Avoid copying db files if not necessary

osxphotos/_version.py

@@ -1,4 +1,4 @@
 """ version info """
 
-__version__ = "0.36.14"
+__version__ = "0.36.15"
 

osxphotos/fileutil.py

@@ -1,6 +1,5 @@
 """ FileUtil class with methods for copy, hardlink, unlink, etc. """
 
-import logging
 import os
 import pathlib
 import stat
@@ -74,7 +73,6 @@ class FileUtilMacOS(FileUtilABC):
         try:
             os.link(src, dest)
         except Exception as e:
-            logging.critical(f"os.link returned error: {e}")
             raise e
 
     @classmethod
@@ -92,7 +90,7 @@ class FileUtilMacOS(FileUtilABC):
         if src is None or dest is None:
             raise ValueError("src and dest must not be None", src, dest)
 
-        if not os.path.isfile(src):
+        if not os.path.exists(src):
             raise FileNotFoundError("src file does not appear to exist", src)
 
         if norsrc:
@@ -104,9 +102,6 @@ class FileUtilMacOS(FileUtilABC):
         try:
             result = subprocess.run(command, check=True, stderr=subprocess.PIPE)
         except subprocess.CalledProcessError as e:
-            logging.critical(
-                f"ditto returned error: {e.returncode} {e.stderr.decode(sys.getfilesystemencoding()).rstrip()}"
-            )
             raise e
 
         return result.returncode

osxphotos/photosdb/photosdb.py

@@ -35,6 +35,7 @@ from .._constants import (
 from .._version import __version__
 from ..albuminfo import AlbumInfo, FolderInfo, ImportInfo
 from ..datetime_utils import datetime_has_tz, datetime_naive_to_local
+from ..fileutil import FileUtil
 from ..personinfo import PersonInfo
 from ..photoinfo import PhotoInfo
 from ..utils import (
@@ -102,7 +103,7 @@ class PhotosDB:
         # tempfile.TemporaryDirectory gets cleaned up when the object does
         self._tempdir = tempfile.TemporaryDirectory(prefix="osxphotos_")
         self._tempdir_name = self._tempdir.name
-
+        
         # set up the data structures used to store all the Photo database info
 
         # TODO: I don't think these keywords flags are actually used
@@ -265,8 +266,11 @@ class PhotosDB:
         # photoanalysisd sometimes maintains this lock even after Photos is closed
         # In those cases, make a temp copy of the file for sqlite3 to read
         if _db_is_locked(self._dbfile):
-            verbose(f"Database locked, creating temporary copy.")
-            self._tmp_db = self._copy_db_file(self._dbfile)
+            try:
+                self._tmp_db = self._link_db_file(self._dbfile)
+            except:
+                verbose(f"Database locked, creating temporary copy.")
+                self._tmp_db = self._copy_db_file(self._dbfile)
 
         self._db_version = get_db_version(self._tmp_db)
 
@@ -281,8 +285,11 @@ class PhotosDB:
                 verbose(f"Processing database {self._dbfile_actual}")
                 # if database is exclusively locked, make a copy of it and use the copy
                 if _db_is_locked(self._dbfile_actual):
-                    verbose(f"Database locked, creating temporary copy.")
-                    self._tmp_db = self._copy_db_file(self._dbfile_actual)
+                    try:
+                        self._tmp_db = self._link_db_file(self._dbfile_actual)
+                    except:
+                        verbose(f"Database locked, creating temporary copy.")
+                        self._tmp_db = self._copy_db_file(self._dbfile_actual)
 
             if _debug():
                 logging.debug(
@@ -546,6 +553,32 @@ class PhotosDB:
 
         return dest_path
 
+    def _link_db_file(self, fname):
+        """ links the sqlite database file to a temp file """
+        """ returns the name of the temp file """
+        """ If sqlite shared memory and write-ahead log files exist, those are copied too """
+        # required because python's sqlite3 implementation can't read a locked file
+        # _, suffix = os.path.splitext(fname)
+        dest_name = dest_path = ""
+        try:
+            dest_name = pathlib.Path(fname).name
+            dest_path = os.path.join(self._tempdir_name, dest_name)
+            FileUtil.hardlink(fname, dest_path)
+            # link write-ahead log and shared memory files (-wal and -shm) files if they exist
+            if os.path.exists(f"{fname}-wal"):
+                FileUtil.hardlink(f"{fname}-wal", f"{dest_path}-wal")
+            if os.path.exists(f"{fname}-shm"):
+                FileUtil.hardlink(f"{fname}-shm", f"{dest_path}-shm")
+        except:
+            print("Error linking " + fname + " to " + dest_path, file=sys.stderr)
+            raise Exception
+
+        if _debug():
+            logging.debug(dest_path)
+
+        return dest_path
+
+
     def _process_database4(self):
         """ process the Photos database to extract info
             works on Photos version <= 4.0 """

tests/tempdiskimage.py

@@ -0,0 +1,114 @@
+""" Create a temporary disk image on MacOS """
+
+import pathlib
+import platform
+import subprocess
+import tempfile
+import time
+
+
+class TempDiskImage:
+    """ Create and mount a temporary disk image """
+
+    def __init__(self, size=100, prefix=None):
+        """ Create and mount a temporary disk image.
+        
+        Args:
+            size: int; size in MB of disk image, default = 100
+            prefix: str; optional prefix to prepend to name of the temporary disk image
+            name: str; name of the mounted volume, default = "TemporaryDiskImage"
+
+        Raises:
+            TypeError if size is not int
+            RunTimeError if not on MacOS
+        """
+        if type(size) != int:
+            raise TypeError("size must be int")
+
+        system = platform.system()
+        if system != "Darwin":
+            raise RuntimeError("TempDiskImage only runs on MacOS")
+
+        self._tempdir = tempfile.TemporaryDirectory()
+        # hacky mktemp: this could create a race condition but unlikely given it's created in a TemporaryDirectory
+        prefix = "TemporaryDiskImage" if prefix is None else prefix
+        volume_name = f"{prefix}_{str(time.time()).replace('.','_')}_{str(time.perf_counter()).replace('.','_')}"
+        image_name = f"{volume_name}.dmg"
+        image_path = pathlib.Path(self._tempdir.name) / image_name
+        hdiutil = subprocess.run(
+            [
+                "/usr/bin/hdiutil",
+                "create",
+                "-size",
+                f"{size}m",
+                "-fs",
+                "HFS+",
+                "-volname",
+                volume_name,
+                image_path,
+            ],
+            check=True,
+            text=True,
+            capture_output=True,
+        )
+
+        if "created" not in hdiutil.stdout:
+            raise OSError(f"Could not create DMG {image_path}")
+
+        self.path = image_path
+        self._mount_point, self.name = self._mount_image(self.path)
+
+    def _mount_image(self, image_path):
+        """ mount a DMG file and return path, returns (mount_point, path) """
+        hdiutil = subprocess.run(
+            ["/usr/bin/hdiutil", "attach", image_path],
+            check=True,
+            text=True,
+            capture_output=True,
+        )
+        mount_point, path = None, None
+        for line in hdiutil.stdout.split("\n"):
+            line = line.strip()
+            if "Apple_HFS" not in line:
+                continue
+            output = line.split()
+            if len(output) < 3:
+                raise ValueError(f"Error mounting disk image {image_path}")
+            mount_point = output[0]
+            path = output[2]
+            break
+        return (mount_point, path)
+
+    def unmount(self):
+        try:
+            if self._mount_point:
+                hdiutil = subprocess.run(
+                    ["/usr/bin/hdiutil", "detach", self._mount_point],
+                    check=True,
+                    text=True,
+                    capture_output=True,
+                )
+                self._mount_point = None
+        except AttributeError:
+            pass
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.unmount()
+        if exc_type:
+            return False
+
+
+if __name__ == "__main__":
+    # Create a temporary disk image, 50mb in size
+    img = TempDiskImage(size=50, prefix="MyDiskImage")
+    # Be sure to unmount it, image will be cleaned up automatically
+    img.unmount()
+
+    # Or use it as a context handler
+    # Default values are 100mb and prefix = "TemporaryDiskImage"
+    with TempDiskImage() as img:
+        print(f"image: {img.path}")
+        print(f"mounted at: {img.name}")

tests/test_link_db.py

@@ -0,0 +1,29 @@
+""" Test PhotosDB._link_db_file """
+
+import pytest
+
+from tempdiskimage import TempDiskImage
+
+PHOTOS_DB = "tests/Test-Movie-5_0.photoslibrary"
+
+def test_link_db(capsys):
+    """ Test that database doesn't get copied when opened """
+    import osxphotos
+    photosdb = osxphotos.PhotosDB(dbfile=PHOTOS_DB, verbose=print)
+    captured = capsys.readouterr()
+    assert "creating temporary copy" not in captured.out
+
+def test_copy_db(capsys):
+    """ Test that database does get copied if on different filesystem """
+    import pathlib
+    import tempfile
+    import osxphotos
+    
+    from osxphotos.fileutil import FileUtil
+    
+    with TempDiskImage(prefix="osxphotos") as tmpimg:
+        newdb = pathlib.Path(tmpimg.name) / pathlib.Path(PHOTOS_DB).name
+        FileUtil.copy(PHOTOS_DB,newdb)
+        photosdb = osxphotos.PhotosDB(dbfile=newdb, verbose=print)
+        captured = capsys.readouterr()
+        assert "creating temporary copy" in captured.out