Logseq syntax: what Pb Notes supports and what an importer has to handle

[shovon]

Groundwork for a Logseq importer. The plan so far is in app/docs/logseq-import.md. This post sorts Logseq's syntax into what Pb Notes already renders and what it doesn't, so we can decide what an import has to do with each piece.

Pb Notes renders block text with react-markdown, plus remark-math and the wikilink plugin in packages/pages/wikilink.ts. There is no remark-gfm. The Logseq side below is from memory, not checked against Logseq's docs, so corrections are welcome.

Works the same in both

Syntax	Notes
[[Page]], #tag, #[[multi word]]	All name the same page.
[label]([[Page]])	Alias link.
((uuid)), [label](((uuid)))	Block refs. They only resolve if the Logseq id:: becomes the Pb Notes block id.
Headings, **bold**, *italic*, `code`, fenced code, quotes, links, 	Plain Markdown.
$inline$, $$block$$	Math, rendered with KaTeX.
Nesting, zooming into a block, journal pages	Journal pages are named YYYY-MM-DD.

Not supported in Pb Notes

Properties (key:: value), on blocks and pages. Logseq uses them for:

• id::, which block refs depend on
• collapsed::, which waits on the collapsed-subtree decision
• alias::, which lets several names point to one page
• title::, tags::, public::, filters::, and any key you invent

Tasks:

• TODO/DOING/DONE/LATER/NOW/WAITING/CANCELED
• [#A] priorities
• SCHEDULED:/DEADLINE: with repeaters
• :LOGBOOK: drawers with CLOCK: lines

All of these would show up as literal text.

Macros:

• {{embed [[page]]}}, {{embed ((uuid))}}
• {{query …}}, and #+BEGIN_QUERY queries
• {{video}}, {{youtube}}, {{tweet}}, {{cloze}}, {{renderer …}}

Org-style blocks: #+BEGIN_NOTE/TIP/WARNING/QUOTE/CENTER.

GFM and other inline formatting:

• tables
• ~~strike~~ Done
• - [ ] checkboxes
• bare URLs turning into links Done
• footnotes
• ==highlight== / ^^highlight^^ Done

Other Logseq features:

• the namespace hierarchy view ([[a/b]] works as a link, but there's no tree of child pages)
• #card flashcards
• whiteboards
• PDF highlight pages (hls__…)
• .org files
• ../assets/ links

Two catches for the importer

1. Journal links won't match. In Logseq's text, journal pages are linked by their display title, such as [[Jan 31st, 2024]], in whatever format the graph's :journal/page-title-format sets. The importer turns the journal files into 2024-01-31 pages, so if text is copied word for word, those links point to pages that don't exist. The importer will have to rewrite them.
2. Page names are case-sensitive in Pb Notes. Logseq treats [[Mira]] and [[mira]] as one page. Pb Notes only ignores case when searching for a title in [[ autocomplete, so a graph that mixes the two would split into two pages. Before the import, we need to pick one: normalise titles on import, or match titles without case.

Open questions

• For each unsupported item: drop it, keep it as plain text, or store it somewhere for later?
• Which of these are worth supporting in Pb Notes itself before an importer ships? GFM (tables, strikethrough, checkboxes) is the cheapest, since it's one remark plugin.
• Case-insensitive titles: a change to Pb Notes, or something only the importer handles?

[shovon]

• id:: is not needed; every block must have an ID, by default, in our ledger-based system
• collapsed::: I'm lukewarm about this
• alias::: useful
• Tasks: does it need to be syntax?
• Org-style blocks: I'll need to spend time thinking about it
• Macros: definitely useful
• Footnotes: a must-have
• Namespace hierarchy: this is the first time that I have heard of it
• #card flashcards: intereseting
• Whiteboards: no
• PDF highlight pages (hls__...): fascinating
• .org files: Does it make sense to support this?
• ../assets/ links: this is going to get tricky.

Regarding journal links won't matching: maybe alias:: could be useful?

I could imagine case-sensitivity being a huge problem. I will need to think about it.