[enc.git] / ua / kitchen / reformat-html.1

#!/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 sleep_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 HTML."
    sleep 3
  fi
  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=${input#\'}
fi
test -f "$input" -a "${input%.html}" != "$input" || sleep_or_exit 1

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

tmp=$(mktemp -t ref.XXXXXX) || exit 1
tmp1=$(mktemp -t ref.XXXXXX) || exit 1
tmp2=$(mktemp -t ref.XXXXXX) || exit 1
trap 'rm -f "$tmp" "$tmp1" "$tmp2"' EXIT

# Don't touch the scripts.
sed -n '/<script/,$p' $input > $tmp1
sed    '/<script/,$d' $input > $tmp

# Clean up extra spaces and tabs; remove blank lines.
sed -i -e 's,[[:space:]]\+, ,g' \
       -e 's,^ ,,' \
       -e 's, $,,' \
       -e 's,> <,><,g' \
       -e '/^$/d' $tmp

# For the language list: fix the commented-out items (broken by po4a).
sed -i -e '/<\/a>$/ {N; s,<\/a>\n<,<\/a><,}' $tmp
sed -i -e '/^<li/ {N; s,>\n<a ,><a ,}' $tmp

# </p> at the end of the line (much better to do it by hand).
sed -i -e '/[>.]$/ {N;s,\n</strong,</strong,}' $tmp
sed -i -e '/[>.]$/ {N;s,\n</p,</p,}' $tmp
sed -i -e '/"$/ {N;s,\n</p,</p,}' $tmp

# <p> and its attributes on the same line.
sed -i -e '/<p$/ {N;s,\n, ,}' $tmp

# Remove LF after opening tags.
for tag in li p strong a h3; do
  sed -i "/<$tag[^>]*>$/ {N; s,\\n,,}" $tmp
done

# Single out paragraphs (fused with <noscript> in one instance), main
# sections, and image links (e.g. infographic) when followed by text.
# Separate truncated "~~~" comment from fused tag.
# Fold img tags; Add a line after footer.
sed -i -e 's,\([^t]>\)<p,\1\n<p,' \
       -e 's,><noscript,>\n<noscript,' \
       -e 's,<!-- ~~,\n<!-- ~~,' \
       -e 's,/></a> \([[:alnum:]]\),/></a>\n\1,' \
       -e 's,~~~[ ]\?[-]\?[-]\?[ ]\?<,~~~\n<,' \
       -e 's, src=",\nsrc=",' \
       -e 's, alt=",\nalt=",' \
       -e 's,<!-- End #footer -->,&\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

# Make the text more readable.
sed -i 's,\(<link[^>]*>\)<,\1\n<,' $tmp
sed -i 's,\(<meta[^>]*>\)<,\1\n<,' $tmp

for tag in p dd li h1 h2 h3 form; do
  sed -i "/<\\/$tag>$/s,$,\\n," $tmp
done
for tag in p dl ul h1 h2 h3 h4 form body; do
  sed -i "/^<$tag/s,^,\\n," $tmp
done
sed -i '/^$/ {N; s,^\n</dl>,</dl>\n,}' $tmp
sed -i '/^$/ {N; s,^\n</ul>,</ul>\n,}' $tmp

# Remove blank lines in menus and image blocks.
sed -i '/^$/ {N; s,^\n<li><img,<li><img,}' $tmp
sed -i '/^$/ {N; s,^\n<li[^>]*><a ,<li><a ,}' $tmp
sed -i '/JS enabled -->/ {N;N; s,\n\n,\n,}' $tmp

# Unwrap the last item of the language list for easier
# replacement.
sed -i '/<li><a/ {N; s,\n\(.*Translation_Guide\), \1,}' $tmp

# Wrap lines.
fmt -s -w 80 $tmp > $tmp2

# Suppress repeated empty lines.
cat -s $tmp2 $tmp1 > $output

sleep_or_exit 0
Commit	Line	Data
5ec82b3f A	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" == "$PPID" && called_from_color_wdiff=1
	27
	28	function sleep_or_exit () {
	29	# turns off interactivity and lets the terminal close normally if the script
	30	# is called from color-wdiff.
	31
	32	if test "$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 HTML."
	37	sleep 3
	38	fi
	39	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=${input#\'}
	49	fi
	50	test -f "$input" -a "${input%.html}" != "$input" \|\| sleep_or_exit 1
	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=$(mktemp -t ref.XXXXXX) \|\| exit 1
	60	tmp1=$(mktemp -t ref.XXXXXX) \|\| exit 1
	61	tmp2=$(mktemp -t ref.XXXXXX) \|\| exit 1
	62	trap 'rm -f "$tmp" "$tmp1" "$tmp2"' EXIT
	63
	64	# Don't touch the scripts.
65	sed -n '/<script/,$p' $input > $tmp1
66	sed '/<script/,$d' $input > $tmp
67
68	# Clean up extra spaces and tabs; remove blank lines.
69	sed -i -e 's,[[:space:]]\+, ,g' \
70	-e 's,^ ,,' \
71	-e 's, $,,' \
72	-e 's,> <,><,g' \
73	-e '/^$/d' $tmp
74
75	# For the language list: fix the commented-out items (broken by po4a).
76	sed -i -e '/<\/a>$/ {N; s,<\/a>\n<,<\/a><,}' $tmp
77	sed -i -e '/^<li/ {N; s,>\n<a ,><a ,}' $tmp
78
79	# </p> at the end of the line (much better to do it by hand).
80	sed -i -e '/[>.]$/ {N;s,\n</strong,</strong,}' $tmp
81	sed -i -e '/[>.]$/ {N;s,\n</p,</p,}' $tmp
82	sed -i -e '/"$/ {N;s,\n</p,</p,}' $tmp
83
84	# <p> and its attributes on the same line.
85	sed -i -e '/<p$/ {N;s,\n, ,}' $tmp
86
87	# Remove LF after opening tags.
88	for tag in li p strong a h3; do
89	sed -i "/<$tag[^>]*>$/ {N; s,\\n,,}" $tmp
90	done
91
92	# Single out paragraphs (fused with <noscript> in one instance), main
93	# sections, and image links (e.g. infographic) when followed by text.
94	# Separate truncated "~~~" comment from fused tag.
95	# Fold img tags; Add a line after footer.
96	sed -i -e 's,\([^t]>\)<p,\1\n<p,' \
97	-e 's,><noscript,>\n<noscript,' \
98	-e 's,<!-- ~~,\n<!-- ~~,' \
99	-e 's,/></a> \([[:alnum:]]\),/></a>\n\1,' \
100	-e 's,~~~[ ]\?[-]\?[-]\?[ ]\?<,~~~\n<,' \
101	-e 's, src=",\nsrc=",' \
102	-e 's, alt=",\nalt=",' \
103	-e 's,<!-- End #footer -->,&\n,' $tmp
104
105	# Fuse header, section and footer with the corresponding div.
106	for tag in header section footer; do
107	sed -i "/^<$tag/ {N; s,\\(<$tag[^>]*>\\)\\n<div>,\\1<div>,}" $tmp
108	sed -i "/^<\\/div>$/ {N; s,<\\/div>\\n\\(<\\/$tag>\\),</div>\\1,}" $tmp
109	done
110
111	# Make the text more readable.
112	sed -i 's,\(<link[^>]*>\)<,\1\n<,' $tmp
113	sed -i 's,\(<meta[^>]*>\)<,\1\n<,' $tmp
114
115	for tag in p dd li h1 h2 h3 form; do
116	sed -i "/<\\/$tag>$/s,$,\\n," $tmp
117	done
118	for tag in p dl ul h1 h2 h3 h4 form body; do
119	sed -i "/^<$tag/s,^,\\n," $tmp
120	done
121	sed -i '/^$/ {N; s,^\n</dl>,</dl>\n,}' $tmp
122	sed -i '/^$/ {N; s,^\n</ul>,</ul>\n,}' $tmp
123
124	# Remove blank lines in menus and image blocks.
125	sed -i '/^$/ {N; s,^\n<li><img,<li><img,}' $tmp
126	sed -i '/^$/ {N; s,^\n<li[^>]*><a ,<li><a ,}' $tmp
127	sed -i '/JS enabled -->/ {N;N; s,\n\n,\n,}' $tmp
128
129	# Unwrap the last item of the language list for easier
130	# replacement.
131	sed -i '/<li><a/ {N; s,\n\(.*Translation_Guide\), \1,}' $tmp
132
133	# Wrap lines.
134	fmt -s -w 80 $tmp > $tmp2
135
136	# Suppress repeated empty lines.
137	cat -s $tmp2 $tmp1 > $output
138
139	sleep_or_exit 0