Parsing a Single Text File into Multiple Files

I am an academic that uses bibliographic management software called Papers. After I highlight my manuscripts Papers allows me to export all my highlights and comments as a single file.

I am trying to figure out how to parse that single file into individual files (each highlight or comment as its own file) using KM. I am stuck, unable to parse the single file in multiple files–I thought I’d ask the group for ideas. Below is a sample of the single exported note that I’m trying to break into pieces.

Peter tried to help me–but I’m a novice with KM and haven’t been able to get the script to work over the last 2 days. I think I need to be told explicitly the name of each command and submenu. Sorry I’m such a neophyte.


  1. Rosen MA, Dietz AS, Yang T, Priebe CE, Pronovost PJ. An integrative framework for sensor-based measurement of teamwork in healthcare. Journal of the American Medical Informatics Association. 2014:amiajnl–2013–002606. doi:10.1136/amiajnl-2013-002606.

p.1: The large, multidisciplinary science of teams has informed the development of measurement systems.7–9 – Highlighted jul 24, 2014

p.1: Breakdowns in teamwork and communication are an independent cause of, and a cross-cutting theme in, many of the system failures leading to patient harm.1–3 – Highlighted jul 24, 2014

p.1: Past reviews of team performance measurement in healthcare indicated a lack of tools with strong theoretical grounding and methodologically rigorous development and validation processes.17 – Highlighted jul 24, 2014

My reply, which is omitted from @Aachoo2’s post, was:

It looks like it is split by multiple line endings (say 2+?) Are they always single lines of text?

  • Read the file into a variable.
  • Add some line endings to the variable (it just makes the process a bit easier).
  • Set a variable Index to 1
  • For Each variable Entry in regular expression substrings “.*[\n\r]{2+}” in the variable
  • Write the variable Entry to file %Variable%Path%-%Variable%Index%
  • Set Variable Index to Calculation Index+1

That’s a start anyway.

I’d love to help. Can you email me one of the files that Papers gives you?

Also, how do you want to tell KM to act on the file? Will you have it selected in the Finder?

  • Michael

Thanks Michael! I’ll send you some of the output files.

To answer your question: For the time being I will select the file in the Finder and have the script work from there. I would like all of the individual notes, once parsed, moved into a folder that has the same name as the original notes file.

I need to append the citation reference to the end of each individual file (the first entry of each full notes file) so it would look something like this:

Highlight or Note:


Citation Reference

Just so you’re aware, I will move the folder containing the parsed files into a Devonthink Database. I use this database to search for relevant snippets when writing grants or manuscripts.

Great. I’ve sent you a macro to see if it works on your files. If it does, I’ll post it here and let my betters nitpick it to make it stronger.

This appears to be a functional solution to Aachoo2's problem. I'd welcome any suggested improvements to the flow, regex, or anything else.

Basically, for each file in the finder's selection (must be a .txt file)

  • it reads the path, base name, and file name to a variable
  • reads the file contents into a variable
  • adds the extra line endings as suggested by Peter-sama
  • matches the reference in the file to a variable
  • removes the reference from the file contents variable (makes regex matching later easier)
  • creates a new folder with the file name
    • and then for each regex match in the file contents (which should match the highlights/notes)
    • writes a new file in the named folder with the highlight/note text, an empty line, and the citation

Potential areas of improvement:

  1. The regex works, but I'm never confident that it is the best way to write it and won't have failure points later.
  2. The reference string is unchanged from the original which is "1. SOME TEXT". Could strip out the number and spaces since it's only ever 1 citation.

Aachoo Papers Parser.kmmacros (7.4 KB)

1 Like

As I’m reading this impressive macro, I found myself saying "Holy ^#@$, you can have KM do something with every selected file in Finder?! Then, “HOLY ^&#@(*%^#@, you can have KM take action on every substring found in a variable?!?!?!”

This is exciting stuff. I really need to get better at regular expressions…!

1 Like

It looks pretty reasonable to me.

Variable naming wise

  • I would have named “temp” -> “filePath”
  • I would have named “filePath” -> “parentPath”
  • I would have named “Variable” -> “fileContents”

%Variable%temp% is the same as %Variable%filePath%/%Variable%fileName%, so I would have used that when reading the file. And since that’s the only use of fileName, that means that variable can be eliminated entirely.

I would be tempted to have:

Set Variable folderPath to %Variable%parentPath%/%Variable%fileBasename%

and use that in the Create New Folder and Write to File actions. That avoids the duplication which is always a potential future problem (what if you decide to name the folder something different, then you have two places to make the change).

The regex looks OK (although its always dependent on how variable the input is, but that is the nature of parsing text). I would have used \n\n+ at the end (so that it would cope with more than one blank line), but thats about it. I’m not really sure what [\s\S] is, it that just “match everything including line endings”? For that I would generally turn on the (?s) flag and then use “.”, but really it doesn’t matter.

So basically, no significant advice, except to be careful with variable naming. There are only two hard things in Computer Science: cache invalidation, naming things, and off-by-one errors.

Thanks for the comments.

[\s\S] was for all white space and non-white space characters.

The purpose of this (via trial and error) was to match both of the following as two separate matches for the way he has his notes:

some text here in a single string followed by two line returns (this is one match)

as well as

a line of text with only one line return
a line of text with only one line return
another line of text with one line return
a final line of text with two line returns (all four lines are one match)

I’m not as knowledgeable about the different flags. I’ll have to study them.

The regex flags were discussed in grutitous detail recently on the YahooGroups mailing list.

The flag options are documented on the ICU regex reference page.

There are four of relevance:

i — case insensitive (often Keyboard Maestro provides a default for this).
x — allows white space in patersn, and #commentotendofline comments.
s — “single line” mode. “.” matches any character instead of any character except eol (\n).
m — “multi line’ mode. “^” and “$” match the start and end of lines within the text.

Generally, you would only use (?x) for very complex regular expressions that you wanted to try to document inline.

You use them like this:

(?i)insensitive (?-i)CaseSenSiTive
\A.*\n(?s)(.*HERE) — matches the entire document (multiple lines) except the first line, until it finds the text.
(?m)^.HERE.$ — matches a single line containing the text.

The flags apply to the regex from that position onwards. You can also use them to apply to specific parts of the regular expression, like this:

(?i:insensitive) (?-i:CaseSenSiTive)

So [\s\S]+? should be the same as (?s:.+?) or even (?s:.)+?

I really appreciate all the help with this macro. Wonderful software and wonderful community!

Although this worked initially, I’m having trouble with the script recognizing new references.

It parses each individual note properly, but instead of appending the correct reference, it appends the same (old) reference left over from another paper. The old reference is not overwritten with the new.

Any ideas?

This is hugely helpful. Please excuse me if this is basic–I’m new to KM–but can I append text to every file as it’s created? I’d like to include the bibtex citekey (from the clipboard) at the end of every one of the newly generated text files. Having a little trouble figuring it out.


Hey Michael,

Every file?

All the time?

From what app?

Please give a more detailed description of your task.

Best Regards,

Sorry. The above macro takes a single text file, parses it into a number of different files and adds them to a new subdirectory. What I would like to do is copy the clipboard to the end of each of those files as they are created.

Hey Michael,

This is a big thread, so to me at least your question lacked enough context to answer.

Knowing you were specifically asking about the other Michael’s macro makes giving a cogent answer straightforward enough.


In the above string:

%CurrentClipboard% is a Text-Token for exactly what it says.

\n is a newline character.

"My-Citation" is an example of literal-text combined with the tokens.

In the Write Text to File step of Michael’s macro you can simply add %CurrentClipboard% and whatever other text or tokens you need to the end.