An Alfred + Bear workflow.

Inspired by alfred-bear, implemented in Go for faster searching, with daily note journaling capabilities.

To search for a note by title, use the bt keyword.

To search a note by its full contents, use the bs keyword.

To search for a note's backlinks, use the bbl keyword.

Here you can see that searching for mcc shows Remote Team Interactions Workbook ~Skelton, Pais which references Team of Teams ~McChrystal.

Selecting that item will open Remote Team Interactions Workbook ~Skelton, Pais.

The btag and bsearch keywords will pass your search into Bear's search bar, thereby doing in-app filtering. All other search commands will open a single note.

If your btitle search fails to find any matching notes, the workflow will give you an option to create a new note w/ a title of the same name.

The captainslog keyword will either open today's log (a note with the title YYYY-MM-DD) or create a new note with that title.

Run freddiebear help journal for details on how to tweak the tag it attaches to new notes.

You can export the text contents of your notes to Markdown files. Specify the directory and we'll create files in the form of <title> (<sha>).md.

Titles aren't unique, so we append a unique ID for each note. This also allows us to track renamed notes.

This can be used in conjunction with the sample script backup.sh -- it exports your notes and attachments and pushes them to GitHub for archiving and a rudimentary form of revision history.

You can create a Graphviz graph of how notes are linked together. The Alfred keyword bg will redirect freddiebear graph to a .dot file, generate a PDF from it, and open the PDF w/ Preview.

Requires graphviz.

This Golang implementation is pretty snappy on my current 5MB database. Most of the performance gains of this implementaion over a Python implementation appear to be reduced startup cost. That said, db.go includes some SQLite3 pragmas that will hopefully keep it snappy as it grows. Sample timing that returns about half the records in the database:

# time freddiebear search --all drip >/dev/null
freddiebear search --all drip > /dev/null  0.00s user 0.00s system 24% cpu 0.029 total

Show tags will generate a list longest-path tags to show as an Alfred item's subtitle.

For example, a note with a tags q and a/b/c will have four tags in the database:

1. q
2. a
3. a/b
4. a/b/c

And we'll only return the terminal/non-intermediate tags (a/b/c and q in this example).

The current implementation uses a naive O(n^2) algorithm, but in practice is quite fast for small sets of tags:

→ go test -bench=.
goos: darwin
goarch: amd64
pkg: github.com/mnadel/freddiebear/util
cpu: Intel(R) Core(TM) i9-9880H CPU @ 2.30GHz
BenchmarkRemoveIntermediatePrefixes-16   4701206               233.7 ns/op

And if you've got an M1, it's 36% faster:

→ go test -bench=.
goos: darwin
goarch: arm64
pkg: github.com/mnadel/freddiebear/util
BenchmarkRemoveIntermediatePrefixes-10           7679737               149.4 ns/op