1
- // CharsetDetectorTest.cs created with MonoDevelop
1
+ // CharsetDetectorTest.cs created with MonoDevelop
2
2
//
3
3
// Author:
4
4
// Rudi Pettazzi <rudi.pettazzi@gmail.com>
@@ -25,9 +25,9 @@ public void TestAscii()
25
25
using ( stream )
26
26
{
27
27
var result = CharsetDetector . DetectFromStream ( stream ) ;
28
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
29
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
30
- Assert . IsFalse ( result . Detected . HasBOM ) ;
28
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
29
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
30
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
31
31
}
32
32
}
33
33
@@ -42,9 +42,9 @@ public async Task TestAsciiAsync()
42
42
using ( stream )
43
43
{
44
44
var result = await CharsetDetector . DetectFromStreamAsync ( stream ) ;
45
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
46
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
47
- Assert . IsFalse ( result . Detected . HasBOM ) ;
45
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
46
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
47
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
48
48
}
49
49
}
50
50
@@ -56,8 +56,8 @@ public void TestAscii_with_HZ_sequence()
56
56
using ( stream )
57
57
{
58
58
var result = CharsetDetector . DetectFromStream ( stream ) ;
59
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
60
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
59
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
60
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
61
61
}
62
62
}
63
63
@@ -69,8 +69,8 @@ public async Task TestAscii_with_HZ_sequenceAsync()
69
69
using ( stream )
70
70
{
71
71
var result = await CharsetDetector . DetectFromStreamAsync ( stream ) ;
72
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
73
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
72
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
73
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
74
74
}
75
75
}
76
76
@@ -98,7 +98,7 @@ public void DetectFromStreamMaxBytes(int? maxBytes, int expectedPosition, int st
98
98
CharsetDetector . DetectFromStream ( stream , maxBytes ) ;
99
99
100
100
// Assert
101
- Assert . AreEqual ( expectedPosition , stream . Position ) ;
101
+ Assert . That ( stream . Position , Is . EqualTo ( expectedPosition ) ) ;
102
102
}
103
103
104
104
[ Test ]
@@ -120,7 +120,7 @@ public async Task DetectFromStreamMaxBytesAsync(int? maxBytes, int expectedPosit
120
120
await CharsetDetector . DetectFromStreamAsync ( stream , maxBytes ) ;
121
121
122
122
// Assert
123
- Assert . AreEqual ( expectedPosition , stream . Position ) ;
123
+ Assert . That ( stream . Position , Is . EqualTo ( expectedPosition ) ) ;
124
124
}
125
125
126
126
[ Test ]
@@ -138,9 +138,9 @@ public void DetectFromByteArray(int offset, int len, string detectedCodepage)
138
138
var result = CharsetDetector . DetectFromBytes ( bytes , offset , len ) ;
139
139
140
140
// Assert
141
- Assert . AreEqual ( detectedCodepage , result . Detected . EncodingName ) ;
142
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
143
- Assert . IsFalse ( result . Detected . HasBOM ) ;
141
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( detectedCodepage ) ) ;
142
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
143
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
144
144
}
145
145
146
146
[ Test ]
@@ -153,9 +153,9 @@ public void TestCaseBomUtf7(byte[] bufferBytes)
153
153
{
154
154
var result = CharsetDetector . DetectFromBytes ( bufferBytes )
155
155
. Detected ;
156
- Assert . AreEqual ( CodepageName . UTF7 , result . EncodingName ) ;
157
- Assert . AreEqual ( 1.0f , result . Confidence ) ;
158
- Assert . IsTrue ( result . HasBOM ) ;
156
+ Assert . That ( result . EncodingName , Is . EqualTo ( CodepageName . UTF7 ) ) ;
157
+ Assert . That ( result . Confidence , Is . EqualTo ( 1.0f ) ) ;
158
+ Assert . That ( result . HasBOM , Is . True ) ;
159
159
}
160
160
161
161
[ Test ]
@@ -164,9 +164,9 @@ public void TestBomGb18030()
164
164
var bufferBytes = new byte [ ] { 0x84 , 0x31 , 0x95 , 0x33 } ;
165
165
var result = CharsetDetector . DetectFromBytes ( bufferBytes )
166
166
. Detected ;
167
- Assert . AreEqual ( CodepageName . GB18030 , result . EncodingName ) ;
168
- Assert . AreEqual ( 1.0f , result . Confidence ) ;
169
- Assert . IsTrue ( result . HasBOM ) ;
167
+ Assert . That ( result . EncodingName , Is . EqualTo ( CodepageName . GB18030 ) ) ;
168
+ Assert . That ( result . Confidence , Is . EqualTo ( 1.0f ) ) ;
169
+ Assert . That ( result . HasBOM , Is . True ) ;
170
170
}
171
171
172
172
[ Test ]
@@ -178,19 +178,19 @@ public void TestUTF8_1()
178
178
"利用案内でどうぞ。" ;
179
179
byte [ ] buf = Encoding . UTF8 . GetBytes ( s ) ;
180
180
var result = CharsetDetector . DetectFromBytes ( buf ) ;
181
- Assert . AreEqual ( CodepageName . UTF8 , result . Detected . EncodingName ) ;
182
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
183
- Assert . IsFalse ( result . Detected . HasBOM ) ;
181
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF8 ) ) ;
182
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
183
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
184
184
}
185
185
186
186
[ Test ]
187
187
public void TestBomUtf8 ( )
188
188
{
189
189
byte [ ] buf = { 0xEF , 0xBB , 0xBF , 0x68 , 0x65 , 0x6C , 0x6C , 0x6F , 0x21 } ;
190
190
var result = CharsetDetector . DetectFromBytes ( buf ) ;
191
- Assert . AreEqual ( CodepageName . UTF8 , result . Detected . EncodingName ) ;
192
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
193
- Assert . IsTrue ( result . Detected . HasBOM ) ;
191
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF8 ) ) ;
192
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
193
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
194
194
}
195
195
196
196
[ Test ]
@@ -199,9 +199,9 @@ public void Test2byteArrayBomUTF16_BE()
199
199
byte [ ] buf = { 0xFE , 0xFF , } ;
200
200
201
201
var result = CharsetDetector . DetectFromBytes ( buf ) ;
202
- Assert . AreEqual ( CodepageName . UTF16_BE , result . Detected . EncodingName ) ;
203
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
204
- Assert . IsTrue ( result . Detected . HasBOM ) ;
202
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF16_BE ) ) ;
203
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
204
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
205
205
}
206
206
207
207
[ Test ]
@@ -210,9 +210,9 @@ public void TestBomUTF16_BE()
210
210
byte [ ] buf = { 0xFE , 0xFF , 0x00 , 0x68 , 0x00 , 0x65 } ;
211
211
212
212
var result = CharsetDetector . DetectFromBytes ( buf ) ;
213
- Assert . AreEqual ( CodepageName . UTF16_BE , result . Detected . EncodingName ) ;
214
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
215
- Assert . IsTrue ( result . Detected . HasBOM ) ;
213
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF16_BE ) ) ;
214
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
215
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
216
216
}
217
217
218
218
[ Test ]
@@ -222,9 +222,9 @@ public void TestBomX_ISO_10646_UCS_4_3412()
222
222
byte [ ] buf = { 0xFE , 0xFF , 0x00 , 0x00 , 0x65 } ;
223
223
224
224
var result = CharsetDetector . DetectFromBytes ( buf ) ;
225
- Assert . AreEqual ( CodepageName . X_ISO_10646_UCS_4_3412 , result . Detected . EncodingName ) ;
226
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
227
- Assert . IsTrue ( result . Detected . HasBOM ) ;
225
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . X_ISO_10646_UCS_4_3412 ) ) ;
226
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
227
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
228
228
}
229
229
230
230
[ Test ]
@@ -234,88 +234,88 @@ public void TestBomX_ISO_10646_UCS_4_2143()
234
234
byte [ ] buf = { 0x00 , 0x00 , 0xFF , 0xFE , 0x00 , 0x65 } ;
235
235
236
236
var result = CharsetDetector . DetectFromBytes ( buf ) ;
237
- Assert . AreEqual ( CodepageName . X_ISO_10646_UCS_4_2143 , result . Detected . EncodingName ) ;
238
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
239
- Assert . IsTrue ( result . Detected . HasBOM ) ;
237
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . X_ISO_10646_UCS_4_2143 ) ) ;
238
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
239
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
240
240
}
241
241
242
242
[ Test ]
243
243
public void Test2byteArrayBomUTF16_LE ( )
244
244
{
245
245
byte [ ] buf = { 0xFF , 0xFE , } ;
246
246
var result = CharsetDetector . DetectFromBytes ( buf ) ;
247
- Assert . AreEqual ( CodepageName . UTF16_LE , result . Detected . EncodingName ) ;
248
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
249
- Assert . IsTrue ( result . Detected . HasBOM ) ;
247
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF16_LE ) ) ;
248
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
249
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
250
250
}
251
251
252
252
[ Test ]
253
253
public void TestBomUTF16_LE ( )
254
254
{
255
255
byte [ ] buf = { 0xFF , 0xFE , 0x68 , 0x00 , 0x65 , 0x00 } ;
256
256
var result = CharsetDetector . DetectFromBytes ( buf ) ;
257
- Assert . AreEqual ( CodepageName . UTF16_LE , result . Detected . EncodingName ) ;
258
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
259
- Assert . IsTrue ( result . Detected . HasBOM ) ;
257
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF16_LE ) ) ;
258
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
259
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
260
260
}
261
261
262
262
[ Test ]
263
263
public void TestBomUTF32_BE ( )
264
264
{
265
265
byte [ ] buf = { 0x00 , 0x00 , 0xFE , 0xFF , 0x00 , 0x00 , 0x00 , 0x68 } ;
266
266
var result = CharsetDetector . DetectFromBytes ( buf ) ;
267
- Assert . AreEqual ( CodepageName . UTF32_BE , result . Detected . EncodingName ) ;
268
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
269
- Assert . IsTrue ( result . Detected . HasBOM ) ;
267
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF32_BE ) ) ;
268
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
269
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
270
270
}
271
271
272
272
[ Test ]
273
273
public void TestBomUTF32_LE ( )
274
274
{
275
275
byte [ ] buf = { 0xFF , 0xFE , 0x00 , 0x00 , 0x68 , 0x00 , 0x00 , 0x00 } ;
276
276
var result = CharsetDetector . DetectFromBytes ( buf ) ;
277
- Assert . AreEqual ( CodepageName . UTF32_LE , result . Detected . EncodingName ) ;
278
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
279
- Assert . IsTrue ( result . Detected . HasBOM ) ;
277
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . UTF32_LE ) ) ;
278
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
279
+ Assert . That ( result . Detected . HasBOM , Is . True ) ;
280
280
}
281
281
282
282
[ Test ]
283
283
public void TestIssue3 ( )
284
284
{
285
285
byte [ ] buf = Encoding . UTF8 . GetBytes ( "3" ) ;
286
286
var result = CharsetDetector . DetectFromBytes ( buf ) ;
287
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
288
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
289
- Assert . IsFalse ( result . Detected . HasBOM ) ;
287
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
288
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
289
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
290
290
}
291
291
292
292
[ Test ]
293
293
public void TestOutOfRange ( )
294
294
{
295
295
byte [ ] buf = Encoding . UTF8 . GetBytes ( "3" ) ;
296
296
var result = CharsetDetector . DetectFromBytes ( buf ) ;
297
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
298
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
299
- Assert . IsFalse ( result . Detected . HasBOM ) ;
297
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
298
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
299
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
300
300
}
301
301
302
302
[ Test ]
303
303
public void TestOutOfRange2 ( )
304
304
{
305
305
byte [ ] buf = Encoding . UTF8 . GetBytes ( "1234567890" ) ;
306
306
var result = CharsetDetector . DetectFromBytes ( buf ) ;
307
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
308
- Assert . AreEqual ( 1.0f , result . Detected . Confidence ) ;
309
- Assert . IsFalse ( result . Detected . HasBOM ) ;
307
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
308
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1.0f ) ) ;
309
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
310
310
}
311
311
312
312
[ Test ]
313
313
public void TestSingleChar ( )
314
314
{
315
315
byte [ ] buf = Encoding . UTF8 . GetBytes ( "3" ) ;
316
316
var result = CharsetDetector . DetectFromBytes ( buf ) ;
317
- Assert . AreEqual ( CodepageName . ASCII , result . Detected . EncodingName ) ;
318
- Assert . AreEqual ( 1 , result . Detected . Confidence ) ;
319
- Assert . IsFalse ( result . Detected . HasBOM ) ;
317
+ Assert . That ( result . Detected . EncodingName , Is . EqualTo ( CodepageName . ASCII ) ) ;
318
+ Assert . That ( result . Detected . Confidence , Is . EqualTo ( 1 ) ) ;
319
+ Assert . That ( result . Detected . HasBOM , Is . False ) ;
320
320
}
321
- }
321
+ }
0 commit comments