# ansi2html Converts old BBS ANSI screens into HTML.
# by Leonard Richardson (http://www.crummy.com/)
# Public domain.
#
# It converts ANSI color codes into HTML styles.
# It converts CP437 characters (aka "IBM Extended ASCII") into HTML entities.
# It puts the whole thing in a preformatted HTML presentation.
#
# Thanks to:
# http://www.fileformat.info/info/unicode/
# http://search.cpan.org/~autrijus/HTML-FromANSI-1.01/
# http://en.wikipedia.org/wiki/Code_page_437
# http://en.wikipedia.org/wiki/ANSI_escape_code
# http://home.claranet.de/xyzzy/ibm850.htm
# http://www.crummy.com/source/software/download/bbs2ansi.pl
# First a dict mapping CP437 characters to numeric HTML entities and Unicode
# names.
# Characters in standard ASCII are not in this dict.
CP437_DICT = {
'\x01' : ("☺", "WHITE SMILING FACE"),
'\x02' : ("☻", "BLACK SMILING FACE"),
'\x03' : ("♥", "BLACK HEART SUIT"),
'\x04' : ("♦", "BLACK DIAMOND SUIT"),
'\x05' : ("♣", "BLACK CLUB SUIT"),
'\x06' : ("♠", "BLACK SPADE SUIT"),
'\x07' : ("•", "BULLET"),
'\x08' : ("◘", "INVERSE BULLET"),
'\x09' : ("○", "WHITE CIRCLE"),
'\x0a' : ("◙", "INVERSE WHITE CIRCLE"),
'\x0b' : ("♂", "MALE SIGN"),
'\x0c' : ("♀", "FEMALE SIGN"),
'\x0d' : ("♪", "EIGHTH NOTE"),
'\x0e' : ("♫", "BEAMED EIGHTH NOTES"),
'\x0f' : ("☼", "WHITE SUN WITH RAYS"),
'\x10' : ("▸", "BLACK RIGHT-POINTING SMALL TRIANGLE"),
'\x11' : ("◂", "BLACK LEFT-POINTING SMALL TRIANGLE"),
'\x12' : ("↕", "UP DOWN ARROW"),
'\x13' : ("‼", "DOUBLE EXCLAMATION MARK"),
'\x14' : ("¶", "PILCROW SIGN"),
'\x15' : ("§", "SECTION SIGN"),
'\x16' : ("▬", "BLACK RECTANGLE"),
'\x17' : ("↨", "UP DOWN ARROW WITH BASE"),
'\x18' : ("↑", "UPWARDS ARROW"),
'\x19' : ("↓", "DOWNWARDS ARROW"),
'\x1a' : ("→", "RIGHTWARDS ARROW"),
'\x1b' : ("←", "LEFTWARDS ARROW"),
'\x1c' : ("∟", "RIGHT ANGLE"),
'\x1d' : ("↔", "LEFT RIGHT ARROW"),
'\x1e' : ("▴", "BLACK UP-POINTING SMALL TRIANGLE"),
'\x1f' : ("▾", "BLACK DOWN-POINTING SMALL TRIANGLE"),
'\x21' : ("!", "EXCLAMATION MARK"),
'\x22' : (""", "QUOTATION MARK"),
'\x23' : ("#", "NUMBER SIGN"),
'\x24' : ("$", "DOLLAR SIGN"),
'\x25' : ("%", "PERCENT SIGN"),
'\x26' : ("&", "AMPERSAND"),
'\x27' : ("'", "APOSTROPHE"),
'\x28' : ("(", "LEFT PARENTHESIS"),
'\x29' : (")", "RIGHT PARENTHESIS"),
'\x2a' : ("*", "ASTERISK"),
'\x2b' : ("+", "PLUS SIGN"),
'\x2c' : (",", "COMMA"),
'\x2d' : ("-", "HYPHEN-MINUS"),
'\x2e' : (".", "FULL STOP"),
'\x2f' : ("/", "SOLIDUS"),
'\x7f' : ("⌂", "HOUSE"),
'\x80' : ("Ç", "LATIN CAPITAL LETTER C WITH CEDILLA"),
'\x81' : ("ü", "LATIN SMALL LETTER U WITH DIAERESIS"),
'\x82' : ("é", "LATIN SMALL LETTER E WITH ACUTE"),
'\x83' : ("â", "LATIN SMALL LETTER A WITH CIRCUMFLEX"),
'\x84' : ("ä", "LATIN SMALL LETTER A WITH DIAERESIS"),
'\x85' : ("à", "LATIN SMALL LETTER A WITH GRAVE"),
'\x86' : ("å", "LATIN SMALL LETTER A WITH RING ABOVE"),
'\x87' : ("ç", "LATIN SMALL LETTER C WITH CEDILLA"),
'\x88' : ("ê", "LATIN SMALL LETTER E WITH CIRCUMFLEX"),
'\x89' : ("ë", "LATIN SMALL LETTER E WITH DIAERESIS"),
'\x8a' : ("è", "LATIN SMALL LETTER E WITH GRAVE"),
'\x8b' : ("ï", "LATIN SMALL LETTER I WITH DIAERESIS"),
'\x8c' : ("î", "LATIN SMALL LETTER I WITH CIRCUMFLEX"),
'\x8d' : ("ì", "LATIN SMALL LETTER I WITH GRAVE"),
'\x8e' : ("Ä", "LATIN CAPITAL LETTER A WITH DIAERESIS"),
'\x8f' : ("Å", "LATIN CAPITAL LETTER A WITH RING ABOVE"),
'\x90' : ("É", "LATIN CAPITAL LETTER E WITH ACUTE"),
'\x91' : ("æ", "LATIN SMALL LETTER AE"),
'\x92' : ("Æ", "LATIN CAPITAL LETTER AE"),
'\x93' : ("ô", "LATIN SMALL LETTER O WITH CIRCUMFLEX"),
'\x94' : ("ö", "LATIN SMALL LETTER O WITH DIAERESIS"),
'\x95' : ("ò", "LATIN SMALL LETTER O WITH GRAVE"),
'\x96' : ("û", "LATIN SMALL LETTER U WITH CIRCUMFLEX"),
'\x97' : ("ù", "LATIN SMALL LETTER U WITH GRAVE"),
'\x98' : ("ÿ", "LATIN SMALL LETTER Y WITH DIAERESIS"),
'\x99' : ("Ö", "LATIN CAPITAL LETTER O WITH DIAERESIS"),
'\x9a' : ("Ü", "LATIN CAPITAL LETTER U WITH DIAERESIS"),
'\x9b' : ("¢", "CENT SIGN"),
'\x9c' : ("£", "POUND SIGN"),
'\x9d' : ("¥", "YEN SIGN"),
'\x9e' : ("₧", "PESETA SIGN"),
'\x9f' : ("ƒ", "LATIN SMALL LETTER F WITH HOOK"),
'\xa0' : ("á", "LATIN SMALL LETTER A WITH ACUTE"),
'\xa1' : ("í", "LATIN SMALL LETTER I WITH ACUTE"),
'\xa2' : ("ó", "LATIN SMALL LETTER O WITH ACUTE"),
'\xa3' : ("ú", "LATIN SMALL LETTER U WITH ACUTE"),
'\xa4' : ("ñ", "LATIN SMALL LETTER N WITH TILDE"),
'\xa5' : ("Ñ", "LATIN CAPITAL LETTER N WITH TILDE"),
'\xa6' : ("ª", "FEMININE ORDINAL INDICATOR"),
'\xa7' : ("º", "MASCULINE ORDINAL INDICATOR"),
'\xa8' : ("¿", "INVERTED QUESTION MARK"),
'\xa9' : ("⌐", "REVERSED NOT SIGN"),
'\xaa' : ("¬", "NOT SIGN"),
'\xab' : ("½", "VULGAR FRACTION ONE HALF"),
'\xac' : ("¼", "VULGAR FRACTION ONE QUARTER"),
'\xad' : ("¡", "INVERTED EXCLAMATION MARK"),
'\xae' : ("«", "LEFT-POINTING DOUBLE ANGLE QUOTATION MARK"),
'\xaf' : ("»", "RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK"),
'\xb0' : ("░", "LIGHT SHADE"),
'\xb1' : ("▒", "MEDIUM SHADE"),
'\xb2' : ("▓", "DARK SHADE"),
'\xb3' : ("│", "BOX DRAWINGS LIGHT VERTICAL"),
'\xb4' : ("┤", "BOX DRAWINGS LIGHT VERTICAL AND LEFT"),
'\xb5' : ("╡", "BOX DRAWINGS VERTICAL SINGLE AND LEFT DOUBLE"),
'\xb6' : ("╢", "BOX DRAWINGS VERTICAL DOUBLE AND LEFT SINGLE"),
'\xb7' : ("╖", "BOX DRAWINGS DOWN DOUBLE AND LEFT SINGLE"),
'\xb8' : ("╕", "BOX DRAWINGS DOWN SINGLE AND LEFT DOUBLE"),
'\xb9' : ("╣", "BOX DRAWINGS DOUBLE VERTICAL AND LEFT"),
'\xba' : ("║", "BOX DRAWINGS DOUBLE VERTICAL"),
'\xbb' : ("╗", "BOX DRAWINGS DOUBLE DOWN AND LEFT"),
'\xbc' : ("╝", "BOX DRAWINGS DOUBLE UP AND LEFT"),
'\xbd' : ("╜", "BOX DRAWINGS UP DOUBLE AND LEFT SINGLE"),
'\xbe' : ("╛", "BOX DRAWINGS UP SINGLE AND LEFT DOUBLE"),
'\xbf' : ("┐", "BOX DRAWINGS LIGHT DOWN AND LEFT"),
'\xc0' : ("└", "BOX DRAWINGS LIGHT UP AND RIGHT"),
'\xc1' : ("┴", "BOX DRAWINGS LIGHT UP AND HORIZONTAL"),
'\xc2' : ("┬", "BOX DRAWINGS LIGHT DOWN AND HORIZONTAL"),
'\xc3' : ("├", "BOX DRAWINGS LIGHT VERTICAL AND RIGHT"),
'\xc4' : ("─", "BOX DRAWINGS LIGHT HORIZONTAL"),
'\xc5' : ("┼", "BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL"),
'\xc6' : ("╞", "BOX DRAWINGS VERTICAL SINGLE AND RIGHT DOUBLE"),
'\xc7' : ("╟", "BOX DRAWINGS VERTICAL DOUBLE AND RIGHT SINGLE"),
'\xc8' : ("╚", "BOX DRAWINGS DOUBLE UP AND RIGHT"),
'\xc9' : ("╔", "BOX DRAWINGS DOUBLE DOWN AND RIGHT"),
'\xca' : ("╩", "BOX DRAWINGS DOUBLE UP AND HORIZONTAL"),
'\xcb' : ("╦", "BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL"),
'\xcc' : ("╠", "BOX DRAWINGS DOUBLE VERTICAL AND RIGHT"),
'\xcd' : ("═", "BOX DRAWINGS DOUBLE HORIZONTAL"),
'\xce' : ("╬", "BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL"),
'\xcf' : ("╧", "BOX DRAWINGS UP SINGLE AND HORIZONTAL DOUBLE"),
'\xd0' : ("╨", "BOX DRAWINGS UP DOUBLE AND HORIZONTAL SINGLE"),
'\xd1' : ("╤", "BOX DRAWINGS DOWN SINGLE AND HORIZONTAL DOUBLE"),
'\xd2' : ("╥", "BOX DRAWINGS DOWN DOUBLE AND HORIZONTAL SINGLE"),
'\xd3' : ("╙", "BOX DRAWINGS UP DOUBLE AND RIGHT SINGLE"),
'\xd4' : ("╘", "BOX DRAWINGS UP SINGLE AND RIGHT DOUBLE"),
'\xd5' : ("╒", "BOX DRAWINGS DOWN SINGLE AND RIGHT DOUBLE"),
'\xd6' : ("╓", "BOX DRAWINGS DOWN DOUBLE AND RIGHT SINGLE"),
'\xd7' : ("╫", "BOX DRAWINGS VERTICAL DOUBLE AND HORIZONTAL SINGLE"),
'\xd8' : ("╪", "BOX DRAWINGS VERTICAL SINGLE AND HORIZONTAL DOUBLE"),
'\xd9' : ("┘", "BOX DRAWINGS LIGHT UP AND LEFT"),
'\xda' : ("┌", "BOX DRAWINGS LIGHT DOWN AND RIGHT"),
'\xdb' : ("█", "FULL BLOCK"),
'\xdc' : ("▄", "LOWER HALF BLOCK"),
'\xdd' : ("▌", "LEFT HALF BLOCK"),
'\xde' : ("▐", "RIGHT HALF BLOCK"),
'\xdf' : ("▀", "UPPER HALF BLOCK"),
'\xe0' : ("α", "GREEK SMALL LETTER ALPHA"),
'\xe1' : ("β", "GREEK SMALL LETTER BETA"),
'\xe2' : ("Γ", "GREEK CAPITAL LETTER GAMMA"),
'\xe3' : ("π", "GREEK SMALL LETTER PI"),
'\xe4' : ("Σ", "GREEK CAPITAL LETTER SIGMA"),
'\xe5' : ("σ", "GREEK SMALL LETTER SIGMA"),
'\xe6' : ("µ", "MICRO SIGN"),
'\xe7' : ("τ", "GREEK SMALL LETTER TAU"),
'\xe8' : ("Φ", "GREEK CAPITAL LETTER PHI"),
'\xe9' : ("Θ", "GREEK CAPITAL LETTER THETA"),
'\xea' : ("Ω", "GREEK CAPITAL LETTER OMEGA"),
'\xeb' : ("δ", "GREEK SMALL LETTER DELTA"),
'\xec' : ("∞", "INFINITY"),
'\xed' : ("∅", "EMPTY SET"),
'\xee' : ("∈", "ELEMENT OF"),
'\xef' : ("∩", "INTERSECTION"),
'\xf0' : ("≡", "IDENTICAL TO"),
'\xf1' : ("±", "PLUS-MINUS SIGN"),
'\xf2' : ("≥", "GREATER-THAN OR EQUAL TO"),
'\xf3' : ("≤", "LESS-THAN OR EQUAL TO"),
'\xf4' : ("⌠", "TOP HALF INTEGRAL"),
'\xf5' : ("⌡", "BOTTOM HALF INTEGRAL"),
'\xf6' : ("÷", "DIVISION SIGN"),
'\xf7' : ("≈", "ALMOST EQUAL TO"),
'\xf8' : ("°", "DEGREE SIGN"),
'\xf9' : ("∙", "BULLET OPERATOR"),
'\xfa' : ("·", "MIDDLE DOT"),
'\xfb' : ("√", "SQUARE ROOT"),
'\xfc' : ("ⁿ", "SUPERSCRIPT LATIN SMALL LETTER N"),
'\xfd' : ("²", "SUPERSCRIPT TWO"),
'\xfe' : ("▪", "SMALL BLACK SQUARE"),
}
# FYI, common alternate choices for Unicode representations of CP437
# characters:
# 0x0f : ("✶", "SIX POINTED BLACK STAR"),
# 0x10 : ("►", "BLACK RIGHT-POINTING POINTER"),
# 0x11 : ("◄", "BLACK LEFT-POINTING POINTER"),
# 0x1e : ("▲", "BLACK UP-POINTING TRIANGLE"),
# 0x1f : ("▼", "BLACK DOWN-POINTING TRIANGLE"),
# 0xfe : ("■", "BLACK SQUARE"),
import copy
import re
import sys
class ANSIDerivedHTML:
ANSI_CODES = re.compile('\x1b\[(([0-9]+;)*)([0-9]+)?(.)')
CP437_CHARS = re.compile('([\x01-\x09\x0b-\x0c\x0e-\x1f\x7f-\xfe])')
# That regex includes a hack that avoids looking at newlines.
DEFAULT_STYLE = { 'foreground' : 7,
'background' : 0,
'bold' : True,
'blink' : False }
HTML_COLORS = ['black', 'darkred', 'darkgreen', '#8b8b00',
'darkblue', 'darkmagenta', 'darkcyan', 'gray']
BOLD_HTML_COLORS = ['dimgray', 'red', 'green', 'yellow',
'blue', 'magenta', 'cyan', 'white']
def __init__(self, text):
self.ignored_commands = []
self.style = copy.copy(self.DEFAULT_STYLE)
self.unbalanced_span = False
self.unbalanced_blink = False
self.converted_text = self.to_html(text)
def __str__(self):
"""Return the converted text."""
return self.converted_text
def to_html(self, text):
"""Replace CP437 characters and process ANSI codes."""
# Get rid of anything between a cursor save and a cursor restore.
text = re.compile("\x1b\[s.*?\x1b\[u", re.DOTALL).sub("", text)
return self.htmlify_cp437(self.htmlify_ansi_color(text))
def preformatted(self):
"""Return the converted text, preformatted for your convenience."""
return (''
% (self.css_style(self.DEFAULT_STYLE), self))
#return '%s
' % self
def htmlify_cp437(self, text):
"""Replace special CP437 characters with HTML entity equivalents."""
return self.CP437_CHARS.sub(self.process_cp437_char, text)
def htmlify_ansi_color(self, text):
"""Replace ANSI color codes with styled tags."""
text = self.ANSI_CODES.sub(self.process_ansi_code, text)
if self.unbalanced_span:
text = text + ''
self.unbalanced_span = False
if self.unbalanced_blink:
text = text + ''
self.unbalanced_blink = False
return text
def process_cp437_char(self, matchobj):
char = matchobj.groups()[0]
return CP437_DICT[char][0]
def process_ansi_code(self, matchobj):
result_parts = []
groups = matchobj.groups()
command = groups[-1]
new_style = copy.copy(self.style)
new_blink = False
args = groups[0].split(';')[:-1]
if groups[-2] is not None:
args.append(groups[-2])
args = [int(arg) for arg in args]
if command == 'C': # Move cursor forward
# Simulate by adding spaces.
distance = groups[-2]
return ' ' * int(distance)
elif command == 'H': # Move cursor to arbitrary place
# In most non-animated ANSIs this is used as a carriage return/
# line feed. We'll change it to a newline.
if args[-1] == 1:
return '\n'
else:
sys.stderr.write("Couldn't handle cursor move to %s,%s\n"
% tuple(args))
elif command == 'm': # Change display properties
for command in args:
command = int(command)
if command == 0:
new_style = copy.copy(self.DEFAULT_STYLE)
elif command == 1:
new_style['bold'] = True
elif command == 2:
new_style['bold'] = False
elif command == 5 or command == 6:
new_style['blink'] = True
elif command == 25:
new_style['blink'] = False
elif command >= 30 and command < 39:
new_style['foreground'] = min(command-30, 7)
elif command == 39:
new_style['foreground'] = self.DEFAULT_STYLE['foreground']
elif command >= 40 and command < 49:
new_style['background'] = min(command-40, 7)
elif command == 49:
new_style['background'] = self.DEFAULT_STYLE['background']
else:
sys.stderr.write("Not sure how to handle SGR param "
'"%s"\n' % command)
return ''
else:
sys.stderr.write("Not sure how to handle command %s\n"%
matchobj.group()[1:])
self.ignored_commands.append(matchobj.group())
return ''
# If any attribute changed, close the tag and start
if (new_style != self.style):
if self.unbalanced_span:
result_parts.append('')
self.unbalanced_span = False
result_parts.append(''
% self.css_style(new_style))
self.unbalanced_span = True
# Update the style.
self.style = new_style
return ''.join(result_parts)
def html_color(self, ansi_color, is_foreground, style=None):
if style is None:
style = self.style
if style['bold'] and is_foreground:
return self.BOLD_HTML_COLORS[ansi_color]
else:
return self.HTML_COLORS[ansi_color]
def css_style(self, style=None):
"""A CSS style for the current color and blink settings."""
if style is None:
style = self.style
blink_style = ''
if style['blink']:
blink_style = '; text-decoration: blink'
return 'color:%s; background-color:%s%s' % (
self.html_color(style['foreground'], True, style),
self.html_color(style['background'], False, style),
blink_style)
def self_test():
# Character test
char_lines = [[]]
for i in range(1, 255):
# Blah, special case for newline chars
if i in [10, 13]:
i = ord('X')
char_lines[-1].append(chr(i))
if i % 16 == 0 and i != 0:
char_lines.append([])
char_lines = [('\x1b[0m' + ''.join(line)) for line in char_lines]
# Color test
color_lines = [[]]
for foreground in range(0,8):
for background in range(0,8):
color_lines[-1].append('\x1b[1;%s;%sm\x03' % (
foreground+30, background+40))
color_lines[-1].append('\x1b[2m\x03')
#Finish off with a blink demo
color_lines[-1].append('\x1b[5m\x03')
color_lines[-1].append('\x1b[0m')
color_lines.append([])
color_lines = [''.join(line) for line in color_lines]
while len(color_lines) < len(char_lines):
color_lines.append('')
text = '\n'.join(['%s %s' % (ch, co)
for ch, co in zip(char_lines, color_lines)])
return ANSIDerivedHTML(text).preformatted()
for file in sys.argv:
ansi = open(file).read()
html_file = file + '.html'
print "%s => %s" % (file, html_file)
html = ANSIDerivedHTML(ansi)
html_handle = open(html_file, 'w')
html_handle.write(html.preformatted())
html_handle.close()