If not in full page, open in full page here: Notion Wrapped | github.com/jesenator/notion-wrapped | pypi.org/project/notion-wrapped

<aside> <img src="/icons/gift_orange.svg" alt="/icons/gift_orange.svg" width="40px" /> Notion Wrapped! (not affiliated with Notion) Given an API key and a starting page, Notion wrapped recurses through every sub-block, tallying them up, noting the words used, the block relationships, and more to create a visual history of your Notion in the form of numbers, plots, word clouds, and network graphs!

</aside>

<aside> <img src="/icons/fireworks_red.svg" alt="/icons/fireworks_red.svg" width="40px" />

Network Graph (animated)

Network graph of my entire Notion - all 96 thousand blocks

• Full-size photo - Click on an image to expand it
• What is this? </aside>

<aside> <img src="/icons/cloud_purple.svg" alt="/icons/cloud_purple.svg" width="40px" />

Word Cloud

Word cloud of the top words in my 1,641,416 word Notion

• See the word cloud without the Notion logo (easier to see the words) </aside>

<aside> <img src="/icons/checklist_yellow.svg" alt="/icons/checklist_yellow.svg" width="40px" />

How to make your own Notion wrapped!

1. Acquire a Notion API key for your Notion Workspace
   • How to
2. Give the API key access to the relevant pages
   • How to
3. Install Python if you don’t yet have it
4. Install the Python library with pip install notion-wrapped (make a Python environment first! - ask chatGPT how if you need help)
5. Run notion-wrapped with the required command line arguments (run notion-wrapped --help for help)
   • Example
6. See your plots and analytics the analytics folder!
7. Share on social media and tag #NotionWrapped and this page (jessewgilbert.com/notion-wrapped)!

Btw: I don’t store or use any of your data!! It is all transient in the Python program

</aside>

<aside> <img src="/icons/calendar-month_green.svg" alt="/icons/calendar-month_green.svg" width="40px" />

Blocks created / edited per day

Somehow there is a block in my Notion workspace created in 1970… Kind of messed up the visualization haha

After filtering out that block (and smoothing). This looks better :)

</aside>

<aside> <img src="/icons/pen_orange.svg" alt="/icons/pen_orange.svg" width="40px" />

Top creators / editors in my workspace

blocks created by them / blocks last edited by them

 - Jesse Gilbert: 85551 / 87865
 - Liam: 692 / 547
 - Notion Calendar: 602 / 0
 - Julia: 554 / 406
 - Lydia: 451 / 408
 - Emma: 392 / 236
 - Vanessa: 145 / 153
 - Zapier: 102 / 0
...

• See more </aside>

<aside> <img src="/icons/mobile_yellow.svg" alt="/icons/mobile_yellow.svg" width="40px" />

Types of blocks

</aside>

<aside> <img src="/icons/metronome_red.svg" alt="/icons/metronome_red.svg" width="40px" />

Blocks created / edited per minute of the day

</aside>

<aside> <img src="/icons/timeline_blue.svg" alt="/icons/timeline_blue.svg" width="40px" />

Other Stats

Total Block Count: 96,328

Total Word Count: 1,641,416

Max Recursion Depth: 23

Time Spent on Notion Estimate: 1454 hours / 60 days

Average blocks created or edited per day: 126.33

Median blocks created or edited per day: 89

Days with zero blocks created or edited: 21

Total number of days using Notion: 1525

Program Execution time: 03:38:19.55

</aside>

⚙️ How it works

<aside> <img src="/icons/sync-reverse_brown.svg" alt="/icons/sync-reverse_brown.svg" width="40px" />

recurse.py class

• Starts with a list of IDs of the top-level pages
• Function recurse gets the children of a certain block and recursively so it calls recurse on each of the direct children
• After my initial implementation, I added more modular functionality to the recurse function in the form of a mapping function and reducing function parameters.
  • Mapping function: This lets you pass an arbitrary function which every block is independently passed into (in addition to specific metadata like the depth of the block and what number child it is)
  • Reducing function: Lets you pass an arbitrary function that “reduces” all blocks to a single value
    • The reducing function does two distinct things to do this, each of which can be specified separately
      1. Applies a function to the current block
      2. Takes the existing output of prior reductions and combines it (with some function) with the output of step 1 or an initial value if the existing output is None.
• Other features
  • Multithreading!!
    • You can specify the number of workers and it it keeps at most that many distinct threads of the recursion running
    • Increases speed by ~4 times, could be SO much faster (1000x or more) if not for API rate limiting by Notion 🥲 (but I don’t blame them)
    • This was hard to make!
  • Caching
    • It can save blocks as it goes to a JSON file and can read through that instead of using the Notion API
    • This was especially useful for testing the big network graphs. Made recursing through my Notion take 10 seconds instead of 3 hours! </aside>

<aside> <img src="/icons/search_pink.svg" alt="/icons/search_pink.svg" width="40px" />

analytics.py class

• If you want to just run the analytics you pass the add_block function from analytics.py as the mapping function in recurse.py.
  • I thought this was a particularly elegant way of structuring these two classes
• Tallies the blocks and creates all of the plots, graphs, word clouds, and statistics you see above.

Terminal output - you can see the structure of the recursion here

</aside>

<aside> <img src="/icons/calendar-month_green.svg" alt="/icons/calendar-month_green.svg" width="40px" />

Network graph with PyVis

• It starts with a twopi layout (powered by Graphviz) which sets up nodes so they are already roughly grouped where they should be, so the physics simulation can converge more easily
• The simulation I’m using with PyVis is the Barnes Hut Algorithm which is a force-directed graph drawing algorithm and is the one I found to end up the prettiest
• It has a notion of gravity between the nodes, repulsion to avoid overlap, and models the edges as springs with adjustable spring lengths, and spring constants </aside>

Getting the animation and high-fidelity photos of the network

• Used the node library puppeteer to load the webpage in headless mode, at a much larger size
• Clicks to start the physics simulation → takes a photo every x seconds for y hours → takes on really high fidelity photo at the end
• A Python program uses FFmpeg to combine the screenshots into a video

Other files

• notion_wrapped.py - makes the classes into an accessible command line tool
• utils.py - various utilities used by both classes
• notion_client.py - all the code for interfacing with the Notion API

Progression of the network over time

Smaller, interactable, and anonymized version of the Notion network graph

https://prismatic-parfait-599ad4.netlify.app/

I ended up printing a poster of it!

I’m really happy how it came out. I love being able to walk up really close to it to see all the individual nodes and connections.