Skip to content

Latest commit

 

History

History
87 lines (59 loc) · 5.02 KB

README.md

File metadata and controls

87 lines (59 loc) · 5.02 KB
Raw

LogSeqToObsidian

LogSeq and Obsidian: are two (awesome) note taking tools.

I: am a flawed human with a penchant for flitting between note taking apps.

These: are a couple of files to help me migrate from LogSeq to Obsidian.

Please: backup all of your files before trying the migration script on your own data. The script is highly experimental.

bonofix-snippet.css

A small CSS snippet to help Obsidian look more like a mix of the (LogSeq Bonofix Theme)[https://github.com/Sansui233/logseq-bonofix-theme] and Typora

Install the Minimal theme, and then paste it in your snippets folder and enable it. Doesn't do much more than (what I consider) the basics - any contributions welcome.

convert_notes.py

A basic Python script that converts LogSeq's markdown files to a style that plays nicer with Obsidian.

Usage

python convert_notes.py --logseq /path/to/logseq/graph --output /path/to/output/folder

Flags:

  • Add the --overwrite_output flag if you want any existing folder at the output path to be overwritten
  • Add the --unindent_once flag if you want all lines to be unindented once. If you do this, the base level of indentation will be paragraph-style text with no bullet points
  • Add --ignore_dot_for_namespaces if you want to ignore the . character in determining namespace hierarchies - default behavior is to treat . characters in filenames as namespace delimiters in some cases
  • Add --convert_tags_to_links if you want to convert #[[long tags]] to [[long tags]] links and #tags to [[tags]] links - default behavior is to convert long tags to #long_tags tags and leave short tags alone
  • Add --tag_prop_to_taglist to convert front matter of the form tags:: value1, #[[value 2]] to Taglinks:: [[value1]], [[value 2]]. That is, the tags in the front matter will be converted to links and named 'Taglinks' instead of 'tags'
  • Add --journal_dashes if you want to use dashes in the filenames for journal pages, eg 2023-08-03.md instead of 2023_08_03.md

Further information

Known assumptions:

  • Dots in the logseq filename are assumed to indicate namespaces
  • < and > characters are assumed to be part of text, and therefore escaped so that they display correctly in Obsidian

What this script does:

  • Creates a folder/subfolder hierarchy based on namespaces, copies notes appropriately, and updates links between notes
  • Links to notes that have not yet been created are replaced with tags
    • Use the --convert_tags_to_links argument, it willl Convert
  • Copies embedded assets into an 'attachments' subfolder under the given note. Resizes embedded images in Obsidian to match any resizing that was done in Logseq
  • Removes block links and block embeds
  • Converts front matter of the title:: My Note format to the format expected by Obsidian (key: value wrapped in triple-hyphen lines)
  • Use --tag_prop_to_taglist to convert a tags:: [[list]] #of #[[tags with spaces]] header into frontmatter with a taglinks property that is a list of links: 
    Taglinks:
- [[list]]
- [[of]]
- "[[tags with spaces]]
  • If a code block has been embedded inside a list, prepends a line - without this, the code block does not display correctly
  • Minor reformatting to prettify notes: escapes < and > characters, replaces 2-4 spaces with a tab, ignores Logseq artefacts like collapsed:: true
  • Use --journal_dashes to convert journal file entries from the format Jan 2,2023.md to the default Obsidian format 2023-01-02.md
  • Any files found with a %3A in the name (html encoded colon character :) to a . character instead.

What this script does not do:

  • Process page properties, and use them for finding namespaces
  • Get file copier to work with subfolders in logseq (right now only copies pages in the base directory)
  • Handle aliases
  • Handle namespaces under journal pages
  • Embed PDF as option
  • Seems like asset names cannot have '%20' in them - is that right?

I might also like to do the below at some point

(That is, none of the below has been done)

First, a note on content reformatting. LogSeq and Obsidian both allow for headers and outlining, but one key difference is that Obsidian lets you fold headings to collapse the content under it. LogSeq doesn't let you do this, so if you want a header to be foldable, you have to indent one level under it. My script assumes that any indents you've made after a header in LogSeq are just for the purpose of folding, and should be undone. Specifically, this is what will happen:

  • All content will be outdented one level (since everything starts off as part of an outline in LogSeq)
  • Then, anything under a H2 header will be outdented another level.
  • Then, anything under a H3 header will be outdented another level.
  • Etc.

As an example, if this is what your LogSeq note looks like:

image-20221111221326446

Then this is how your Obsidian note will end up being structured:

image-20221111231753576