                thumbnail_path = None
 
             if EXPORTER_ARCHIVE_NAME in record:
                 archive_file = record[EXPORTER_ARCHIVE_NAME]
@@ -212,19 +215,21 @@ class Command(BaseCommand):
 
                 shutil.copy2(document_path, document.source_path)
 
-                if thumbnail_path.suffix in {".png", ".PNG"}:
-                    run_convert(
-                        density=300,
-                        scale="500x5000>",
-                        alpha="remove",
-                        strip=True,
-                        trim=False,
-                        auto_orient=True,
-                        input_file=f"{thumbnail_path}[0]",
-                        output_file=str(document.thumbnail_path),
-                    )
-                else:
-                    shutil.copy2(thumbnail_path, document.thumbnail_path)
+                if thumbnail_path:
+                    if thumbnail_path.suffix in {".png", ".PNG"}:
+                        run_convert(
+                            density=300,
+                            scale="500x5000>",
+                            alpha="remove",
+                            strip=True,
+                            trim=False,
+                            auto_orient=True,
+                            input_file=f"{thumbnail_path}[0]",
+                            output_file=str(document.thumbnail_path),
+                        )
+                    else:
+                        shutil.copy2(thumbnail_path, document.thumbnail_path)
+
                 if archive_path:
                     create_source_path_directory(document.archive_path)
                     # TODO: this assumes that the export is valid and

src/documents/tests/test_management_exporter.py

@@ -102,6 +102,8 @@ class TestExportImport(DirectoriesMixin, TestCase):
         use_filename_format=False,
         compare_checksums=False,
         delete=False,
+        no_archive=False,
+        no_thumbnail=False,
     ):
         args = ["document_exporter", self.target]
         if use_filename_format:
@@ -110,6 +112,10 @@ class TestExportImport(DirectoriesMixin, TestCase):
             args += ["--compare-checksums"]
         if delete:
             args += ["--delete"]
+        if no_archive:
+            args += ["--no-archive"]
+        if no_thumbnail:
+            args += ["--no-thumbnail"]
 
         call_command(*args)
 
@@ -497,3 +503,63 @@ class TestExportImport(DirectoriesMixin, TestCase):
                 call_command(*args)
 
                 self.assertEqual("That path doesn't appear to be writable", str(e))
+
+    def test_no_archive(self):
+        shutil.rmtree(os.path.join(self.dirs.media_dir, "documents"))
+        shutil.copytree(
+            os.path.join(os.path.dirname(__file__), "samples", "documents"),
+            os.path.join(self.dirs.media_dir, "documents"),
+        )
+
+        manifest = self._do_export()
+        has_archive = False
+        for element in manifest:
+            if element["model"] == "documents.document":
+                has_archive = (
+                    has_archive or document_exporter.EXPORTER_ARCHIVE_NAME in element
+                )
+        self.assertTrue(has_archive)
+
+        has_archive = False
+        manifest = self._do_export(no_archive=True)
+        for element in manifest:
+            if element["model"] == "documents.document":
+                has_archive = (
+                    has_archive or document_exporter.EXPORTER_ARCHIVE_NAME in element
+                )
+        self.assertFalse(has_archive)
+
+        with paperless_environment() as dirs:
+            call_command("document_importer", self.target)
+            self.assertEqual(Document.objects.count(), 4)
+
+    def test_no_thumbnail(self):
+        shutil.rmtree(os.path.join(self.dirs.media_dir, "documents"))
+        shutil.copytree(
+            os.path.join(os.path.dirname(__file__), "samples", "documents"),
+            os.path.join(self.dirs.media_dir, "documents"),
+        )
+
+        manifest = self._do_export()
+        has_thumbnail = False
+        for element in manifest:
+            if element["model"] == "documents.document":
+                has_thumbnail = (
+                    has_thumbnail
+                    or document_exporter.EXPORTER_THUMBNAIL_NAME in element
+                )
+        self.assertTrue(has_thumbnail)
+
+        has_thumbnail = False
+        manifest = self._do_export(no_thumbnail=True)
+        for element in manifest:
+            if element["model"] == "documents.document":
+                has_thumbnail = (
+                    has_thumbnail
+                    or document_exporter.EXPORTER_THUMBNAIL_NAME in element
+                )
+        self.assertFalse(has_thumbnail)
+
+        with paperless_environment() as dirs:
+            call_command("document_importer", self.target)
+            self.assertEqual(Document.objects.count(), 4)