Job Stories for Flowershow

Refs ideas/flowershow

What we want in terms of rendering and behavior

I'm going to work through here the job stories for what we want to be able to do with concrete examples where possible.

Job story list

• Auto-extracting display relevant metadata in markdown Auto-handling various cases of title e.g. as h1, file name, in title frontmatter etc so that … a) render page nicely b) generate SEO info automatically for your pages (based on attributes above for example) c) create listings
  • Display the date from front matter metadata if exists
  • Show last modified date based on Git commit history or file metadata
  • Extract and display the primary image in a structured way e.g. Use the first image in the document as the default if no front matter image is set
  • auto-extract a description if not set …
• Listing Pages e.g.
  • Generate an index page for blog posts
  • Create a directory listing for structured content (e.g. projects, datasets) - see https://www.portaljs.com/blog/create-a-simple-catalog-of-anything-using-markdown
• Get a list of all the files (for showing in sidebar)
• Rendering Data Files
  • Render a CSV file as a table
  • Render an Excel file as a table or structured data display
• Display a data package with metadata
• Embedding Data Previews
  • Embed a preview of a CSV file within another Markdown document
• Show a data file with or without its schema
• Link/Network Graphs
  • Display backlinks for internal references
  • Show forward links to related documents
  • Render a visual graph of document relationships
• Full text search
• Image optimization

Future

• Direct upload to r2 (especially of large files)

What bothers rufus today …

• Lack of search …
• Inability to do directories and blog sections
  • a common workflow for rufus is to edit on github itself and then i have to create post and then go to indx page and add link …
• Lack of good SEO preview (or default preview image from flowershow/datahub)
  • would like auto selection or even auto generation …
• Link / network graphs woudl be nice …
• ![](path to csv) and get a table preview of a csv file
• I would love to dump a data file on flowershow or datahub and just get a table and a search interface (e.g. sql to it …)

Be able to make minor-ish changes / improvements easily (without hassling ola)

Job stories (detailed)

Title Handling

• Extract title from the first H1 tag if no front matter title is provided
• Use the title from the front matter if available
• Handle cases where no title exists (use file name)

Creating a directory / catalog of other files

Examples: https://developmentalspaces.org/examples https://secondrenaissance.net/ecosystem/pip and https://www.portaljs.com/blog/create-a-simple-catalog-of-anything-using-markdown

When I have a list of files e.g. in a folder or selected by some query I want to display a list of these in another page so that i can have a listing of e.g. my blog posts, my ideas, my movies, my books read and more …

In simplest case: take all files in folder X and display "cards" for each item. For it to work each item in folder X should have: title, description, image (optionally author(s))

<Listing folder="blog" format="card" />

This should render like the blog posts on portaljs.org

Other rendering options (we can take inspiration from dataview):

• Bulleted list
• Table

In more complex case want to provide a query to Listing and use results of that for the listing (again ala dataview).

Extras

• Customization of the fields displayed
• Pagination (do we handle that …) – my sense is that for most minimal use cases this is not a problem especially if we optimize images.

Blog post display

When i create a blog post entry (with layout)

🅿️ Parking lot

• Dig out my notes about creating a directory and why markdown is good for this

Plan of Work

• List out all the job stories i can think of 🚧2025-03-10
• Prioritize the ones i want to detail
• Work on detailing those

Architectural implications

Take the title case … we don't want the frontend having to figure out all the options. We want that auto-computed and the frontend can just grab the title that has been computed and use that.

# My Title

xxxx

=>

{
  title: "My Title",
  content:  "xxxx",
  _raw: "# My Title\n\nxxxx"
}

NB: at least 2 things should happen here …

• extraction of h1 content to
• removal of h1 from the content so that when we render body of page that is dealt with …