#!/usr/local/bin/perl -w
#
# Jeffrey Friedl            		襸ե꡼ 
# Omron Corporation			ʳ
# Nagaokakyoshi, Japan			617Ĺ
#
# jfriedl@nff.ncl.omron.co.jp
#
# This work is placed under the terms of the GNU General Purpose Licence
# (the "GNU Copyleft").
#

print <<END_OF_TEXT;
/*
 *
 *  This file generated from (and by) the file "$0".
 *
 */

#define S const unsigned char *

END_OF_TEXT

$/ = '';	## set record seperator to be double-newline.
$count = 0;

$linenum = &linenumbr_of_data;
while (<DATA>)
{
	++$count;

	($flag, $help, $usage, $regex, $execution, $empty, $other) =
	        split(/\s*\n\s*/);

        ##
	## We expect exactly $flag .. $execution. We expect $empty to be
	## defined but null. If $empty is undefined or non-null, or
	## or $other is defined, the count out of split was wrong.
	if (!defined($empty) || $empty ne '' || defined($other)) {
	    print qq/\n#error was bad record in "$0" at line $linenum.\n/;
	    die qq/$0:$linenum: bad command record.\n/;
	}

	$help =~ s/["\\]/\\$&/g;
	$help = ($help eq "#") ? "0" : qq/"$help"/;
	$usage =~ s/["\\]/\\$&/g;
	$usage = ($usage eq "#") ? "0" : qq/"$usage"/;

	$regex =~ s/\s+/\\s*/g;		# turn whitespace into a regex
	$regex = '^\\s*'.$regex.'\\s*$' if $regex !~ m/^\^/ && $regex !~ m/\$\$/;
	$regex =~ s/["\\]/\\$&/g;	# protect quotes and backslashes

	if ($execution eq "skip") {
           $fun = "0";
        } elsif ($execution =~ m/^(\w+)\(\);?$/) {
	   $fun = $1;
        } else {
	    $execution =~ s/\\(\d)/cmd_paren[$1-1]/g;
	    $fun = "_func${count}_";
	    print "\n/* generated from \"$0\", record at line $linenum */\n";
	    print "static int $fun(void)\n{\n  return $execution;\n}\n";
	}
	push(@entries, join("\n",
	    
	    qq#\n    /* generated from "$0" record at line $linenum*/#,
	    qq#    {#,
            qq#        $flag,#,
	    qq#        (S)$usage,#,
	    qq#        (S)$help,#,
	    qq#        (S)"$regex",#,
	    qq#         $fun,#,
	    qq#    }#));

	$linenum += tr/\n/\n/;
}
print "\nstatic struct command command[] = {\n", join(",\n",@entries), "};\n";

sub linenumbr_of_data { __LINE__ + 2; } ## make sure I'm just before __END__
__END__
CMD_FILE_ONLY /* comment line */
#
#
^\s*#
skip

CMD_FILE_ONLY /* blank line */
#
#
^\s*$
skip

CMD_FILE_ONLY /* search */
#
#
^\s*([+!=].*)
cmd_do_search(\1)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns automatic romaji conversion on or off, or reports current status.
[default] autokana [on|1|off|0]
default autokana> (on|1|off|0)?
cmd_set_local_autokana_flag(\1)

CMD_GENERAL
#
#
default autokana> (on|1|off|0)?
cmd_set_default_autokana_flag(\1)

CMD_GENERAL
clear the screen
clear|cls
(clear|cls)
cmd_clear()

CMD_GENERAL
set the this-is-a-command character
cmdchar ['bytechar']
cmd(char)? ('(\a)')?
cmd_cmdchar(\3)

CMD_GENERAL
Turns command debugging on or off, or reports current status.
command debug [on|1|off|0]
c(om(mand)?)?> debug> (on|1|off|0)?
cmd_set_default_cmd_debug_flag(\3)

CMD_GENERAL|CMD_LOAD_RELATED
Combines previously-loaded slots to one new slot.
combine> ["name"] [num =] num [num....]
comb(o|ine) ((["'])(.*)\3 =?)? (\d+ \+?=)? ((#? \d+ ,? )+)
cmd_combine(\4, \5 ? atoi(\5) : -1, \6)

CMD_GENERAL
Turns debugging on or off, or reports current status.
debug [on|1|off|0]
debug> (on|1|off|0)?
cmd_set_default_debug_flag(\1)


CMD_GENERAL
describes the encodings for the character(s) indicated
describe "string"|character|[kuten|euc|jis|sjis|ascii]code
describe> (['"])(.*)\1
cmd_describe_raw(\2);

CMD_GENERAL
#
#
describe> (\A+|[!-~])
cmd_describe_raw(\1);


CMD_GENERAL
#
#
describe> (kuten> )?(\d\d\d\d)
cmd_describe_kuten(\2);

CMD_GENERAL
#
#
describe ascii (\\|0)([0-7]+)
cmd_describe_ascii(8, \2);

CMD_GENERAL
#
#
describe ascii (\d+)
cmd_describe_ascii(10, \1);

CMD_GENERAL
#
#
describe ascii (0[xX])?([0-9a-f][0-9a-f])
cmd_describe_ascii(16, \1);


CMD_GENERAL
#
#
describe> (sjis>|jis>|euc>)? (0[xX])?(<[0-9a-f][0-9a-f][0-9a-f][0-9a-f]>)
cmd_describe_encoding(\1, \3);

CMD_GENERAL
#
#
describe> (\$@|\$B|\$@\$B)?(([!-~][!-~])+)(\(\$@|\$B|\$@\$B|\$\(D|\(J|\(H|\(B|\(I)?
cmd_describe_jis_string(\2)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns display of matching lines on or off.
[default] display [on|1|off|0]
display> (on|1|off|0)?
cmd_set_local_display_flag(\1)

CMD_GENERAL
#
#
default display> (on|1|off|0)?
cmd_set_default_display_flag(\1)

CMD_GENERAL|CMD_ENCODING_RELATED
set the input/output encoding-method
encoding (euc|jis|sjis)
encod(e|ing)> (euc|jis|sjis)?
cmd_encoding(\2)

CMD_GENERAL
list what files are loaded into what slots.
files|slots [-]
(slot|file)s?> (-|-?(help|long))?
cmd_list_files(\2 ? 1 : 0)

CMD_GENERAL|CMD_LOAD_RELATED
set the filter for the selected file.
filter ["name"] [!] /regex/[i]
filter> ("([^"]*)")? (!?) (\S)(.+)\4(i?)
cmd_filter(\5, \2, \3[0] == '!', \6[0] == 'i')

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
turn the filter (for the selected file) on or off, or report its status
filter [on|1|off|0]
filter> (on|1|off|0)?
cmd_toggle_filter(\1)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns case folding on or off, or reports current status.
[default] fold [on|1|off|0]
fold> (on|1|off|0)?
cmd_set_local_fold_flag(\1)

CMD_GENERAL
#
#
default fold> (on|1|off|0)?
cmd_set_default_fold_flag(\1)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns fuzzification on or off, or reports current status.
[default] fuzz [on|1|off|0]
fuzz> (on|1|off|0)?
cmd_set_local_fuzz_flag(\1)

CMD_GENERAL
#
#
default fuzz> (on|1|off|0)?
cmd_set_default_fuzz_flag(\1)

CMD_GENERAL
list help
help [string]
help (.*\S)?
cmd_help(\1)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns highlighting on or off, or reports current status.
[default] highlight [on|1|off|0]
h(igh)?l(ight)?> (on|1|off|0)?
cmd_set_local_highlight_flag(\3)

CMD_GENERAL
Sets the highlighting style (bold, inverse[standout], or to a given HTML tag)
highlight [bold|inverse|blink|under|<___>|black|red|green|yellow|blue|purple|cyan|white]
h(igh)?l(ight)?> (style)? (red|green|yellow|blue|purple|cyan|white|bold|inverse|standout|blink|under|\<([a-zA-Z]+)\>)?
cmd_set_highlighting_style(\4)

CMD_GENERAL
if {expr} command
Runs command only if EXPR is true.
if {([^}]*)} (.*)
cmd_if(\1,\2)

CMD_GENERAL
#
#
default h(igh)?l(ight)?> (on|1|off|0)?
cmd_set_default_highlight_flag(\3)

CMD_GENERAL|CMD_ENCODING_RELATED
report or set the input encoding-method for 8-bit bytes (JIS always OK)
input encoding [euc|sjis]
input (encoding)?> (euc|sjis)?
cmd_input_encoding(\2)

CMD_GENERAL
set the maximum number of lines to print during any one command.
limit [ <value> ]
limit (=? (\d+))?
cmd_set_limit(\2)

CMD_GENERAL|CMD_LOAD_RELATED
load a file (and read or compute its index as needed)
load "file"
(fast)?load> (-?(now)|-?(w(hen)?n(eeded)?))? (['"]?)(\S+)\7
((\3||\4) ? warn("flags no longer need for 'load' command\n") : 0), cmd_load((const char *)\8);

CMD_GENERAL
log output to a file
log [- | [+]"file"]
log> ((-|<off>)|((<to>)? (\+)? (['"]?)(.+)\6))?
cmd_log(\2 ? 1 : 0, \5 ? 1 : 0, \7)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
sets the modify regular expression and replacement for the selected file.
modify /regex/replace/[ig]
modify> (\S)(.+)\1(.*)\1([ig]?)
cmd_modify(\2,\3, \4[0] == 'i' || \4[1] == 'i', \4[0] == 'g' || \4[1] == 'g')

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
sets the modify filter on or off (for the indicated file)
modify [on|1|off|0]
modify> (on|1|off|0)?
cmd_toggle_modify(\1)

CMD_GENERAL
prints a message to the screen
msg ....
msg (.*)
cmd_msg(\1)

CMD_GENERAL|CMD_ENCODING_RELATED
report or set the output encoding-method
output encoding [euc|sjis|jis|...]
output> (encoding>)? (euc|sjis|jis-?(78|83|90)?(-(ascii|roman))?)?((<(212|no212|hwk|nohwk|foldhwk|disp|nodisp|code|mark)>|[-,\s]+)*)
cmd_output_encoding(\2, \3, \5, \6)

CMD_GENERAL
configure (width x height) or toggle the output pager
pager [boolean | [W x] H]
pager> ((on|1|off|0)|(([1-9]\d* [,x])? ([1-9]\d*)))?
cmd_pager(\2, \4, \5)

CMD_GENERAL|CMD_NEEDS_SLOT
set the prompt format string
[local] prompt "string"
local prompt> ((['"])(.+)\2)?
cmd_set_prompt(1, \3)

CMD_GENERAL
#
#
prompt> ((['"])(.+)\2)?
cmd_set_prompt(0, \3)

CMD_GENERAL
Turns regex debugging on or off, or reports current status.
regex debug [on|1|off|0]
r(egex)? debug> (on|1|off|0)?
cmd_set_default_regex_debug_flag(\2)

CMD_GENERAL
set the number of elided lines to remember for the "show" command.
saved list size [ <value> ]
saved? (list)? (size|len(gth)?)? (=? (\d+))?
cmd_list_size(\5)

CMD_GENERAL|CMD_LOAD_RELATED
sets the default file
select [ num | name | . ]
select> ((\.)|(#? (\d+))|((["']?)(.+)\6))?
cmd_select(\4, (const char *)\7, \2 ? 1 : 0)

CMD_GENERAL
show the lines filtered by the last search, if any
show
show
cmd_show()

CMD_GENERAL
load commands from a file
source "filename"
source> (['"]?)(.+)\1
cmd_source((const char *)\2)

CMD_GENERAL
sets the spinner to move each <value> lines checked (0 to disable)
spinner [ <value> ]
spinner (=? (\d+))?
cmd_set_spinner(\2)

CMD_GENERAL
reports stats about the last search
stats
stats?
cmd_stats()

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
set, toggle, or report the tag for the slot.
tag [boolean] ["string"]
tag> (on|1|off|0)? ((['"])(.*)\3)?
cmd_tag(\1, \4)

CMD_GENERAL
Turns verbosity on or off, or reports current status.
verbose [on|1|off|0]
verbose> (on|1|off|0)?
set_verbose_flag(\1)

CMD_GENERAL
report the version number
version
version
cmd_version()

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Turns word-preference mode on or off, or reports current status.
[default] word [on|1|off|0]
word> (on|1|off|0)?
cmd_set_local_word_flag(\1)

CMD_GENERAL
#
#
default word> (on|1|off|0)?
cmd_set_default_word_flag(\1)

CMD_GENERAL|CMD_LOAD_RELATED|CMD_NEEDS_SLOT
Selects wildcard "glob" patterns instead of regular expression patterns.
[default] wildcard [on|1|off|0]
(glob|wild(card)?)> (on|1|off|0)?
cmd_set_local_glob_flag(\3)

CMD_GENERAL
#
#
default (glob|wild(card)?)> (on|1|off|0)?
cmd_set_default_glob_flag(\3)

CMD_GENERAL
Exits the program
exit|quit|bye|leave|done
(exit|quit|bye|leave|done)
cmd_exit()

CMD_GENERAL
#
#
(load|describe|source)
cmd_error("expecting argument", \1)

CMD_GENERAL
#
#
(default|local)? (autokana|cmdchar|debug|describe|encoding|files?|filter|fold|fuzz|help|highlight (style)?|input (encoding)?|limit|load|modify|output (encoding)?|pager|prompt|r(egex)? debug|saved? (list)? (size|len(gth)?)?|select|show|source|spinner|stats?|verbose|version|word|exit|quit|bye|leave|done|combine|combo|tag).*
cmd_error("argument error", \2)