#!/bin/ksh
#
# a2html.sh
#
# Copyright (C) 2014 Taiji Yamada <taiji@aihara.co.jp>
#
me="$(basename "$0")"
usage(){
  cat <<EOF
$me - converts plain text to HTML like AsciiDoc
usage:
	$me - < pathname
	$me pathname
author(s):
	Taiji Yamada <taiji@aihara.co.jp>
EOF
}
sed --posix < /dev/null > /dev/null 2>&1 && sed_flags='-r --posix' || sed_flags=-E
html_lang="$(echo "$LC_ALL" | sed 's/_.*//')"
case "$LC_ALL" in
ja_JP.eucJP)	cat="iconv -f euc-jp -t utf-8"	;;
*)		cat="cat"	;;
esac

atext2htext_body(){
  LC_ALL=C sed $sed_flags -ne '
/^./,/^[-=~^+]{4,}(\[\[.+\]\])?$/{
  N
  /^(.+)\n[-=~^+]{4,}(\[\[.+\]\])?$/{
    s/^(.+)\n={4,}(\[\[.+\]\])?$/= \1 =\2/
    s/^(.+)\n-{4,}(\[\[.+\]\])?$/== \1 ==\2/
    s/^(.+)\n~{4,}(\[\[.+\]\])?$/=== \1 ===\2/
    s/^(.+)\n\^{4,}(\[\[.+\]\])?$/==== \1 ====\2/
    s/^(.+)\n\+{4,}(\[\[.+\]\])?$/===== \1 =====\2/
    p
    d
  }
}
P
D
' | LC_ALL=C sed $sed_flags -ne '
:first
s|^={4,}$|<hr/>|p
t last
s|^-{4,}$|<hr/>|p
t last
/^(.*) <{4,}$/{
  s|^(.*) <{4,}$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^(.*)$|<div style="text-align: left;">\1</div>|
  b no_attribute_p_last
}
/^<{4,} (.*) >{4,}$/{
  s|^<{4,} (.*) >{4,}$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^(.*)$|<div style="text-align: center;">\1</div>|
  b no_attribute_p_last
}
/^>{4,} (.*)$/{
  s|^>{4,} (.*)$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^(.*)$|<div style="text-align: right;">\1</div>|
  b no_attribute_p_last
}
/^= (.*) =\[\[(.+)\]\]$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^= (.*) =\[\[(.+)\]\]$|<h1 id="\2">\1</h1>|
  b no_attribute_p_last
}
/^== (.*) ==\[\[(.+)\]\]$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^== (.*) ==\[\[(.+)\]\]$|<h2 id="\2">\1</h2>|
  b no_attribute_p_last
}
/^=== (.*) ===\[\[(.+)\]\]$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^=== (.*) ===\[\[(.+)\]\]$|<h3 id="\2">\1</h3>|
  b no_attribute_p_last
}
/^==== (.*) ====\[\[(.+)\]\]$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^==== (.*) ====\[\[(.+)\]\]$|<h4 id="\2">\1</h4>|
  b no_attribute_p_last
}
/^===== (.*) =====\[\[(.+)\]\]$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^===== (.*) =====\[\[(.+)\]\]$|<h5 id="\2">\1</h5>|
  b no_attribute_p_last
}
/^= (.*) =$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^= (.*) =$|<h1>\1</h1>|
  b p_last
}
/^== (.*) ==$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^== (.*) ==$|<h2>\1</h2>|
  b p_last
}
/^=== (.*) ===$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^=== (.*) ===$|<h3>\1</h3>|
  b p_last
}
/^==== (.*) ====$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^==== (.*) ====$|<h4>\1</h4>|
  b p_last
}
/^===== (.*) =====$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^===== (.*) =====$|<h5>\1</h5>|
  b p_last
}
/^= (.*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^= (.*)$|<h1>\1</h1>|
  b p_last
}
/^== (.*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^== (.*)$|<h2>\1</h2>|
  b p_last
}
/^=== (.*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^=== (.*)$|<h3>\1</h3>|
  b p_last
}
/^==== (.*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^==== (.*)$|<h4>\1</h4>|
  b p_last
}
/^===== (.*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^===== (.*)$|<h5>\1</h5>|
  b p_last
}
/^\/{4,} (.*) \/{4,}/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^\/{4,} (.*) \/{4,}$|<address>\1</address>|
  b p_last
}
/^\.([^ .].*)$/{
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^\.([^ .].*)$|<div style="font-weight: bolder; font-size: larger;">\1</div>|
  b no_attribute_p_last
}
/^[	]*\* (.*)$/,/^$/{
  /^$/{
    x
    /<ul>$/b close_ul
    b last
  }
  /^[	]*\* (.*)$/{
    x
    /<ul>$/!b start_ul
    x
  }
  s|^[	]*\* (.*)$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^.*$|  <li>&</li>|
  b p_last
:start_ul
  s/$/<ul>/
  x
  i\
<ul>
  s|^[	]*\* (.*)$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^.*$|  <li>&</li>|
  b p_last
:close_ul
  i\
</ul>
  s/<ul>$//
  x
  b last
}
/^[	]*\. (.*)$/,/^$/{
  /^$/{
    x
    /<ol>$/b close_ol
    b last
  }
  /^[	]*\. (.*)$/{
    x
    /<ol>$/!b start_ol
    x
  }
  s|^[	]*\. (.*)$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^.*$|  <li>&</li>|
  b p_last
:start_ol
  s/$/<ol>/
  x
  i\
<ol>
  s|^[	]*\. (.*)$|\1|
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|^.*$|  <li>&</li>|
  b p_last
:close_ol
  i\
</ol>
  s/<ol>$//
  x
  b last
}
/^[	]*\|[-=]{4,}$/,/^[	]*\|[-=]{4,}$/{
  /^[	]*\|[-=]{4,}$/{
    x
    /<table>$/b close_table
    b start_table
  }
  s/^[	]*\|//
  s/$/	/
  s/[	]+/	/g
  s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
  s|("[^"]+")|<samp>\1</samp>|g
  s|([^	][^	]*)	|<td>\1</td>|g
  s|<td> ([^	]*) </td>|<td style="white-space: nowrap; text-align: center;">\1</td>|g
  s|<td> ([^	]*)</td>|<td style="white-space: nowrap; text-align: right;">\1</td>|g
  s|<td>([^	]*) </td>|<td style="white-space: nowrap; text-align: left;">\1</td>|g
  s|^.*$|<tr>&</tr>|
  b no_attribute_p_last
:start_table
  s/$/<table>/
  x
  i\
<table>\
<tbody>
  b last
:close_table
  i\
</tbody>\
</table>
  s/<table>$//
  x
  b last
}
/^[	]*-{4,}$/,/^[	]*-{4,}$/{
  /^[	]*-{4,}$/{
    x
    /<pre>$/b close_pre
    b start_pre
  }
  b no_tag_p_last
:start_pre
  s/$/<pre>/
  x
  i\
<pre>
  b last
:close_pre
  i\
</pre>
  s/<pre>$//
  x
  b last
}
/^.+$/,/^$/{
  /^$/{
    x
    /<p>$/b close_p
    b last
  }
  /^	([^-	\.\*].*)/{
    x
    /<p>$/!{
      s/$/<p>/
      i\
<p>
    }
    x
    s/^(	.*)/\1/
    s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
    s|^.*$|<samp>&</samp><br/>|g
    p
    b last
  }
  /.+$/{
    x
    /<p>$/!b start_p
    x
  }
  b no_tag_p_last
:start_p
  s/$/<p>/
  x
  i\
<p>
  b no_tag_p_last
:close_p
  i\
<!--/p-->
  s/<p>$//
  x
  b last
}
:no_tag_p_last
s/&/\&amp;/g;s/</\&lt;/g;s/>/\&gt;/g
:p_last
s|("[^"]+")|<samp>\1</samp>|g
:no_attribute_p_last
s|``([^`'"'"']+)'"''"'|\&ldquo;\1\&rdquo;|g
s|('"'"'[^'"'"']+'"'"')|<code>\1</code>|g
s|(`[^`]+`)|<kbd>\1</kbd>|g
s|\/\/([^\/]+)\/\/|<i>\1</i>|g
s|\*\*([^\*]+)\*\*|<b>\1</b>|g
s|\%\%([^\%]+)\%\%|<tt>\1</tt>|g

s|\[\*([^\*]+)\*\]\[\[,([^]]*)\]\]|<strong style="\2">\1</strong>|g
s|\[-(.+)-\]\[\[,([^]]*)\]\]|<strike style="\2">\1</strike>|g
s|\[\+(.+)\+\]\[\[,([^]]*)\]\]|<ins style="\2">\1</ins>|g
s|\[%([^%]+)%\]\[\[,([^]]*)\]\]|<code style="\2">\1</code>|g
s|\[\$([^$]+)\$\]\[\[,([^]]*)\]\]|<var style="\2">\1</var>|g
s|\[\?([^?]+)\?\]\[\[,([^]]*)\]\]|<abbr style="\2">\1</abbr>|g
s|\[\!([^!]+)\!([^!]+)\!\]\[\[,([^]]*)\]\]|<a href="\2" style="\2">\1</a>|g
s|\[\!([^!]+)\!\]\[\[,([^]]*)\]\]|<a href="\1" style="\2">\1</a>|g
s|\[\#([^#]+)\#\]\[\[,([^]]*)\]\]|<code style="\2">\1</code>|g
s|\[\@([^@]+)\@\]\[\[,([^]]*)\]\]|<kbd style="\2">\1</kbd>|g
s|\[\^([^\^]+)\^\]\[\[,([^]]*)\]\]|<sup style="\2">\1</sup>|g
s|\[\~([^\~]+)\~\]\[\[,([^]]*)\]\]|<sub style="\2">\1</sub>|g
s|\[_([^_]+)_\]\[\[,([^]]*)\]\]|<u style="\2">\1</u>|g
s|\[/([^/]+)/\]\[\[,([^]]*)\]\]|<em style="\2">\1</em>|g
s|\["([^"]+)"\]\[\[,([^]]*)\]\]|<cite style="\2">\1</cite>|g
s|\[\(([^)]+)\)\]\[\[,([^]]*)\]\]|<div style="\2">\1</div>|g

s|\[\*([^\*]+)\*\]|<strong>\1</strong>|g
s|\[-(.+)-\]|<strike>\1</strike>|g
s|\[\+(.+)\+\]|<ins>\1</ins>|g
s|\[%([^%]+)%\]|<code>\1</code>|g
s|\[\$([^$]+)\$\]|<var>\1</var>|g
s|\[\?([^?]+)\?\]|<abbr>\1</abbr>|g
s|\[\!([^!]+)\!([^!]+)\!\]|<a href="\2">\1</a>|g
s|\[\!([^!]+)\!\]|<a href="\1">\1</a>|g
s|\[;([^;]+);([^;]+);\]|<img src="\1" \2 alt="\1"/>|g
s|\[;([^;]+);\]|<img src="\1" alt="\1"/>|g
s|\[\#([^#]+)\#\]|<code>\1</code>|g
s|\[\@([^@]+)\@\]|<kbd>\1</kbd>|g
s|\[\^([^\^]+)\^\]|<sup>\1</sup>|g
s|\[\~([^\~]+)\~\]|<sub>\1</sub>|g
s|\[_([^_]+)_\]|<u>\1</u>|g
s|\[/([^/]+)/\]|<em>\1</em>|g
s|\["([^"]+)"\]|<cite>\1</cite>|g
s|\[\(([^)]+)\)\]|<div>\1</div>|g

s|\[\{(.*)\}\]|<!--\1-->|g
s|(mailto:[^[]+)\[([^]]+)\]|<a href="\1">\2</a>|g
s|([-[:alpha:]]+://[^[]+)\[([^]]+)\]|<a href="\1">\2</a>|g
s|([-[:alpha:]]+://[^[]+)\[\]|<a href="\1"><tt>\1</tt></a>|g
p
:last
${
  x
:close
  /<(ul|ol|pre|p)>$/{
    /<ul>$/{
      s/<ul>$//
      i\
</ul>
    }
    /<ol>$/{
      s/<ol>$//
      i\
</ol>
    }
    /<pre>$/{
      s/<pre>$//
      i\
</pre>
    }
    /<p>$/{
      s/<p>$//
      i\
<!--/p-->
    }
    b close
  }
  q
}
'
}
atext2htext_mathjax=false
atext2htext(){
  file="$1"
  cat <<EOF
<!DOCTYPE html>
<html lang="$html_lang">
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
<title>$1</title>
EOF
$atext2htext_mathjax && cat <<EOF
EOF
cat <<EOF
</head>
<body>
EOF
  $cat "$1" | atext2htext_body
  cat <<EOF
</body>
</html>
EOF
}
while [ "$1" != "" ]; do
  case "$1" in
  -h|--help)	usage; exit 0		;;
  --mathjax)	atext2htext_mathjax=:	;;
  -)		atext2htext_body; exit	;;
  *)		atext2htext "$1"; exit	;;
  esac
  shift
done