#!/usr/bin/perl
# Script for extracting documentation out of header files
# (c) 2008 Michal Vaner <vorner@ucw.cz>

use strict;
use warnings;

my( $inname, $outname, $depname, $basedir, $defdump ) = @ARGV;
if( defined $inname ) {
	open IN, $inname or die "Could not read $inname ($!)\n";
} else {
	open IN, "<&STDIN" or die "Could not read stdin ($!)\n";
}
if( defined $outname ) {
	open OUT, ">$outname" or die "Could not write $outname ($!)\n";
} else {
	open OUT, ">&STDOUT" or die "Could not write to stdout ($!)\n";
}
my $hasdump;
if( defined $defdump ) {
	open DUMP, ">$defdump" or die "Could not write definition dump $defdump ($!)\n";
	$hasdump = 1;
}

# Function to guess type of statement
sub detect( $ ) {
	( $_ ) = @_;
	# typedef struct|enum { something } name;
	return( $1, 1, $2, "typedef $1 { ... } $2;" ) if /^\s*typedef\s+(struct|enum)\s*{.*}\s*(\w+)\s*;\s*$/s;
	# struct|enum name { something };
	return( $1, 1, $2, $_ ) if /^\s*(struct|enum)\s+(\w+)\s*;\s*$/s;
	my $l = length;
	s/\n.*//s;
	# struct|enum name {
	#   something
	# };
	return( $1, 0, $2, $_ ) if /(struct|enum)\s+(\w+)\s+{/;
	return( 'def', 0, $1, $_ ) if /#define\s+(\w+)/;
	if( $l > length ) {
		warn( "Unknown multiline statement $_\n" );
		return( '', 0, $_, $_ );
	}
	# typedef type (*function_type)(params);
	return( 'type', 1, $2, $_ ) if /^\s*typedef[^()]+?(\(\s*?\*\s*?)?(\w+)(\s*\))?\s*\(.*\)/;
	# type (*function_var)(params);
	return( 'var', 1, $1, $_ ) if /^.*?\(\*(\w+)\)\(.*\)/;
	# type function(name);
	return( 'fun', 1, $2, $1 ) if /^(.*?(\w+)\([^{]*\)[^{]*)/;
	# typedef something name;
	return( 'type', 1, $1, $_ ) if /^\s*typedef.*?(\w+);/;
	# type name;
	return( 'var', 1, $1, $_ ) if /\s\**(\w+);/;
	warn( "Unknown statement $_\n" );
	return( '', 0, $_, $_ );
}

my @deps;
my $id = 0;

sub formatNote( $$ ) {
	my( $head, $comment ) = @_;
	$head =~ s/(\S)[ ]+/$1 /g;
	print OUT "\n";
	print OUT "''''\n";
	chomp $head;
	my( $type, $semicolon, $name, $oneline ) = detect( $head );
	# Just few transformations of the result
	$oneline =~ s/\s+$//;
	$oneline =~ s/;?$/;/ if( $semicolon );
	$head =~ s/;?\s*$/;/ if( $semicolon );
	$head =~ s/(\s|,|\()(\.\.\.)/$1\\$2/g; # Do not convert tripple dot into ellipsis
	print OUT "[[${type}_$name]]\n";
	$head = $oneline if $type eq 'fun';#Remove { from inline functions
	# Remove the generic hack markup
	$head =~ s/_OPEN_PAREN_/(/g;
	$head =~ s/_CLOSE_PAREN_/)/g;
	print OUT "..................\n";
	print OUT "$head\n";
	print OUT "..................\n\n";
	if( $hasdump ) {
		$oneline =~ s/_OPEN_PAREN_/(/g;
		$oneline =~ s/_CLOSE_PAREN_/)/g;
		my $symname = $type.'_'.$name;
		$name =~ s/_OPEN_PAREN_/(/g;
		$name =~ s/_CLOSE_PAREN_/)/g;
		print DUMP "$outname,$symname,$type,$name,$oneline\n";
		$id ++;
	}
	$comment =~ s/_OPEN_PAREN_/(/g;
	$comment =~ s/_CLOSE_PAREN_/)/g;
	$comment =~ s/_GENERIC_LINK_\|([^|]+)\|([^|]+)\|/${1}_OPEN_PAREN_${2}_CLOSE_PAREN_/g;
	print OUT "$comment\n\n";
}

sub process( $$ ) {
	my( $file, $prefixes ) = @_;
	open FILE, $file or die "Could nod read $file ($!)\n";
	my $line;
	my $active;
	my $verbatim;
	my $buff;
	my $head;
	my $struct;
	my $def;
	my $sdepth;
	while( defined( $line = <FILE> ) ) {
		chomp $line;
		# Generic macro hack - replaces the parenthesis so it is valid identifier
		$line =~ s/$_\(([^()]+)\)/${_}_OPEN_PAREN_${1}_CLOSE_PAREN_/g foreach @{$prefixes};
		if( $def ) {
			$head .= "\n".$line;
			$line =~ s/(\/\*.*?\*\/|\/\/.*)//g;
			if( $line !~ /\\\s*$/ ) {
				formatNote( $head, $buff );
				$def = 0;
				$buff = $head = undef;
			}
		} elsif( $struct ) {
			$head .= "\n".$line;
			my $cp = $line;
			$sdepth += ($cp =~ tr/{//);
			$sdepth -= ($cp =~ tr/}//);
			if( !$sdepth ) {
				formatNote( $head, $buff );
				$struct = 0;
				$buff = undef;
				$head = undef;
			}
		} elsif( $verbatim ) {
			if( $line =~ /\*\// ) {
				$verbatim = 0;
				print OUT "\n";
			} else {
				$line =~ s/^\s*\* ?//;
				print OUT "$line\n";
			}
		} elsif( $active ) {
			if( $line =~ /\*\// ) {
				$active = 0;
			} else {
				$line =~ s/^\s*\* ?//;
				$buff .= "$line\n";
			}
		} else {
			if( ( $line =~ /\S/ ) && ( defined $buff ) ) {
				if( $line =~ /^\s*#define.*\\(\s*(\/\/.*|\/\*.*?\*\/|))*/ ) {
					$head = $line;
					$def = 1;
				} elsif( $line =~ /\(/ || $line !~ /{/ || $line =~ /^\s*#define/ ) {
					$_ = $line;
					s/^\s*\s?//;
					s/\/\/.*//;
					s/\/\*.*?\*\///gs;
					formatNote( $_, $buff );
					$head = undef;
					$buff = undef;
				} else {
					$head = $line;
					$struct = $sdepth = 1;
				}
			} elsif( ( $buff ) = ( $line =~ /\/\*\*\*(.*)\*\*\*\// ) ) {
				$buff =~ s/^\s?//;
				print OUT "$buff\n\n";
				$buff = undef;
			} elsif( ( $buff ) = ( $line =~ /^\s*\/\*\*(.*)\*\*\// ) ) {
				$buff =~ s/^\s*//;
				$buff .= "\n";
			} elsif( ( $head, $buff ) = ( $line =~ /^(.*)\/\*\*(.*)\*\*\// ) ) {
				$buff =~ s/^\s*//;
				$buff =~ s/\s*$//;
				if( $head =~ /\(/ || $head !~ /{/ || $head =~/}/ ) {
					$head =~ s/^\s*//;
					$head =~ s/\/\*.*?\*\///gs;
					formatNote( $head, $buff );
					$head = undef;
					$buff = undef;
				} else {
					$struct = $sdepth = 1;
				}
			} elsif( $line =~ /\/\*\*\*/ ) {
				$verbatim = 1;
			} elsif( $line =~ /\/\*\*/ ) {
				$active = 1;
			}
		}
	}
	close FILE;
}

my $line;
while( defined( $line = <IN> ) ) {
	chomp $line;
	my $prefixes;
	if( my( $fname, $prefixes ) = ( $line =~ /^!!\s*(\S+)(.*)/ ) ) {
		$fname = "$basedir/$fname" if( ( $fname !~ /^\// ) && defined $basedir );
		process( $fname, [ ( map( {
			my( $result ) = /^\s*(.*\S)\s*$/;
			$result;
		} ( split /,/, $prefixes ) ) ) ] );
		push @deps, $fname;
	} else {
		print OUT "$line\n";
	}
}

if( defined $depname ) {
	open DEP, ">>$depname" or die "Could not write dep file $depname ($!)\n";
	print DEP "$outname:";
	print DEP " $_" foreach( @deps );
	print DEP "\n";
	if( $hasdump ) {
		print DEP "$defdump:";
		print DEP " $_" foreach( @deps );
		print DEP "\n";
	}
	close DEP;
}

close IN;
close OUT;
close DUMP;