feat(receipts): import scanned purchases into Notion

[MaximilianMauroner]

Entering purchases from paper receipts into Notion takes manual transcription and often loses product, price, category, or date data.

Proposed change

Read a receipt image, extract its purchase data, let the user correct the result, and write the confirmed items to the configured Notion database.

Acceptance criteria

• The importer extracts merchant, date, line items, prices, and total when present.
• Users can edit extracted values before import.
• Imports support categories and tags.
• Failed or uncertain fields are marked for review.
• The same receipt is not imported twice.

[MaximilianMauroner]

Implementing a full-featured computer vision aided invoice reader and parser with Notion integration in Rust would be a significant undertaking that would require a significant amount of time and effort. Here are some high-level steps that could be taken to implement such a system:

Choose a Rust computer vision library/framework: There are several Rust libraries available that can help you with computer vision tasks such as image processing, object detection, and machine learning. Some popular Rust computer vision libraries/frameworks include OpenCV-rust, RustyCV, and TensorFlow Rust. Choose one that is suitable for your needs.

Train a machine learning model: In order to identify and extract purchase details from an invoice, a machine learning model would need to be trained on a dataset of labeled invoices. This dataset would need to be annotated with information such as product name, price, and other relevant details. Once the machine learning model is trained, it can be used to process new invoices and extract the relevant data.

Integrate with Notion API: Notion has a public API that can be used to programmatically create, read, update, and delete data in a Notion database. You would need to authenticate your application with Notion and use their API to insert the extracted purchase details into the appropriate Notion database.

Build a user interface: To make the application user-friendly, you would need to build a user interface that allows the user to take a picture of an invoice, view the extracted purchase details, edit or confirm the details, and insert them into the Notion database.

Test and iterate: Once the application is built, it would need to be thoroughly tested to ensure that it is functioning correctly and meeting the user's needs. Iterate on the application based on user feedback and refine the machine learning model as needed.

Note that this is a high-level overview of what would be involved in implementing a computer vision aided invoice reader and parser with Notion integration in Rust. The actual implementation would require more detailed planning and coding work.