ocean.text.entities.HtmlEntitySet source code

1 /*******************************************************************************
2 
3     Html entities. Built on the basis of the xml entities.
4 
5     Copyright:
6         Copyright (c) 2009-2016 dunnhumby Germany GmbH.
7         All rights reserved.
8 
9     License:
10         Boost Software License Version 1.0. See LICENSE_BOOST.txt for details.
11         Alternatively, this file may be distributed under the terms of the Tango
12         3-Clause BSD License (see LICENSE_BSD.txt for details).
13 
14 *******************************************************************************/
15 
16 module ocean.text.entities.HtmlEntitySet;
17 
18 
19 
20 import ocean.text.entities.model.IEntitySet;
21 
22 import ocean.text.entities.XmlEntitySet;
23 
24 import ocean.meta.types.Qualifiers;
25 
26 /*******************************************************************************
27 
28     Html entity set class - extends XmlEntitySet
29 
30 *******************************************************************************/
31 
32 public class HtmlEntitySet : XmlEntitySet
33 {
34     /***************************************************************************
35 
36         This alias.
37 
38     ***************************************************************************/
39 
40     public alias typeof(this) This;
41 
42 
43     /***************************************************************************
44 
45         ISO 8859-1 (Latin 1) character entities. Uses the basic xml entities,
46         plus the ISO8859_1_extra entities defined below.
47 
48     ***************************************************************************/
49 
50     protected static immutable Entity[] ISO8859_1 = xml_entities ~ ISO8859_1_extra;
51 
52 
53     /***************************************************************************
54 
55         All html entities. Union of ISO 8859-1 (Latin 1) and -15 (Latin 9)
56         character entities.
57 
58     ***************************************************************************/
59 
60     public static immutable Entity[] html_entities = ISO8859_1 ~ ISO8859_15_extra;
61 
62 
63     /***************************************************************************
64 
65         Extra characters of ISO 8859-1 (Latin 1) (in addition to the xml
66         entities)
67 
68     ***************************************************************************/
69 
70     protected static immutable Entity[] ISO8859_1_extra =
71     [
72         {"nbsp",   0x00A0}, // ' '
73         {"iexcl",  0x00A1}, // '¡'
74         {"cent",   0x00A2}, // '¢'
75         {"pound",  0x00A3}, // '£'
76         {"curren", 0x00A4}, // '¤'
77         {"yen",    0x00A5}, // '¥'
78         {"brvbar", 0x00A6}, // '¦'
79         {"sect",   0x00A7}, // '§'
80         {"uml",    0x00A8}, // '¨'
81         {"copy",   0x00A9}, // '©'
82         {"ordf",   0x00AA}, // 'ª'
83         {"laquo",  0x00AB}, // '«'
84         {"not",    0x00AC}, // '¬'
85         {"shy",    0x00AD}, // ''
86         {"reg",    0x00AE}, // '®'
87         {"macr",   0x00AF}, // '¯'
88         {"deg",    0x00B0}, // '°'
89         {"plusmn", 0x00B1}, // '±'
90         {"sup2",   0x00B2}, // '²'
91         {"sup3",   0x00B3}, // '³'
92         {"acute",  0x00B4}, // '´'
93         {"micro",  0x00B5}, // 'µ'
94         {"para",   0x00B6}, // '¶'
95         {"middot", 0x00B7}, // '·'
96         {"cedil",  0x00B8}, // '¸'
97         {"sup1",   0x00B9}, // '¹'
98         {"ordm",   0x00BA}, // 'º'
99         {"raquo",  0x00BB}, // '»'
100         {"frac14", 0x00BC}, // '¼'
101         {"frac12", 0x00BD}, // '½'
102         {"frac34", 0x00BE}, // '¾'
103         {"iquest", 0x00BF}, // '¿'
104         {"Agrave", 0x00C0}, // 'À'
105         {"Aacute", 0x00C1}, // 'Á'
106         {"Acirc",  0x00C2}, // 'Â'
107         {"Atilde", 0x00C3}, // 'Ã'
108         {"Auml",   0x00C4}, // 'Ä'
109         {"Aring",  0x00C5}, // 'Å'
110         {"AElig",  0x00C6}, // 'Æ'
111         {"Ccedil", 0x00C7}, // 'Ç'
112         {"Egrave", 0x00C8}, // 'È'
113         {"Eacute", 0x00C9}, // 'É'
114         {"Ecirc",  0x00CA}, // 'Ê'
115         {"Euml",   0x00CB}, // 'Ë'
116         {"Igrave", 0x00CC}, // 'Ì'
117         {"Iacute", 0x00CD}, // 'Í'
118         {"Icirc",  0x00CE}, // 'Î'
119         {"Iuml",   0x00CF}, // 'Ï'
120         {"ETH",    0x00D0}, // 'Ð'
121         {"Ntilde", 0x00D1}, // 'Ñ'
122         {"Ograve", 0x00D2}, // 'Ò'
123         {"Oacute", 0x00D3}, // 'Ó'
124         {"Ocirc",  0x00D4}, // 'Ô'
125         {"Otilde", 0x00D5}, // 'Õ'
126         {"Ouml",   0x00D6}, // 'Ö'
127         {"times",  0x00D7}, // '×'
128         {"Oslash", 0x00D8}, // 'Ø'
129         {"Ugrave", 0x00D9}, // 'Ù'
130         {"Uacute", 0x00DA}, // 'Ú'
131         {"Ucirc",  0x00DB}, // 'Û'
132         {"Uuml",   0x00DC}, // 'Ü'
133         {"Yacute", 0x00DD}, // 'Ý'
134         {"THORN",  0x00DE}, // 'Þ'
135         {"szlig",  0x00DF}, // 'ß'
136         {"agrave", 0x00E0}, // 'à'
137         {"aacute", 0x00E1}, // 'á'
138         {"acirc",  0x00E2}, // 'â'
139         {"atilde", 0x00E3}, // 'ã'
140         {"auml",   0x00E4}, // 'ä'
141         {"aring",  0x00E5}, // 'å'
142         {"aelig",  0x00E6}, // 'æ'
143         {"ccedil", 0x00E7}, // 'ç'
144         {"egrave", 0x00E8}, // 'è'
145         {"eacute", 0x00E9}, // 'é'
146         {"ecirc",  0x00EA}, // 'ê'
147         {"euml",   0x00EB}, // 'ë'
148         {"igrave", 0x00EC}, // 'ì'
149         {"iacute", 0x00ED}, // 'í'
150         {"icirc",  0x00EE}, // 'î'
151         {"iuml",   0x00EF}, // 'ï'
152         {"eth",    0x00F0}, // 'ð'
153         {"ntilde", 0x00F1}, // 'ñ'
154         {"ograve", 0x00F2}, // 'ò'
155         {"oacute", 0x00F3}, // 'ó'
156         {"ocirc",  0x00F4}, // 'ô'
157         {"otilde", 0x00F5}, // 'õ'
158         {"ouml",   0x00F6}, // 'ö'
159         {"divide", 0x00F7}, // '÷'
160         {"oslash", 0x00F8}, // 'ø'
161         {"ugrave", 0x00F9}, // 'ù'
162         {"uacute", 0x00FA}, // 'ú'
163         {"ucirc",  0x00FB}, // 'û'
164         {"uuml",   0x00FC}, // 'ü'
165         {"yacute", 0x00FD}, // 'ý'
166         {"yuml",   0x00FF}, // 'ÿ'
167         {"thorn",  0x00FE}, // 'þ'
168     ];
169 
170 
171     /***************************************************************************
172 
173         Extra characters of ISO 8859-15 (Latin 9) which are not in ISO 8859-1
174         (Latin 1)
175 
176     ***************************************************************************/
177 
178     protected static immutable Entity[] ISO8859_15_extra =
179     [
180         {"OElig",  0x0152}, // 'Œ'
181         {"OElig",  0x0153}, // 'œ'
182         {"Scaron", 0x0160}, // 'Š'
183         {"scaron", 0x0161}, // 'š'
184         {"Zcaron", 0x017D}, // 'Ž'
185         {"zcaron", 0x017E}, // 'ž'
186         {"euro",   0x20AC}, // '€'
187     ];
188 
189 
190     /***************************************************************************
191 
192         Returns the list of entities.
193 
194     ***************************************************************************/
195 
196     public override const(Entity)[] entities ( )
197     {
198         return This.html_entities;
199     }
200 }
201