#!/bin/bash

# page-reformat: reformat the original pages of emailselfdefense.fsf.org

# cd to the directory where the script is located, then:
#      $ ./page-reformat [path to the HTML file]
# You can also drag-and-drop the file into the terminal.
# The result has extension .html.html to avoid writing over the original page.

# If you want to wrap the text, uncomment line 97 and comment out line 94.


cp $1 tmp

# Remove javascript, which shouldn't be reformated.
sed -i '/jquery-1.11.0.min.js/,$d' tmp

# Remove leading and trailing spaces/tabs.
sed -i 's,\t, ,g' tmp
sed -i 's,^ *,,' tmp
sed -i 's, *$,,' tmp

# Remove LF after </a>.
sed -i '/<\/a>$/ {N; s,<\/a>\n\([^<]\),<\/a>\1,}' tmp

# One string per paragraph, header or list item.
for tag in li p strong a h3; do
  sed -i "/<$tag[^>]*>$/ {N; s,\\n, ,}" tmp
done
for tag in a strong; do
  sed -i "/<\\/$tag>$/ {N; s,\\n, ,}" tmp
done
# This command may need to be repeated. Adjust the number of repeats. This
# could be done by looping back to a sed marker, but a while loop seems
# quicker.
i=0
while (( i < 2 )); do
  sed -i '/[^<>]$/ {N; s,\([^<>]\)\n,\1 ,}' tmp
  let i=i+1
done

sed -i '/ \/>$/ {N; s,\( \/>\)\n,\1 ,}' tmp
sed -i '/ <a[^>]*>$/ {N; s,\(<a[^>]*>\)\n\([^<]\),\1 \2,}' tmp

# Make sure there is only one paragraph per string. This command may need to
# be repeated. Adjust the number of repeats.
i=0
while (( i < 2 )); do
  sed -i 's,</p>\(.\+\)$,</p>\n\1,' tmp
  let i=i+1
done

# Single out the tags which include p (will also work for pre).
sed -i 's,\(.\)<p,\1\n<p,' tmp

# Single-out input meta and link.
for tag in input meta link link; do
  sed -i "s,> <$tag,>\n<$tag," tmp
done
# Remove leading and trailing spaces, double spaces and blank lines.
sed -i 's,^ *,,' tmp
sed -i 's, *$,,' tmp
sed -i 's,  , ,g' tmp
sed -i '/^$/d' tmp

# Fuse comment with </p>.
sed -i '/<\/p>$/ {N;s,\n\(<!-- [^~]\),\1,}' tmp

# Separate truncated "~~~" comment from fused tag.
sed -i 's,~~~[ ]\?[-]\?[-]\?[ ]\?<,~~~\n<,' tmp

# Fuse header, section and footer with the corresponding div.
for tag in header section footer; do
  sed -i "/^<$tag/ {N; s,\\(<$tag[^>]*>\\)\\n<div>,\\1<div>,}" tmp
  sed -i "/^<\\/div>$/ {N; s,<\\/div>\\n\\(<\\/$tag>\\),</div>\\1,}" tmp
done

# Add LF before main sections and commented-out parts.
sed -i 's,<!-- ~~,\n<!-- ~~,' tmp
sed -i '/COMMENTED OUT/ s,^,\n,' tmp

# Make the text more readable.
for tag in p h1 h2 h3 h4 dl title form; do
  sed -i "s,<$tag,\\n&," tmp
done
for tag in p dl ul h1 h2 h3 h4 title head footer form script; do
  sed -i "/<\\/$tag>/s,$,\\n," tmp
done
sed -i '/<\/dd>/ {N; s,</dd>\n<dt,</dd>\n\n<dt,}' tmp
sed -i '/<\/dt>/ {N; s,</dt>\n<dd,</dt>\n\n<dd,}' tmp
sed -i 's,</p></span>$,</p>\n</span>,' tmp

sed -i 's, alt=,\nalt=,g' tmp
sed -i 's, | , |\n,g' tmp
mv tmp $1.html

# Wrap the text.
#fmt -s -w 95 tmp > $1.html

# Remove extra LFs, if any.
sed -i ':a /^$/ {N; s,\n$,,; ba}' $1.html
sed -i ':a /^\n*$/ {$d; N; ba}' $1.html

rm -f tmp