[WIP] atomic write of mets.xml, fix #278 (#285)

[WIP] atomic write of mets.xml, fix #278

CHANGELOG.md

@@ -8,6 +8,7 @@ Versioned according to [Semantic Versioning](http://semver.org/).
 Changed:
 
   * workspace bagger will create files with extension
+  * `save_mets` is atomic now, #278, #285
 
 ## [1.0.0b15] - 2019-08-14
 

ocrd/ocrd/workspace.py

@@ -5,6 +5,7 @@
 import cv2
 from PIL import Image
 import numpy as np
+from atomicwrites import atomic_write
 
 from ocrd_models import OcrdMets, OcrdExif
 from ocrd_utils import getLogger, is_local_filename, abspath, pushd_popd
@@ -170,8 +171,8 @@ def save_mets(self):
         log.info("Saving mets '%s'" % self.mets_target)
         if self.automatic_backup:
             WorkspaceBackupManager(self).add()
-        with open(self.mets_target, 'wb') as f:
-            f.write(self.mets.to_xml(xmllint=True))
+        with atomic_write(self.mets_target, overwrite=True) 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

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

ocrd/requirements.txt

@@ -9,3 +9,4 @@ opencv-python-headless
 Flask
 jsonschema
 pyyaml
+atomicwrites >= 1.3.0