1 package Thrasher
::HTMLNormalize
::Entities
;
6 our @EXPORT = qw(html_entity);
8 # This file simply provides a method which returns a hash of
9 # entities to their numeric mapping in Unicode. This allows
10 # the use of extended HTML entities in browsers that don't
11 # necessarily support them.
12 # This may not be entirely necessary, but you have to be careful
13 # with entities in XHTML; an illegal entity makes for an illegal
14 # XML file. It's just easier and safer to normalize these things.
15 # This is take from Python 2.5's htmlentitydefs, converted with
16 # a bit of emacsro-fu.
18 my %name2codepoint = (
19 AElig
=> 0x00c6, # latin capital letter AE = latin capital ligature AE, U+00C6 ISOlat1
20 Aacute
=> 0x00c1, # latin capital letter A with acute, U+00C1 ISOlat1
21 Acirc
=> 0x00c2, # latin capital letter A with circumflex, U+00C2 ISOlat1
22 Agrave
=> 0x00c0, # latin capital letter A with grave = latin capital letter A grave, U+00C0 ISOlat1
23 Alpha
=> 0x0391, # greek capital letter alpha, U+0391
24 Aring
=> 0x00c5, # latin capital letter A with ring above = latin capital letter A ring, U+00C5 ISOlat1
25 Atilde
=> 0x00c3, # latin capital letter A with tilde, U+00C3 ISOlat1
26 Auml
=> 0x00c4, # latin capital letter A with diaeresis, U+00C4 ISOlat1
27 Beta
=> 0x0392, # greek capital letter beta, U+0392
28 Ccedil
=> 0x00c7, # latin capital letter C with cedilla, U+00C7 ISOlat1
29 Chi
=> 0x03a7, # greek capital letter chi, U+03A7
30 Dagger
=> 0x2021, # double dagger, U+2021 ISOpub
31 Delta
=> 0x0394, # greek capital letter delta, U+0394 ISOgrk3
32 ETH
=> 0x00d0, # latin capital letter ETH, U+00D0 ISOlat1
33 Eacute
=> 0x00c9, # latin capital letter E with acute, U+00C9 ISOlat1
34 Ecirc
=> 0x00ca, # latin capital letter E with circumflex, U+00CA ISOlat1
35 Egrave
=> 0x00c8, # latin capital letter E with grave, U+00C8 ISOlat1
36 Epsilon
=> 0x0395, # greek capital letter epsilon, U+0395
37 Eta
=> 0x0397, # greek capital letter eta, U+0397
38 Euml
=> 0x00cb, # latin capital letter E with diaeresis, U+00CB ISOlat1
39 Gamma
=> 0x0393, # greek capital letter gamma, U+0393 ISOgrk3
40 Iacute
=> 0x00cd, # latin capital letter I with acute, U+00CD ISOlat1
41 Icirc
=> 0x00ce, # latin capital letter I with circumflex, U+00CE ISOlat1
42 Igrave
=> 0x00cc, # latin capital letter I with grave, U+00CC ISOlat1
43 Iota
=> 0x0399, # greek capital letter iota, U+0399
44 Iuml
=> 0x00cf, # latin capital letter I with diaeresis, U+00CF ISOlat1
45 Kappa
=> 0x039a, # greek capital letter kappa, U+039A
46 Lambda
=> 0x039b, # greek capital letter lambda, U+039B ISOgrk3
47 Mu
=> 0x039c, # greek capital letter mu, U+039C
48 Ntilde
=> 0x00d1, # latin capital letter N with tilde, U+00D1 ISOlat1
49 Nu
=> 0x039d, # greek capital letter nu, U+039D
50 OElig
=> 0x0152, # latin capital ligature OE, U+0152 ISOlat2
51 Oacute
=> 0x00d3, # latin capital letter O with acute, U+00D3 ISOlat1
52 Ocirc
=> 0x00d4, # latin capital letter O with circumflex, U+00D4 ISOlat1
53 Ograve
=> 0x00d2, # latin capital letter O with grave, U+00D2 ISOlat1
54 Omega
=> 0x03a9, # greek capital letter omega, U+03A9 ISOgrk3
55 Omicron
=> 0x039f, # greek capital letter omicron, U+039F
56 Oslash
=> 0x00d8, # latin capital letter O with stroke = latin capital letter O slash, U+00D8 ISOlat1
57 Otilde
=> 0x00d5, # latin capital letter O with tilde, U+00D5 ISOlat1
58 Ouml
=> 0x00d6, # latin capital letter O with diaeresis, U+00D6 ISOlat1
59 Phi
=> 0x03a6, # greek capital letter phi, U+03A6 ISOgrk3
60 Pi
=> 0x03a0, # greek capital letter pi, U+03A0 ISOgrk3
61 Prime
=> 0x2033, # double prime = seconds = inches, U+2033 ISOtech
62 Psi
=> 0x03a8, # greek capital letter psi, U+03A8 ISOgrk3
63 Rho
=> 0x03a1, # greek capital letter rho, U+03A1
64 Scaron
=> 0x0160, # latin capital letter S with caron, U+0160 ISOlat2
65 Sigma
=> 0x03a3, # greek capital letter sigma, U+03A3 ISOgrk3
66 THORN
=> 0x00de, # latin capital letter THORN, U+00DE ISOlat1
67 Tau
=> 0x03a4, # greek capital letter tau, U+03A4
68 Theta
=> 0x0398, # greek capital letter theta, U+0398 ISOgrk3
69 Uacute
=> 0x00da, # latin capital letter U with acute, U+00DA ISOlat1
70 Ucirc
=> 0x00db, # latin capital letter U with circumflex, U+00DB ISOlat1
71 Ugrave
=> 0x00d9, # latin capital letter U with grave, U+00D9 ISOlat1
72 Upsilon
=> 0x03a5, # greek capital letter upsilon, U+03A5 ISOgrk3
73 Uuml
=> 0x00dc, # latin capital letter U with diaeresis, U+00DC ISOlat1
74 Xi
=> 0x039e, # greek capital letter xi, U+039E ISOgrk3
75 Yacute
=> 0x00dd, # latin capital letter Y with acute, U+00DD ISOlat1
76 Yuml
=> 0x0178, # latin capital letter Y with diaeresis, U+0178 ISOlat2
77 Zeta
=> 0x0396, # greek capital letter zeta, U+0396
78 aacute
=> 0x00e1, # latin small letter a with acute, U+00E1 ISOlat1
79 acirc
=> 0x00e2, # latin small letter a with circumflex, U+00E2 ISOlat1
80 acute
=> 0x00b4, # acute accent = spacing acute, U+00B4 ISOdia
81 aelig
=> 0x00e6, # latin small letter ae = latin small ligature ae, U+00E6 ISOlat1
82 agrave
=> 0x00e0, # latin small letter a with grave = latin small letter a grave, U+00E0 ISOlat1
83 alefsym
=> 0x2135, # alef symbol = first transfinite cardinal, U+2135 NEW
84 alpha
=> 0x03b1, # greek small letter alpha, U+03B1 ISOgrk3
85 amp
=> 0x0026, # ampersand, U+0026 ISOnum
86 and => 0x2227, # logical and = wedge, U+2227 ISOtech
87 ang
=> 0x2220, # angle, U+2220 ISOamso
88 aring
=> 0x00e5, # latin small letter a with ring above = latin small letter a ring, U+00E5 ISOlat1
89 asymp
=> 0x2248, # almost equal to = asymptotic to, U+2248 ISOamsr
90 atilde
=> 0x00e3, # latin small letter a with tilde, U+00E3 ISOlat1
91 auml
=> 0x00e4, # latin small letter a with diaeresis, U+00E4 ISOlat1
92 bdquo
=> 0x201e, # double low-9 quotation mark, U+201E NEW
93 beta
=> 0x03b2, # greek small letter beta, U+03B2 ISOgrk3
94 brvbar
=> 0x00a6, # broken bar = broken vertical bar, U+00A6 ISOnum
95 bull
=> 0x2022, # bullet = black small circle, U+2022 ISOpub
96 cap
=> 0x2229, # intersection = cap, U+2229 ISOtech
97 ccedil
=> 0x00e7, # latin small letter c with cedilla, U+00E7 ISOlat1
98 cedil
=> 0x00b8, # cedilla = spacing cedilla, U+00B8 ISOdia
99 cent
=> 0x00a2, # cent sign, U+00A2 ISOnum
100 chi
=> 0x03c7, # greek small letter chi, U+03C7 ISOgrk3
101 circ
=> 0x02c6, # modifier letter circumflex accent, U+02C6 ISOpub
102 clubs
=> 0x2663, # black club suit = shamrock, U+2663 ISOpub
103 cong
=> 0x2245, # approximately equal to, U+2245 ISOtech
104 copy
=> 0x00a9, # copyright sign, U+00A9 ISOnum
105 crarr
=> 0x21b5, # downwards arrow with corner leftwards = carriage return, U+21B5 NEW
106 cup
=> 0x222a, # union = cup, U+222A ISOtech
107 curren
=> 0x00a4, # currency sign, U+00A4 ISOnum
108 dArr
=> 0x21d3, # downwards double arrow, U+21D3 ISOamsa
109 dagger
=> 0x2020, # dagger, U+2020 ISOpub
110 darr
=> 0x2193, # downwards arrow, U+2193 ISOnum
111 deg
=> 0x00b0, # degree sign, U+00B0 ISOnum
112 delta
=> 0x03b4, # greek small letter delta, U+03B4 ISOgrk3
113 diams
=> 0x2666, # black diamond suit, U+2666 ISOpub
114 divide
=> 0x00f7, # division sign, U+00F7 ISOnum
115 eacute
=> 0x00e9, # latin small letter e with acute, U+00E9 ISOlat1
116 ecirc
=> 0x00ea, # latin small letter e with circumflex, U+00EA ISOlat1
117 egrave
=> 0x00e8, # latin small letter e with grave, U+00E8 ISOlat1
118 empty
=> 0x2205, # empty set = null set = diameter, U+2205 ISOamso
119 emsp
=> 0x2003, # em space, U+2003 ISOpub
120 ensp
=> 0x2002, # en space, U+2002 ISOpub
121 epsilon
=> 0x03b5, # greek small letter epsilon, U+03B5 ISOgrk3
122 equiv
=> 0x2261, # identical to, U+2261 ISOtech
123 eta
=> 0x03b7, # greek small letter eta, U+03B7 ISOgrk3
124 eth
=> 0x00f0, # latin small letter eth, U+00F0 ISOlat1
125 euml
=> 0x00eb, # latin small letter e with diaeresis, U+00EB ISOlat1
126 euro
=> 0x20ac, # euro sign, U+20AC NEW
127 exist
=> 0x2203, # there exists, U+2203 ISOtech
128 fnof
=> 0x0192, # latin small f with hook = function = florin, U+0192 ISOtech
129 forall
=> 0x2200, # for all, U+2200 ISOtech
130 frac12
=> 0x00bd, # vulgar fraction one half = fraction one half, U+00BD ISOnum
131 frac14
=> 0x00bc, # vulgar fraction one quarter = fraction one quarter, U+00BC ISOnum
132 frac34
=> 0x00be, # vulgar fraction three quarters = fraction three quarters, U+00BE ISOnum
133 frasl
=> 0x2044, # fraction slash, U+2044 NEW
134 gamma
=> 0x03b3, # greek small letter gamma, U+03B3 ISOgrk3
135 ge => 0x2265, # greater-than or equal to, U+2265 ISOtech
136 gt => 0x003e, # greater-than sign, U+003E ISOnum
137 hArr
=> 0x21d4, # left right double arrow, U+21D4 ISOamsa
138 harr
=> 0x2194, # left right arrow, U+2194 ISOamsa
139 hearts
=> 0x2665, # black heart suit = valentine, U+2665 ISOpub
140 hellip
=> 0x2026, # horizontal ellipsis = three dot leader, U+2026 ISOpub
141 iacute
=> 0x00ed, # latin small letter i with acute, U+00ED ISOlat1
142 icirc
=> 0x00ee, # latin small letter i with circumflex, U+00EE ISOlat1
143 iexcl
=> 0x00a1, # inverted exclamation mark, U+00A1 ISOnum
144 igrave
=> 0x00ec, # latin small letter i with grave, U+00EC ISOlat1
145 image
=> 0x2111, # blackletter capital I = imaginary part, U+2111 ISOamso
146 infin
=> 0x221e, # infinity, U+221E ISOtech
147 int => 0x222b, # integral, U+222B ISOtech
148 iota
=> 0x03b9, # greek small letter iota, U+03B9 ISOgrk3
149 iquest
=> 0x00bf, # inverted question mark = turned question mark, U+00BF ISOnum
150 isin
=> 0x2208, # element of, U+2208 ISOtech
151 iuml
=> 0x00ef, # latin small letter i with diaeresis, U+00EF ISOlat1
152 kappa
=> 0x03ba, # greek small letter kappa, U+03BA ISOgrk3
153 lArr
=> 0x21d0, # leftwards double arrow, U+21D0 ISOtech
154 lambda
=> 0x03bb, # greek small letter lambda, U+03BB ISOgrk3
155 lang
=> 0x2329, # left-pointing angle bracket = bra, U+2329 ISOtech
156 laquo
=> 0x00ab, # left-pointing double angle quotation mark = left pointing guillemet, U+00AB ISOnum
157 larr
=> 0x2190, # leftwards arrow, U+2190 ISOnum
158 lceil
=> 0x2308, # left ceiling = apl upstile, U+2308 ISOamsc
159 ldquo
=> 0x201c, # left double quotation mark, U+201C ISOnum
160 le => 0x2264, # less-than or equal to, U+2264 ISOtech
161 lfloor
=> 0x230a, # left floor = apl downstile, U+230A ISOamsc
162 lowast
=> 0x2217, # asterisk operator, U+2217 ISOtech
163 loz
=> 0x25ca, # lozenge, U+25CA ISOpub
164 lrm
=> 0x200e, # left-to-right mark, U+200E NEW RFC 2070
165 lsaquo
=> 0x2039, # single left-pointing angle quotation mark, U+2039 ISO proposed
166 lsquo
=> 0x2018, # left single quotation mark, U+2018 ISOnum
167 lt => 0x003c, # less-than sign, U+003C ISOnum
168 macr
=> 0x00af, # macron = spacing macron = overline = APL overbar, U+00AF ISOdia
169 mdash
=> 0x2014, # em dash, U+2014 ISOpub
170 micro
=> 0x00b5, # micro sign, U+00B5 ISOnum
171 middot
=> 0x00b7, # middle dot = Georgian comma = Greek middle dot, U+00B7 ISOnum
172 minus
=> 0x2212, # minus sign, U+2212 ISOtech
173 mu
=> 0x03bc, # greek small letter mu, U+03BC ISOgrk3
174 nabla
=> 0x2207, # nabla = backward difference, U+2207 ISOtech
175 nbsp
=> 0x00a0, # no-break space = non-breaking space, U+00A0 ISOnum
176 ndash
=> 0x2013, # en dash, U+2013 ISOpub
177 ne => 0x2260, # not equal to, U+2260 ISOtech
178 ni
=> 0x220b, # contains as member, U+220B ISOtech
179 not => 0x00ac, # not sign, U+00AC ISOnum
180 notin
=> 0x2209, # not an element of, U+2209 ISOtech
181 nsub
=> 0x2284, # not a subset of, U+2284 ISOamsn
182 ntilde
=> 0x00f1, # latin small letter n with tilde, U+00F1 ISOlat1
183 nu
=> 0x03bd, # greek small letter nu, U+03BD ISOgrk3
184 oacute
=> 0x00f3, # latin small letter o with acute, U+00F3 ISOlat1
185 ocirc
=> 0x00f4, # latin small letter o with circumflex, U+00F4 ISOlat1
186 oelig
=> 0x0153, # latin small ligature oe, U+0153 ISOlat2
187 ograve
=> 0x00f2, # latin small letter o with grave, U+00F2 ISOlat1
188 oline
=> 0x203e, # overline = spacing overscore, U+203E NEW
189 omega
=> 0x03c9, # greek small letter omega, U+03C9 ISOgrk3
190 omicron
=> 0x03bf, # greek small letter omicron, U+03BF NEW
191 oplus
=> 0x2295, # circled plus = direct sum, U+2295 ISOamsb
192 or => 0x2228, # logical or = vee, U+2228 ISOtech
193 ordf
=> 0x00aa, # feminine ordinal indicator, U+00AA ISOnum
194 ordm
=> 0x00ba, # masculine ordinal indicator, U+00BA ISOnum
195 oslash
=> 0x00f8, # latin small letter o with stroke, = latin small letter o slash, U+00F8 ISOlat1
196 otilde
=> 0x00f5, # latin small letter o with tilde, U+00F5 ISOlat1
197 otimes
=> 0x2297, # circled times = vector product, U+2297 ISOamsb
198 ouml
=> 0x00f6, # latin small letter o with diaeresis, U+00F6 ISOlat1
199 para
=> 0x00b6, # pilcrow sign = paragraph sign, U+00B6 ISOnum
200 part
=> 0x2202, # partial differential, U+2202 ISOtech
201 permil
=> 0x2030, # per mille sign, U+2030 ISOtech
202 perp
=> 0x22a5, # up tack = orthogonal to = perpendicular, U+22A5 ISOtech
203 phi
=> 0x03c6, # greek small letter phi, U+03C6 ISOgrk3
204 pi
=> 0x03c0, # greek small letter pi, U+03C0 ISOgrk3
205 piv
=> 0x03d6, # greek pi symbol, U+03D6 ISOgrk3
206 plusmn
=> 0x00b1, # plus-minus sign = plus-or-minus sign, U+00B1 ISOnum
207 pound
=> 0x00a3, # pound sign, U+00A3 ISOnum
208 prime
=> 0x2032, # prime = minutes = feet, U+2032 ISOtech
209 prod
=> 0x220f, # n-ary product = product sign, U+220F ISOamsb
210 prop
=> 0x221d, # proportional to, U+221D ISOtech
211 psi
=> 0x03c8, # greek small letter psi, U+03C8 ISOgrk3
212 quot
=> 0x0022, # quotation mark = APL quote, U+0022 ISOnum
213 rArr
=> 0x21d2, # rightwards double arrow, U+21D2 ISOtech
214 radic
=> 0x221a, # square root = radical sign, U+221A ISOtech
215 rang
=> 0x232a, # right-pointing angle bracket = ket, U+232A ISOtech
216 raquo
=> 0x00bb, # right-pointing double angle quotation mark = right pointing guillemet, U+00BB ISOnum
217 rarr
=> 0x2192, # rightwards arrow, U+2192 ISOnum
218 rceil
=> 0x2309, # right ceiling, U+2309 ISOamsc
219 rdquo
=> 0x201d, # right double quotation mark, U+201D ISOnum
220 real
=> 0x211c, # blackletter capital R = real part symbol, U+211C ISOamso
221 reg
=> 0x00ae, # registered sign = registered trade mark sign, U+00AE ISOnum
222 rfloor
=> 0x230b, # right floor, U+230B ISOamsc
223 rho
=> 0x03c1, # greek small letter rho, U+03C1 ISOgrk3
224 rlm
=> 0x200f, # right-to-left mark, U+200F NEW RFC 2070
225 rsaquo
=> 0x203a, # single right-pointing angle quotation mark, U+203A ISO proposed
226 rsquo
=> 0x2019, # right single quotation mark, U+2019 ISOnum
227 sbquo
=> 0x201a, # single low-9 quotation mark, U+201A NEW
228 scaron
=> 0x0161, # latin small letter s with caron, U+0161 ISOlat2
229 sdot
=> 0x22c5, # dot operator, U+22C5 ISOamsb
230 sect
=> 0x00a7, # section sign, U+00A7 ISOnum
231 shy
=> 0x00ad, # soft hyphen = discretionary hyphen, U+00AD ISOnum
232 sigma
=> 0x03c3, # greek small letter sigma, U+03C3 ISOgrk3
233 sigmaf
=> 0x03c2, # greek small letter final sigma, U+03C2 ISOgrk3
234 sim
=> 0x223c, # tilde operator = varies with = similar to, U+223C ISOtech
235 spades
=> 0x2660, # black spade suit, U+2660 ISOpub
236 sub => 0x2282, # subset of, U+2282 ISOtech
237 sube
=> 0x2286, # subset of or equal to, U+2286 ISOtech
238 sum
=> 0x2211, # n-ary sumation, U+2211 ISOamsb
239 sup
=> 0x2283, # superset of, U+2283 ISOtech
240 sup1
=> 0x00b9, # superscript one = superscript digit one, U+00B9 ISOnum
241 sup2
=> 0x00b2, # superscript two = superscript digit two = squared, U+00B2 ISOnum
242 sup3
=> 0x00b3, # superscript three = superscript digit three = cubed, U+00B3 ISOnum
243 supe
=> 0x2287, # superset of or equal to, U+2287 ISOtech
244 szlig
=> 0x00df, # latin small letter sharp s = ess-zed, U+00DF ISOlat1
245 tau
=> 0x03c4, # greek small letter tau, U+03C4 ISOgrk3
246 there4
=> 0x2234, # therefore, U+2234 ISOtech
247 theta
=> 0x03b8, # greek small letter theta, U+03B8 ISOgrk3
248 thetasym
=> 0x03d1, # greek small letter theta symbol, U+03D1 NEW
249 thinsp
=> 0x2009, # thin space, U+2009 ISOpub
250 thorn
=> 0x00fe, # latin small letter thorn with, U+00FE ISOlat1
251 tilde
=> 0x02dc, # small tilde, U+02DC ISOdia
252 times => 0x00d7, # multiplication sign, U+00D7 ISOnum
253 trade
=> 0x2122, # trade mark sign, U+2122 ISOnum
254 uArr
=> 0x21d1, # upwards double arrow, U+21D1 ISOamsa
255 uacute
=> 0x00fa, # latin small letter u with acute, U+00FA ISOlat1
256 uarr
=> 0x2191, # upwards arrow, U+2191 ISOnum
257 ucirc
=> 0x00fb, # latin small letter u with circumflex, U+00FB ISOlat1
258 ugrave
=> 0x00f9, # latin small letter u with grave, U+00F9 ISOlat1
259 uml
=> 0x00a8, # diaeresis = spacing diaeresis, U+00A8 ISOdia
260 upsih
=> 0x03d2, # greek upsilon with hook symbol, U+03D2 NEW
261 upsilon
=> 0x03c5, # greek small letter upsilon, U+03C5 ISOgrk3
262 uuml
=> 0x00fc, # latin small letter u with diaeresis, U+00FC ISOlat1
263 weierp
=> 0x2118, # script capital P = power set = Weierstrass p, U+2118 ISOamso
264 xi
=> 0x03be, # greek small letter xi, U+03BE ISOgrk3
265 yacute
=> 0x00fd, # latin small letter y with acute, U+00FD ISOlat1
266 yen
=> 0x00a5, # yen sign = yuan sign, U+00A5 ISOnum
267 yuml
=> 0x00ff, # latin small letter y with diaeresis, U+00FF ISOlat1
268 zeta
=> 0x03b6, # greek small letter zeta, U+03B6 ISOgrk3
269 zwj
=> 0x200d, # zero width joiner, U+200D NEW RFC 2070
270 zwnj
=> 0x200c, # zero width non-joiner, U+200C NEW RFC 2070
273 sub html_entity
{ $name2codepoint{$_[0]} }
