RewriteNotes.md

Rewrite Notes

Notes about the challenges and important changes during the major rewrite.

• Used an IMHO good IDE - Visual Studio Code with the official Python module and Pylance - which has great code quality (potentially unininitalized variables, unused variables, ...) and refactoring features (renaming, find usages, ...). There is also an excellent TODO Tree extension.
• Code developed and tested on Python 3.11.4 x64 for Windows.
• Introduced static typing where I thought of it especially functions and classes. This makes functions interfaces and their intentions clear, where code breaks them and what is expected to be passed around where.
• Documented functions as much as I could to also improve my understanding of the code. This also led to function renames to express their purpose more clearly.
• Partitioned code into functions to modularize it.
• Spread out functions in modules and separate files especially long functions.
• Switched to Context Managers (with) where appropriate to avoid potentially leaking resources on unexpected exceptions.
• Switched to the often more elegant pathlib library and Path where fesasible. Only os.walk() has no alternative yet.
• Used fallback= in configparser getters to be more resilient and allow minimal configuration files. Default values match the contents of the original config.ini from the repository.
• Renamed utilise_duplicate_threshold to use_duplicate_threshold but code retains compatibility with older config.ini files.
• Introduced use_suffix option in config.ini. If I have a folder for all my Fansly downloads, I do not see the point of suffixing every creator's subfolder with _fansly. This especially makes no sense any more since the rewritten code does not need to parse out the creator's folder from an arbitrary path in odd places. I did, however, retain the previous behavior - it defaults to True.
• Having the program version in a config file (config.ini) makes no sense and is potentially dangerous. The program version should (and is now) in a proper file heading block and can be read from there. Versioning a config file might make sense in some cases but all the changes to config structure so far can easily be handled by the config read/validation functions without reliance on versioning.
• While reworking delete_deprecated_files() I found a bug - os.path.splitext() includes the full path to the file name thus the in must have always failed. (See https://docs.python.org/3/library/os.path.html - "root + ext == path")
• I corrected the ratios for rich's TextColumn/BarColumn: They are ints now, I assumed 2/5 thus 1; 5 (were: 0.355; 2).
• Switched to Enums (StrEnum) for download_mode and download_type to be explicit and prevent magical string values floating around
• I made all hard errors raise and introduced an interactive flag (also as config.ini option) to bypass any Press <ENTER> to continue prompts. Thus you can now automate/schedule it using Task Scheduler on Windows or cron on Linux/UNIX or just have it run through while you are away. What is more, this also helps multiple user processing - an account info error due to invalid user name should not prevent all other users from being downloaded and can be caught in the user loop.
• There also now distinct program exit codes to facilitate automation.
• There is now a prompt_on_exit setting. This might seem redundant to interactive but helps for semi-automated runs (interactive=False) letting the computer run and when coming back wanting to see what happened meanwhile in the console window. For a truly automated/scheduled experience you need to set both interactive and prompt_on_exit to False.
• I made - hopefully - arg parsing foolproof by using a separate temporary config_args.ini for that scenario. Thus the validation functions, which may alter and save the config, are prevented from overwriting the users' original configuration as are other cases where save_config_or_raise() might be called.
• What's to difficult to test for me is the self-updating functionality. I rewrote it to the best of my knowledge but since it references the original repo and such, only accepted PR and time will tell.
• Renamed use_suffix to use_folder_suffix to be more clear and better convey meaning.
• All config.ini options have finally been implemented as command-line arguments.
• Since everything is now properly defaulted and handled (I hope) you are now able to run Fansly Downloader with empty config files or no config.ini at all :D You just need the executable and are good to go.
• Als worked around an issue in Firefox token retrieval where differently encoded DBs woult throw sqlite3.OperationalError.
• Fansly Downloader now also logs output to file, at least all stuff going through loguru. Log-rollover is at 1 MiB size and keeping the last 5 files.

-- prof79