fr/kitchen/reformat-html

   1 #!/bin/bash
   2
   3 # NAME
   4 #    reformat-html - reformat HTML files from emailselfdefense.fsf.org
   5
   6 # SYNOPSIS
   7 #    reformat-html /PATH/TO/NAME.html
   8
   9 # GRAPHIC INTERFACE HOWTO
  10 #    * Launch the script by double-clicking on it; a terminal will open.
  11 #    * At the prompt, drag and drop the input file into the terminal.
  12 #
  13 #    Alternatively (in Gnome, KDE, XFCE, etc.)
  14 #    * create a launcher for the application menu;
  15 #    * launch the script from the contextual menu of the HTML file.
  16 #
  17 #    The reformatted file is created in the directory where the input file
  18 #    resides, and its name is NAME-r.html.
  19
  20 #==============================================================================
  21
  22 set -e
  23
  24 # Test whether the script is called from color-wdiff
  25 p=$(pidof -x color-wdiff) || true
  26 test "$p" == "$PPID" && called_from_color_wdiff=1
  27
  28 function close_or_exit () {
  29 # turns off interactivity and lets the terminal close normally if the script
  30 # is called from color-wdiff.
  31
  32 if test "$called_from_color_wdiff" == "1"; then
  33   exit $1
  34 else
  35   if test "$1" == "1"; then
  36     echo -e 1>&2 "\n!!! $input doesn't exist or is not an HTML."
  37   fi
  38   echo -e '\n*** Close the terminal window or press Return.'; read OK
  39   test -z "$OK" && exit $1
  40 fi
  41 }
  42
  43 # Get a valid HTML as input.
  44 input=$1
  45 if test ! -f "$input" -o ! -s "$input"; then
  46   echo -e "\n*** reformat-html - Please enter the HTML file."
  47   read input
  48   input=${input%\'}; input=${input#\'}
  49   test -f "$input" -a "${input%.html}" != "$input" || close_or_exit 1
  50 fi
  51
  52 # Define the output file.
  53 if test "$called_from_color_wdiff" == "1"; then
  54   output=$2
  55 else
  56   output=${input%.html}-r.html
  57 fi
  58
  59 tmp=$(mktemp -t ref.XXXXXX) || close_or_exit 1
  60 trap "rm -f $tmp" EXIT
  61
  62 cp $input $tmp
  63
  64 # Remove javascript, which shouldn't be reformatted, leading and trailing
  65 # spaces/tabs, multiple spaces, LF after </a> and <li>.
  66 sed -i -e '/jquery-1.11.0.min.js/,$d' \
  67        -e 's,\t, ,g'  \
  68        -e 's,^ *,,'   \
  69        -e 's,  *, ,g' \
  70        -e 's, *$,,'   $tmp
  71 sed -i -e '/<\/a>$/ {N; s,<\/a>\n<,<\/a> <,}' $tmp
  72 sed -i -e '/^<li/ {N; s,>\n<a ,> <a ,}' $tmp
  73
  74 # One string per paragraph, header or list item.
  75 for tag in li p strong a h3; do
  76   sed -i "/<$tag[^>]*>$/ {N; s,\\n, ,}" $tmp
  77 done
  78 for tag in a strong; do
  79   sed -i "/<\\/$tag>$/ {N; s,\\n, ,}" $tmp
  80 done
  81 # This command may need to be repeated. Adjust the number of repeats. This
  82 # could be done by looping back to a sed marker, but a while loop seems
  83 # quicker.
  84 i=0
  85 while (( i < 2 )); do
  86   sed -i '/[^<>]$/ {N; s,\([^<>]\)\n,\1 ,}' $tmp
  87   let i=i+1
  88 done
  89 sed -i -e '/ \/>$/ {N; s,\( \/>\)\n,\1 ,}' \
  90        -e '/ <a[^>]*>$/ {N; s,\(<a[^>]*>\)\n\([^<]\),\1 \2,}' $tmp
  91
  92 # Make sure there is only one paragraph per string. This command may need to
  93 # be repeated. Adjust the number of repeats.
  94 i=0
  95 while (( i < 2 )); do
  96   sed -i 's,</p>\(.\+\)$,</p>\n\1,' $tmp
  97   let i=i+1
  98 done
  99
 100 # Single out the tags which include p (will also work for pre).
 101 sed -i 's,\(.\)<p,\1\n<p,' $tmp
 102
 103 # Single-out input meta and link.
 104 for tag in input meta link link; do
 105   sed -i "s,> <$tag,>\n<$tag," $tmp
 106 done
 107
 108 # Remove leading and trailing spaces, double spaces and blank lines.
 109 # Fuse comment with </p>; separate truncated "~~~" comment from fused tag.
 110 sed -i -e 's,^ *,,'  \
 111        -e 's, *$,,'  \
 112        -e 's,  , ,g' \
 113        -e '/^$/d'    \
 114        -e '/<\/p>$/ {N;s,\n\(<!-- [^~]\),\1,}' \
 115        -e 's,~~~[ ]\?[-]\?[-]\?[ ]\?<,~~~\n<,' $tmp
 116
 117 # Fuse header, section and footer with the corresponding div.
 118 for tag in header section footer; do
 119   sed -i "/^<$tag/ {N; s,\\(<$tag[^>]*>\\)\\n<div>,\\1<div>,}" $tmp
 120   sed -i "/^<\\/div>$/ {N; s,<\\/div>\\n\\(<\\/$tag>\\),</div>\\1,}" $tmp
 121 done
 122
 123 # Add LF before main sections and commented-out parts.
 124 sed -i 's,<!-- ~~,\n<!-- ~~,' $tmp
 125 sed -i '/COMMENTED OUT/ s,^,\n,' $tmp
 126
 127 # Make the text more readable.
 128 for tag in p h1 h2 h3 h4 dl title form; do
 129   sed -i "s,<$tag,\\n&," $tmp
 130 done
 131 for tag in p dl ul h1 h2 h3 h4 title head footer form script; do
 132   sed -i "/<\\/$tag>/s,$,\\n," $tmp
 133 done
 134 sed -i '/<\/dd>/ {N; s,</dd>\n<dt,</dd>\n\n<dt,}' $tmp
 135 sed -i '/<\/dt>/ {N; s,</dt>\n<dd,</dt>\n\n<dd,}' $tmp
 136 sed -i -e 's,</p></span>$,</p>\n</span>,' \
 137        -e 's, alt=,\nalt=,g' \
 138        -e 's, | , |\n,g' $tmp
 139
 140 # Remove extra LFs, if any.
 141 sed -i ':a /^$/ {N; s,\n$,,; ba}' $tmp
 142 sed -i ':a /^\n*$/ {$d; N; ba}' $tmp
 143
 144 # Wrap the text.
 145 fmt -s -w 95 $tmp > $output
 146
 147 close_or_exit 0