Clean up LibreOffice FODP files

compress_spans.sh

#!/bin/bash
# There are many times OpenOffice / LibreOffice will put in useless spans.

# What it does:
#  Finds a span with a style that has the format <capital letter><1 or more digits>
#  Finds the close tag and then looks to see if the NEXT span has the same style defined
#  If they match, the first end and the second begin are stripped
#    (This is why it has to loop repeatedly, a third won't be seen unless a fourth - they
#     would match as well)
[ "$1" == '' ] && echo "No parameters given: try '$0 *' or '$0 *fodp'"

for f in "$@"; do
  [ -f $f ] || { echo "Skipping non-file '$f'"; continue; }
  [ ${f##*.} == "fodp" ] || { echo "Skipping non-fodp '$f'"; continue; }
  NEW=1
  FIRST=$(stat -c %s $f)
  echo -n "Processing $f: "
  while [ $((ORIG-NEW)) -ne 0 ]; do
    ORIG=$(stat -c %s $f)
    perl -i -pe 's|<text:span text:style-name="([A-Z]\d+)">([^<]*?)</text:span><text:span text:style-name="\1">([^<]*?)</text:span>|<text:span text:style-name="\1">\2\3</text:span>|g' $f
    NEW=$(stat -c %s $f)
  done
  DELTA=$((FIRST-NEW))
  if [ $DELTA -ne 0 ]; then
    echo "stripped $((FIRST-NEW)) bytes"
    # git diff --color=always --stat=,1 $f | cut -f2- -d'|' | head -1
  else
    echo "(no change)"
  fi
done

Seriously... FODP files change so much when you just hit "save."

-      <text:p text:style-name="P21"><text:span text:style-name="T11">Devic</text:span><text:span text:style-name="T11">e </text:span><text:span text:style-name="T11">Suppo</text:span><text:span text:style-name="T11">rt </text:span><text:span text:style-name="T11">Stack</text:span></text:p>
+      <text:p text:style-name="P21"><text:span text:style-name="T11">Device Support Stack</text:span></text:p>

Processing Extra_Device_Support_Development.fodp: stripped 34264 bytes