(Draft & Call for feedback) Revamp of product addition in a wizard/pipeline form

[teolemon]

I have started working on two things that are a revamp of product addition in the form of a pipeline as well as packaging input as part of the time-sensitive ADEME project. This is a grand vision that is and can be broken into smaller parts that are less intimidating.

You can see the whole pipeline in Figma to get a bird eye view of this draft, but please, let's discuss that thoroughly (ideally in a video call) before doing anything.

Steps of the pipeline

Before entering product addition in the scan screen

• Handle the various completion rates on the scan card with custom screens
  • a screen for products that don't exist in any of the Open x Facts projects
    • 
    • This implies at the very minimum being able to change the base URL on the fly
    • This also implies substantially modifying the behaviour of product addition
  • a screen for products that have neither Nutri-Score, nor Eco-Score, nor category
    • Some products might have a category that does not allow for a Nutri-Score nor an Eco-Score
  • a screen when one of your mandatory criteria is missing e.g. gluten
    • We can have deep links from attributes in the extended product page (including the new contribution attributes: Extend the attributes system to add a section for contribution openfoodfacts-server#7608)
      • 
    • We can have a scan card button to open a specific completion pipeline
• Add a screen to mention to which project the product should be added to

Before entering product addition in the product page

• Increasing the roads to editing with Add more actions to panels openfoodfacts-server#6987 (will work and use the current granular system, which we will keep, with better indicators of completion status for each section)
  • 
• Adding a button to open the full pipeline mode
• Adding button to open score-oriented pipelines (for more on that, keep reading)

About the long size of the wizard

• The idea is that in product completion mode, some of the following sections will be hidden if they are already completed in a satisfactory manner.
• The idea is that in product addition mode, all of the following sections will be shown, unless the category makes them unecessary (eg the fishing zones/method screen will be hidden for 95% of products)
• We could call product completion mode with arguments (nutriscore, nutriscore+ecoscore, nutriscore refresh…) and show/hide screens based on this+completion status
• We could make this pipeline a separate package that 3rd party apps could integrate (and call with aforementionned arguments)
• In the future, we could even start driving this pipeline from the server (eg the Eco-Score label list could be fetched remotely, the list of screens needed for a refresh could be provided by the server, based on the freshness of fields, for non food products, we could drive specific screen based on the product category, for food product we could add some specific Folksonomy Engine based screens, for food products, we could customize the nutrition table based on the product category, eg add cocoa for chocolate, minerals for waters…)
• I am not sure about editing mode, since people might want the choice of random access, as opposed to being forced through a pipeline
• By any mean, we should not deprecate the random access mode and views we currently have because
  • some people will like them better
  • they will still be leveraged by Knowledge Panel actions ("Click to complete origins")

How to start

• Let's create a new pipeline that can be enabled from DEV mode, and provides a second button on the scan card when the product is missing

Photo addition wizard

• Revamp initial photo acquisition with server-side cropping and the possibility to add a photo in other languages
  • 
  • Add an option to mention that there is no ingredients or nutrition or packaging or origins on the packaging (requires server-side fix for packaging, ingredients, and origins, already available for nutrition)
    • 
    • 
    • 
• Send the whole image and perform server-side cropping so that we get more of the packaging (useful for machine learning)
  • Add a warning that "The whole image is sent to the server, and not just the crop. Feel free to retake the photo if there are problematic elements in the background"
  • Add client side (using ML Kit) segmentation of the product
• Add an onboarding screen for the packaging photo
• Add an option to add a photo of origins mentioned on the packaging #3297
  • Make photo addition multilingual
    • 

Basic details input wizard

• Simplify the basic details input by removing brand and weight fields, and offering Robotoff insight fetching and validation instead (with possibly a fallback ?)
  • 
  • 
  • Simplify the basic details input by using Robotoff predictions #3279
• Make basic details inputs multilingual
  • (Edit) Make it possible to edit the ingredients, product name, and packaging text in several languages #2422
  • On a French device, with French and France as settings, product_name is saved as product_name_en #3042

Category input wizard

• Add Robotoff category insight validation within the product addition pipeline
  • 
• Add a category selection screen in the pipeline in case no category has been validated from Robotoff
  • We could set on the most granular category available if we do have a category, but no Eco-Score, with an option to tell that the category is already very specific to move on to the next step
    • We could probably add a server side toggle for that, but that seems error prone.

Ingredients input wizard

• Overview
  • 
• Add a container screen for ingredient extraction and ingredient origin input
  • 
• Make the ingredients extraction screen multilingual
  • Make it possible to edit the ingredients in several languages #2564
• In the product addition wizard, auto-extract ingredients and add a prompt to check them ("I have checked it", "Skip")
  • 
• Add server side support for no ingredient list status: Allow to mention that a field is complete openfoodfacts-server#7612
• Add a button to mention no ingredients are visible

Origins of ingredients structure input

• Moved to Implement full/guided ingredients origins input (tracker) #517
  • Add a one-time NOVA + ingredient analysis explainer after ingredients input + showing the points normally granted by the openfoodfacts-events points system (blocked at least on Send edit events to openfoodfacts-events openfoodfacts-server#6259, otherwise, we would just pretend the points were granted)
    • 

Nutrition input wizard

• Moved to Nutrition input (Tracker) #511

Eco-Score wizard (at this point all data should be already here)

Eco-Score Labels input wizard

• Moved to Allow to positively confirm the presence or absence of all Eco-Score labels #3295
• we could put labels earlier as some people are likely to drop because of nutrition input

Packaging

• Moved to Implement guided packaging input (wizard-like) #515

  • Add an Eco-Score thank you screen + showing the points normally granted by the openfoodfacts-events points system (blocked at least on Send edit events to openfoodfacts-events openfoodfacts-server#6259, otherwise, we would just pretend the points were granted)
    • 

[teolemon]

Should I move this to a Google Document (for easy and granular comments)? do we comment on Figma ? Do we comment here ?

[M123-dev]

Adding my comment from slack here to keep all infos central.

I skimmed over it and I it looks great, though before we can start with such a big task we should stabilize the current rough edges of the app.

[AshAman999]

Looking at the designs and a new flow, it would take some time to implement and make it run smoothly, so as @M123-dev said, we have to make sure the current version on the play store and app stores are okay to go , So in the meantime, the user's don't face any problems

[monsieurtanuki]

@teolemon That's a lot.
I agree with @M123-dev and @AshAman999, the priority now would be to have a stable version, e.g. with a quiet new release every week.

[anshpathania7]

Looks pretty good, though is this gonna be a separate module? or it will be written directly inside smooth ?

[teolemon]

I'm going to try to extract low hanging fruits.