Adde from_date and to_date options to limit the processed files (#202)

* Added from_date and to_date options to limit the processed files * Removed a leftover debug print * Added info for this feature in the readme. Replaced "_" with "-" in the arguments names --------- Co-authored-by: Marco Dondero <[email protected]> Co-authored-by: Ivan Dokov <[email protected]>
ivandokov · Nov 13, 2023 · 3f85d38 · 3f85d38
1 parent 83f2d88
commit 3f85d38
Show file tree

Hide file tree

Showing 4 changed files with 137 additions and 3 deletions.
diff --git a/phockup.py b/phockup.py
@@ -296,6 +296,26 @@ def parse_args(args=sys.argv[1:]):
         files by an additional level after sorting by date.
         """
     )
+
+    parser.add_argument(
+        '--from-date',
+        type=str,
+        default=None,
+        help="""\
+Limit the operations to the files that are newer than --from-date (inclusive).
+The date must be specified in format YYYY-MM-DD. Files with unknown date won't be skipped.
+"""
+    )
+
+    parser.add_argument(
+        '--to-date',
+        type=str,
+        default=None,
+        help="""\
+Limit the operations to the files that are older than --to-date (inclusive).
+The date must be specified in format YYYY-MM-DD. Files with unknown date won't be skipped.
+"""
+    )
 
     return parser.parse_args(args)
 
@@ -344,7 +364,9 @@ def main(options):
         no_date_dir=options.no_date_dir,
         skip_unknown=options.skip_unknown,
         output_prefix=options.output_prefix,
-        output_suffix=options.output_suffix
+        output_suffix=options.output_suffix,
+        from_date=options.from_date,
+        to_date=options.to_date
     )
 
 

diff --git a/readme.md b/readme.md
@@ -220,6 +220,29 @@ saving to the same central respository.
 
 The two options above can be used to help sort/store images
 
+#### Limit files processed by date
+`--from-date` flag can be used to limit the operations to the files that are newer than the provided date (inclusive).
+The date must be specified in format YYYY-MM-DD. Files with unknown date won't be skipped.
+
+For example:
+```
+phockup ~/Pictures/DCIM/NIKOND40 ~/Pictures/sorted --from-date="2017-01-02"
+```
+`--to-date` flag can be used to limit the operations to the files that are older than the provided date (inclusive).
+The date must be specified in format YYYY-MM-DD. Files with unknown date won't be skipped.
+
+For example:
+```
+phockup ~/Pictures/DCIM/NIKOND40 ~/Pictures/sorted --to-date="2017-01-02"
+```  
+
+`--from-date` and `--to-date` can be combined for better control over the files that are processed.
+
+For example:
+```
+phockup ~/Pictures/DCIM/NIKOND40 ~/Pictures/sorted --from-date="2017-01-02" --to-date="2017-01-03"
+``` 
+
 ### Missing date information in EXIF
 If any of the photos does not have date information you can use the `-r | --regex` option to specify date format for date extraction from filenames:
 ```

diff --git a/src/phockup.py b/src/phockup.py
@@ -55,6 +55,14 @@ def __init__(self, input_dir, output_dir, **args):
         self.max_depth = args.get('max_depth', -1)
         # default to concurrency of one to retain existing behavior
         self.max_concurrency = args.get("max_concurrency", 1)
+
+        self.from_date = args.get("from_date", None)
+        self.to_date = args.get("to_date", None)
+        if self.from_date is not None:
+            self.from_date = Date.strptime(f"{self.from_date} 00:00:00", "%Y-%m-%d %H:%M:%S")
+        if self.to_date is not None:
+            self.to_date = Date.strptime(f"{self.to_date} 23:59:59", "%Y-%m-%d %H:%M:%S")
+
         if self.max_concurrency > 1:
             logger.info(f"Using {self.max_concurrency} workers to process files.")
 
@@ -249,7 +257,7 @@ def process_file(self, filename):
 
         progress = f'{filename}'
 
-        output, target_file_name, target_file_path, target_file_type = self.get_file_name_and_path(filename)
+        output, target_file_name, target_file_path, target_file_type, file_date = self.get_file_name_and_path(filename)
         suffix = 1
         target_file = target_file_path
 
@@ -261,6 +269,7 @@ def process_file(self, filename):
                 logger.info(progress)
                 break
 
+            date_unknown = file_date == None or output.endswith(self.no_date_dir)
             if self.skip_unknown and output.endswith(self.no_date_dir):
                 # Skip files that didn't generate a path from EXIF data
                 progress = f"{progress} => skipped, unknown date EXIF information for '{target_file_name}'"
@@ -270,6 +279,22 @@ def process_file(self, filename):
                 logger.info(progress)
                 break
 
+            if not date_unknown:
+                skip = False
+                if type(file_date) is dict:
+                    file_date = file_date["date"]
+                if self.from_date is not None and file_date < self.from_date:
+                    progress = f"{progress} => {filename} skipped: date {file_date} is older than --from-date {self.from_date}"
+                    skip = True
+                if self.to_date is not None and file_date > self.to_date:
+                    progress = f"{progress} => {filename} skipped: date {file_date} is newer than --to-date {self.to_date}"
+                    skip = True
+                if skip:
+                    if self.progress:
+                        self.pbar.write(progress)
+                    logger.info(progress)
+                    break
+
             if os.path.isfile(target_file):
                 if filename != target_file and filecmp.cmp(filename, target_file, shallow=False):
                     progress = f'{progress} => skipped, duplicated file {target_file}'
@@ -330,6 +355,7 @@ def get_file_name_and_path(self, filename):
         if exif_data and 'MIMEType' in exif_data:
             target_file_type = self.get_file_type(exif_data['MIMEType'])
 
+        date = None
         if target_file_type in ['image', 'video']:
             date = Date(filename).from_exif(exif_data, self.timestamp, self.date_regex,
                                             self.date_field)
@@ -342,7 +368,7 @@ def get_file_name_and_path(self, filename):
             target_file_name = os.path.basename(filename)
 
         target_file_path = os.path.sep.join([output, target_file_name])
-        return output, target_file_name, target_file_path, target_file_type
+        return output, target_file_name, target_file_path, target_file_type, date
 
     def process_xmp(self, original_filename, file_name, suffix, output):
         """

diff --git a/tests/test_phockup.py b/tests/test_phockup.py
@@ -473,3 +473,66 @@ def test_skip_unknown():
     assert len([name for name in os.listdir(dir4) if
                 os.path.isfile(os.path.join(dir4, name))]) == 1
     shutil.rmtree('output', ignore_errors=True)
+
+def test_from_date():
+    shutil.rmtree('output', ignore_errors=True)
+    Phockup('input', 'output', from_date="2017-10-06")
+    dir1 = 'output/2017/01/01'
+    dir2 = 'output/2017/10/06'
+    dir3 = 'output/unknown'
+    dir4 = 'output/2018/01/01/'
+    assert os.path.isdir(dir1)
+    assert os.path.isdir(dir2)
+    assert os.path.isdir(dir3)
+    assert os.path.isdir(dir4)
+    assert len([name for name in os.listdir(dir1) if
+                os.path.isfile(os.path.join(dir1, name))]) == 0
+    assert len([name for name in os.listdir(dir2) if
+                os.path.isfile(os.path.join(dir2, name))]) == 1
+    assert len([name for name in os.listdir(dir3) if
+                os.path.isfile(os.path.join(dir3, name))]) == 1
+    assert len([name for name in os.listdir(dir4) if
+                os.path.isfile(os.path.join(dir4, name))]) == 1
+    shutil.rmtree('output', ignore_errors=True)
+
+def test_to_date():
+    shutil.rmtree('output', ignore_errors=True)
+    Phockup('input', 'output', to_date="2017-10-06", progress=True)
+    dir1 = 'output/2017/01/01'
+    dir2 = 'output/2017/10/06'
+    dir3 = 'output/unknown'
+    dir4 = 'output/2018/01/01/'
+    assert os.path.isdir(dir1)
+    assert os.path.isdir(dir2)
+    assert os.path.isdir(dir3)
+    assert os.path.isdir(dir4)
+    assert len([name for name in os.listdir(dir1) if
+                os.path.isfile(os.path.join(dir1, name))]) == 3
+    assert len([name for name in os.listdir(dir2) if
+                os.path.isfile(os.path.join(dir2, name))]) == 1
+    assert len([name for name in os.listdir(dir3) if
+                os.path.isfile(os.path.join(dir3, name))]) == 1
+    assert len([name for name in os.listdir(dir4) if
+                os.path.isfile(os.path.join(dir4, name))]) == 0
+    shutil.rmtree('output', ignore_errors=True)
+
+def test_from_date_to_date():
+    shutil.rmtree('output', ignore_errors=True)
+    Phockup('input', 'output', to_date="2017-10-06", from_date="2017-01-02", progress=True)
+    dir1 = 'output/2017/01/01'
+    dir2 = 'output/2017/10/06'
+    dir3 = 'output/unknown'
+    dir4 = 'output/2018/01/01/'
+    assert os.path.isdir(dir1)
+    assert os.path.isdir(dir2)
+    assert os.path.isdir(dir3)
+    assert os.path.isdir(dir4)
+    assert len([name for name in os.listdir(dir1) if
+                os.path.isfile(os.path.join(dir1, name))]) == 0
+    assert len([name for name in os.listdir(dir2) if
+                os.path.isfile(os.path.join(dir2, name))]) == 1
+    assert len([name for name in os.listdir(dir3) if
+                os.path.isfile(os.path.join(dir3, name))]) == 1
+    assert len([name for name in os.listdir(dir4) if
+                os.path.isfile(os.path.join(dir4, name))]) == 0
+    shutil.rmtree('output', ignore_errors=True)