Spbased

NOTE! This project is still a work in progress. There are some features still left on the table before a 1.0 release.

Spbased is a content agnostic spased repetition tool. It only knows about the notion of a generic "review item", an object with an id, parameters for the spased repetition algorithm, and some json data. The full schema can be seen below.

CREATE TABLE IF NOT EXISTS item (
    id INTEGER PRIMARY KEY,
    maturity TEXT NOT NULL DEFAULT "New",
    stability REAL NOT NULL DEFAULT 0.0,                      -- sra parameter. The number of days since last review date until probability of recal reaches 90%
    difficulty REAL NOT NULL DEFAULT 0.0,                     -- sra parameter. Number between 1 and 10. Meausure of item difficulty
    last_review_date TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP, -- sra parameter. Date in iso8601
    n_reviews INTEGER NOT NULL DEFAULT 0,                     -- sra parameter. Number of times we've review and given the review a score.
    n_lapses INTEGER NOT NULL DEFAULT 0,                      -- sra parameter. Number of times we've failed to recall the item.
    model TEXT NOT NULL,                                      -- the model, tells us how data is to be interpreted
    data TEXT NOT NULL,                                       -- json data
    updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,       -- metadata
    created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP        -- metadata
);

It is built as a command line tool, to be wrapped in custom scripts for more advanced or streamlined usecases.

Quick rundown of vocabulary

• "review item" - the flashcard,prompt,task etc. being scheduled in a spased maner.
• "review" - (in the context of spbased) given a review item that is due, launch the program that can interpret it, review it and return a measure of how well it went.
• "review item model" - or just 'model', the specific type of review item, and the format of its data

Command line usage

Pre filter expression language

The commandline query commands (review command included) can be passed a --pre-filter flag which then takes a small domainspecific language as an argument.

The language allows for a simple combination of comparison and logical operations. Comparisons between some review_item field and value which may then be combined with logical operations.

• --pre-filter 'id==3'
• --pre-filter "maturity=='Young'"

Supported values include strings, integers, floats and booleans.

Some special handling is taken on fields which represent time. Here care is taken such that the value is a valid point in time. Here we may add semantic descriptions of time in the future (e.g "three days ago")

Post filter

Most commands that return a json result can also be passed a --post-filter flag which takes a jmespath expression. This can be done to further extract any wanted data.

> spbasedctl items query --pre-filter "maturity='Young'" --post-filter "[*].id"

whose output may be

[1, 2, 42, 4]

Examples

This repo also contains a few example scripts that showcase how the tool can be used.

TODO:

• edit tags on item
  • simple add_tags and remove_tags (if exists)
    • create query that adds tags to item
    • create query that removes tags from item
    • add docstring to flags
• some way to filter on tags needs to be implemented
  • simple include_tags and excludes_tags NO domainspecific language just for the sake of it
    • map include tags and exclude tags to sql
• way more tests!
• add more scripts
  • add_flashcard
  • edit_flashcard
  • delete_flashcard
  • review_flashcard
• start using the application!

Wish list:

• spbased 'open this program with this metadata'

  • This would allow the content agnostic core to become really powerful
  • json content of an item could include a program field
  • thought really just what I'm doing currently

• spased reading

  • add a list of pdfs, webpages

Description