Added unit test for charset detection

5 files changed, 22 insertions, 0 deletions

diff --git a/unittest/data/charset-detection/input/big5.txt b/unittest/data/charset-detection/input/big5.txt
new file mode 100644
index 00000000..59db9548
--- /dev/null
+++ b/unittest/data/charset-detection/input/big5.txt
@@ -0,0 +1 @@
+羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ羉砰いゅ
\ No newline at end of file
diff --git a/unittest/data/charset-detection/input/gb18030.txt b/unittest/data/charset-detection/input/gb18030.txt
new file mode 100644
index 00000000..962df87d
--- /dev/null
+++ b/unittest/data/charset-detection/input/gb18030.txt
@@ -0,0 +1 @@
+简体中文简体中文简体中文简体中文简体中文简体中文简体中文简体中文简体中文简体中文简体中文
\ No newline at end of file
diff --git a/unittest/data/charset-detection/input/shift_jis.txt b/unittest/data/charset-detection/input/shift_jis.txt
new file mode 100644
index 00000000..a580281d
--- /dev/null
+++ b/unittest/data/charset-detection/input/shift_jis.txt
@@ -0,0 +1 @@
+擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅擔杮岅
diff --git a/unittest/data/charset-detection/input/utf-8.txt b/unittest/data/charset-detection/input/utf-8.txt
new file mode 100644
index 00000000..cd66b082
--- /dev/null
+++ b/unittest/data/charset-detection/input/utf-8.txt
@@ -0,0 +1 @@
+姹夊瓧婕㈠瓧绲变竴绶ㄧ⒓钀湅纰�
diff --git a/unittest/unittest.mm b/unittest/unittest.mm
index ee543766..b125231f 100644
--- a/unittest/unittest.mm
+++ b/unittest/unittest.mm
@@ -86,6 +86,7 @@
     NSString * _parserPath;
     NSString * _builderOutputPath;
     NSString * _parserOutputPath;
+    NSString * _charsetDetectionPath;
 }
 
 - (void)setUp {
@@ -96,6 +97,7 @@
     _builderOutputPath = [_mainPath stringByAppendingPathComponent:@"builder/output"];
     _parserPath = [_mainPath stringByAppendingPathComponent:@"parser/input"];
     _parserOutputPath = [_mainPath stringByAppendingPathComponent:@"parser/output"];
+    _charsetDetectionPath = [_mainPath stringByAppendingPathComponent:@"charset-detection"];
 }
 
 - (void)tearDown {
@@ -192,4 +194,20 @@
     }
 }
 
+- (void)testCharsetDetection {
+    NSArray * list = [[NSFileManager defaultManager] subpathsAtPath:_charsetDetectionPath];
+    for(NSString * name in list) {
+        NSString * path = [_charsetDetectionPath stringByAppendingPathComponent:name];
+        BOOL isDirectory = NO;
+        [[NSFileManager defaultManager] fileExistsAtPath:path isDirectory:&isDirectory];
+        if (isDirectory) {
+            continue;
+        }
+        NSData * data = [NSData dataWithContentsOfFile:path];
+        NSString * charset = MCO_TO_OBJC([data mco_mcData]->charsetWithFilteredHTML(false));
+        charset = [charset lowercaseString];
+        XCTAssertEqualObjects([[name lastPathComponent] stringByDeletingPathExtension], charset);
+    }
+}
+
 @end