ocean.text.entities.model.IEntitySet source code

1 /*******************************************************************************
2 
3     An abstract class encapsulating a set of entities for en/decoding. A typical
4     example is the various html entities which are required to be encoded, for
5     example:
6 
7         '&' should be encoded as "&amp;"
8 
9     The class should be implemented, and the entities() methods made to return
10     the list of entities to be handled.
11 
12     Copyright:
13         Copyright (c) 2009-2016 dunnhumby Germany GmbH.
14         All rights reserved.
15 
16     License:
17         Boost Software License Version 1.0. See LICENSE_BOOST.txt for details.
18         Alternatively, this file may be distributed under the terms of the Tango
19         3-Clause BSD License (see LICENSE_BSD.txt for details).
20 
21 *******************************************************************************/
22 
23 module ocean.text.entities.model.IEntitySet;
24 
25 
26 
27 
28 import ocean.text.utf.UtfString : InvalidUnicode, utf_match;
29 
30 import ocean.transition;
31 
32 /*******************************************************************************
33 
34     Abstract entity set class.
35 
36 *******************************************************************************/
37 
38 public abstract class IEntitySet
39 {
40     /***************************************************************************
41 
42         An entity. Simply a tuple of a name and a unicode value (eg "amp", '&').
43 
44     ***************************************************************************/
45 
46     public struct Entity
47     {
48         istring name;
49         dchar unicode;
50     }
51 
52 
53     /***************************************************************************
54 
55         Abstract method to return the list of entities.
56 
57     ***************************************************************************/
58 
59     public abstract Const!(Entity)[] entities ( );
60 
61 
62     /***************************************************************************
63 
64         Abstract method to get the encoded form of an entity.
65 
66     ***************************************************************************/
67 
68     abstract public char[] getEncodedEntity ( dchar unicode, ref char[] output );
69 
70 
71     /***************************************************************************
72 
73         Gets the unicode character associated with the passed name.
74 
75         Params:
76             Char = character type of name
77             name = name to check
78 
79         Returns:
80             unicode corresponding to name, or InvalidUnicode if name is not in
81             the entity list
82 
83     ***************************************************************************/
84 
85     public dchar getUnicode ( Char ) ( Char[] name )
86     {
87         foreach ( check_name, unicode; this )
88         {
89             if ( utf_match(name, check_name) )
90             {
91                 return unicode;
92             }
93         }
94 
95         return InvalidUnicode;
96     }
97 
98 
99     /***************************************************************************
100 
101         Gets the name associated with the passed unicode character.
102 
103         Params:
104             unicode = unicode value to check
105 
106         Returns:
107             name corresponding to unicode, or "" if unicode is not in the entity
108             list
109 
110     ***************************************************************************/
111 
112     public istring getName ( dchar unicode )
113     {
114         foreach ( name, check_unicode; this )
115         {
116             if ( check_unicode == unicode )
117             {
118                 return name;
119             }
120         }
121 
122         return "";
123     }
124 
125 
126     /***************************************************************************
127 
128         Checks whether the passed name is in the list of entities.
129 
130         Params:
131             name = name to check
132 
133         Returns:
134             true if name is an entity
135 
136     ***************************************************************************/
137 
138     public bool opIn_r ( char[] name )
139     {
140         foreach ( ref entity; this.entities )
141         {
142             if ( utf_match(name, entity.name) )
143             {
144                 return true;
145             }
146         }
147 
148         return false;
149     }
150 
151 
152     /***************************************************************************
153 
154         Checks whether the passed name is in the list of entities.
155 
156         Params:
157             name = name to check
158 
159         Returns:
160             true if name is an entity
161 
162     ***************************************************************************/
163 
164     public bool opIn_r ( wchar[] name )
165     {
166         foreach ( ref entity; this.entities )
167         {
168             if ( utf_match(name, entity.name) )
169             {
170                 return true;
171             }
172         }
173 
174         return false;
175     }
176 
177 
178     /***************************************************************************
179 
180         Checks whether the passed name is in the list of entities.
181 
182         Params:
183             name = name to check
184 
185         Returns:
186             true if name is an entity
187 
188     ***************************************************************************/
189 
190     public bool opIn_r ( dchar[] name )
191     {
192         foreach ( ref entity; this.entities )
193         {
194             if ( utf_match(name, entity.name) )
195             {
196                 return true;
197             }
198         }
199 
200         return false;
201     }
202 
203 
204     /***************************************************************************
205 
206         Checks whether the passed unicode is in the list of entities.
207 
208         Params:
209             unicode = unicode value to check
210 
211         Returns:
212             true if unicode is an entity
213 
214     ***************************************************************************/
215 
216     public bool opIn_r ( dchar unicode )
217     {
218         foreach ( ref entity; this.entities )
219         {
220             if ( entity.unicode == unicode )
221             {
222                 return true;
223             }
224         }
225 
226         return false;
227     }
228 
229 
230     /***************************************************************************
231 
232         Checks whether the passed unicode is in the list of entities.
233 
234         Params:
235             unicode = unicode value to check
236 
237         Returns:
238             true if unicode is an entity
239 
240     ***************************************************************************/
241 
242     public bool opIn_r ( wchar unicode )
243     {
244         return (cast(dchar)unicode) in this;
245     }
246 
247 
248     /***************************************************************************
249 
250         Checks whether the passed unicode is in the list of entities.
251 
252         Params:
253             unicode = unicode value to check
254 
255         Returns:
256             true if unicode is an entity
257 
258     ***************************************************************************/
259 
260     public bool opIn_r ( char unicode )
261     {
262         return (cast(dchar)unicode) in this;
263     }
264 
265 
266     /***************************************************************************
267 
268         foreach iterator over the list of entities.
269 
270         foreach arguments exposed:
271             char[] name = entity name
272             dchar unicode = entity unicode value
273 
274 
275     ***************************************************************************/
276 
277     public int opApply ( scope int delegate ( ref Const!(istring), ref Const!(dchar) ) dg )
278     {
279         int res;
280         foreach ( ref entity; this.entities )
281         {
282             res = dg(entity.name, entity.unicode);
283             if ( res )
284             {
285                 break;
286             }
287         }
288 
289         return res;
290     }
291 }
292