Working on issue 206

osxphotos/fileutil.py

@@ -34,7 +34,12 @@ class FileUtilABC(ABC):
 
     @classmethod
     @abstractmethod
-    def cmp_sig(cls, file1, file2):
+    def cmp(cls, file1, file2, mtime1 = None):
+        pass
+
+    @classmethod
+    @abstractmethod
+    def cmp_file_sig(cls, file1, file2):
         pass
 
     @classmethod
@@ -114,11 +119,32 @@ class FileUtilMacOS(FileUtilABC):
         os.utime(path, times)
 
     @classmethod
-    def cmp_sig(cls, f1, s2):
+    def cmp(cls, f1, f2, mtime1 = None):
+        """Does shallow compare (file signatures) of f1 to file f2.
+        Arguments:
+        f1 --  File name
+        f2 -- File name
+        mtime1 -- optional, pass alternate file modification timestamp for f1; will be converted to int
+
+        Return value:
+        True if the file signatures as returned by stat are the same, False otherwise. 
+        Does not do a byte-by-byte comparison.
+        """
+
+        s1 = cls._sig(os.stat(f1))
+        if mtime1 is not None:
+            s1 = (s1[0], s1[1], int(mtime1))
+        s2 = cls._sig(os.stat(f2))
+        if s1[0] != stat.S_IFREG or s2[0] != stat.S_IFREG:
+            return False
+        return s1 == s2
+
+    @classmethod
+    def cmp_file_sig(cls, f1, s2):
         """Compare file f1 to signature s2.
         Arguments:
         f1 --  File name
-        s2 -- stats as returned by sig
+        s2 -- stats as returned by _sig
 
         Return value:
         True if the files are the same, False otherwise.
@@ -140,7 +166,12 @@ class FileUtilMacOS(FileUtilABC):
 
     @staticmethod
     def _sig(st):
-        return (stat.S_IFMT(st.st_mode), st.st_size, st.st_mtime)
+        """ return tuple of (mode, size, mtime) of file based on os.stat
+            Args:
+                st: os.stat signature
+        """
+        # use int(st.st_mtime) because ditto does not copy fractional portion of mtime
+        return (stat.S_IFMT(st.st_mode), st.st_size, int(st.st_mtime))
 
 
 class FileUtil(FileUtilMacOS):
@@ -151,8 +182,8 @@ class FileUtil(FileUtilMacOS):
 
 class FileUtilNoOp(FileUtil):
     """ No-Op implementation of FileUtil for testing / dry-run mode
-        all methods with exception of cmp_sig and file_cmp are no-op
+        all methods with exception of cmp, cmp_file_sig and file_cmp are no-op
-        cmp_sig functions as FileUtil.cmp_sig does
+        cmp and cmp_file_sig functions as FileUtil methods do
         file_cmp returns mock data
     """
 

osxphotos/photoinfo/_photoinfo_export.py

@@ -11,7 +11,6 @@
 
 # TODO: should this be its own PhotoExporter class?
 
-import filecmp
 import glob
 import json
 import logging
@@ -484,7 +483,7 @@ def export2(
         if update and dest.exists():
             # destination exists, check to see if destination is the right UUID
             dest_uuid = export_db.get_uuid_for_file(dest)
-            if dest_uuid is None and filecmp.cmp(src, dest):
+            if dest_uuid is None and fileutil.cmp(src, dest):
                 # might be exporting into a pre-ExportDB folder or the DB got deleted
                 logging.debug(
                     f"Found matching file with blank uuid: {self.uuid}, {dest}"
@@ -516,7 +515,7 @@ def export2(
                         dest = pathlib.Path(file_)
                         found_match = True
                         break
-                    elif dest_uuid is None and filecmp.cmp(src, file_):
+                    elif dest_uuid is None and fileutil.cmp(src, file_):
                         # files match, update the UUID
                         logging.debug(
                             f"Found matching file with blank uuid: {self.uuid}, {file_}"
@@ -836,21 +835,33 @@ def _export_photo(
         op_desc = "export_by_copying"
 
     if not update:
-        # not update, do the the hardlink
+        # not update, export the file
-        logging.debug(f"Not update: {op_desc} linking file {src} {dest}")
+        logging.debug(f"Exporting file with {op_desc} {src} {dest}")
         exported_files.append(dest_str)
-    else: #updating
+    else:  # updating
         if not dest_exists:
             # update, destination doesn't exist (new file)
             logging.debug(f"Update: exporting new file with {op_desc} {src} {dest}")
             update_new_files.append(dest_str)
         else:
             # update, destination exists, but we might not need to replace it...
-            if ((    export_as_hardlink and     dest.samefile(src)) or
+            if touch_file:
-                (not export_as_hardlink and not dest.samefile(src) and (
+                sig_cmp = fileutil.cmp(src, dest, mtime1=self.date.timestamp())
-                    (    exiftool and fileutil.cmp_sig(dest_str, export_db.get_stat_exif_for_file(dest_str))) or
+            else:
-                    (not exiftool and filecmp.cmp(src, dest)))
+                sig_cmp = fileutil.cmp(src, dest)
-                )):
+            if (export_as_hardlink and dest.samefile(src)) or (
+                not export_as_hardlink
+                and not dest.samefile(src)
+                and (
+                    (
+                        exiftool
+                        and fileutil.cmp_file_sig(
+                            dest_str, export_db.get_stat_exif_for_file(dest_str)
+                        )
+                    )
+                    or (not exiftool and sig_cmp)
+                )
+            ):
                 # destination exists but its signature is "identical"
                 logging.debug(f"Update: skipping identical original files {src} {dest}")
                 # call set_stat because code can reach this spot if no export DB but exporting a RAW or live photo
@@ -859,13 +870,17 @@ def _export_photo(
                 update_skipped_files.append(dest_str)
             else:
                 # destination exists but is different
-                logging.debug(f"Update: removing existing file prior to {op_desc} {src} {dest}")
+                logging.debug(
+                    f"Update: removing existing file prior to {op_desc} {src} {dest}"
+                )
                 update_updated_files.append(dest_str)
 
     if not update_skipped_files:
         if dest_exists and (update or overwrite):
             # need to remove the destination first
-            logging.debug(f"Update: removing existing file prior to export_as_hardlink {src} {dest}")
+            logging.debug(
+                f"Update: removing existing file prior to export_as_hardlink {src} {dest}"
+            )
             # dest.unlink()
             fileutil.unlink(dest)
         if export_as_hardlink:
@@ -873,7 +888,7 @@ def _export_photo(
         else:
             fileutil.copy(src, dest_str, norsrc=no_xattr)
         if touch_file:
-            ts=self.date.timestamp()
+            ts = self.date.timestamp()
             fileutil.utime(dest, (ts, ts))
 
     export_db.set_data(
@@ -886,7 +901,11 @@ def _export_photo(
     )
 
     return ExportResults(
-        exported_files + update_new_files + update_updated_files, update_new_files, update_updated_files, update_skipped_files, []
+        exported_files + update_new_files + update_updated_files,
+        update_new_files,
+        update_updated_files,
+        update_skipped_files,
+        [],
     )