archived old versions of the web pages

[enc.git] / old / ru / kitchen / reformat-html

#!/bin/bash

# NAME
#    reformat-html - reformat HTML files from emailselfdefense.fsf.org

# SYNOPSIS
#    reformat-html /PATH/TO/NAME.html

# GRAPHIC INTERFACE HOWTO
#    * Launch the script by double-clicking on it; a terminal will open.
#    * At the prompt, drag and drop the input file into the terminal.
#
#    Alternatively (in Gnome, KDE, XFCE, etc.)
#    * create a launcher for the application menu;
#    * launch the script from the contextual menu of the HTML file.
#
#    The reformatted file is created in the directory where the input file
#    resides, and its name is NAME-r.html.

#==============================================================================

set -e

# Test whether the script is called from color-wdiff
p=$(pidof -x color-wdiff) || true
test "$p" == "$PPID" && called_from_color_wdiff=1

function close_or_exit () {
# turns off interactivity and lets the terminal close normally if the script
# is called from color-wdiff.

if test "$called_from_color_wdiff" == "1"; then
  exit $1
else
  if test "$1" == "1"; then
    echo -e 1>&2 "\n!!! $input doesn't exist or is not an HTML."
  fi
  echo -e '\n*** Close the terminal window or press Return.'; read OK
  test -z "$OK" && exit $1
fi
}

# Get a valid HTML as input.
input=$1
if test ! -f "$input" -o ! -s "$input"; then
  echo -e "\n*** reformat-html - Please enter the HTML file."
  read input
  input=${input%\'}; input=${input#\'}
  test -f "$input" -a "${input%.html}" != "$input" || close_or_exit 1
fi

# Define the output file.
if test "$called_from_color_wdiff" == "1"; then
  output=$2
else
  output=${input%.html}-r.html
fi

tmp=$(mktemp -t ref.XXXXXX) || close_or_exit 1
trap "rm -f $tmp" EXIT

cp $input $tmp

# Remove javascript, which shouldn't be reformatted, leading and trailing
# spaces/tabs, multiple spaces, LF after </a> and <li>.
sed -i -e '/jquery-1.11.0.min.js/,$d' \
       -e 's,\t, ,g'  \
       -e 's,^ *,,'   \
       -e 's,  *, ,g' \
       -e 's, *$,,'   $tmp
sed -i -e '/<\/a>$/ {N; s,<\/a>\n<,<\/a> <,}' $tmp
sed -i -e '/^<li/ {N; s,>\n<a ,> <a ,}' $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 ,}' $tmp
  let i=i+1
done
sed -i -e '/ \/>$/ {N; s,\( \/>\)\n,\1 ,}' \
       -e '/ <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.
# Fuse comment with </p>; separate truncated "~~~" comment from fused tag.
sed -i -e 's,^ *,,'  \
       -e 's, *$,,'  \
       -e 's,  , ,g' \
       -e '/^$/d'    \
       -e '/<\/p>$/ {N;s,\n\(<!-- [^~]\),\1,}' \
       -e '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,}' $tmp
sed -i -e 's,</p></span>$,</p>\n</span>,' \
       -e 's, alt=,\nalt=,g' \
       -e 's, | , |\n,g' $tmp

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

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

close_or_exit 0