Options to hide directories or anonymize the username in output logs (#7

)

* --hide-dirs option

* --anonymize-user-dir option

* Update usage image in README

* update help screenshot

CHANGELOG.md

@@ -1,4 +1,6 @@
 # NEXT RELEASE
+* New `--hide-dirs` option
+* New `--anonymize-user-dir` option
 * Bump `PyPDF` to `5.0.1`
 * More default sorting rules
 

clown_sort/config.py

@@ -31,6 +31,7 @@ class Config:
     leave_in_place: bool = False
     screenshots_only: bool = True
     delete_originals: bool = False
+    hide_dirs: bool = False
     rescan_sorted: bool = False
     yes_overwrite: bool = False
     print_as_parsed: bool = False
@@ -50,9 +51,11 @@ def configure(cls):
             cls.enable_debug_mode()
 
         Config.filename_regex = re.compile(args.filename_regex)
+        Config.anonymize_user_dir = True if args.anonymize_user_dir else False
+        Config.hide_dirs = True if args.hide_dirs else False
+        Config.delete_originals = True if args.delete_originals else False
         Config.leave_in_place = True if args.leave_in_place else False
         Config.only_if_match = True if args.only_if_match else False
-        Config.delete_originals = True if args.delete_originals else False
         Config.rescan_sorted = True if args.rescan_sorted else False
         Config.yes_overwrite = True if args.yes_overwrite else False
 

clown_sort/files/sortable_file.py

@@ -20,10 +20,11 @@
 from clown_sort.filename_extractor import FilenameExtractor
 from clown_sort.lib.rule_match import RuleMatch
 from clown_sort.sort_selector import process_file_with_popup
-from clown_sort.util.filesystem_helper import copy_file_creation_time
+from clown_sort.util.filesystem_helper import copy_file_creation_time, loggable_filename
 from clown_sort.util.logging import log
-from clown_sort.util.rich_helper import (bullet_text, comma_join, console, copying_file_log_message,
-     indented_bullet, mild_warning, moving_file_log_message, print_dim_bullet, print_headline, stderr_console)
+from clown_sort.util.rich_helper import (bullet_text, comma_join, console,
+     copying_file_log_message, indented_bullet, mild_warning, moving_file_log_message,
+     print_dim_bullet, stderr_console)
 
 MAX_EXTRACTION_LENGTH = 4096
 NOT_MOVING_FILE = "Not moving file to processed dir because it's"
@@ -197,7 +198,9 @@ def _extracted_text_panel(self) -> Panel:
         return Panel(self._extracted_str(MAX_EXTRACTION_LENGTH), expand=True, style='dim')
 
     def _filename_panel(self) -> Panel:
-        return Panel(str(self.file_path), expand=False, style='bright_white reverse')
+        """Panelized version of the filename for display."""
+        filename = loggable_filename(self.file_path, Config)
+        return Panel(filename, expand=False, style='bright_white reverse')
 
     def _log_copy_file(self, destination_path: Path, match: Optional[re.Match] = None) -> None:
         """Log info about a file copy."""
@@ -209,15 +212,17 @@ def _log_copy_file(self, destination_path: Path, match: Optional[re.Match] = Non
 
         if destination_path.parent == Config.destination_dir:
             console.print(indented_bullet(log_msg.append('root sorted dir...')))
-        else:
-            log_msg.append(str(destination_path.parent), style='sort_destination')
+            return
+
+        dirname = loggable_filename(destination_path.parent, Config)
+        log_msg.append(str(dirname), style='sort_destination')
 
-            if match is not None:
-                log_msg.append(f" (matched '", style='dim')
-                log_msg.append(match.group(0).strip(), style='magenta dim')
-                log_msg.append("')", style='dim')
+        if match is not None:
+            log_msg.append(f" (matched '", style='dim')
+            log_msg.append(match.group(0).strip(), style='magenta dim')
+            log_msg.append("')", style='dim')
 
-            console.print(indented_bullet(log_msg))
+        console.print(indented_bullet(log_msg))
 
     def _move_to_processed_dir(self) -> None:
         """Relocate the original file to the [SCREENSHOTS_DIR]/Processed/ folder."""

clown_sort/sorting_rules/crypto.csv

@@ -1,7 +1,7 @@
 folder,regex
 1mdb,1mdb|jho[-\s.]*low|Victor[-\s.]*Hoo
 3ac,\b3ac\b|Three[-\s.]*Arrows[-\s.]*Capital|su[-\s.]*zhu|zhu[-\s.]*su|kyle[-\s.]*davies|OPNX|Kyle[-\s.]*Davies
-A16Z,a16z|Andreessen|Packym|Arianna[-\s1]*Simpson|Ben[-\s.]*Horowitz|\bAXS\b|\bAli[-\s.]*Yahya|Bitclout
+A16Z,a16z|Andreessen|Packym|Arianna[-\s1]*Simpson|B(en)?[-\s.]*Horowitz|\bAXS\b|\bAli[-\s.]*Yahya|Bitclout
 Aave,\b(Aave|Stani|Kulechov)\b
 AAX,\bAAX(Exchange)?\b
 AEX,\bAEX(Global)?\b
@@ -53,7 +53,7 @@ Changelly,Changelly
 China,fentanyl|\bchin(a|ese)|Hong[-\s.]*Kong|BANK[-\s.]*OF[-\s.]*COMMUNICATIONS|Wan[-\s.]*Kuok|Broken[-\s.]*Tooth|baiyun91586947
 Circle,[@#]circle|usdc|circle[-\s.]*internet|disparte|\ballaire|jerallaire|Macromedia
 CMCC Global,CMCC[-\s.]*Global
-Coinbase,Coinbase|\$COIN\b|brian[-\s.]*armstrong|\b(Aggarwal|grewal|remitly|Venovate)\b|Asiff[-\s.]*Hirji|\bbalaji|Fred[-\s.]*Ehrsam|\$RELY\b|Emilie[-\s.]*Choi\b|BuildOnBase|Digital[-\s.]*Wealth|Asiff[-\s.]*Hirji|Keystone[-\s.]*Capital
+Coinbase,Coinbase|\$COIN\b|brian[-\s.]*armstrong|\b(Aggarwal|grewal|remitly|Venovate)\b|Asiff[-\s.]*Hirji|\bbalaji|Fred[-\s.]*Ehrsam|\$RELY\b|Emilie[-\s.]*Choi\b|BuildOnBase|Digital[-\s.]*Wealth|Asiff[-\s.]*Hirji|Keystone[-\s.]*Capital|jesse[-\s.]*pollak
 CoinField,\b(CoinField|CFC|Manticore|FlaxNFT|Lightman)\b|Sologenic|\bBob[-\s.]\bRas\b
 Coinflex,Mark[-\s.]*Lamb|Roger[-\s.]*Ver|coinflex|Doug[-\s.]*Polk|OPNX|flexusd|Leslie[-\s.]*Lamb|dougpolkvids
 CoinLoan,coinloan|Faliushin\b

clown_sort/util/argument_parser.py

@@ -67,6 +67,12 @@
 parser.add_argument('-y', '--yes-overwrite', action='store_true',
                     help='skip confirmation prompt and always overwrite if a file with the same name already exists')
 
+parser.add_argument('--hide-dirs', action='store_true',
+                    help='hide directories in log output (only show basenames)')
+
+parser.add_argument('--anonymize-user-dir', action='store_true',
+                    help='anonymize the user directory in log output')
+
 parser.add_argument('--rescan-sorted', action='store_true',
                     help="rescan already sorted files (useful if you updated your sorting rules)")
 

clown_sort/util/filesystem_helper.py

@@ -7,6 +7,7 @@
 import stat
 import time
 from datetime import datetime
+from getpass import getuser
 from os import path
 from pathlib import Path
 from typing import List, Optional, Union
@@ -26,6 +27,9 @@
 SORTABLE_FILE_EXTENSIONS = IMAGE_FILE_EXTENSIONS + [PDF_EXTENSION, '.mov']
 MAC_SCREENSHOT_TIMESTAMP_FORMAT = '%Y-%m-%d at %I.%M.%S %p'
 
+ANONYMIZED_USERNAME = 'uzor'
+CURRENT_USERNAME = getuser()
+
 
 def files_in_dir(dir: Union[os.PathLike, str], with_extname: Optional[str] = None) -> List[str]:
     """Paths for non-hidden, non-directory files, optionally ending in 'with_extname'."""
@@ -127,6 +131,21 @@ def extract_timestamp_from_filename(filename: str) -> datetime:
     return datetime.strptime(match.group(1), MAC_SCREENSHOT_TIMESTAMP_FORMAT)
 
 
+def loggable_filename(file_path: Path, config: 'Config') -> str:
+    """Return a loggable version of the filename."""
+    if config.hide_dirs:
+        return Path(file_path).name
+
+    filename = str(file_path)
+
+    if config.anonymize_user_dir:
+        username_str = f"{path.sep}{CURRENT_USERNAME}{path.sep}"
+        anonuser_str = f"{path.sep}{ANONYMIZED_USERNAME}{path.sep}"
+        return filename.replace(username_str, anonuser_str)
+    else:
+        return filename
+
+
 def _non_hidden_files_in_dir(dir: Union[os.PathLike, str]) -> List[str]:
     return [path.join(dir, file) for file in os.listdir(dir) if not file.startswith('.')]