Convert docx files

[rjkantor]

I am trying to convert docx files to text prior to loading into BC4 under Ubuntu 13.10. On load, I am using this conversion command (below). I am unable to get the resulting filename.txt to load since it is not part of the conversion statement. How do I load the resultant converted txt file rather than the docx?

/usr/bin/soffice --headless --convert-to txt:Text %s

Thank you,
Rob

[Aaron]

Hello,

You would need to wrap the conversion into another, larger script which then takes the converted text file and places it into the %t variable. Where does your script currently put the target/converted .txt file? Can you then write that into a %t target?

[rjkantor]

Aaron -

/usr/bin/soffice --headless --convert-to txt:Text %s

the txt in the command above tells the soffice to convert the named %s and change the extension to .txt. So the util manages the output file from the designation above.

In general, how do I assign a value to the %t. Do you have an example?

Rob

[Aaron]

Hi Rob,

I installed OpenOffice to play around with this. The --convert-to accepts an output directory, but not a target file. We would need a target file somehow.

The surrounding script would look in your --convert-to --outdir /bcconvert/*.txt
then assuming the directory is emptied otherwise:
cat *.temp >> $2

where $2 points back to %t as the script. script.bash %s %t from BC3's external conversion. My own Bash scripting skills are not quite up to par to write the whole conversion myself. If there was a way to make soffice pass to a specific filename instead of directory, we could use that more directly.

[rjkantor]

new extension variable

Is it possible to add a new variable which allows one to specify the new extension (txt) of the same filename?

It seems I can echo out the resultant destination file, but I don't know how to assign it back to the %t which is not set until after my script would run. It is a chicken/egg issue. I can run my script but it needs to know the %t prior to execution.

If you supported the destination extension you can then look for the source name with the new extension.

Rob

[Aaron]

Thanks for the suggestion. I'll add this to our wishlist and alternatively see if one of our developers can come up with a better wrapper script. If you are able to create one, please do let us know and we'll take a look at it.

[dr_barnowl]

Make the file move to %t

We can't pass a new value of %t back to bcompare, so we have to make the output come to %t

#!/bin/sh
#
# docx-to-txt
#
# Converts MS Office Open XML (MOO-XML) DOCX to txt using soffice
#
# Moves output from a temporary batch folder to target
#
# Adrian Wilkins 2014 : licensed free for any use
#
# params 
#
# $1 = source file
# $2 = target file

SOURCE=$1
TARGET=$2

# Make some folders
OUTPUTFOLDER=$(mktemp -d)
TEMPHOME=$(mktemp -d)

# Set a new HOME variable for this process or soffice will refuse to start another instance
# This means this fails silently if you have a document window open
HOME=$TEMPHOME

# Convert
soffice --headless --convert-to txt:Text --outdir "$OUTPUTFOLDER" "$SOURCE" 

# Rename converted file to supplied target location
OUTFILE=$(ls $OUTPUTFOLDER)
mv "$OUTPUTFOLDER/$OUTFILE" "$TARGET"

rm -rf $TEMPHOME $OUTPUTFOLDER

[Aaron]

Thanks for the script! We'll take a look at this and see if we can incorporate it or something similar onto our Formats page.

[rjkantor]

Thanks this worked well.