relikd/RSXML2
1
0
Fork 0
Code Issues Pull Requests Releases Activity

Added all bugfixes from current state of RSParser. Except HTML entities are still decoded for title, abstract and body.

Browse Source
This commit is contained in:
relikd
2019-01-12 23:28:48 +01:00
parent 2732573003
commit ca9452bc18
5 changed files with 27 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
RSXML/NSString+RSXML.h
View File

@@ -26,7 +26,7 @@
@interface NSString (RSXML) @interface NSString (RSXML)
- (NSString *)rs_stringByDecodingHTMLEntities; - (NSString *)rsxml_stringByDecodingHTMLEntities;
- (nonnull NSString *)rsxml_md5HashString; - (nonnull NSString *)rsxml_md5HashString;
- (nullable NSString *)absoluteURLWithBase:(nonnull NSURL *)baseURL; - (nullable NSString *)absoluteURLWithBase:(nonnull NSURL *)baseURL;

4
RSXML/NSString+RSXML.m
View File

@@ -61,7 +61,7 @@
return self; return self;
} }
- (NSString *)rs_stringByDecodingHTMLEntities { - (NSString *)rsxml_stringByDecodingHTMLEntities {
@autoreleasepool { @autoreleasepool {
@@ -315,7 +315,7 @@ static NSDictionary *RSEntitiesDictionary(void) {
@"#255": @"ÿ", @"#255": @"ÿ",
@"#32": @" ", @"#32": @" ",
@"#34": @"\"", @"#34": @"\"",
@"#39": @"", @"#39": @"'",
@"#8194": @" ", @"#8194": @" ",
@"#8195": @" ", @"#8195": @" ",
@"#8211": @"-", @"#8211": @"-",

18
RSXML/RSAtomParser.m
View File

@@ -223,12 +223,8 @@ static NSString *kRelatedValue = @"related";
else if (EqualBytes(localName, "source", 6)) { else if (EqualBytes(localName, "source", 6)) {
self.parsingSource = NO; self.parsingSource = NO;
} }
return; else if (isArticle && EqualBytes(localName, "issued", 6)) { // Atom 0.3 date
case 8: self.currentArticle.datePublished = [self dateFromCharacters:SAXParser.currentCharacters];;
if (!self.parsingArticle && !self.parsingSource && self.parsedFeed.subtitle.length == 0) {
if (EqualBytes(localName, "subtitle", 8)) {
self.parsedFeed.subtitle = SAXParser.currentStringWithTrimmedWhitespace;
}
} }
return; return;
case 7: case 7:
@@ -244,6 +240,16 @@ static NSString *kRelatedValue = @"related";
} }
} }
return; return;
case 8:
if (!self.parsingArticle && !self.parsingSource && self.parsedFeed.subtitle.length == 0) {
if (EqualBytes(localName, "subtitle", 8)) {
self.parsedFeed.subtitle = SAXParser.currentStringWithTrimmedWhitespace;
}
}
else if (isArticle && EqualBytes(localName, "modified", 8)) { // Atom 0.3 date
self.currentArticle.dateModified = [self dateFromCharacters:SAXParser.currentCharacters];;
}
return;
case 9: case 9:
if (isArticle && EqualBytes(localName, "published", 9)) { if (isArticle && EqualBytes(localName, "published", 9)) {
self.currentArticle.datePublished = [self dateFromCharacters:SAXParser.currentCharacters]; self.currentArticle.datePublished = [self dateFromCharacters:SAXParser.currentCharacters];

2
RSXML/RSFeedParser.m
View File

@@ -52,7 +52,7 @@
/// @return currentString by removing HTML encoded entities. /// @return currentString by removing HTML encoded entities.
- (NSString *)decodeHTMLEntities:(NSString *)str { - (NSString *)decodeHTMLEntities:(NSString *)str {
return [str rs_stringByDecodingHTMLEntities]; return [str rsxml_stringByDecodingHTMLEntities];
} }
@end @end

25
RSXMLTests/RSEntityTests.m
View File

@@ -58,37 +58,32 @@
} }
- (void)testNotEntities { - (void)testNotEntities {
NSString *s = @"&&\t\nFoo & Bar &0; Baz & 1238 4948 More things &foobar;&"; NSString *s = @"&&\t\nFoo & Bar &0; Baz & 1238 4948 More things &foobar;&";
NSString *result = [s rs_stringByDecodingHTMLEntities]; XCTAssertEqualObjects([s rs_stringByDecodingHTMLEntities], s);
XCTAssertEqualObjects(result, s);
} }
- (void)testURLs { - (void)testURLs {
NSString *s = @"http://www.nytimes.com/2015/09/05/us/at-west-point-annual-pillow-fight-becomes-weaponized.html?mwrsm=Email&_r=1&pagewanted=all";
NSString *urlString = @"http://www.nytimes.com/2015/09/05/us/at-west-point-annual-pillow-fight-becomes-weaponized.html?mwrsm=Email&_r=1&pagewanted=all";
NSString *expectedResult = @"http://www.nytimes.com/2015/09/05/us/at-west-point-annual-pillow-fight-becomes-weaponized.html?mwrsm=Email&_r=1&pagewanted=all"; NSString *expectedResult = @"http://www.nytimes.com/2015/09/05/us/at-west-point-annual-pillow-fight-becomes-weaponized.html?mwrsm=Email&_r=1&pagewanted=all";
XCTAssertEqualObjects([s rs_stringByDecodingHTMLEntities], expectedResult);
NSString *result = [urlString rs_stringByDecodingHTMLEntities];
XCTAssertEqualObjects(result, expectedResult);
} }
- (void)testEntityPlusWhitespace { - (void)testEntityPlusWhitespace {
NSString *s = @"∞ Permalink"; NSString *s = @"∞ Permalink";
NSString *expectedResult = @"∞ Permalink"; NSString *expectedResult = @"∞ Permalink";
XCTAssertEqualObjects([s rs_stringByDecodingHTMLEntities], expectedResult);
NSString *result = [s rs_stringByDecodingHTMLEntities];
XCTAssertEqualObjects(result, expectedResult);
} }
- (void)testNonBreakingSpace { - (void)testNonBreakingSpace {
NSString *s = @"   -- just some spaces"; NSString *s = @"   -- just some spaces";
NSString *expectedResult = [NSString stringWithFormat:@"%C%C -- just some spaces", 160, 160]; NSString *expectedResult = [NSString stringWithFormat:@"%C%C -- just some spaces", 160, 160];
XCTAssertEqualObjects([s rs_stringByDecodingHTMLEntities], expectedResult);
}
NSString *result = [s rs_stringByDecodingHTMLEntities]; - (void)test39encoding {
XCTAssertEqualObjects(result, expectedResult); NSString *s = @"These are the times that try men's souls.";
NSString *expectedResult = @"These are the times that try men's souls.";
XCTAssertEqualObjects([s rs_stringByDecodingHTMLEntities], expectedResult);
} }
@end @end