wrap atomic_write preserving permissions of mets.xml

ocrd/ocrd/workspace.py

@@ -5,12 +5,12 @@
 import cv2
 from PIL import Image
 import numpy as np
-from atomicwrites import atomic_write
 from deprecated.sphinx import deprecated
 
 from ocrd_models import OcrdMets, OcrdExif, OcrdFile
 from ocrd_models.ocrd_page import parse
 from ocrd_utils import (
+    atomic_write,
     getLogger,
     image_from_polygon,
     coordinates_of_segment,
@@ -252,7 +252,7 @@ def save_mets(self):
         log.info("Saving mets '%s'", self.mets_target)
         if self.automatic_backup:
             WorkspaceBackupManager(self).add()
-        with atomic_write(self.mets_target, overwrite=True) as f:
+        with atomic_write(self.mets_target) as f:
             f.write(self.mets.to_xml(xmllint=True).decode('utf-8'))
 
     def resolve_image_exif(self, image_url):

ocrd/ocrd/workspace_backup.py

@@ -6,12 +6,10 @@
 import hashlib
 
 from ocrd_models import OcrdMets
-from ocrd_utils import getLogger
+from ocrd_utils import getLogger, atomic_write
 
 from .constants import BACKUP_DIR
 
-from atomicwrites import atomic_write
-
 def _chksum(s):
     return hashlib.sha256(s).hexdigest()
 
@@ -85,7 +83,7 @@ def add(self):
             mets_file = join(d, 'mets.xml')
             log.info("Backing up to %s" % mets_file)
             makedirs(d)
-            with atomic_write(mets_file, overwrite=True) as f:
+            with atomic_write(mets_file) as f:
                 f.write(mets_str.decode('utf-8'))
         return chksum
 

ocrd/requirements.txt

@@ -7,5 +7,4 @@ opencv-python-headless
 Flask
 jsonschema
 pyyaml
-atomicwrites >= 1.3.0
 Deprecated == 1.2.0

ocrd_utils/ocrd_utils/__init__.py

@@ -128,6 +128,7 @@
 
 from .os import (
     abspath,
+    atomic_write,
     pushd_popd,
     unzip_file_to_dir)
 

ocrd_utils/ocrd_utils/os.py

@@ -5,12 +5,14 @@
     'abspath',
     'pushd_popd',
     'unzip_file_to_dir',
+    'atomic_write',
 ]
 
+from atomicwrites import atomic_write as atomic_write_, AtomicWriter
 from tempfile import TemporaryDirectory
 import contextlib
-from os import getcwd, chdir
-import os.path
+from os import getcwd, chdir, stat, fchmod, umask
+from os.path import exists, abspath as abspath_
 
 from zipfile import ZipFile
 
@@ -22,7 +24,7 @@ def abspath(url):
     """
     if url.startswith('file://'):
         url = url[len('file://'):]
-    return os.path.abspath(url)
+    return abspath_(url)
 
 @contextlib.contextmanager
 def pushd_popd(newcwd=None, tempdir=False):
@@ -54,3 +56,24 @@ def unzip_file_to_dir(path_to_zip, output_directory):
     z.close()
 
 
+
+# ht @pabs3
+# https://github.com/untitaker/python-atomicwrites/issues/42
+class AtomicWriterPerms(AtomicWriter):
+    def get_fileobject(self, **kwargs):
+        f = super().get_fileobject(**kwargs)
+        try:
+            mode = stat(self._path).st_mode
+        except FileNotFoundError:
+            # Creating a new file, emulate what os.open() does
+            mask = umask(0)
+            umask(mask)
+            mode = 0o664 & ~mask
+        fd = f.fileno()
+        fchmod(fd, mode)
+        return f
+
+@contextlib.contextmanager
+def atomic_write(fpath):
+    with atomic_write_(fpath, writer_cls=AtomicWriterPerms, overwrite=True) as f:
+        yield f

ocrd_utils/requirements.txt

@@ -2,3 +2,4 @@ Pillow >= 7.2.0
 # All current versions of TensorFlow require numpy < 1.19.0,
 # so make sure that now newer version is installed here.
 numpy >= 1.17.0, < 1.19.0
+atomicwrites >= 1.3.0

tests/test_workspace.py

@@ -1,4 +1,5 @@
-from os import walk
+from os import walk, stat, chmod, umask
+from stat import filemode
 from subprocess import run, PIPE
 from os.path import join, exists, abspath, basename, dirname
 from tempfile import TemporaryDirectory, mkdtemp
@@ -10,7 +11,7 @@
 from tests.base import TestCase, assets, main, copy_of_directory
 
 from ocrd_models.ocrd_page import parseString
-from ocrd_utils import pushd_popd
+from ocrd_utils import pushd_popd, initLogging
 from ocrd.resolver import Resolver
 from ocrd.workspace import Workspace
 
@@ -278,6 +279,20 @@ def test_image_from_page_basic(self):
             img, info, exif = ws.image_from_page(pcgts.get_Page(), page_id='PHYS_0017')
             self.assertEquals(info['features'], 'binarized,clipped')
 
+    def test_mets_permissions(self):
+        with TemporaryDirectory() as tempdir:
+            ws = self.resolver.workspace_from_nothing(tempdir)
+            ws.save_mets()
+            mets_path = join(ws.directory, 'mets.xml')
+            mask = umask(0)
+            umask(mask)
+            assert (stat(mets_path).st_mode) == 0o100664 & ~mask
+            chmod(mets_path, 0o777)
+            ws.save_mets()
+            assert filemode(stat(mets_path).st_mode) == '-rwxrwxrwx'
+
+
+
 
 if __name__ == '__main__':
     main(__file__)