Nutrition Information

[dfrankland]

It'd be great to specify the nutrition of each ingredient and have that be tabulated to the total nutrition and per serving nutrition.

[edibotopic]

This would be great. If calculations from input macronutrient data (protein, lipid, carbs) are possible in the program then they could be converted to approximate caloric content using Atwater factors.

[pubmania]

While I am unsure if I would want the cooklang itself to have the ability to Parse nutrition data, it can be achieved for smaller projects fairly easily as I have done on my small site using the nutritionix api.

They are quite generous on limits for free use.

The advantage of this approach in my opinion is one does not have to manually enter the nutrition info on the recipe. Secondly, information from API is far more detailed than we could manually capture anyway, allowing for a better standardised visualisation of nutritional info as can be seen below:

[chrbala]

I'm currently working on building out my own API like this. I could potentially be interested in vending something to cooklang apps if there's interest.

[dfego]

https://support.cronometer.com/hc/en-us/articles/360018239472-Data-Sources

This app uses a few data sets. It looks like at least the USDA one is available to download and use.

[pubmania]

It looks like at least the USDA one is available to download and use.

The USDA data is available as API directly from official sources: https://fdc.nal.usda.gov/api-guide.html

[dubadub]

Yeah, definitely this was one of my biggest motivators to start Cooklang. It's paradoxical that while the nutrition balance is so important for the health, we have very little feedback on how we're doing.

Slowly we're getting there. But first we need to figure out different units and ingredient density.

[azlekov]

Hey @dubadub unfortunately the links here

Slowly we're getting there. But first we need to figure out different units and ingredient density.

are not working anymore. Is there any update on Nutrition Information?

I was thinking to use the metadata to put there all Nutrition Information, but looks like most of the parsers supports just string as a value, so I cannot group them. My idea was:

>> nutrition facts: [calories: 221kcal, carbohydrates: 52g, protein: 6g, fat: 1g, cholesterol: 2mg, sodium: 60mg, fiber: 4g, sugar: 41g]

Something which can be easily be parseable to dictionary.

The approach where in the meta I put all the facts in a separate item will require some additional work to group them for nice visualization later.

>> calories: 221kcal
>> carbohydrates: 52g
>> protein: 6g
>> fat: 1g
>> cholesterol: 2mg
>> sodium: 60mg
>> fiber: 4g
>> sugar: 41g

[dubadub]

Hey @dubadub unfortunately the links here
are not working anymore. Is there any update on Nutrition Information?

No updates yet, here is one of the links #51 (changed after moving from issues to discussions), second one is gone.

In long-term I think there will be two approaches for nutrition facts:

1. Generating estimates based on ingredients data (with some optimism that we can make computers do their job)
2. Using canonical metadata like you mentioned. Canonical keys for this not defined yet. You can share them with me or update this PR https://github.com/cooklang/spec/pull/69/files#diff-683343bdf93f55ed3cada86151abb8051282e1936e58d4e0a04beca95dff6e51R142-R152 to include this into Roadmap. After that we can update CLI and app to display this data nicely and include it into main spec.

[Cabalist]

@dubadub To your point on optimism this looks to be an interesting service on a small scale:
https://developer.edamam.com/edamam-nutrition-api

I don't think it is super relevant from a sustainable open source project standpoint but having a connector I could write that would pass it .cook files, get json back and then inserting it into the .cook files (in a canonical way) seems like something that just requires the 'canonical way' to be decided on. :)

It returns something like this:

{
    'yield': 6,
    'calories': 4453,
    'cautions': ['SULFITES', 'FODMAP'],
    'dietLabels': [],
    'totalWeight': 1839.536434375,
    'healthLabels': None,
    'totalNutrients': {
        'K': {
            'unit': 'mg',
            'label': 'Potassium, K',
            'quantity': 1578.6452949687496
        },
        'P': {
            'unit': 'mg',
            'label': 'Phosphorus, P',
            'quantity': 2115.28593625
        },
        'CA': {
            'unit': 'mg',
            'label': 'Calcium, Ca',
            'quantity': 3018.609738755267
        },
        'FE': {
            'unit': 'mg',
            'label': 'Iron, Fe',
            'quantity': 32.106330479852524
        },
        'MG': {'unit': 'mg', 'label': 'Magnesium, Mg', 'quantity': 233.821823},
        'NA': {
            'unit': 'mg',
            'label': 'Sodium, Na',
            'quantity': 6249.6787184375
        },
        'ZN': {
            'unit': 'mg',
            'label': 'Zinc, Zn',
            'quantity': 14.591718511249999
        },
        'FAT': {
            'unit': 'g',
            'label': 'Total lipid (fat)',
            'quantity': 248.98952886858143
        },
        'NIA': {'unit': 'mg', 'label': 'Niacin', 'quantity': 30.612364862375},
        'FAMS': {
            'unit': 'g',
            'label': 'Fatty acids, total monounsaturated',
            'quantity': 64.44136919228126
        },
        'FAPU': {
            'unit': 'g',
            'label': 'Fatty acids, total polyunsaturated',
            'quantity': 10.320202799843749
        },
        'RIBF': {
            'unit': 'mg',
            'label': 'Riboflavin',
            'quantity': 3.537692636937499
        },
        'THIA': {
            'unit': 'mg',
            'label': 'Thiamin',
            'quantity': 4.44658187165625
        },
        'VITC': {
            'unit': 'mg',
            'label': 'Vitamin C, total ascorbic acid',
            'quantity': 41.238846681249996
        },
        'VITD': {
            'unit': 'µg',
            'label': 'Vitamin D (D2 + D3)',
            'quantity': 1.5057465500000002
        },
        'CHOLE': {
            'unit': 'mg',
            'label': 'Cholesterol',
            'quantity': 301.26293072120785
        },
        'FASAT': {
            'unit': 'g',
            'label': 'Fatty acids, total saturated',
            'quantity': 72.11352240053547
        },
        'FATRN': {
            'unit': 'g',
            'label': 'Fatty acids, total trans',
            'quantity': 0
        },
        'FIBTG': {
            'unit': 'g',
            'label': 'Fiber, total dietary',
            'quantity': 23.323686719276683
        },
        'FOLAC': {
            'unit': 'µg',
            'label': 'Folic acid',
            'quantity': 654.8739841875
        },
        'FOLFD': {
            'unit': 'µg',
            'label': 'Folate, food',
            'quantity': 383.87059521875
        },
        'SUGAR': {
            'unit': 'g',
            'label': 'Sugars, total',
            'quantity': 27.4451861503125
        },
        'VITK1': {
            'unit': 'µg',
            'label': 'Vitamin K (phylloquinone)',
            'quantity': 99.60489525312502
        },
        'WATER': {'unit': 'g', 'label': 'Water', 'quantity': 831.965346323125},
        'CHOCDF': {
            'unit': 'g',
            'label': 'Carbohydrate, by difference',
            'quantity': 392.25124516902036
        },
        'FOLDFE': {
            'unit': 'µg',
            'label': 'Folate, DFE',
            'quantity': 1498.00685403125
        },
        'PROCNT': {
            'unit': 'g',
            'label': 'Protein',
            'quantity': 156.83993407281253
        },
        'TOCPHA': {
            'unit': 'mg',
            'label': 'Vitamin E (alpha-tocopherol)',
            'quantity': 11.041630610937501
        },
        'VITB12': {
            'unit': 'µg',
            'label': 'Vitamin B-12',
            'quantity': 7.3615287725
        },
        'VITB6A': {
            'unit': 'mg',
            'label': 'Vitamin B-6',
            'quantity': 0.7601271169375
        },
        'VITA_RAE': {
            'unit': 'µg',
            'label': 'Vitamin A, RAE',
            'quantity': 801.9469298125
        },
        'ENERC_KCAL': {
            'unit': 'kcal',
            'label': 'Energy',
            'quantity': 4453.754064853932
        },
        'SUGAR.added': {
            'unit': 'g',
            'label': 'Sugars, added',
            'quantity': 10.479000000000001
        }
    }
}

[dfego]

It's been quite some time since this was discussed. Are there any concrete steps you might suggest someone could take to help push this forward? I'm willing to spend some time trying to contribute!

[dubadub]

Yeah, I was thinking about key/value because I don't see any need to query collections of values from the database. In template where we use it we know the ingredient and know what kind of data we need. Hence jq syntax is just too much

[dubadub]

Well, the returned value can still be array or hashmap though

[dubadub]

What do you mean by scriptable? I'm familiar with code that works on templates (e.g. like https://pkg.go.dev/text/template). Regardless of how complex the input is, theoretically the parser should be able to take a template and plaintext data as input and produce output without need to recompile. But maybe I'm missing something!

Looks interesting, definitely can be an option and make things easier. I'm wondering if rust supports something similar.

[dfego]

I'm currently working on this a bit and running into some interesting challenges since I'm making it generic. Apparently YAML is very permissive in what it allows in, for example, key names. From what I can tell you can put basically any character imaginable in there. That creates the challenge of how to specify a string of keys (e.g. market.cost.tesco) in a string form with some kind of delimiter on our end, since market.cost.tesco could literally be the name of a key.

A more correct way to handle this is to have the interface explicitly take in a vector of relevant strings, i.e. vec!["market", "cost", "tesco"]. If I were to do that, you could enforce in the cooklang code what the format looks like and code to that.

tl;dr YAML permissive, creates ambiguity in the example use case. I'm currently moving towards making the actual API slightly more correct and generic, and leaving converting something market.cost.tesco into vec!["market", "cost", "tesco"] as an exercise for the user of the library, i.e. cooklang.

Just using this as a forum to dump some of my thoughts on the matter and also point out the direction I'm moving in case you have any issues!

[dubadub]

Interesting observation. Yes, good point, let's accept vectors. Are you on our Discord server? we may discuss things there and others can chip in?