Etiquette is a tag-based file organization system with a web interface, built with Flask and SQLite3. Tag-based systems solve problems that a traditional folder hierarchy can't: *which folder should a file go in if it equally belongs in both?* and *how do I make my files searchable without littering the filenames themselves with keywords?*
Etiquette is unique because *the tags themselves are hierarchical*. By tagging one of your vacation photos with the `family.parents.dad` tag, it will automatically appear in searches for `family.parents` and `family` as well. A traditional folder system, here called albums, is available to bundle files that always belong together without creating a bespoke tag to represent that bundle. Regardless, the files on disk are never modified.
- The repository you're looking at right now is `D:\Git\Etiquette`. The toplevel `etiquette` folder is the main package. We want this package to be a child of our existing lib directory.
- The easiest way to find your lib path is `python -c "import os; print(os)"`.
- The repository you're looking at right now is `D:\Git\Etiquette`. The toplevel `etiquette` folder is the main package.
The pythonpath points to directories that *contain* the packages you need to import, not to the packages themselves. Therefore we point to the repository.
This only applies to the current cmd session. To make it permanent, use Windows's Environment Variable editor or the `setx` command. The editor is easier to use.
- Run `python -i etiquette_repl_launch.py` to launch the Python interpreter with the PhotoDB pre-loaded into a variable called `P`. Try things like `P.new_photo` or `P.digest_directory`.
- Note: Do not `cd` into the frontends folder. Stay wherever you want the photodb to be created, and start the frontend by specifying full file path of the launch file.
1. Use the PYTHONPATH technique to make `etiquette` and the flask `backend` both importable. Symlinking into the lib is not as convenient here because the server relies on the static files and jinja templates relative to the code's location.
The Pythonpath points to directories that *contain* the packages you need to import, not to the packages themselves. Therefore we point to the etiquette and etiquette_flask repositories.
You may notice that Etiquette doesn't have a version number anywhere. That's because I don't think it's ready for one. I am using this project to learn and practice, and breaking changes are very common.
The Etiquette library is designed to be usable through a variety of interfaces. The Flask interface is my primary focus and does, admittedly, have an influence on the design of the backend, but other interfaces should have no trouble integrating with Etiquette. Every folder here is essentially a completely separate project which imports etiquette just like any other dependency.
- Currently, the Jinja templates are having a tangling influence on the backend objects, because Jinja cannot import my other modules like bytestring, but it can access the methods of the objects I pass into the template. As a result, the objects have excess helper methods. Consider making them into Jinja filters instead. Which is also kind of ugly but will move that pollution out of the backend at least.
- Perhaps instead of actually deleting objects, they should just have a `deleted` flag, to make easy restoration possible. Also consider regrouping the children of restored Groupables if those children haven't already been reassigned somewhere else.
- Add a new table to store permanent history of add/remove of tags on photos, so that accidents or trolling can be reversed.
- Extension currently does not believe in the override filename. On one hand this is kind of good because if they override the name to have no extension, we can still provide a downloadable file with the correct extension by remembering it. But on the other hand it does break the illusion of override_filename.
- When batch fetching objects, consider whether or not a NoSuch should be raised. Perhaps a warningbag should be used.
- Find a way to batch the fetching of photo tags in a way that isn't super ugly (e.g. on an album page, the photos themselves are batched, but then the `photo.get_tags()` on each one is not. In order to batch this we would have to have a separate function that fetches a whole bunch of tags and assigns them to the photo object).
- Consider using executemany for some of the batch operations.
Here are some thoughts about the kinds of features that need to exist within the permission system. I don't know how I'll actually manage it just yet. Possibly a `permissions` table in the database with `user_id | permission` where `permission` is some reliably-formatted string.
- Preventing logged out users from viewing any page except root and /login.
