html2db.pl

web51的开发库

#! /usr/bin/perl
#
#  Convert html (text) source to a51 asm (db) source
#

$verinfo = "1.01";

sub usage {
  print STDERR "usage: $ego [-cpurom] [-cpueeprom ] [-eeprom]\n";
  print STDERR "         [-html] [-plain] [-jpeg] [-gif] [-png] [-app] [-noheader]\n";
  print STDERR "         [-index VALUE] [-dir DIR_SEGMENT] [-body BODY_SEGMENT]\n";
  print STDERR "         [-offset VALUE] [-noconverthtml] [-appendlf] [-keeplf]\n";
  print STDERR "         [-binutils] [-version] [-help] file\n";
  exit 1;
}

sub version {
  print STDERR "$ego version $verinfo\n";
  exit 1;
}

sub printDB {
  local ($file, $co) = @_;
  $co =~ s/\\'/'/;
  $co =~ s/\r//;
  if (length $co gt 0) {
    if (length $co eq 1) {
      if ($co eq "'") {
	if ($binutils) { print $file "\t.byte\t\''''\n"; }
        else { print $file "\tDB\t\''''\n"; }
      } else{
	if ($binutils) { print $file "\t.byte\t\'$co\'\n"; }
        else { print $file "\tDB\t\'$co\'\n"; }
      }
    } else {
      if ($binutils) {
	$co =~ s/"/\\\"/g;
        print $file "\t.ascii\t\"$co\"\n";
      } else {
        $co =~ s/"/",22h,"/g;
        print $file "\tDB\t\"$co\"\n";
      }
    }
  }
  if ($appendlf) {
    if ($binutils) { print $file "\t.byte\t0x0A\n"; }
    else { print $file "\tDB\t0Ah\n"; }
  }
}

sub process_define {
  local ($file, $co) = @_;
  if ($binutils eq 0) {
    if ($co =~ /^#:C[(]([_a-zA-Z0-9]+)[)]/) {
#external code
      $co = $1;
      print $file "extrn\tCODE($co)\n";
    } elsif ($co =~ /^#:D[(]([_a-zA-Z0-9]+)[)]/) {
#external data
      $co = $1;
      print $file "extrn\tDATA($co)\n";
    } elsif ($co =~ /^#:X[(]([_a-zA-Z0-9]+)[)]/) {
#external data
      $co = $1;
      print $file "extrn\tXDATA($co)\n";
    } elsif ($co =~ /^#:Y[(]([_a-zA-Z0-9]+)[)]/) {
#external data
      $co = $1;
      print $file "extrn\tIDATA($co)\n";
    } elsif ($co =~ /^#:N[(]([_a-zA-Z0-9]+)[)]/) {
#external number
      $co = $1;
      print $file "extrn\tNUMBER($co)\n";
    } elsif ($co =~ /^#:H[(]([_a-zA-Z0-9]+)[)]/) {
#header - directory
      $directory = $1;
    } elsif ($co =~ /^#:I[(]([_a-zA-Z0-9]+)[)]/) {
#index - directory
      $indexdirectory = $1;
    } elsif ($co =~ /^#:B[(]([_a-zA-Z0-9]+)[)]/) {
#body
      $body = $1;
    }
  } else {
    if ($co =~ /^#:H[(]([_a-zA-Z0-9]+)[)]/) {
#header - directory
      $directory = $1;
    } elsif ($co =~ /^#:I[(]([_a-zA-Z0-9]+)[)]/) {
#index - directory
      $indexdirectory = $1;
    } elsif ($co =~ /^#:B[(]([_a-zA-Z0-9]+)[)]/) {
#body
      $body = $1;
    }
  }
}


sub printBYTE {
  local ($file, $co) = @_;
#print STDERR "$co ";
#    $co =~ !s/^('\w')$/low($1)/;
    if ($binutils) {
#     $co = "0xFF & $co";
#     $co =~ s/^0xFF & ('\w')$/$1/;
    } else {
      $co = "low($co)";
      $co =~ s/^low\(('\w')\)$/$1/;
    }
    print $file "$co";
#print STDERR "$co\n";
}

sub process_cgi_call {
  local ($file, $str) = @_;
#CGI call 'mac.cgi'
  while ($str =~ /^((\\')|[^']*)'([_a-z0-9]+)\.cgi'(.*)$/i) {
    $substr1 = $1;	#([^']*)
    $substr2 = $3;	#([_a-z0-9]+)
    $str     = $4;	#(.*)
    &printDB ($file, $substr1);
    if ($binutils) { print $file "\t.ascii\t\"'$substr2.cgi'\"\n"; }
    else { print $file "\tDB\t\"'$substr2.cgi'\"\n"; }
  }
#CGI call 'mac.cgi?X'
  while ($str =~ /^((\\')|[^']*)'([_a-z0-9]+)\.cgi[?]('\w'|[_a-z0-9]+)'(.*)$/i) {
    $substr1 = $1;	#([^']*)
    $substr2 = $3;	#([_a-z0-9]+\.cgi)
    $substr3 = $4;	#([_a-z0-9]+)
    $str     = $5;	#(.*)
    &printDB ($file, $substr1);
    if ($binutils) { print $file "\t.ascii\t\"'$substr2.cgi?\"\n\t.byte\t"; }
    else { print $file "\tDB\t\"'$substr2.cgi?\","; }
    &printBYTE ($file ,$substr3);
    print $file "\n";
  }
#CGI call '?mac.cgi'text1{text2}
  while ($str =~ /^((\\')|[^']*)'[?]([_a-z0-9]+)\.cgi'(.*)$/i) {
    $substr1 = $1;	#([^?]*)
    $substr2 = $3;	#([_a-z0-9]+)
    $str     = $4;	#(.*)
    &printDB ($file, $substr1);
    if ($binutils) { print $file "\t.ascii\t\"'?$substr2.cgi'\"\n"; }
    else { print $file "\tDB\t\"'?$substr2.cgi'\"\n"; }
  }
#CGI call '?mac.cgi?X'text1{text2}
  while ($str =~ /^((\\')|[^']*)'[?]([_a-z0-9]+)\.cgi[?]('\w'|[_a-z0-9]+)'(.*)$/i) {
    $substr1 = $1;	#([^?]*)
    $substr2 = $3;	#([_a-z0-9]+\.cgi)
    $substr3 = $4;	#([_a-z0-9]+)
    $str     = $5;	#(.*)
    &printDB ($file, $substr1);
    if ($binutils) { print $file "\t.ascii\t\"'?$substr2.cgi?\"\n\t.byte\t"; }
    else { print $file "\tDB\t\"'?$substr2.cgi?\","; }
    &printBYTE ($file ,$substr3);
    print $file "\n";
  }
  return ($str);
}

$ego = $0;
$ego =~ s/.*[\/\\]//;

$rommode = 0;
$binmode = 99;
$convert = 1;
$appendlf = 0;
$keeplf = 0;
$binutils = 0;
$directory = "";
$body = "";
$indexdirectory = "";
$offset = "";
while ($ARGV[0] =~ /^-/ && $ARGV[0] ne '-') {
    $opt = shift;
    if ($opt eq '-cpurom') {
        $rommode = 1;
	if ($binutils eq 0) {
          if ($directory eq "") { $directory = "CPU_DIR"; }
          if ($body eq "") { $body = "CPU_FILES"; }
        } else {
          if ($directory eq "") { $directory = "cpu_dir"; }
          if ($body eq "") { $body = "cpu_files"; }
        }
    } elsif ($opt eq '-cpueeprom') {
        $rommode = 2;
	if ($binutils eq 0) {
          if ($directory eq "") { $directory = "EE_DIR"; }
          if ($body eq "") { $body = "EE_FILES"; }
	  if ($offset eq "") { $offset = "0E000h"; } #-2000h
        } else {
          if ($directory eq "") { $directory = "ee_dir"; }
          if ($body eq "") { $body = "ee_files"; }
        }
    } elsif ($opt eq '-eeprom') {
        $rommode = 3;
	if ($binutils eq 0) {
          if ($directory eq "") { $directory = "I2C_DIR"; }
          if ($body eq "") { $body = "I2C_FILES"; }
	  if ($offset eq "") { $offset = "8000h"; }
        } else {
          if ($directory eq "") { $directory = "i2c_dir"; }
          if ($body eq "") { $body = "i2c_files"; }
	  if ($offset eq "") { $offset = "0x8000"; }
        }
    } elsif ($opt eq '-html') {
        $binmode = 1;		#text/html
    } elsif ($opt eq '-plain') {
        $binmode = 2;		#text/plain
    } elsif ($opt eq '-jpeg') {
        $binmode = 3;		#image/jpeg
    } elsif ($opt eq '-gif') {
        $binmode = 4;		#image/gif
    } elsif ($opt eq '-png') {
        $binmode = 5;		#image/png
    } elsif ($opt eq '-app') {
        $binmode = 6;		#application/octet-stream
    } elsif ($opt eq '-noheader') {
        $binmode = 0;		#without header
    } elsif ($opt eq '-noconverthtml') {
        $convert = 0;		#no convert html
    } elsif ($opt eq '-appendlf') {
        $appendlf = 1;		#don't purge NL from html code - mode 1
    } elsif ($opt eq '-keeplf') {
        $keeplf = 1;		#don't purge NL from html code - mode 2
    } elsif ($opt eq '-dir') {
        $opt = shift;
        $directory = $opt;
    } elsif ($opt eq '-body') {
        $opt = shift;
        $body = $opt;
    } elsif ($opt eq '-index') {
        $opt = shift;
        $indexdirectory = $opt;
    } elsif ($opt eq '-offset') {
        $opt = shift;
        $offset = $opt;
    } elsif ($opt eq '-version') {
        &version;
    } elsif ($opt eq '-help') {
        &usage;
    } elsif ($opt eq '-binutils') {
        $binutils = 1;
    } else {
        &usage;
    }
}

$l0 = length $ARGV[0];
$name = $ARGV[0];

if ($l0 eq 0) { &usage; }

@names = split (/\./, $name);
if (scalar @names gt 1) {
  $filetype = pop @names;
  $filetype =~ tr/a-z/A-Z/;
  $oname = join (".", @names);
} else {
  $filetype = "999";
  $oname = $names[0];
}

if ($binmode eq 99) {
    if ($filetype eq 'HTML') {
        $binmode = 1;
    } elsif ($filetype eq 'HTM') {
        $binmode = 1;
    } elsif ($filetype eq 'JS') {
        $binmode = 1;
    } elsif ($filetype eq 'TXT') {
        $binmode = 2;
    } elsif ($filetype eq 'JPG') {
        $binmode = 3;
    } elsif ($filetype eq 'JPEG') {
        $binmode = 3;
    } elsif ($filetype eq 'GIF') {
        $binmode = 4;
    } elsif ($filetype eq 'PNG') {
        $binmode = 5;
    } elsif ($filetype eq 'CLASS') {
        $binmode = 6;
    } elsif ($filetype eq 'CLA') {
        $binmode = 6;
    }
}

if ($binutils) { $oname = $oname.".asm"; }
else { $oname = $oname.".a51"; }

$NAME = "<".$name;
unless (open NAME) {
    print STDERR "Missing input file $name\n";
    return;
}

$ONAME = ">".$oname;
unless (open ONAME) {
    print STDERR "Cannot open output file $oname\n";
    return;
}

$prolog = 1;

if ($binmode eq 99) {
  print STDERR "WARNING !!! missing -html/plain/jpeg/gif/png/app, using html mode\n";
  $binmode =1;
}
if ((($binmode eq 2) || ($binmode eq 1)) && ($convert ne 0)) {
  while (<NAME>) {
    $str = $_;
    $str =~ s/[\n]/\r/g;	# map mew-line to CR
    if ($str =~ /^#:[BCDHINXY][(][_a-zA-Z0-9]+[)]/) {
      &process_define (ONAME, $str);
      $str = "";
    } else {
      if ($prolog eq 1) {
        if ($binutils) {
          print ONAME "\n\t.section\t$directory, #alloc\n\n";
          print ONAME "\t.word\t'", join ("'+'", split (//, $name)), "'\n";
        } else {
          print ONAME "\n$directory\tSEGMENT\tCODE\n\tRSEG\t$directory\n\n";
          print ONAME "\tDW\t'", join ("'+'", split (//, $name)), "'\n";
        }
        if ($binmode eq 1) {
	  if ($binutils) { print ONAME "\t.byte\t1\t;text/html\n"; }
          else { print ONAME "\tDB\t1\t;text/html\n"; }
        } else {
	  if ($binutils) { print ONAME "\t.byte\t2\t;text/plain\n"; }
          else { print ONAME "\tDB\t2\t;text/plain\n"; }
        }
        if($offset eq "") {
	  if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0\n\t.word\tend0\n\n"; }
          else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0\n\tDW\tend0\n\n"; }
	} else {
	  if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0 + $offset\n\t.word\tend0 + $offset\n\n"; }
          else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0 + $offset\n\tDW\tend0 + $offset\n\n"; }
	}
        if ($indexdirectory ne "") {
	  if ($binutils) { print ONAME "\t.word\t$indexdirectory\n"; }
          else { print ONAME "\tDW\t$indexdirectory\n"; }
          if ($binmode eq 1) {
	    if ($binutils) { print ONAME "\t.byte\t1\t;text/html\n"; }
            else { print ONAME "\tDB\t1\t;text/html\n"; }
          } else {
	    if ($binutils) { print ONAME "\t.byte\t2\t;text/plain\n"; }
            else { print ONAME "\tDB\t2\t;text/plain\n"; }
          }
          if($offset eq "") {
	    if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0\n\t.word\tend0\n\n"; }
            else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0\n\tDW\tend0\n\n"; }
          } else {
	    if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0 + $offset\n\t.word\tend0 + $offset\n\n"; }
            else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0 + $offset\n\tDW\tend0 + $offset\n\n"; }
          }
        }
	if ($binutils) { print ONAME "\t.section\t$body, #alloc\n\n"; }
        else { print ONAME "$body\tSEGMENT\tCODE\n\tRSEG\t$body\n\n"; }
        print ONAME "begin0:\n";
        $prolog = 0;
      }
      while (1) {
        ($retstr) = &process_cgi_call (ONAME, $str);
        last if ($retstr eq $str);
        $str = $retstr;
      }
    }
    &printDB (ONAME, $str);
    if ($keeplf) {
      if ($binutils) { print ONAME "\t.byte\t0x0A\n"; }
      else { print ONAME "\tDB\t0Ah\n"; }
      }
    }
  if ($binutils) { print ONAME "end0:\n\t;; end\n"; }
  else { print ONAME "end0:\n\tend\n"; }
} else {
  if ($binutils) {
    print ONAME "\n\t.section\t$directory, #alloc\n\n";
    print ONAME "\t.word\t'", join ("'+'", split (//, $name)), "'\n";
  } else {
    print ONAME "\n$directory\tSEGMENT\tCODE\n\tRSEG\t$directory\n\n";
    print ONAME "\tDW\t'", join ("'+'", split (//, $name)), "'\n";
  }
  if ($binutils) {
    if ($binmode eq 3) {
      print ONAME "\t.byte\t3\t;image/jpeg\n";
    } elsif ($binmode eq 4) {
      print ONAME "\t.byte\t4\t;image/gif\n";
    } elsif ($binmode eq 5) {
      print ONAME "\t.byte\t5\t;image/png\n";
    } elsif ($binmode eq 6) {
      print ONAME "\t.byte\t6\t;application/octet-stream\n";
    } elsif ($binmode eq 0) {
      print ONAME "\t.byte\t0\t;without header\n";
    }
  } else {
    if ($binmode eq 3) {
      print ONAME "\tDB\t3\t;image/jpeg\n";
    } elsif ($binmode eq 4) {
      print ONAME "\tDB\t4\t;image/gif\n";
    } elsif ($binmode eq 5) {
      print ONAME "\tDB\t5\t;image/png\n";
    } elsif ($binmode eq 6) {
      print ONAME "\tDB\t6\t;application/octet-stream\n";
    } elsif ($binmode eq 0) {
      print ONAME "\tDB\t0\t;without header\n";
    }
  }
  if($offset eq "") {
    if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0\n\t.word\tend0\n\n"; }
    else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0\n\tDW\tend0\n\n"; }
  } else {
    if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0 + $offset\n\t.word\tend0 + $offset\n\n"; }
    else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0 + $offset\n\tDW\tend0 + $offset\n\n"; }
  }
  if ($indexdirectory ne "") {
    if ($binutils) { print ONAME "\t.word\t$indexdirectory\n"; }
    else { print ONAME "\tDW\t$indexdirectory\n"; }
    if ($binutils) {
      if ($binmode eq 2) {
        print ONAME "\t.byte\t2\t;text/plain\n";
      } elsif ($binmode eq 3) {
        print ONAME "\t.byte\t3\t;image/jpeg\n";
      } elsif ($binmode eq 4) {
        print ONAME "\t.byte\t4\t;image/gif\n";
      } elsif ($binmode eq 5) {
        print ONAME "\t.byte\t5\t;image/png\n";
      } elsif ($binmode eq 6) {
        print ONAME "\t.byte\t6\t;application/octet-stream\n";
      }
    } else {
      if ($binmode eq 2) {
        print ONAME "\tDB\t2\t;text/plain\n";
      } elsif ($binmode eq 3) {
        print ONAME "\tDB\t3\t;image/jpeg\n";
      } elsif ($binmode eq 4) {
        print ONAME "\tDB\t4\t;image/gif\n";
      } elsif ($binmode eq 5) {
        print ONAME "\tDB\t5\t;image/png\n";
      } elsif ($binmode eq 6) {
        print ONAME "\tDB\t6\t;application/octet-stream\n";
      }
    }
    if($offset eq "") {
      if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0\n\t.word\tend0\n\n"; }
      else { print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0\n\tDW\tend0\n\n"; }
    } else {
      if ($binutils) { print ONAME "\t.byte\t0\t;reserved\n\t.word\tbegin0 + $offset\n\t.word\tend0 + $offset\n\n"; }
      else {print ONAME "\tDB\t0\t;reserved\n\tDW\tbegin0 + $offset\n\tDW\tend0 + $offset\n\n"; }
    }
  }
  if ($binutils) { print ONAME "\t.section\t$body, #alloc\n\n"; }
  else { print ONAME "$body\tSEGMENT\tCODE\n\tRSEG\t$body\n\n"; }
  print ONAME "begin0:\n";
  binmode(NAME) ;
  while (1) {
    sysread(NAME,$buf,8);
    if (length($buf) != 0) {
      if ($binutils) { print ONAME "\t.byte\t"; }
      else { print ONAME "\tDB\t"; }
      for ($i=0 ; $i<(length($buf) - 1) ; $i++)  {
	if ($binutils) { printf ONAME "0x%02X, ",unpack("C",substr($buf,$i,1)) ; }
        else { printf ONAME "0%02Xh, ",unpack("C",substr($buf,$i,1)) ; }
      }
      if ($binutils) { printf ONAME "0x%02X\n",unpack("C",substr($buf,$i,1)) ; }
      else { printf ONAME "0%02Xh\n",unpack("C",substr($buf,$i,1)) ; }
    } else { last; }
  }
  if ($binutils) { print ONAME "end0:\n\t;; end\n"; }
  else { print ONAME "end0:\n\tend\n"; }
}
close NAME;
close ONAME;