odt2md

#!/bin/bash
# Script used to convert odt file to markdown file.

#shopt -o -s xtrace

PANDOC="pandoc"
ODT2HTML="odt2html"

# Function to diaply a warning to the user.
function alertwarning(){
  tty -s
  if [ "0" == "$?" ]; then
    echo "$1"
  else
    zenity --warning --text="$1"
  fi
}

# Increase header level +1
read -r -d '' HEADERS <<'EOF'
from pandocfilters import toJSONFilter, Header

def headers(key, value, format, meta):
  if key == 'Header':
      return Header(value[0] + 1, value[1], value[2])

if __name__ == "__main__":
  toJSONFilter(headers)
EOF

# Extract embeded images
read -r -d '' IMAGES <<'EOF'
import sys
import base64
from pandocfilters import toJSONFilter, Image
try:
   from StringIO import StringIO
except ImportError:
   from io import StringIO

count=1
basename=sys.argv[1]

def images(key, value, format, meta):
  global count
  global basename
  if key == 'Image' and value[1][0].startswith('data:'):
    src = value[1][0]
    start = src.find(',')
    ext = src[src.find('/')+1:src.find(';')]
    filename = '%s-%s.%s' % (basename, count, ext)
    with open(filename, 'w') as f:
      base64.decode(StringIO(src[start:]), f)
    value[1][0] = filename
    # Increase image count
    count=count+1
    return Image(value[0],value[1])

if __name__ == "__main__":
  toJSONFilter(images)
EOF

# Replace LineBreak from Table and BulletList by LBTOK
read -r -d '' LINEBREAKS <<'EOF'
from pandocfilters import walk, toJSONFilter, Str, Table, Plain, BulletList

def replace_linebreaks(key, value, format, meta):
  # Replace Line break by token
  if key == 'LineBreak':
    return Str('LBTOK')

def replace_linebreaks_from(key, value, format, meta):
  # Keep last line break.
  if key == 'Plain':
    return Plain(walk(value[:-1], replace_linebreaks, format, meta) + [value[-1]])

def linebreaks(key, value, format, meta):
  if key == 'Table':
      return Table(value[0], value[1], value[2], value[3], walk(value[4], replace_linebreaks_from, format, meta))
  if key == 'BulletList':
    return BulletList(walk(value, replace_linebreaks, format, meta))

if __name__ == "__main__":
  toJSONFilter(linebreaks)
EOF

# Style Div / Para
read -r -d '' DIVSTYLES <<'EOF'
from pandocfilters import toJSONFilter, Div, Str, LineBreak

def divstyles(key, value, format, meta):
  if key == 'Div' and len(value[0][1])>0:
    # Create string similare to {: .text-muted }
    style = '{: ' + ' '.join(['.' + x for x in value[0][1]]) + '}'
    # Remove attributes from originals
    value[0][1] = []
    # Try to add our style string to div content.
    if value[1][-1]['t'] == 'Plain':
      value[1][-1]['c'].append(LineBreak())
      value[1][-1]['c'].append(Str(style))
    else:
        value[1].append(Str(style))
    return Div(*value)

if __name__ == "__main__":
  toJSONFilter(divstyles)
EOF

# Remove western from Header
read -r -d '' WESTERNS <<'EOF'
from pandocfilters import toJSONFilter, Header, Div, CodeBlock

def filter_attrs(attrs):
  """Remove western attributes. Also remove `-western` suffix."""
  return [x.replace('-western', '')
          for x in attrs
          if x != 'western']

def remove_westerns(key, value, format, meta):
  # Header Int Attr [Inline]
  if key == 'Header' and len(value[1][1]) > 0:
    value[1][1] = filter_attrs(value[1][1])
    return Header(*value)

  # Div Attr [Block]
  if key == 'Div' and len(value[0][1]) > 0:
    value[0][1] = filter_attrs(value[0][1])
    return Div(*value)

  # CodeBlock Attr String
  if key == 'CodeBlock' and len(value[0][1]) > 0:
    value[0][1] = filter_attrs(value[0][1])
    return CodeBlock(*value)

if __name__ == "__main__":
  toJSONFilter(remove_westerns)
EOF


# Check if pandoc is available
if ! (command -v "$PANDOC" >/dev/null 2>&1); then
  alertwarning "$PANDOC is missing!"
  exit 2
fi

# Check if odt2html exists
if ! (command -v "$ODT2HTML" >/dev/null 2>&1); then
  alertwarning "$ODT2HTML is missing!"
  exit 2
fi

# Loop on every arguments
for F in "$@"; do

  # Convert file to HTML
  TFILE="/tmp/$(basename $0).$$.html"
  "$ODT2HTML" --output="$TFILE" --format=html "$F"

  # Fix <li> issues
  lead='^<[ou]l>$'
  tail='^<\/[ou]l>$'
  sed -i.bak "/$lead/,/$tail/{s#<li/>##g}" "$TFILE"
  sed -i.bak "/$lead/,/$tail/{s/<p/<li/g}" "$TFILE"
  sed -i.bak "/$lead/,/$tail/{s/<\/p>/<\/li>/g}" "$TFILE"
  sed -i.bak "/$lead/,/$tail/{s/<br \/>//g}" "$TFILE"
  
  # Fix <table> issues - remove <p> from <table>
  lead='^<table'
  tail='^<\/table>$'
  sed -i.bak "/$lead/,/$tail/{s/<p[^>]*><br\/>//g}" "$TFILE"
  sed -i.bak "/$lead/,/$tail/{s/<p[^>]*>//g}" "$TFILE"
  sed -i.bak "/$lead/,/$tail/{s/<\/p>/<br \/>/g}" "$TFILE"
  #sed -i.bak "/$lead/,/$tail/{s/<br \/>//g}" "$TFILE"

  # Convert <p> to <div> to keep attributes
  sed -i.bak "s/<p /<div /g" "$TFILE"
  sed -i.bak "s/<p>/<div>/g" "$TFILE"
  sed -i.bak "s/<\/p>/<\/div>/g" "$TFILE"

  # Convert file to markdown.
  MDFILE=${1%.*}.md
  "$PANDOC" --from=html --to=json "$TFILE" | \
    python -c "$WESTERNS" | \
    python -c "$HEADERS" | \
    python -c "$LINEBREAKS" | \
    python -c "$DIVSTYLES" | \
    python -c "$IMAGES" "${1%.*}" | \
    "$PANDOC" --standalone --from=json --to=markdown_github+yaml_metadata_block --output="$MDFILE"

  # Remove {.western}
  #sed -i.bak 's/{.western[^}]*}//g' "$MDFILE"

  # Replace `...` to `---`
  sed -i.bak 's#^\.\.\.$#---#g' "$MDFILE"

  # Change keywords to tags.
  sed -i.bak -r "s#^keywords: '?(.*)'?#tags: [\1]#g" "$MDFILE"

  # Change ``` to ~~~
  sed -i.bak 's#```#~~~#g' "$MDFILE"

  # Fix Table lineBreak
  sed -i.bak -r 's#LBTOK#<br/>#g' "$MDFILE"

  # Remove temporary file
  rm "$TFILE"
  rm "$TFILE.bak"
  rm "$MDFILE.bak"

done