File enblend-enfuse-html2xhtml of Package enblend-enfuse
#! /bin/sh
#! /bin/dash -x
# This file is part of Enblend.
# Licence details can be found in the file COPYING.
# name: html2xhtml
# synopsis: Convert the HTML output of makeinfo(1) to valid XHTML
# author: Dr. Christoph L. Spiel
# shell version: bash-3.2.39, dash-0.5.4
readonly global__command=`basename $0`
global__meta_key_val=
readonly global__sed=${SED:-sed}
readonly global__tidy=${TIDY:-tidy}
global__tidy_flags=""
readonly global__xmllint=${XMLLINT:-xmllint}
global__xmllint_flags="--valid --noblanks --noent -nonet --nsclean --format --encode iso-8859-1"
failwith_error ()
{
echo "$global__command: $@" 1>&2
exit 1
}
convert_with_tidy ()
{
# We add the system identifier with sed(1) after tidy(1). See
# function postprocess_tidy_output. Also see
# http://www.w3.org/QA/2002/04/valid-dtd-list.html
# for a list of recommended doctype declarations.
$global__tidy \
--add-xml-decl yes \
--doctype '"-//W3C//DTD XHTML 1.1 plus MathML 2.0 plus SVG 1.1//EN"' \
--file /dev/null \
--new-blocklevel-tags "math, mfenced, mfrac, mover, mrow, msqrt, msub, msup, mtable, mtd, mtr, munder, munderover" \
--new-empty-tags mspace \
--new-inline-tags "mathinline, mfencedinline, mi, mn, mo, mrowinline, msqrtinline, msubinline, msupinline, mtext" \
--sort-attributes alpha \
--output-encoding latin1 \
--output-xhtml yes \
--preserve-entities yes \
--quiet yes \
--wrap 0 \
$global__tidy_flags \
"$@"
}
# We must postprocess tidy(1)'s output, because 1/ it still contains
# elements that make an XML parser puke and 2/ we have intentionally
# introduced pseudo MathML tags to coerce tidy(1) into formatting them
# inline; these tags must be reverted.
postprocess_tidy_output ()
{
if test -z "$lazy__meta_tag" && test -n "$global__meta_key_val"; then
lazy__meta_tag=`echo $global__meta_key_val | \
$global__sed -e 's#;#\n#g' | \
$global__sed -e 's#\([^,]*\),\(.*\)#<meta name="\1" content="\2" />#'`
fi
$global__sed \
-e '1,9s#""#"http://www.w3.org/2002/04/xhtml-math-svg/xhtml-math-svg.dtd"#' \
\
-e "s#^<head>#<head>\n$lazy__meta_tag#" \
\
-e 's#<html\([^>]*\) lang="[^"]*"\([^>]*\)#<html\1\2#' \
-e 's#<ol[^>]*>#<ol>#' \
-e 's#<t\([dh]\)\([^>]*\)width="[^"]*"\([^>]*\)#<t\1\2\3#' \
-e 's#<ul\([^>]*\)compact="[^"]*"\([^>]*\)#<ul\1\2#' \
\
-e 's#<\(/*\)\(math\|mfenced\|mrow\|msqrt\|msub\|msup\)inline#<\1\2#g'
}
# Makeinfo(1) generates indirect-reference files for each float.
# Their filenames start with "Table_" or "Figure_*", which is how we
# distinguish them from the html files making up the body.
is_indirect_reference_file ()
{
local filename=$1
test -z "${filename%Table_*}" || test -z "${filename%Figure_*}"
}
# Convert all html files iside the given directory.
convert_directory ()
{
local html_directory=$1
local xhtml_directory=`basename $html_directory .html`.xhtml
mkdir -p $xhtml_directory
(
cd $html_directory || failwith_error "cannot cd to \"$html_directory\""
ls -1 *.html | \
$global__sed -ne 's#\(.*\)\.html#s|\1\\.html|\1.xhtml|g#p' > ,fix-file-refs.sed
)
printf '' > $html_directory/,fix-indir.sed
for html_file in $html_directory/*.html; do
local html_base=`basename $html_file`
if is_indirect_reference_file $html_base; then
url=`$global__sed -ne 's#^.*url=\([^"]*\)".*$#\1#p' < $html_file`
target_file=`echo $url | $global__sed -e 's/#.*$//'`
echo "s|$html_base|$target_file|" >> $html_directory/,fix-indir.sed
fi
done
for html_file in $html_directory/*.html; do
local html_base=`basename $html_file`
if ! is_indirect_reference_file $html_base; then
local xhtml_file=$xhtml_directory/`basename $html_base .html`.xhtml
convert_with_tidy $html_file | \
postprocess_tidy_output | \
$global__sed \
-f $html_directory/,fix-indir.sed \
-f $html_directory/,fix-file-refs.sed \
-e 's|<a\([^>]*\)name="[^"]*"\([^>]*\)|<a\1\2|g' \
-e 's|\("[^/"]*\)\.html\([#"]\)|\1.xhtml\2|' | \
$global__xmllint $global__xmllint_flags --output $xhtml_file -
test $? -eq 0 || { rm -f $xhtml_file; exit 1; }
fi
done
}
# Convert a single standalone html file.
convert_file ()
{
local html_file=$1
local xhtml_file=`basename $html_file .html`.xhtml
convert_with_tidy $html_file | \
postprocess_tidy_output | \
$global__sed -e 's|<a\([^>]*\)name=\([^>]*\)|<a\1id=\2|g' | \
$global__xmllint $global__xmllint_flags --output $xhtml_file -
test $? -eq 0 || { rm -f $xhtml_file; exit 1; }
}
convert_html_to_xhtml ()
{
local html=$1
if test -d "$html"; then
convert_directory $html
else
convert_file $html
fi
}
# We prefer to check for all binaries before we get an unpleasant
# surprise in one of the pipes.
check_binaries ()
{
for v in global__sed global__tidy global__xmllint; do
eval x=\$$v
test -z "$x" && failwith_error "variable \"${v#global__}\" is not defined"
$x --version > /dev/null 2>&1 || \
failwith_error "cannot execute \"$x\" for \"${v#global__}\""
done
}
parse_options ()
{
: $((n=0))
while test -n "$1"; do
case "$1" in
--meta=*)
if test -z "$global__meta_key_val"; then
global__meta_key_val="${1#--meta=}"
else
global__meta_key_val="$global__meta_key_val;${1#--meta=}"
fi
;;
-h | --help)
show_help
exit 0
;;
--tidy-flags=*)
global__tidy_flags="$global__tidy_flags ${1#--tidy-flags=}"
;;
--xmllint-flags=*)
global__xmllint_flags="$global__xmllint_flags ${1#--xmllint-flags=}"
;;
--)
shift
: $((n=n+1))
break 2
;;
--*)
failwith_error "unknown long option \"$1\""
;;
-*)
failwith_error "unknown short option \"$1\""
;;
*)
break 2
esac
shift
: $((n=n+1))
done
return $n
}
show_help ()
{
printf %s "\
Usage: $global__command_name [OPTIONS] [HTML...]
Convert HTML files or directories to XHTML.
Options:
--meta=NAME,CONTENT add meta tags with NAME and CONTENT
--tidy-flags=FLAGS set flags for tidy(1)
--xmllint-flags=FLAGS set flags for xmllint(1)
-h, --help show this help screen
Influential environment variables:
SED name of sed binary [default: \"$global__sed\"]
TIDY name of tidy binary [default: \"$global__tidy\"]
XMLLINT name of xmllint binary [default: \"$global__xmllint\"]
"
}
main ()
{
parse_options "$@"
shift $? # shift away all options
check_binaries
for x in "$@"; do
convert_html_to_xhtml "$x"
done
}
main "$@"
exit 0
