Hi all,

I'm trying to work way above my expertise and understanding, so you'll have to forgive me if I'm not able to express my goals as clearly as necessary.

Basically, here's what I'd like to do:

In my syllabus TBX file (shared he before), I have agents that collect reading notes and lesson plans and then an export template that spits the whole thing out for me to print nicely and use as notes in class.

The template is an envelope-style one. The agent gets this template:

Code:

^value($DisplayName)^
^text^
^if(^get(ChildCount)^)^
^children(LessonPlanItem)^ 

While the "LessonPlanItem" merely consists of Code:

^text^ 

This results in me having to open up the export file in a word processor and apply light formatting.

I've been experimenting with running the whole output through Pandoc (essentially similar to a Markdown script). I couldn't figure out how to do that, mostly because I'm not really all that proficient at command line stuff--I just cut and paste from examples and cross my fingers!

I did, however, manage to cobble together a solution via Automator. I created a workflow that:

1. Gets specified finder file (the export file created by TBX)
2. Runs shell script:
Code:

PATH=$HOME/.cabal/bin:/usr/local/bin:/usr/texbin:$PATH

for file in "$@"
do
    output=${file%%.*}.docx
    pandoc --smart "$file" -o "$output"
done

echo "$output"
 

PS: I copied this code from elsewhere, so I only have a limited understanding of how or why it works. I understand all the Pandoc options (e.g., --smart sets all the typography to smart quotes) but not the scripting.
3. Opens Finder item (which results in MS Word opening the newly created "export file.docx" in Word.

This lets me apply markdown formatting in TBX and end up with a properly formatted Word file--no more manual messing with things. I'm guessing I may have been able to create properly formatted HTML pages all along, but at this point I'm more comfortable with the whole Markdown-->Word processor workflow.

At any rate, I'm wondering whether I've added an unnecessary step by using Automator. Is there some way to call the Pandoc process through the runCommand and include that in the template itself, so that TBX itself will spit out the docx file?

$HTMLExportCommand contains a string that specifies a command line to be applied to your exported file.  So this will do exactly what you want.  (Sorry to be terse; this internet connection in Kolkata times out quickly!)

Your CL is expecting an input file 'test.txt' that hasn't yet been created. The $HTMLExportCommand runs after all other activity and just before the output file is written.

I think you need to re-write your CL to use the shell's stdin as the input instead of 'test.txt'. Perhaps one of the more CL-savvy folk here can fill in the gaps?

My hunch (I'm outside my expertise area here) is that pandoc may be installed in a location/configuration such that when called from TB, it's not recognised.  IIRC, the pwd (current directory) for TB CL calls is your account's home directory (i.e. ~). Using Terminal, from your home directory can you run a pandoc test successfully? What you're trying to confirm is that if pandoc is called from your ~ directory that it can read (specimen) stdin. If that works, your code should work from TB.

My experience with other open-source CL tools is the creator assumes users have the necessary knowledge re file names/locations/permissions. For instance, you might need to include the path to pandoc in your CL. From your quote it looks like there's a pandoc mailing list. If so that might also be a good place to ask.

(Side note, Mark B's currently at a conference in India & on a slow/flaky connection for a few days, but it might be a good idea to folow up via support)

On my machine its /usr/local/bin/pandoc, but same result (zero kb file) with:

/usr/local/bin/pandoc --smart -o test.docx

Will follow up with support sometime in the near future. Not urgent. Just playing. Could be useful for future project though.

pandoc's documentation (I don't have the app itself) indicates that if you need to use stdout - e.g. in the $HTMLExportCommand context - you can't use DOCX ouptut format as that doesn't support stdout:

Quote:


If the aim is an MS Word DOCX document you may need to use pandoc after exporting from TB, as you were doing previously. I may be misunderstanding the docs, so I'd suggest asking on pandocs mailing list. The possible ambiguity is whether you can get DOCX formatted output via stdout (it may be, for instance, that you can but can't specify the file name/path).

Neither TB not pandoc are doing anything wrong here, it's just that their design assumptions don't mesh. If you set your syllabus TBX(s) to export to a particular folder, you could set that OS folder up with a folder script or droplet making the running of the pandoc transform fairly simple to trigger.

Aha. This looks to be the problem. I'll investigate more. DOCX is not really necessary, but it let me set a reference file style sheet to get things styled up automatically. Not sure if RTF supports that. Will continue to look into it and report back.

Quote:


This seems like the best advice and easiest to implement, given the weird restrictions from Pandoc on DOCX or ODT, which both support reference style sheets.

If I recall $HTMLExportCommand runs on the note being exported - IOW setting it a template has no effect. If you want lots of notes to use the same command, seed it via a prototype.

A good while back, in v5, I needed to use a complex sed call in an export command.  As you only get a small input box in Get Info of as a ket attribute, i made a code note and made that not'e rule set the other note's $HTMLExportCommand.

As for the wider pandoc process, I'm afraid it's a rabbit hole I've not the spare time to explore for a while - in case you were hoping - but likely someone else with a need will step forward. No antipathy, just not enough time.