Skip to content

Latest commit

 

History

History
443 lines (329 loc) · 16.8 KB

README.rst

File metadata and controls

443 lines (329 loc) · 16.8 KB

Price And Nutrition Tracking System

Note for Current Users - DB Changes when Upgrading

Changes to the database schema have been made recently (2020-04-13) and when upgrading from older versions you will need to migrate your database by running './manage.py migrate' - see the "Schema Changes" subsection below for more details.

Introduction

PANTS is a self-hosted, open-source nutrition tracker and a tool for nutritional data analysis of ingredients and recipes. It can be run on your own computer or as a multiuser web service (e.g. for use by a dietician/trainer/researcher and their clients).

As well as tracking daily calories etc, PANTS is designed to make it easy to compare and optimize recipes which form a regular part of your diet; a key feature is the ability for recipes to be components of other recipes.

For example a dough recipe and a filling recipe can both be components in a pie recipe; you can clone the pie recipe with alternate fillings to compare the nutritional values of the alternatives. Any changes to the dough recipe will be reflected in the data for all of the pies. These "meta-recipes" can also be used to provide an "average" breakfast/lunch/snack/etc for meal and diet planning.

PANTS is under active, daily use by the author so updates should be fairly frequent. On the other hand, the code tends to be quick and dirty as new things get added because the author wanted to use them ASAP while cooking his dinner.

It is currently not recommended for non-technical users; basic familiarity with Django is useful.

Use Cases

In particular PANTS is designed for:

  • Tracking the calories and other nutrients in your food, especially things cooked frequently.
  • Storing a list of recipes, and determining the cost and nutritional values of a recipe. This includes recipes which are used in other recipes (recursively) e.g. a dough recipe and a filling recipe can both be components in a pie recipe.
  • Comparing the cost and nutritional values of different ingredients or recipes (e.g. "What food has the most protein-per-dollar?", "Which of these recipes has the least calories but is still high in fibre?").
  • Long term dietary planning - working out which ingredients or recipes you want to include regularly in your diet, based on cost and other factors.

PANTS is not ideal for:

  • Tracking takeaway/restaraunt meals or prepackaged snack foods (rather than stuff you prepare yourself from basic ingredients)
  • Determing the best place to buy a shopping list of items (cost is not designed to be updated regularly or automatically, it is there for long-term planning).
  • Comparing the cost of ingredients across different regions.

Overview and Features

There are 4 basic sections of the system:

Ingredients
Generic ingredients, like "Rolled Oats", "Green Split Peas", "Skim Milk", not any particular brand or store. You can compare relative data such as Protein per Joule or Fibre per Dollar on the ingredients pages.
Recipes
A collection of ingredients and/or other recipes, showing the combined nutritional value and cost of each serve of the recipe, as well as the relative per-calorie and per-dollar values.
Diary
Record of food intake; compares the total to your target and also shows how much each food contributed to your total for the time period.
Targets
Minimum and Maximum nutritional and cost values you are aiming to reach each day. As well as daily targets you can also set targets for particular meals (e.g. a pre/post workout meal), or special days.

Some example screenshots can be seen at https://github.com/dylanleigh/PriceAndNutritionTrackingSystem/tree/master/screenshots

Other Feature Notes and Tips

Tagging

Ingredients and recipes can be given multiple tags for easier searching and analysis; this makes it much easier to (for example) compare the fibre per calorie in different vegetables or the calories per serve of different dessert.

Recipes can also be "flagged" to mark them as untested, tested and working, requiring further improvement, outdated, etc.

Repurposing Recursive Recipes

As recipes can include other recipes, this can be used for other analysis and planning purposes.

For example. an "average breakfast" meta-recipe which just contains one of all the other breakfast recipes; divided by the number of serves this provides an average breakfast which can be used for planning. This can be combined with other "typical meal" meta-recipes to make an "average day" overview which can be used as a meal plan.

This can also be used as a sort of "variables" in other recipes, e.g. if you sometimes use normal flour or gluten free flour, a "flour" recipe can be created which can be used to toggle between them with one change which effectively toggles the ingredient in multiple recipes at once.

PANTS doesn't make assumptions or guesses
It is preferable to show no data instead of wrong data. If an ingredient has something missing (e.g. no fibre listed, no prices), any derived statistics will also be missing (e.g. no fibre-per-kj, or no protein-per-$ if there is no price). This also means that recipes which use that ingredient will not show a value for the sum of fibre in that recipe until all ingredients have that data entered in.
PANTS doesn't assume everyone sleeps at midnight
Diary shows breakdown of nutrients by both calendar day and 24-hour periods, so it can be used by shift workers or those with irregular sleep cycles.
Recipes update; Diary doesn't change
Diary entries are "crystallized" (future changes to a recipe do not affect past entries). On the other hand, changes to an ingredient/recipe immediately show up in any recipes which use them.
Once-off Diary entries
Diary entries do not have to be linked to a specific ingredient/recipe, one off diary entries can be created with manual nutritional data e.g. when going out for the night and you can only guess how many calories are in dinner.
Per-user and global data
Ingredients and Recipes added by the admin are visible to all users (but not editable by them); Normal users can also enter in their own Recipes and Ingredients, which only they have access to (sharing may be added in a later version).
Progress/Percentage bars
Once your default target is set, it will be used to compare against ingredients/recipes you view so you can see how much % of your daily intake that recipe/ingredient will satisfy. On most pages, Green/Yellow/Red bars indicate how much of the daily target's minimum/maximum are accounted for; Purple progress bars are used to show percentage out of the current total, or amount compared to the highest value in a list of recipes/ingredients.
Micronutrients
All Australian standard nutritional data is stored (e.g. sodium and saturated fat) but not everything is shown in all views by default. There was support for micronutrients such as individual amino acids which was removed as part of a DB schema change but this is planned to be readded in a more stable way (see roadmap for details).

Installation

git clone https://github.com/dylanleigh/PriceAndNutritionTrackingSystem.git
cd PriceAndNutritionTrackingSystem

VirtualEnv

It is highly recommended that PANTS is installed in a virtual environment, and comes with a requirements.txt for pip:

python3 -m venv env
source env/bin/activate

Requirements

pip install -r requirements.txt

These will be installed automatically via pip:

  • Django 2+
  • Django-extensions 1.8+
  • Django REST Framework
  • Django-filter

Setup 1: Secret Key

The environment variable "PANTS_DJANGO_SECRET_KEY" needs to be set for PANTS/Django to start.

The exact place to set this will depend on the OS and environment you are using. For hosted infrastructure such as AWS or Heroku this can be set in the instance settings; for local installs you can set it as part of the virtualenv activation script or as a variable on the account of the user who will be running it:

echo export PANTS_DJANGO_SECRET_KEY='968af690a7bcca77c9261e395885af77bc661d1c' >> ~/.profile

You can generate an appropriate secret key from the SHA1 of any randomly chosen phrase or file:

$ echo blahblahblah | sha1sum
968af690a7bcca77c9261e395885af77bc661d1c  -
$ export PANTS_DJANGO_SECRET_KEY='968af690a7bcca77c9261e395885af77bc661d1c'
Warning
The Django Secret Key is used to generate session tokens and other cryptographically important things. Keeping it in an environment variable makes it easier to have seperate, secure secrets on different installations. If someone knows the secret key it may be possible to login as admin users and mess with things, so keep it secret.

Setup 2: Migrations and Admin User

Finally you will need to run initial migrations and create an admin user who can log in and create the initial ingredients, recipes etc:

cd pants
./manage.py migrate
./manage.py createsuperuser

Starting the Server

To run the server locally and access it via a browser:

./manage.py runserver

Sample Ingredient Data

The author's ingredient data (about 200 ingredients as of 2019) can be imported from a fixture with this command:

./manage.py loaddata fixtures/pants-ingredient-fixture.json

This command should only be run on an empty/new database, to avoid overwriting any entries you have already created

Starting data entry

You will need to log in as an admin user (at http://127.0.0.1:8000/adminbackend/ ) to start creating initial ingredients, and then recipes which use those ingredients (and recipes which use those recipes...)

The about page will show some basic DB stats, including the count of ingredients which are missing nutritional data and other potential issues.

No ingredients/recipes need to be created to start using the diary (although every entry will have to have all its data added manually if there are no recipes or ingredients to use).

Adding Non-Super Users

TODO this needs to be documented for API consumers.

API

This is a work in progress as of 2020-04-13. Documentation will go here when it's implemented for all major models. It is located at /api/1/ (i.e. http://127.0.0.1:8000/api/1/ on a local server).

History

PANTS grew out of a spreadsheet I was using in early 2017 to do nutritional analysis of different foods, looking for the best ratios of protein and fibre to calories and cost.

I wanted to add recipes which combined different ingredients and this became so cumbersome I realised it would be easier to do in a DB and started the project in Django, importing the initial set of ingredients from the spreadsheet. Soon I also realised since I was entering in all my recipes here it would also be easier if I used it as my daily calorie counter and added that as well.

Schema Changes

For all changes mentioned here, your database must be migrated by running the following commands:

git pull
./manage.py migrate

No further user input or manual conversion should be required. The details below are mostly for background.

2020-04-13

Recipe and Ingredient can now be linked to an "owner" (user) - user created recipes and ingredients through the API will be owned by that user. Only the logged in user can see/edit things they own.

"Global" recipes/ingredients with no owner are visible to everyone, and only editable by admin (i.e. no change from previous versions)

2020-04-02
Each Recipe and Ingredient may now have an "Introduction" and "Notes" - these are freeform text fields that are simply displayed at the start/end of the detail page for the recipe or ingredient.
2020-03-20 (v0.93)

Following on from yesterday's changes, Price has now been fully detached from Product. This update also changes Prices to require an Ingredient set (during the migration, this was optional to allow data to be migrated automatically).

If there are errors applying this migration it is probably because there are Price objects which don't have an Ingredient. The last version should have converted all the old ones automatically, and converted any new ones created in the admin when they were saved. However, if by some chance you have any corrupt prices not linked to an ingredient, these will have to be deleted for the migration to work.

The product model still exists, but is now deprecated; it has no current purpose except to associate brand names with ingredients. If you don't care about that, products can be all safely deleted via the admin interface (use the checkbox to "select all" and then drop-down action box to "delete selected"). They should not be any performance effects from leaving them there, as no calculations use products anymore.

2020-03-19 (v0.92)

Prices are changing from being attached to a Product to directly being attached to Ingredient, to simplify both the user interface and the code.

As of this version, Price is attached to both Product and Ingredient. Ingredient will be set automatically from the Product.

Future versions will make Price settable via the Ingredient section of the admin interface rather than Product.

2020-03-11

Recipe Flags added. These differ from Tags in two ways - each recipe can have only one flag, but flags are much more visible (being shown in lists etc).

The intended use case is to mark recipes which are OK for general use to differentiate them from recipes which aren't working and need further changes and testing, or outdated recipes no longer recommended. However, they can be used for whatever the admin wants.

Also, tags for Recipe and Ingredient can now have a brief text description which is shown in list view when that tag is selected.

2019-09-07 (v0.91)

Recipe Components now have separate "servings" and weight" to bring them in line with the way all other models work (previously, "weight" was interpreted as number of serves if connected to a recipe).

This fixes various issues, including data entry errors from overloading one field to have two meanings and allows a lot of the code between ingredient/recipe/diary to be simplified.

Existing recipes will be converted to this new system by recipes/migrations/0018_auto_20190908_0152.py when the migrate command is run.

Removed Features

Products (partially)

After the nutrient model was merged into Ingredient, Product lost it's ability to have separate nutrient data, and it was just a redundant way of linking prices to ingredients, which is now done directly.

The product model still exists in the admin, but currently has no purpose except to associate brand names with an ingredient. It may be used again in the future for storing further data about a specific branded product.

Collections
Never properly implemented; need for this is reduced by heavier use of tags, creative use of recursive recipes (e.g. a "daily meal plan" as a "recipe"), better frontend comparison tools and CSV export to spreadsheet for doing analysis there.
Plots
Recipes/Ingredients now have a CSV export button, use that to create charts externally via a spreadsheet.
Amino Acids
The original design could handle detailed micronutrient stats (including individual amino acids, minerals, EFAs, fibre types) but when the nutrients object was merged into ingredient this was dropped. It was rarely used, but may be readded when the nutrient_data class/cache system is reworked to be less kludgy - see TODOs

Developer Notes

As mentioned earlier the code contains many crufty bits because many features were added quickly when immediately required.

In particular, sets of "nutrition data" are often passed around as a dict with a few specific sets of keys (specified in settings) and there is an ongoing project to convert this to a class that manages it in a sane way, handling all comparisons, additions and per-weight calculations sensibly. A lot of future work is on hold pending this tech debt cleanup to be completed.

See the todo list below for more details.

Bits Useful for Other Projects

  • Recipe/Ingredient have a very simple but effective CSV export view.
  • There are convenient templatetags to do division, combined min/max percentage display and generate a little CSS bar chart (most tabular data uses them, see the screenshots for examples).

Roadmap, Todos and Issues

See https://github.com/dylanleigh/PriceAndNutritionTrackingSystem/blob/master/TODO.rst

Authors

Dylan Leigh 2017-2019