commands.extract.extractors - limit sanitized filenames to 20 charact…

…ers and use unique prefixes

Fix #748

digiarch/commands/extract/extractors/extractor_msg.py

@@ -123,7 +123,7 @@ def extract(self) -> list[tuple[Path, Path]]:
                 if isinstance(attachment, (Message, MessageSigned)):
                     name: str = (attachment.filename or "").strip() or (attachment.subject or "").strip()
                     name = name.strip() or f"attachment-{n}"
-                    name_sanitized: str = sanitize_filename(name).strip("_") or f"attachment-{n}"
+                    name_sanitized: str = sanitize_filename(name, 20, True).strip("_") or f"attachment-{n}"
                     attachment.export(tmp_dir / name_sanitized)
                     files.append((name_sanitized, name))
                 elif isinstance(attachment.data, bytes):
@@ -133,7 +133,7 @@ def extract(self) -> list[tuple[Path, Path]]:
                         else attachment.longFilename or ""
                     )
                     name = name.strip() or f"attachment-{n}"
-                    name_sanitized: str = sanitize_filename(name).strip("_") or f"attachment-{n}"
+                    name_sanitized: str = sanitize_filename(name, 20, True).strip("_") or f"attachment-{n}"
                     with tmp_dir.joinpath(name_sanitized).open("wb") as fh:
                         fh.write(attachment.data or b"")
                     files.append((name_sanitized, name))

digiarch/commands/extract/extractors/extractor_patool.py

@@ -5,6 +5,7 @@
 from patoolib import extract_archive
 from patoolib.util import PatoolError
 
+from digiarch.common import sanitize_filename
 from digiarch.common import sanitize_path
 from digiarch.common import TempDir
 
@@ -62,6 +63,7 @@ def extract(self) -> list[tuple[Path, Path]]:
 
                 for path in find_files(tmp_dir):
                     path_sanitized: Path = extract_folder / sanitize_path(path.relative_to(tmp_dir))
+                    path_sanitized = path_sanitized.with_name(sanitize_filename(path_sanitized.name, 20, True))
                     while path_sanitized.exists():
                         path_sanitized = path_sanitized.with_name("_" + path_sanitized.name)
                     path_sanitized.parent.mkdir(parents=True, exist_ok=True)

digiarch/commands/extract/extractors/extractor_tnef.py

@@ -25,7 +25,7 @@ def extract(self) -> list[tuple[Path, Path]]:
         with TempDir(self.file.root) as tmp_dir:
             for attachment in tnef.attachments:
                 name: str = attachment.long_filename() or attachment.name
-                path: Path = tmp_dir.joinpath(sanitize_filename(name))
+                path: Path = tmp_dir.joinpath(sanitize_filename(name, 20, True))
                 with path.open("wb") as oh:
                     oh.write(attachment.data)
                 files.append((path.name, name))

digiarch/commands/extract/extractors/extractor_zip.py

@@ -4,6 +4,7 @@
 from zipfile import LargeZipFile
 from zipfile import ZipFile
 
+from digiarch.common import sanitize_filename
 from digiarch.common import sanitize_path
 from digiarch.common import TempDir
 
@@ -37,6 +38,7 @@ def extract(self) -> list[tuple[Path, Path]]:
 
                         path_original: Path = Path(zf.extract(member, tmp_extract))
                         path_final: Path = tmp_final.joinpath(sanitize_path(path_original.relative_to(tmp_extract)))
+                        path_final = path_final.with_name(sanitize_filename(path_final.name, 20, True))
                         while path_final.exists():
                             path_final = path_final.with_name("_" + path_final.name)