Initial revision

diff --git a/Lib/htmlentitydefs.py b/Lib/htmlentitydefs.py
new file mode 100644 (file)
index 0000000..36ae8df
--- /dev/null
+++ b/Lib/htmlentitydefs.py
@@ -0,0 +1,101 @@
+# Proposed entity definitions for HTML, taken from
+# http://www.w3.org/hypertext/WWW/MarkUp/html-spec/html-spec_14.html
+
+entitydefs = {
+    'nbsp':    chr(160),       # no-break space
+    'iexcl':   chr(161),       # inverted exclamation mark
+    'cent':    chr(162),       # cent sign
+    'pound':   chr(163),       # pound sterling sign
+    'curren':  chr(164),       # general currency sign
+    'yen':     chr(165),       # yen sign
+    'brvbar':  chr(166),       # broken (vertical) bar
+    'sect':    chr(167),       # section sign
+    'uml':     chr(168),       # umlaut (dieresis)
+    'copy':    chr(169),       # copyright sign
+    'ordf':    chr(170),       # ordinal indicator, feminine
+    'laquo':   chr(171),       # angle quotation mark, left
+    'not':     chr(172),       # not sign
+    'shy':     chr(173),       # soft hyphen
+    'reg':     chr(174),       # registered sign
+    'macr':    chr(175),       # macron
+    'deg':     chr(176),       # degree sign
+    'plusmn':  chr(177),       # plus-or-minus sign
+    'sup2':    chr(178),       # superscript two
+    'sup3':    chr(179),       # superscript three
+    'acute':   chr(180),       # acute accent
+    'micro':   chr(181),       # micro sign
+    'para':    chr(182),       # pilcrow (paragraph sign)
+    'middot':  chr(183),       # middle dot
+    'cedil':   chr(184),       # cedilla
+    'sup1':    chr(185),       # superscript one
+    'ordm':    chr(186),       # ordinal indicator, masculine
+    'raquo':   chr(187),       # angle quotation mark, right
+    'frac14':  chr(188),       # fraction one-quarter
+    'frac12':  chr(189),       # fraction one-half
+    'frac34':  chr(190),       # fraction three-quarters
+    'iquest':  chr(191),       # inverted question mark
+    'Agrave':  chr(192),       # capital A, grave accent
+    'Aacute':  chr(193),       # capital A, acute accent
+    'Acirc':   chr(194),       # capital A, circumflex accent
+    'Atilde':  chr(195),       # capital A, tilde
+    'Auml':    chr(196),       # capital A, dieresis or umlaut mark
+    'Aring':   chr(197),       # capital A, ring
+    'AElig':   chr(198),       # capital AE diphthong (ligature)
+    'Ccedil':  chr(199),       # capital C, cedilla
+    'Egrave':  chr(200),       # capital E, grave accent
+    'Eacute':  chr(201),       # capital E, acute accent
+    'Ecirc':   chr(202),       # capital E, circumflex accent
+    'Euml':    chr(203),       # capital E, dieresis or umlaut mark
+    'Igrave':  chr(204),       # capital I, grave accent
+    'Iacute':  chr(205),       # capital I, acute accent
+    'Icirc':   chr(206),       # capital I, circumflex accent
+    'Iuml':    chr(207),       # capital I, dieresis or umlaut mark
+    'ETH':     chr(208),       # capital Eth, Icelandic
+    'Ntilde':  chr(209),       # capital N, tilde
+    'Ograve':  chr(210),       # capital O, grave accent
+    'Oacute':  chr(211),       # capital O, acute accent
+    'Ocirc':   chr(212),       # capital O, circumflex accent
+    'Otilde':  chr(213),       # capital O, tilde
+    'Ouml':    chr(214),       # capital O, dieresis or umlaut mark
+    'times':   chr(215),       # multiply sign
+    'Oslash':  chr(216),       # capital O, slash
+    'Ugrave':  chr(217),       # capital U, grave accent
+    'Uacute':  chr(218),       # capital U, acute accent
+    'Ucirc':   chr(219),       # capital U, circumflex accent
+    'Uuml':    chr(220),       # capital U, dieresis or umlaut mark
+    'Yacute':  chr(221),       # capital Y, acute accent
+    'THORN':   chr(222),       # capital THORN, Icelandic
+    'szlig':   chr(223),       # small sharp s, German (sz ligature)
+    'agrave':  chr(224),       # small a, grave accent
+    'aacute':  chr(225),       # small a, acute accent
+    'acirc':   chr(226),       # small a, circumflex accent
+    'atilde':  chr(227),       # small a, tilde
+    'auml':    chr(228),       # small a, dieresis or umlaut mark
+    'aring':   chr(229),       # small a, ring
+    'aelig':   chr(230),       # small ae diphthong (ligature)
+    'ccedil':  chr(231),       # small c, cedilla
+    'egrave':  chr(232),       # small e, grave accent
+    'eacute':  chr(233),       # small e, acute accent
+    'ecirc':   chr(234),       # small e, circumflex accent
+    'euml':    chr(235),       # small e, dieresis or umlaut mark
+    'igrave':  chr(236),       # small i, grave accent
+    'iacute':  chr(237),       # small i, acute accent
+    'icirc':   chr(238),       # small i, circumflex accent
+    'iuml':    chr(239),       # small i, dieresis or umlaut mark
+    'eth':     chr(240),       # small eth, Icelandic
+    'ntilde':  chr(241),       # small n, tilde
+    'ograve':  chr(242),       # small o, grave accent
+    'oacute':  chr(243),       # small o, acute accent
+    'ocirc':   chr(244),       # small o, circumflex accent
+    'otilde':  chr(245),       # small o, tilde
+    'ouml':    chr(246),       # small o, dieresis or umlaut mark
+    'divide':  chr(247),       # divide sign
+    'oslash':  chr(248),       # small o, slash
+    'ugrave':  chr(249),       # small u, grave accent
+    'uacute':  chr(250),       # small u, acute accent
+    'ucirc':   chr(251),       # small u, circumflex accent
+    'uuml':    chr(252),       # small u, dieresis or umlaut mark
+    'yacute':  chr(253),       # small y, acute accent
+    'thorn':   chr(254),       # small thorn, Icelandic
+    'yuml':    chr(255),       # small y, dieresis or umlaut mark
+}