Start work on turning RSXML.framework into RSParser.framework.

This commit is contained in:
Brent Simmons 2017-06-20 21:18:46 -07:00
parent cf77f19276
commit b00b2f39d7
47 changed files with 5241 additions and 0 deletions

View File

@ -0,0 +1,24 @@
//
// FeedParser.h
// RSXML
//
// Created by Brent Simmons on 7/12/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSParsedFeed;
@class RSXMLData;
@protocol FeedParser <NSObject>
+ (BOOL)canParseFeed:(RSXMLData * _Nonnull)xmlData;
- (nonnull instancetype)initWithXMLData:(RSXMLData * _Nonnull)xmlData;
- (nullable RSParsedFeed *)parseFeed:(NSError * _Nullable * _Nullable)error;
@end

View File

@ -0,0 +1,28 @@
//
// RSFeedParser.h
// RSXML
//
// Created by Brent Simmons on 1/4/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import "FeedParser.h"
// If you have a feed and dont know or care what it is (RSS or Atom),
// then call RSParseFeed or RSParseFeedSync.
@class RSXMLData;
@class RSParsedFeed;
NS_ASSUME_NONNULL_BEGIN
BOOL RSCanParseFeed(RSXMLData *xmlData);
typedef void (^RSParsedFeedBlock)(RSParsedFeed * _Nullable parsedFeed, NSError * _Nullable error);
// callback is called on main queue.
void RSParseFeed(RSXMLData *xmlData, RSParsedFeedBlock callback);
RSParsedFeed * _Nullable RSParseFeedSync(RSXMLData *xmlData, NSError * _Nullable * _Nullable error);
NS_ASSUME_NONNULL_END

View File

@ -0,0 +1,216 @@
//
// FeedParser.m
// RSXML
//
// Created by Brent Simmons on 1/4/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import "RSFeedParser.h"
#import "FeedParser.h"
#import "RSXMLData.h"
#import "RSRSSParser.h"
#import "RSAtomParser.h"
static NSArray *parserClasses(void) {
static NSArray *gParserClasses = nil;
static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^{
gParserClasses = @[[RSRSSParser class], [RSAtomParser class]];
});
return gParserClasses;
}
static BOOL feedMayBeParseable(RSXMLData *xmlData) {
/*Sanity checks.*/
if (!xmlData.data) {
return NO;
}
/*TODO: check size, type, etc.*/
return YES;
}
static BOOL optimisticCanParseRSSData(const char *bytes, NSUInteger numberOfBytes);
static BOOL optimisticCanParseAtomData(const char *bytes, NSUInteger numberOfBytes);
static BOOL optimisticCanParseRDF(const char *bytes, NSUInteger numberOfBytes);
static BOOL dataIsProbablyHTML(const char *bytes, NSUInteger numberOfBytes);
static BOOL dataIsSomeWeirdException(const char *bytes, NSUInteger numberOfBytes);
static BOOL dataHasLeftCaret(const char *bytes, NSUInteger numberOfBytes);
static const NSUInteger maxNumberOfBytesToSearch = 4096;
static const NSUInteger minNumberOfBytesToSearch = 20;
static Class parserClassForXMLData(RSXMLData *xmlData) {
if (!feedMayBeParseable(xmlData)) {
return nil;
}
// TODO: check for things like images and movies and return nil.
const char *bytes = xmlData.data.bytes;
NSUInteger numberOfBytes = xmlData.data.length;
if (numberOfBytes > minNumberOfBytesToSearch) {
if (numberOfBytes > maxNumberOfBytesToSearch) {
numberOfBytes = maxNumberOfBytesToSearch;
}
if (!dataHasLeftCaret(bytes, numberOfBytes)) {
return nil;
}
if (optimisticCanParseRSSData(bytes, numberOfBytes)) {
return [RSRSSParser class];
}
if (optimisticCanParseAtomData(bytes, numberOfBytes)) {
return [RSAtomParser class];
}
if (optimisticCanParseRDF(bytes, numberOfBytes)) {
return nil; //TODO: parse RDF feeds
}
if (dataIsProbablyHTML(bytes, numberOfBytes)) {
return nil;
}
if (dataIsSomeWeirdException(bytes, numberOfBytes)) {
return nil;
}
}
for (Class parserClass in parserClasses()) {
if ([parserClass canParseFeed:xmlData]) {
return [[parserClass alloc] initWithXMLData:xmlData];
}
}
return nil;
}
static id<FeedParser> parserForXMLData(RSXMLData *xmlData) {
Class parserClass = parserClassForXMLData(xmlData);
if (!parserClass) {
return nil;
}
return [[parserClass alloc] initWithXMLData:xmlData];
}
static BOOL canParseXMLData(RSXMLData *xmlData) {
return parserClassForXMLData(xmlData) != nil;
}
static BOOL didFindString(const char *string, const char *bytes, NSUInteger numberOfBytes) {
char *foundString = strnstr(bytes, string, numberOfBytes);
return foundString != NULL;
}
static BOOL dataHasLeftCaret(const char *bytes, NSUInteger numberOfBytes) {
return didFindString("<", bytes, numberOfBytes);
}
static BOOL dataIsProbablyHTML(const char *bytes, NSUInteger numberOfBytes) {
// Wont catch every single case, which is fine.
if (didFindString("<html", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("<body", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("doctype html", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("DOCTYPE html", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("DOCTYPE HTML", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("<meta", bytes, numberOfBytes)) {
return YES;
}
if (didFindString("<HTML", bytes, numberOfBytes)) {
return YES;
}
return NO;
}
static BOOL dataIsSomeWeirdException(const char *bytes, NSUInteger numberOfBytes) {
if (didFindString("<errors xmlns='http://schemas.google", bytes, numberOfBytes)) {
return YES;
}
return NO;
}
static BOOL optimisticCanParseRDF(const char *bytes, NSUInteger numberOfBytes) {
return didFindString("<rdf:RDF", bytes, numberOfBytes);
}
static BOOL optimisticCanParseRSSData(const char *bytes, NSUInteger numberOfBytes) {
if (!didFindString("<rss", bytes, numberOfBytes)) {
return NO;
}
return didFindString("<channel", bytes, numberOfBytes);
}
static BOOL optimisticCanParseAtomData(const char *bytes, NSUInteger numberOfBytes) {
return didFindString("<feed", bytes, numberOfBytes);
}
static void callCallback(RSParsedFeedBlock callback, RSParsedFeed *parsedFeed, NSError *error) {
dispatch_async(dispatch_get_main_queue(), ^{
@autoreleasepool {
if (callback) {
callback(parsedFeed, error);
}
}
});
}
#pragma mark - API
BOOL RSCanParseFeed(RSXMLData *xmlData) {
return canParseXMLData(xmlData);
}
void RSParseFeed(RSXMLData *xmlData, RSParsedFeedBlock callback) {
dispatch_async(dispatch_get_global_queue(QOS_CLASS_UTILITY, 0), ^{
NSError *error = nil;
RSParsedFeed *parsedFeed = RSParseFeedSync(xmlData, &error);
callCallback(callback, parsedFeed, error);
});
}
RSParsedFeed *RSParseFeedSync(RSXMLData *xmlData, NSError **error) {
id<FeedParser> parser = parserForXMLData(xmlData);
return [parser parseFeed:error];
}

View File

@ -0,0 +1,32 @@
//
// RSParsedArticle.h
// RSXML
//
// Created by Brent Simmons on 12/6/14.
// Copyright (c) 2014 Ranchero Software LLC. All rights reserved.
//
@import Foundation;
@interface RSParsedArticle : NSObject
- (nonnull instancetype)initWithFeedURL:(NSString * _Nonnull)feedURL;
@property (nonatomic, readonly, nonnull) NSString *feedURL;
@property (nonatomic, nonnull) NSString *articleID; //Calculated. Don't get until other properties have been set.
@property (nonatomic, nullable) NSString *guid;
@property (nonatomic, nullable) NSString *title;
@property (nonatomic, nullable) NSString *body;
@property (nonatomic, nullable) NSString *link;
@property (nonatomic, nullable) NSString *permalink;
@property (nonatomic, nullable) NSString *author;
@property (nonatomic, nullable) NSDate *datePublished;
@property (nonatomic, nullable) NSDate *dateModified;
@property (nonatomic, nonnull) NSDate *dateParsed;
- (void)calculateArticleID; // Optimization. Call after all properties have been set. Call on a background thread.
@end

View File

@ -0,0 +1,101 @@
//
// RSParsedArticle.m
// RSXML
//
// Created by Brent Simmons on 12/6/14.
// Copyright (c) 2014 Ranchero Software LLC. All rights reserved.
//
#import "RSParsedArticle.h"
#import "RSXMLInternal.h"
@implementation RSParsedArticle
#pragma mark - Init
- (instancetype)initWithFeedURL:(NSString *)feedURL {
NSParameterAssert(feedURL != nil);
self = [super init];
if (!self) {
return nil;
}
_feedURL = feedURL;
_dateParsed = [NSDate date];
return self;
}
#pragma mark - Accessors
- (NSString *)articleID {
if (!_articleID) {
_articleID = self.calculatedUniqueID;
}
return _articleID;
}
- (NSString *)calculatedUniqueID {
/*guid+feedID, or a combination of properties when no guid. Then hash the result.
In general, feeds should have guids. When they don't, re-runs are very likely,
because there's no other 100% reliable way to determine identity.*/
NSMutableString *s = [NSMutableString stringWithString:@""];
NSString *datePublishedTimeStampString = nil;
if (self.datePublished) {
datePublishedTimeStampString = [NSString stringWithFormat:@"%.0f", self.datePublished.timeIntervalSince1970];
}
if (!RSXMLStringIsEmpty(self.guid)) {
[s appendString:self.guid];
}
else if (!RSXMLStringIsEmpty(self.link) && self.datePublished != nil) {
[s appendString:self.link];
[s appendString:datePublishedTimeStampString];
}
else if (!RSXMLStringIsEmpty(self.title) && self.datePublished != nil) {
[s appendString:self.title];
[s appendString:datePublishedTimeStampString];
}
else if (self.datePublished != nil) {
[s appendString:datePublishedTimeStampString];
}
else if (!RSXMLStringIsEmpty(self.link)) {
[s appendString:self.link];
}
else if (!RSXMLStringIsEmpty(self.title)) {
[s appendString:self.title];
}
else if (!RSXMLStringIsEmpty(self.body)) {
[s appendString:self.body];
}
NSAssert(!RSXMLStringIsEmpty(self.feedURL), nil);
[s appendString:self.feedURL];
return [s rsxml_md5HashString];
}
- (void)calculateArticleID {
(void)self.articleID;
}
@end

View File

@ -0,0 +1,22 @@
//
// RSParsedFeed.h
// RSXML
//
// Created by Brent Simmons on 7/12/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSParsedArticle;
@interface RSParsedFeed : NSObject
- (nonnull instancetype)initWithURLString:(NSString * _Nonnull)urlString title:(NSString * _Nullable)title link:(NSString * _Nullable)link articles:(NSArray <RSParsedArticle *>* _Nonnull)articles;
@property (nonatomic, readonly, nonnull) NSString *urlString;
@property (nonatomic, readonly, nullable) NSString *title;
@property (nonatomic, readonly, nullable) NSString *link;
@property (nonatomic, readonly, nonnull) NSSet <RSParsedArticle *>*articles;
@end

View File

@ -0,0 +1,29 @@
//
// RSParsedFeed.m
// RSXML
//
// Created by Brent Simmons on 7/12/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
#import "RSParsedFeed.h"
@implementation RSParsedFeed
- (instancetype)initWithURLString:(NSString *)urlString title:(NSString *)title link:(NSString *)link articles:(NSSet *)articles {
self = [super init];
if (!self) {
return nil;
}
_urlString = urlString;
_title = title;
_link = link;
_articles = articles;
return self;
}
@end

View File

@ -0,0 +1,31 @@
//
// RSHTMLLinkParser.h
// RSXML
//
// Created by Brent Simmons on 8/7/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
/*Returns all <a href="some_url">some_text</a> as RSHTMLLink object array.*/
@class RSXMLData;
@class RSHTMLLink;
@interface RSHTMLLinkParser : NSObject
+ (NSArray <RSHTMLLink *> *)htmlLinksWithData:(RSXMLData *)xmlData;
@end
@interface RSHTMLLink : NSObject
// Any of these, even urlString, may be nil, because HTML can be bad.
@property (nonatomic, readonly) NSString *urlString; //absolute
@property (nonatomic, readonly) NSString *text;
@property (nonatomic, readonly) NSString *title; //title attribute inside anchor tag
@end

View File

@ -0,0 +1,151 @@
//
// RSHTMLLinkParser.m
// RSXML
//
// Created by Brent Simmons on 8/7/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import <libxml/xmlstring.h>
#import "RSHTMLLinkParser.h"
#import "RSSAXHTMLParser.h"
#import "RSSAXParser.h"
#import "RSXMLData.h"
#import "RSXMLInternal.h"
@interface RSHTMLLinkParser() <RSSAXHTMLParserDelegate>
@property (nonatomic, readonly) NSMutableArray *links;
@property (nonatomic, readonly) RSXMLData *xmlData;
@property (nonatomic, readonly) NSMutableArray *dictionaries;
@property (nonatomic, readonly) NSURL *baseURL;
@end
@interface RSHTMLLink()
@property (nonatomic, readwrite) NSString *urlString; //absolute
@property (nonatomic, readwrite) NSString *text;
@property (nonatomic, readwrite) NSString *title; //title attribute inside anchor tag
@end
@implementation RSHTMLLinkParser
#pragma mark - Class Methods
+ (NSArray *)htmlLinksWithData:(RSXMLData *)xmlData {
RSHTMLLinkParser *parser = [[self alloc] initWithXMLData:xmlData];
return parser.links;
}
#pragma mark - Init
- (instancetype)initWithXMLData:(RSXMLData *)xmlData {
NSParameterAssert(xmlData.data);
NSParameterAssert(xmlData.urlString);
self = [super init];
if (!self) {
return nil;
}
_links = [NSMutableArray new];
_xmlData = xmlData;
_dictionaries = [NSMutableArray new];
_baseURL = [NSURL URLWithString:xmlData.urlString];
[self parse];
return self;
}
#pragma mark - Parse
- (void)parse {
RSSAXHTMLParser *parser = [[RSSAXHTMLParser alloc] initWithDelegate:self];
[parser parseData:self.xmlData.data];
[parser finishParsing];
}
- (RSHTMLLink *)currentLink {
return self.links.lastObject;
}
static NSString *kHrefKey = @"href";
- (NSString *)urlStringFromDictionary:(NSDictionary *)d {
NSString *href = [d rsxml_objectForCaseInsensitiveKey:kHrefKey];
if (!href) {
return nil;
}
NSURL *absoluteURL = [NSURL URLWithString:href relativeToURL:self.baseURL];
return absoluteURL.absoluteString;
}
static NSString *kTitleKey = @"title";
- (NSString *)titleFromDictionary:(NSDictionary *)d {
return [d rsxml_objectForCaseInsensitiveKey:kTitleKey];
}
- (void)handleLinkAttributes:(NSDictionary *)d {
RSHTMLLink *link = self.currentLink;
link.urlString = [self urlStringFromDictionary:d];
link.title = [self titleFromDictionary:d];
}
static const char *kAnchor = "a";
static const NSInteger kAnchorLength = 2;
- (void)saxParser:(RSSAXHTMLParser *)SAXParser XMLStartElement:(const xmlChar *)localName attributes:(const xmlChar **)attributes {
if (!RSSAXEqualTags(localName, kAnchor, kAnchorLength)) {
return;
}
RSHTMLLink *link = [RSHTMLLink new];
[self.links addObject:link];
NSDictionary *d = [SAXParser attributesDictionary:attributes];
if (!RSXMLIsEmpty(d)) {
[self handleLinkAttributes:d];
}
[SAXParser beginStoringCharacters];
}
- (void)saxParser:(RSSAXParser *)SAXParser XMLEndElement:(const xmlChar *)localName {
if (!RSSAXEqualTags(localName, kAnchor, kAnchorLength)) {
return;
}
self.currentLink.text = SAXParser.currentStringWithTrimmedWhitespace;
}
@end
@implementation RSHTMLLink
@end

View File

@ -0,0 +1,45 @@
//
// RSHTMLMetadata.h
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSHTMLMetadataFeedLink;
@class RSHTMLMetadataAppleTouchIcon;
@interface RSHTMLMetadata : NSObject
- (instancetype)initWithURLString:(NSString *)urlString dictionaries:(NSArray <NSDictionary *> *)dictionaries;
@property (nonatomic, readonly) NSString *baseURLString;
@property (nonatomic, readonly) NSArray <NSDictionary *> *dictionaries;
@property (nonatomic, readonly) NSString *faviconLink;
@property (nonatomic, readonly) NSArray <RSHTMLMetadataAppleTouchIcon *> *appleTouchIcons;
@property (nonatomic, readonly) NSArray <RSHTMLMetadataFeedLink *> *feedLinks;
@end
@interface RSHTMLMetadataAppleTouchIcon : NSObject
@property (nonatomic, readonly) NSString *rel;
@property (nonatomic, readonly) NSString *sizes;
@property (nonatomic, readonly) NSString *urlString; // Absolute.
@end
@interface RSHTMLMetadataFeedLink : NSObject
@property (nonatomic, readonly) NSString *title;
@property (nonatomic, readonly) NSString *type;
@property (nonatomic, readonly) NSString *urlString; // Absolute.
@end

View File

@ -0,0 +1,245 @@
//
// RSHTMLMetadata.m
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSHTMLMetadata.h"
#import "RSXMLInternal.h"
static NSString *urlStringFromDictionary(NSDictionary *d);
static NSString *absoluteURLStringWithRelativeURLString(NSString *relativeURLString, NSString *baseURLString);
static NSString *absoluteURLStringWithDictionary(NSDictionary *d, NSString *baseURLString);
static NSArray *objectsOfClassWithDictionaries(Class class, NSArray *dictionaries, NSString *baseURLString);
static NSString *relValue(NSDictionary *d);
static BOOL typeIsFeedType(NSString *type);
static NSString *kShortcutIconRelValue = @"shortcut icon";
static NSString *kHrefKey = @"href";
static NSString *kSrcKey = @"src";
static NSString *kAppleTouchIconValue = @"apple-touch-icon";
static NSString *kAppleTouchIconPrecomposedValue = @"apple-touch-icon-precomposed";
static NSString *kSizesKey = @"sizes";
static NSString *kTitleKey = @"title";
static NSString *kRelKey = @"rel";
static NSString *kAlternateKey = @"alternate";
static NSString *kRSSSuffix = @"/rss+xml";
static NSString *kAtomSuffix = @"/atom+xml";
static NSString *kTypeKey = @"type";
@interface RSHTMLMetadataAppleTouchIcon ()
- (instancetype)initWithDictionary:(NSDictionary *)d baseURLString:(NSString *)baseURLString;
@end
@interface RSHTMLMetadataFeedLink ()
- (instancetype)initWithDictionary:(NSDictionary *)d baseURLString:(NSString *)baseURLString;
@end
@implementation RSHTMLMetadata
#pragma mark - Init
- (instancetype)initWithURLString:(NSString *)urlString dictionaries:(NSArray <NSDictionary *> *)dictionaries {
self = [super init];
if (!self) {
return nil;
}
_baseURLString = urlString;
_dictionaries = dictionaries;
_faviconLink = [self resolvedLinkFromFirstDictionaryWithMatchingRel:kShortcutIconRelValue];
NSArray *appleTouchIconDictionaries = [self appleTouchIconDictionaries];
_appleTouchIcons = objectsOfClassWithDictionaries([RSHTMLMetadataAppleTouchIcon class], appleTouchIconDictionaries, urlString);
NSArray *feedLinkDictionaries = [self feedLinkDictionaries];
_feedLinks = objectsOfClassWithDictionaries([RSHTMLMetadataFeedLink class], feedLinkDictionaries, urlString);
return self;
}
#pragma mark - Private
- (NSDictionary *)firstDictionaryWithMatchingRel:(NSString *)valueToMatch {
// Case-insensitive.
for (NSDictionary *oneDictionary in self.dictionaries) {
NSString *oneRelValue = relValue(oneDictionary);
if (oneRelValue && [oneRelValue compare:valueToMatch options:NSCaseInsensitiveSearch] == NSOrderedSame) {
return oneDictionary;
}
}
return nil;
}
- (NSArray *)appleTouchIconDictionaries {
NSMutableArray *dictionaries = [NSMutableArray new];
for (NSDictionary *oneDictionary in self.dictionaries) {
NSString *oneRelValue = relValue(oneDictionary).lowercaseString;
if ([oneRelValue isEqualToString:kAppleTouchIconValue] || [oneRelValue isEqualToString:kAppleTouchIconPrecomposedValue]) {
[dictionaries addObject:oneDictionary];
}
}
return dictionaries;
}
- (NSArray *)feedLinkDictionaries {
NSMutableArray *dictionaries = [NSMutableArray new];
for (NSDictionary *oneDictionary in self.dictionaries) {
NSString *oneRelValue = relValue(oneDictionary).lowercaseString;
if (![oneRelValue isEqualToString:kAlternateKey]) {
continue;
}
NSString *oneType = [oneDictionary rsxml_objectForCaseInsensitiveKey:kTypeKey];
if (!typeIsFeedType(oneType)) {
continue;
}
if (RSXMLStringIsEmpty(urlStringFromDictionary(oneDictionary))) {
continue;
}
[dictionaries addObject:oneDictionary];
}
return dictionaries;
}
- (NSString *)resolvedLinkFromFirstDictionaryWithMatchingRel:(NSString *)relValue {
NSDictionary *d = [self firstDictionaryWithMatchingRel:relValue];
return absoluteURLStringWithDictionary(d, self.baseURLString);
}
@end
static NSString *relValue(NSDictionary *d) {
return [d rsxml_objectForCaseInsensitiveKey:kRelKey];
}
static NSString *urlStringFromDictionary(NSDictionary *d) {
NSString *urlString = [d rsxml_objectForCaseInsensitiveKey:kHrefKey];
if (urlString) {
return urlString;
}
return [d rsxml_objectForCaseInsensitiveKey:kSrcKey];
}
static NSString *absoluteURLStringWithRelativeURLString(NSString *relativeURLString, NSString *baseURLString) {
NSURL *url = [NSURL URLWithString:baseURLString];
if (!url) {
return nil;
}
NSURL *absoluteURL = [NSURL URLWithString:relativeURLString relativeToURL:url];
return absoluteURL.absoluteString;
}
static NSString *absoluteURLStringWithDictionary(NSDictionary *d, NSString *baseURLString) {
NSString *urlString = urlStringFromDictionary(d);
if (RSXMLStringIsEmpty(urlString)) {
return nil;
}
return absoluteURLStringWithRelativeURLString(urlString, baseURLString);
}
static NSArray *objectsOfClassWithDictionaries(Class class, NSArray *dictionaries, NSString *baseURLString) {
NSMutableArray *objects = [NSMutableArray new];
for (NSDictionary *oneDictionary in dictionaries) {
id oneObject = [[class alloc] initWithDictionary:oneDictionary baseURLString:baseURLString];
if (oneObject) {
[objects addObject:oneObject];
}
}
return [objects copy];
}
static BOOL typeIsFeedType(NSString *type) {
type = type.lowercaseString;
return [type hasSuffix:kRSSSuffix] || [type hasSuffix:kAtomSuffix];
}
@implementation RSHTMLMetadataAppleTouchIcon
- (instancetype)initWithDictionary:(NSDictionary *)d baseURLString:(NSString *)baseURLString {
self = [super init];
if (!self) {
return nil;
}
_urlString = absoluteURLStringWithDictionary(d, baseURLString);
_sizes = [d rsxml_objectForCaseInsensitiveKey:kSizesKey];
_rel = [d rsxml_objectForCaseInsensitiveKey:kRelKey];
return self;
}
@end
@implementation RSHTMLMetadataFeedLink
- (instancetype)initWithDictionary:(NSDictionary *)d baseURLString:(NSString *)baseURLString {
self = [super init];
if (!self) {
return nil;
}
_urlString = absoluteURLStringWithDictionary(d, baseURLString);
_title = [d rsxml_objectForCaseInsensitiveKey:kTitleKey];
_type = [d rsxml_objectForCaseInsensitiveKey:kTypeKey];
return self;
}
@end

View File

@ -0,0 +1,28 @@
//
// RSHTMLMetadataParser.h
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSHTMLMetadata;
@class RSXMLData;
NS_ASSUME_NONNULL_BEGIN
@interface RSHTMLMetadataParser : NSObject
+ (RSHTMLMetadata *)HTMLMetadataWithXMLData:(RSXMLData *)xmlData;
- (instancetype)initWithXMLData:(RSXMLData *)xmlData;
@property (nonatomic, readonly) RSHTMLMetadata *metadata;
@end
NS_ASSUME_NONNULL_END

View File

@ -0,0 +1,129 @@
//
// RSHTMLMetadataParser.m
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import <libxml/xmlstring.h>
#import "RSHTMLMetadataParser.h"
#import "RSXMLData.h"
#import "RSHTMLMetadata.h"
#import "RSSAXHTMLParser.h"
#import "RSSAXParser.h"
#import "RSXMLInternal.h"
@interface RSHTMLMetadataParser () <RSSAXHTMLParserDelegate>
@property (nonatomic, readonly) RSXMLData *xmlData;
@property (nonatomic, readwrite) RSHTMLMetadata *metadata;
@property (nonatomic) NSMutableArray *dictionaries;
@property (nonatomic) BOOL didFinishParsing;
@end
@implementation RSHTMLMetadataParser
#pragma mark - Class Methods
+ (RSHTMLMetadata *)HTMLMetadataWithXMLData:(RSXMLData *)xmlData {
RSHTMLMetadataParser *parser = [[self alloc] initWithXMLData:xmlData];
return parser.metadata;
}
#pragma mark - Init
- (instancetype)initWithXMLData:(RSXMLData *)xmlData {
NSParameterAssert(xmlData.data);
NSParameterAssert(xmlData.urlString);
self = [super init];
if (!self) {
return nil;
}
_xmlData = xmlData;
_dictionaries = [NSMutableArray new];
[self parse];
return self;
}
#pragma mark - Parse
- (void)parse {
RSSAXHTMLParser *parser = [[RSSAXHTMLParser alloc] initWithDelegate:self];
[parser parseData:self.xmlData.data];
[parser finishParsing];
self.metadata = [[RSHTMLMetadata alloc] initWithURLString:self.xmlData.urlString dictionaries:[self.dictionaries copy]];
}
static NSString *kHrefKey = @"href";
static NSString *kSrcKey = @"src";
static NSString *kRelKey = @"rel";
- (NSString *)linkForDictionary:(NSDictionary *)d {
NSString *link = [d rsxml_objectForCaseInsensitiveKey:kHrefKey];
if (link) {
return link;
}
return [d rsxml_objectForCaseInsensitiveKey:kSrcKey];
}
- (void)handleLinkAttributes:(NSDictionary *)d {
if (RSXMLStringIsEmpty([d rsxml_objectForCaseInsensitiveKey:kRelKey])) {
return;
}
if (RSXMLStringIsEmpty([self linkForDictionary:d])) {
return;
}
[self.dictionaries addObject:d];
}
#pragma mark - RSSAXHTMLParserDelegate
static const char *kBody = "body";
static const NSInteger kBodyLength = 5;
static const char *kLink = "link";
static const NSInteger kLinkLength = 5;
- (void)saxParser:(RSSAXHTMLParser *)SAXParser XMLStartElement:(const xmlChar *)localName attributes:(const xmlChar **)attributes {
if (self.didFinishParsing) {
return;
}
if (RSSAXEqualTags(localName, kBody, kBodyLength)) {
self.didFinishParsing = YES;
return;
}
if (!RSSAXEqualTags(localName, kLink, kLinkLength)) {
return;
}
NSDictionary *d = [SAXParser attributesDictionary:attributes];
if (!RSXMLIsEmpty(d)) {
[self handleLinkAttributes:d];
}
}
@end

View File

@ -0,0 +1,49 @@
//
// RSSAXHTMLParser.h
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSSAXHTMLParser;
@protocol RSSAXHTMLParserDelegate <NSObject>
@optional
- (void)saxParser:(RSSAXHTMLParser *)SAXParser XMLStartElement:(const unsigned char *)localName attributes:(const unsigned char **)attributes;
- (void)saxParser:(RSSAXHTMLParser *)SAXParser XMLEndElement:(const unsigned char *)localName;
- (void)saxParser:(RSSAXHTMLParser *)SAXParser XMLCharactersFound:(const unsigned char *)characters length:(NSUInteger)length;
- (void)saxParserDidReachEndOfDocument:(RSSAXHTMLParser *)SAXParser; // If canceled, may not get called (but might).
@end
@interface RSSAXHTMLParser : NSObject
- (instancetype)initWithDelegate:(id<RSSAXHTMLParserDelegate>)delegate;
- (void)parseData:(NSData *)data;
- (void)parseBytes:(const void *)bytes numberOfBytes:(NSUInteger)numberOfBytes;
- (void)finishParsing;
- (void)cancel;
@property (nonatomic, strong, readonly) NSData *currentCharacters; // nil if not storing characters. UTF-8 encoded.
@property (nonatomic, strong, readonly) NSString *currentString; // Convenience to get string version of currentCharacters.
@property (nonatomic, strong, readonly) NSString *currentStringWithTrimmedWhitespace;
- (void)beginStoringCharacters; // Delegate can call from XMLStartElement. Characters will be available in XMLEndElement as currentCharacters property. Storing characters is stopped after each XMLEndElement.
// Delegate can call from within XMLStartElement.
- (NSDictionary *)attributesDictionary:(const unsigned char **)attributes;
@end

View File

@ -0,0 +1,315 @@
//
// RSSAXHTMLParser.m
// RSXML
//
// Created by Brent Simmons on 3/6/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSSAXHTMLParser.h"
#import "RSSAXParser.h"
#import <libxml/tree.h>
#import <libxml/xmlstring.h>
#import <libxml/HTMLparser.h>
#import "RSXMLInternal.h"
@interface RSSAXHTMLParser ()
@property (nonatomic) id<RSSAXHTMLParserDelegate> delegate;
@property (nonatomic, assign) htmlParserCtxtPtr context;
@property (nonatomic, assign) BOOL storingCharacters;
@property (nonatomic) NSMutableData *characters;
@property (nonatomic) BOOL delegateRespondsToStartElementMethod;
@property (nonatomic) BOOL delegateRespondsToEndElementMethod;
@property (nonatomic) BOOL delegateRespondsToCharactersFoundMethod;
@property (nonatomic) BOOL delegateRespondsToEndOfDocumentMethod;
@end
@implementation RSSAXHTMLParser
+ (void)initialize {
RSSAXInitLibXMLParser();
}
#pragma mark - Init
- (instancetype)initWithDelegate:(id<RSSAXHTMLParserDelegate>)delegate {
self = [super init];
if (self == nil)
return nil;
_delegate = delegate;
if ([_delegate respondsToSelector:@selector(saxParser:XMLStartElement:attributes:)]) {
_delegateRespondsToStartElementMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:XMLEndElement:)]) {
_delegateRespondsToEndElementMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:XMLCharactersFound:length:)]) {
_delegateRespondsToCharactersFoundMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParserDidReachEndOfDocument:)]) {
_delegateRespondsToEndOfDocumentMethod = YES;
}
return self;
}
#pragma mark - Dealloc
- (void)dealloc {
if (_context != nil) {
htmlFreeParserCtxt(_context);
_context = nil;
}
_delegate = nil;
}
#pragma mark - API
static xmlSAXHandler saxHandlerStruct;
- (void)parseData:(NSData *)data {
[self parseBytes:data.bytes numberOfBytes:data.length];
}
- (void)parseBytes:(const void *)bytes numberOfBytes:(NSUInteger)numberOfBytes {
if (self.context == nil) {
xmlCharEncoding characterEncoding = xmlDetectCharEncoding(bytes, (int)numberOfBytes);
self.context = htmlCreatePushParserCtxt(&saxHandlerStruct, (__bridge void *)self, nil, 0, nil, characterEncoding);
htmlCtxtUseOptions(self.context, XML_PARSE_RECOVER | XML_PARSE_NONET | HTML_PARSE_COMPACT);
}
@autoreleasepool {
htmlParseChunk(self.context, (const char *)bytes, (int)numberOfBytes, 0);
}
}
- (void)finishParsing {
NSAssert(self.context != nil, nil);
if (self.context == nil)
return;
@autoreleasepool {
htmlParseChunk(self.context, nil, 0, 1);
htmlFreeParserCtxt(self.context);
self.context = nil;
self.characters = nil;
}
}
- (void)cancel {
@autoreleasepool {
xmlStopParser(self.context);
}
}
- (void)beginStoringCharacters {
self.storingCharacters = YES;
self.characters = [NSMutableData new];
}
- (void)endStoringCharacters {
self.storingCharacters = NO;
self.characters = nil;
}
- (NSData *)currentCharacters {
if (!self.storingCharacters) {
return nil;
}
return self.characters;
}
- (NSString *)currentString {
NSData *d = self.currentCharacters;
if (RSXMLIsEmpty(d)) {
return nil;
}
return [[NSString alloc] initWithData:d encoding:NSUTF8StringEncoding];
}
- (NSString *)currentStringWithTrimmedWhitespace {
return [self.currentString stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]];
}
#pragma mark - Attributes Dictionary
- (NSDictionary *)attributesDictionary:(const xmlChar **)attributes {
if (!attributes) {
return nil;
}
NSMutableDictionary *d = [NSMutableDictionary new];
NSInteger ix = 0;
NSString *currentKey = nil;
while (true) {
const xmlChar *oneAttribute = attributes[ix];
ix++;
if (!currentKey && !oneAttribute) {
break;
}
if (!currentKey) {
currentKey = [NSString stringWithUTF8String:(const char *)oneAttribute];
}
else {
NSString *value = nil;
if (oneAttribute) {
value = [NSString stringWithUTF8String:(const char *)oneAttribute];
}
d[currentKey] = value ? value : @"";
currentKey = nil;
}
}
return [d copy];
}
#pragma mark - Callbacks
- (void)xmlEndDocument {
@autoreleasepool {
if (self.delegateRespondsToEndOfDocumentMethod) {
[self.delegate saxParserDidReachEndOfDocument:self];
}
[self endStoringCharacters];
}
}
- (void)xmlCharactersFound:(const xmlChar *)ch length:(NSUInteger)length {
@autoreleasepool {
if (self.storingCharacters) {
[self.characters appendBytes:(const void *)ch length:length];
}
if (self.delegateRespondsToCharactersFoundMethod) {
[self.delegate saxParser:self XMLCharactersFound:ch length:length];
}
}
}
- (void)xmlStartElement:(const xmlChar *)localName attributes:(const xmlChar **)attributes {
@autoreleasepool {
if (self.delegateRespondsToStartElementMethod) {
[self.delegate saxParser:self XMLStartElement:localName attributes:attributes];
}
}
}
- (void)xmlEndElement:(const xmlChar *)localName {
@autoreleasepool {
if (self.delegateRespondsToEndElementMethod) {
[self.delegate saxParser:self XMLEndElement:localName];
}
[self endStoringCharacters];
}
}
@end
static void startElementSAX(void *context, const xmlChar *localname, const xmlChar **attributes) {
[(__bridge RSSAXHTMLParser *)context xmlStartElement:localname attributes:attributes];
}
static void endElementSAX(void *context, const xmlChar *localname) {
[(__bridge RSSAXHTMLParser *)context xmlEndElement:localname];
}
static void charactersFoundSAX(void *context, const xmlChar *ch, int len) {
[(__bridge RSSAXHTMLParser *)context xmlCharactersFound:ch length:(NSUInteger)len];
}
static void endDocumentSAX(void *context) {
[(__bridge RSSAXHTMLParser *)context xmlEndDocument];
}
static htmlSAXHandler saxHandlerStruct = {
nil, /* internalSubset */
nil, /* isStandalone */
nil, /* hasInternalSubset */
nil, /* hasExternalSubset */
nil, /* resolveEntity */
nil, /* getEntity */
nil, /* entityDecl */
nil, /* notationDecl */
nil, /* attributeDecl */
nil, /* elementDecl */
nil, /* unparsedEntityDecl */
nil, /* setDocumentLocator */
nil, /* startDocument */
endDocumentSAX, /* endDocument */
startElementSAX, /* startElement*/
endElementSAX, /* endElement */
nil, /* reference */
charactersFoundSAX, /* characters */
nil, /* ignorableWhitespace */
nil, /* processingInstruction */
nil, /* comment */
nil, /* warning */
nil, /* error */
nil, /* fatalError //: unused error() get all the errors */
nil, /* getParameterEntity */
nil, /* cdataBlock */
nil, /* externalSubset */
XML_SAX2_MAGIC,
nil,
nil, /* startElementNs */
nil, /* endElementNs */
nil /* serror */
};

View File

@ -0,0 +1,26 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>FMWK</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>$(CURRENT_PROJECT_VERSION)</string>
<key>NSHumanReadableCopyright</key>
<string>Copyright © 2017 Ranchero Software, LLC. All rights reserved.</string>
<key>NSPrincipalClass</key>
<string></string>
</dict>
</plist>

View File

@ -0,0 +1,40 @@
//
// RSParser.h
// RSParser
//
// Created by Brent Simmons on 6/20/17.
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
#import <RSXML/RSSAXParser.h>
#import <RSXML/RSXMLData.h>
#import <RSXML/RSFeedParser.h>
#import <RSXML/FeedParser.h>
#import <RSXML/RSAtomParser.h>
#import <RSXML/RSRSSParser.h>
#import <RSXML/RSParsedFeed.h>
#import <RSXML/RSParsedArticle.h>
#import <RSXML/RSOPMLParser.h>
#import <RSXML/RSOPMLDocument.h>
#import <RSXML/RSOPMLItem.h>
#import <RSXML/RSOPMLAttributes.h>
#import <RSXML/RSOPMLFeedSpecifier.h>
#import <RSXML/RSXMLError.h>
#import <RSXML/NSString+RSXML.h>
#import <RSXML/RSDateParser.h>
// HTML
#import <RSXML/RSSAXHTMLParser.h>
#import <RSXML/RSHTMLMetadataParser.h>
#import <RSXML/RSHTMLMetadata.h>
#import <RSXML/RSHTMLLinkParser.h>

View File

@ -0,0 +1,420 @@
// !$*UTF8*$!
{
archiveVersion = 1;
classes = {
};
objectVersion = 48;
objects = {
/* Begin PBXBuildFile section */
84D81BDC1EFA28E700652332 /* RSParser.h in Headers */ = {isa = PBXBuildFile; fileRef = 84D81BDA1EFA28E700652332 /* RSParser.h */; settings = {ATTRIBUTES = (Public, ); }; };
84FF5F8E1EFA285800C15A01 /* RSParser.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 84FF5F841EFA285800C15A01 /* RSParser.framework */; };
84FF5F931EFA285800C15A01 /* RSParserTests.swift in Sources */ = {isa = PBXBuildFile; fileRef = 84FF5F921EFA285800C15A01 /* RSParserTests.swift */; };
/* End PBXBuildFile section */
/* Begin PBXContainerItemProxy section */
84FF5F8F1EFA285800C15A01 /* PBXContainerItemProxy */ = {
isa = PBXContainerItemProxy;
containerPortal = 84FF5F7B1EFA285800C15A01 /* Project object */;
proxyType = 1;
remoteGlobalIDString = 84FF5F831EFA285800C15A01;
remoteInfo = RSParser;
};
/* End PBXContainerItemProxy section */
/* Begin PBXFileReference section */
84D81BD91EFA28E700652332 /* Info.plist */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
84D81BDA1EFA28E700652332 /* RSParser.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = RSParser.h; sourceTree = "<group>"; };
84FF5F841EFA285800C15A01 /* RSParser.framework */ = {isa = PBXFileReference; explicitFileType = wrapper.framework; includeInIndex = 0; path = RSParser.framework; sourceTree = BUILT_PRODUCTS_DIR; };
84FF5F8D1EFA285800C15A01 /* RSParserTests.xctest */ = {isa = PBXFileReference; explicitFileType = wrapper.cfbundle; includeInIndex = 0; path = RSParserTests.xctest; sourceTree = BUILT_PRODUCTS_DIR; };
84FF5F921EFA285800C15A01 /* RSParserTests.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = RSParserTests.swift; sourceTree = "<group>"; };
84FF5F941EFA285800C15A01 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
/* End PBXFileReference section */
/* Begin PBXFrameworksBuildPhase section */
84FF5F801EFA285800C15A01 /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
84FF5F8A1EFA285800C15A01 /* Frameworks */ = {
isa = PBXFrameworksBuildPhase;
buildActionMask = 2147483647;
files = (
84FF5F8E1EFA285800C15A01 /* RSParser.framework in Frameworks */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXFrameworksBuildPhase section */
/* Begin PBXGroup section */
84FF5F7A1EFA285800C15A01 = {
isa = PBXGroup;
children = (
84D81BDA1EFA28E700652332 /* RSParser.h */,
84D81BD91EFA28E700652332 /* Info.plist */,
84FF5F911EFA285800C15A01 /* RSParserTests */,
84FF5F851EFA285800C15A01 /* Products */,
);
sourceTree = "<group>";
};
84FF5F851EFA285800C15A01 /* Products */ = {
isa = PBXGroup;
children = (
84FF5F841EFA285800C15A01 /* RSParser.framework */,
84FF5F8D1EFA285800C15A01 /* RSParserTests.xctest */,
);
name = Products;
sourceTree = "<group>";
};
84FF5F911EFA285800C15A01 /* RSParserTests */ = {
isa = PBXGroup;
children = (
84FF5F921EFA285800C15A01 /* RSParserTests.swift */,
84FF5F941EFA285800C15A01 /* Info.plist */,
);
path = RSParserTests;
sourceTree = "<group>";
};
/* End PBXGroup section */
/* Begin PBXHeadersBuildPhase section */
84FF5F811EFA285800C15A01 /* Headers */ = {
isa = PBXHeadersBuildPhase;
buildActionMask = 2147483647;
files = (
84D81BDC1EFA28E700652332 /* RSParser.h in Headers */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXHeadersBuildPhase section */
/* Begin PBXNativeTarget section */
84FF5F831EFA285800C15A01 /* RSParser */ = {
isa = PBXNativeTarget;
buildConfigurationList = 84FF5F981EFA285800C15A01 /* Build configuration list for PBXNativeTarget "RSParser" */;
buildPhases = (
84FF5F7F1EFA285800C15A01 /* Sources */,
84FF5F801EFA285800C15A01 /* Frameworks */,
84FF5F811EFA285800C15A01 /* Headers */,
84FF5F821EFA285800C15A01 /* Resources */,
);
buildRules = (
);
dependencies = (
);
name = RSParser;
productName = RSParser;
productReference = 84FF5F841EFA285800C15A01 /* RSParser.framework */;
productType = "com.apple.product-type.framework";
};
84FF5F8C1EFA285800C15A01 /* RSParserTests */ = {
isa = PBXNativeTarget;
buildConfigurationList = 84FF5F9B1EFA285800C15A01 /* Build configuration list for PBXNativeTarget "RSParserTests" */;
buildPhases = (
84FF5F891EFA285800C15A01 /* Sources */,
84FF5F8A1EFA285800C15A01 /* Frameworks */,
84FF5F8B1EFA285800C15A01 /* Resources */,
);
buildRules = (
);
dependencies = (
84FF5F901EFA285800C15A01 /* PBXTargetDependency */,
);
name = RSParserTests;
productName = RSParserTests;
productReference = 84FF5F8D1EFA285800C15A01 /* RSParserTests.xctest */;
productType = "com.apple.product-type.bundle.unit-test";
};
/* End PBXNativeTarget section */
/* Begin PBXProject section */
84FF5F7B1EFA285800C15A01 /* Project object */ = {
isa = PBXProject;
attributes = {
LastSwiftUpdateCheck = 0900;
LastUpgradeCheck = 0900;
ORGANIZATIONNAME = "Ranchero Software, LLC";
TargetAttributes = {
84FF5F831EFA285800C15A01 = {
CreatedOnToolsVersion = 9.0;
};
84FF5F8C1EFA285800C15A01 = {
CreatedOnToolsVersion = 9.0;
};
};
};
buildConfigurationList = 84FF5F7E1EFA285800C15A01 /* Build configuration list for PBXProject "RSParser" */;
compatibilityVersion = "Xcode 8.0";
developmentRegion = en;
hasScannedForEncodings = 0;
knownRegions = (
en,
);
mainGroup = 84FF5F7A1EFA285800C15A01;
productRefGroup = 84FF5F851EFA285800C15A01 /* Products */;
projectDirPath = "";
projectRoot = "";
targets = (
84FF5F831EFA285800C15A01 /* RSParser */,
84FF5F8C1EFA285800C15A01 /* RSParserTests */,
);
};
/* End PBXProject section */
/* Begin PBXResourcesBuildPhase section */
84FF5F821EFA285800C15A01 /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
84FF5F8B1EFA285800C15A01 /* Resources */ = {
isa = PBXResourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXResourcesBuildPhase section */
/* Begin PBXSourcesBuildPhase section */
84FF5F7F1EFA285800C15A01 /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
);
runOnlyForDeploymentPostprocessing = 0;
};
84FF5F891EFA285800C15A01 /* Sources */ = {
isa = PBXSourcesBuildPhase;
buildActionMask = 2147483647;
files = (
84FF5F931EFA285800C15A01 /* RSParserTests.swift in Sources */,
);
runOnlyForDeploymentPostprocessing = 0;
};
/* End PBXSourcesBuildPhase section */
/* Begin PBXTargetDependency section */
84FF5F901EFA285800C15A01 /* PBXTargetDependency */ = {
isa = PBXTargetDependency;
target = 84FF5F831EFA285800C15A01 /* RSParser */;
targetProxy = 84FF5F8F1EFA285800C15A01 /* PBXContainerItemProxy */;
};
/* End PBXTargetDependency section */
/* Begin XCBuildConfiguration section */
84FF5F961EFA285800C15A01 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
CODE_SIGN_IDENTITY = "Mac Developer";
COPY_PHASE_STRIP = NO;
CURRENT_PROJECT_VERSION = 1;
DEBUG_INFORMATION_FORMAT = dwarf;
ENABLE_STRICT_OBJC_MSGSEND = YES;
ENABLE_TESTABILITY = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_DYNAMIC_NO_PIC = NO;
GCC_NO_COMMON_BLOCKS = YES;
GCC_OPTIMIZATION_LEVEL = 0;
GCC_PREPROCESSOR_DEFINITIONS = (
"DEBUG=1",
"$(inherited)",
);
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
MACOSX_DEPLOYMENT_TARGET = 10.12;
MTL_ENABLE_DEBUG_INFO = YES;
ONLY_ACTIVE_ARCH = YES;
SDKROOT = macosx;
SWIFT_ACTIVE_COMPILATION_CONDITIONS = DEBUG;
SWIFT_OPTIMIZATION_LEVEL = "-Onone";
VERSIONING_SYSTEM = "apple-generic";
VERSION_INFO_PREFIX = "";
};
name = Debug;
};
84FF5F971EFA285800C15A01 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
ALWAYS_SEARCH_USER_PATHS = NO;
CLANG_ANALYZER_NONNULL = YES;
CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
CLANG_CXX_LANGUAGE_STANDARD = "gnu++14";
CLANG_CXX_LIBRARY = "libc++";
CLANG_ENABLE_MODULES = YES;
CLANG_ENABLE_OBJC_ARC = YES;
CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
CLANG_WARN_BOOL_CONVERSION = YES;
CLANG_WARN_COMMA = YES;
CLANG_WARN_CONSTANT_CONVERSION = YES;
CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
CLANG_WARN_EMPTY_BODY = YES;
CLANG_WARN_ENUM_CONVERSION = YES;
CLANG_WARN_INFINITE_RECURSION = YES;
CLANG_WARN_INT_CONVERSION = YES;
CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
CLANG_WARN_STRICT_PROTOTYPES = YES;
CLANG_WARN_SUSPICIOUS_MOVE = YES;
CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
CLANG_WARN_UNREACHABLE_CODE = YES;
CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
CODE_SIGN_IDENTITY = "Mac Developer";
COPY_PHASE_STRIP = NO;
CURRENT_PROJECT_VERSION = 1;
DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
ENABLE_NS_ASSERTIONS = NO;
ENABLE_STRICT_OBJC_MSGSEND = YES;
GCC_C_LANGUAGE_STANDARD = gnu11;
GCC_NO_COMMON_BLOCKS = YES;
GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
GCC_WARN_UNDECLARED_SELECTOR = YES;
GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
GCC_WARN_UNUSED_FUNCTION = YES;
GCC_WARN_UNUSED_VARIABLE = YES;
MACOSX_DEPLOYMENT_TARGET = 10.12;
MTL_ENABLE_DEBUG_INFO = NO;
SDKROOT = macosx;
SWIFT_OPTIMIZATION_LEVEL = "-Owholemodule";
VERSIONING_SYSTEM = "apple-generic";
VERSION_INFO_PREFIX = "";
};
name = Release;
};
84FF5F991EFA285800C15A01 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
CODE_SIGN_IDENTITY = "";
COMBINE_HIDPI_IMAGES = YES;
DEFINES_MODULE = YES;
DEVELOPMENT_TEAM = M8L2WTLA8W;
DYLIB_COMPATIBILITY_VERSION = 1;
DYLIB_CURRENT_VERSION = 1;
DYLIB_INSTALL_NAME_BASE = "@rpath";
FRAMEWORK_VERSION = A;
INFOPLIST_FILE = RSParser/Info.plist;
INSTALL_PATH = "$(LOCAL_LIBRARY_DIR)/Frameworks";
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/../Frameworks @loader_path/Frameworks";
PRODUCT_BUNDLE_IDENTIFIER = com.ranchero.RSParser;
PRODUCT_NAME = "$(TARGET_NAME:c99extidentifier)";
SKIP_INSTALL = YES;
SWIFT_VERSION = 4.0;
};
name = Debug;
};
84FF5F9A1EFA285800C15A01 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
CODE_SIGN_IDENTITY = "";
COMBINE_HIDPI_IMAGES = YES;
DEFINES_MODULE = YES;
DEVELOPMENT_TEAM = M8L2WTLA8W;
DYLIB_COMPATIBILITY_VERSION = 1;
DYLIB_CURRENT_VERSION = 1;
DYLIB_INSTALL_NAME_BASE = "@rpath";
FRAMEWORK_VERSION = A;
INFOPLIST_FILE = RSParser/Info.plist;
INSTALL_PATH = "$(LOCAL_LIBRARY_DIR)/Frameworks";
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/../Frameworks @loader_path/Frameworks";
PRODUCT_BUNDLE_IDENTIFIER = com.ranchero.RSParser;
PRODUCT_NAME = "$(TARGET_NAME:c99extidentifier)";
SKIP_INSTALL = YES;
SWIFT_VERSION = 4.0;
};
name = Release;
};
84FF5F9C1EFA285800C15A01 /* Debug */ = {
isa = XCBuildConfiguration;
buildSettings = {
COMBINE_HIDPI_IMAGES = YES;
DEVELOPMENT_TEAM = M8L2WTLA8W;
INFOPLIST_FILE = RSParserTests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/../Frameworks @loader_path/../Frameworks";
PRODUCT_BUNDLE_IDENTIFIER = com.ranchero.RSParserTests;
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 4.0;
};
name = Debug;
};
84FF5F9D1EFA285800C15A01 /* Release */ = {
isa = XCBuildConfiguration;
buildSettings = {
COMBINE_HIDPI_IMAGES = YES;
DEVELOPMENT_TEAM = M8L2WTLA8W;
INFOPLIST_FILE = RSParserTests/Info.plist;
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/../Frameworks @loader_path/../Frameworks";
PRODUCT_BUNDLE_IDENTIFIER = com.ranchero.RSParserTests;
PRODUCT_NAME = "$(TARGET_NAME)";
SWIFT_VERSION = 4.0;
};
name = Release;
};
/* End XCBuildConfiguration section */
/* Begin XCConfigurationList section */
84FF5F7E1EFA285800C15A01 /* Build configuration list for PBXProject "RSParser" */ = {
isa = XCConfigurationList;
buildConfigurations = (
84FF5F961EFA285800C15A01 /* Debug */,
84FF5F971EFA285800C15A01 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
84FF5F981EFA285800C15A01 /* Build configuration list for PBXNativeTarget "RSParser" */ = {
isa = XCConfigurationList;
buildConfigurations = (
84FF5F991EFA285800C15A01 /* Debug */,
84FF5F9A1EFA285800C15A01 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
84FF5F9B1EFA285800C15A01 /* Build configuration list for PBXNativeTarget "RSParserTests" */ = {
isa = XCConfigurationList;
buildConfigurations = (
84FF5F9C1EFA285800C15A01 /* Debug */,
84FF5F9D1EFA285800C15A01 /* Release */,
);
defaultConfigurationIsVisible = 0;
defaultConfigurationName = Release;
};
/* End XCConfigurationList section */
};
rootObject = 84FF5F7B1EFA285800C15A01 /* Project object */;
}

View File

@ -0,0 +1,7 @@
<?xml version="1.0" encoding="UTF-8"?>
<Workspace
version = "1.0">
<FileRef
location = "self:RSParser.xcodeproj">
</FileRef>
</Workspace>

View File

@ -0,0 +1,22 @@
//
// RSXMLData.h
// RSXML
//
// Created by Brent Simmons on 8/24/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
NS_ASSUME_NONNULL_BEGIN
@interface RSXMLData : NSObject
- (instancetype)initWithData:(NSData *)data urlString:(NSString *)urlString;
@property (nonatomic, readonly) NSData *data;
@property (nonatomic, readonly) NSString *urlString;
@end
NS_ASSUME_NONNULL_END

View File

@ -0,0 +1,28 @@
//
// RSXMLData.m
// RSXML
//
// Created by Brent Simmons on 8/24/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
#import "RSXMLData.h"
@implementation RSXMLData
- (instancetype)initWithData:(NSData *)data urlString:(NSString *)urlString {
self = [super init];
if (!self) {
return nil;
}
_data = data;
_urlString = urlString;
return self;
}
@end

View File

@ -0,0 +1,19 @@
//
// RSXMLError.h
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
extern NSString *RSXMLErrorDomain;
typedef NS_ENUM(NSInteger, RSXMLErrorCode) {
RSXMLErrorCodeDataIsWrongFormat = 1024
};
NSError *RSOPMLWrongFormatError(NSString *fileName);

View File

@ -0,0 +1,22 @@
//
// RSXMLError.m
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSXMLError.h"
NSString *RSXMLErrorDomain = @"com.ranchero.RSXML";
NSError *RSOPMLWrongFormatError(NSString *fileName) {
NSString *localizedDescriptionFormatString = NSLocalizedString(@"The file %@ cant be parsed because its not an OPML file.", @"OPML wrong format");
NSString *localizedDescription = [NSString stringWithFormat:localizedDescriptionFormatString, fileName];
NSString *localizedFailureString = NSLocalizedString(@"The file is not an OPML file.", @"OPML wrong format");
NSDictionary *userInfo = @{NSLocalizedDescriptionKey: localizedDescription, NSLocalizedFailureReasonErrorKey: localizedFailureString};
return [[NSError alloc] initWithDomain:RSXMLErrorDomain code:RSXMLErrorCodeDataIsWrongFormat userInfo:userInfo];
}

View File

@ -0,0 +1,22 @@
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>CFBundleDevelopmentRegion</key>
<string>$(DEVELOPMENT_LANGUAGE)</string>
<key>CFBundleExecutable</key>
<string>$(EXECUTABLE_NAME)</string>
<key>CFBundleIdentifier</key>
<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
<key>CFBundleInfoDictionaryVersion</key>
<string>6.0</string>
<key>CFBundleName</key>
<string>$(PRODUCT_NAME)</string>
<key>CFBundlePackageType</key>
<string>BNDL</string>
<key>CFBundleShortVersionString</key>
<string>1.0</string>
<key>CFBundleVersion</key>
<string>1</string>
</dict>
</plist>

View File

@ -0,0 +1,36 @@
//
// RSParserTests.swift
// RSParserTests
//
// Created by Brent Simmons on 6/20/17.
// Copyright © 2017 Ranchero Software, LLC. All rights reserved.
//
import XCTest
@testable import RSParser
class RSParserTests: XCTestCase {
override func setUp() {
super.setUp()
// Put setup code here. This method is called before the invocation of each test method in the class.
}
override func tearDown() {
// Put teardown code here. This method is called after the invocation of each test method in the class.
super.tearDown()
}
func testExample() {
// This is an example of a functional test case.
// Use XCTAssert and related functions to verify your tests produce the correct results.
}
func testPerformanceExample() {
// This is an example of a performance test case.
self.measure {
// Put the code you want to measure the time of here.
}
}
}

View File

@ -0,0 +1,16 @@
//
// NSString+RSXML.h
// RSXML
//
// Created by Brent Simmons on 9/25/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@interface NSString (RSXML)
- (NSString *)rs_stringByDecodingHTMLEntities;
@end

View File

@ -0,0 +1,408 @@
//
// NSString+RSXML.m
// RSXML
//
// Created by Brent Simmons on 9/25/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
#import "NSString+RSXML.h"
@interface NSScanner (RSXML)
- (BOOL)rs_scanEntityValue:(NSString * _Nullable * _Nullable)decodedEntity;
@end
@implementation NSString (RSXML)
- (NSString *)rs_stringByDecodingHTMLEntities {
@autoreleasepool {
NSScanner *scanner = [[NSScanner alloc] initWithString:self];
scanner.charactersToBeSkipped = nil;
NSMutableString *result = [[NSMutableString alloc] init];
while (true) {
NSString *scannedString = nil;
if ([scanner scanUpToString:@"&" intoString:&scannedString]) {
[result appendString:scannedString];
}
if (scanner.isAtEnd) {
break;
}
NSUInteger savedScanLocation = scanner.scanLocation;
NSString *decodedEntity = nil;
if ([scanner rs_scanEntityValue:&decodedEntity]) {
[result appendString:decodedEntity];
}
else {
[result appendString:@"&"];
scanner.scanLocation = savedScanLocation + 1;
}
if (scanner.isAtEnd) {
break;
}
}
if ([self isEqualToString:result]) {
return self;
}
return [result copy];
}
}
static NSDictionary *RSEntitiesDictionary(void);
static NSString *RSXMLStringWithValue(unichar value);
- (NSString * _Nullable)rs_stringByDecodingEntity {
// self may or may not have outer & and ; characters.
NSMutableString *s = [self mutableCopy];
if ([s hasPrefix:@"&"]) {
[s deleteCharactersInRange:NSMakeRange(0, 1)];
}
if ([s hasSuffix:@";"]) {
[s deleteCharactersInRange:NSMakeRange(s.length - 1, 1)];
}
NSDictionary *entitiesDictionary = RSEntitiesDictionary();
NSString *decodedEntity = entitiesDictionary[self];
if (decodedEntity) {
return decodedEntity;
}
if ([s hasPrefix:@"#x"]) { // Hex
NSScanner *scanner = [[NSScanner alloc] initWithString:s];
scanner.charactersToBeSkipped = [NSCharacterSet characterSetWithCharactersInString:@"#x"];
unsigned int hexValue = 0;
if ([scanner scanHexInt:&hexValue]) {
return RSXMLStringWithValue((unichar)hexValue);
}
return nil;
}
else if ([s hasPrefix:@"#"]) {
[s deleteCharactersInRange:NSMakeRange(0, 1)];
NSInteger value = s.integerValue;
if (value < 1) {
return nil;
}
return RSXMLStringWithValue((unichar)value);
}
return nil;
}
@end
@implementation NSScanner (RSXML)
- (BOOL)rs_scanEntityValue:(NSString * _Nullable * _Nullable)decodedEntity {
NSString *s = self.string;
NSUInteger initialScanLocation = self.scanLocation;
static NSUInteger maxEntityLength = 20; // Its probably smaller, but this is just for sanity.
while (true) {
unichar ch = [s characterAtIndex:self.scanLocation];
if ([NSCharacterSet.whitespaceAndNewlineCharacterSet characterIsMember:ch]) {
break;
}
if (ch == ';') {
if (!decodedEntity) {
return YES;
}
NSString *rawEntity = [s substringWithRange:NSMakeRange(initialScanLocation + 1, (self.scanLocation - initialScanLocation) - 1)];
*decodedEntity = [rawEntity rs_stringByDecodingEntity];
self.scanLocation = self.scanLocation + 1;
return *decodedEntity != nil;
}
self.scanLocation = self.scanLocation + 1;
if (self.scanLocation - initialScanLocation > maxEntityLength) {
break;
}
if (self.isAtEnd) {
break;
}
}
return NO;
}
@end
static NSString *RSXMLStringWithValue(unichar value) {
return [[NSString alloc] initWithFormat:@"%C", value];
}
static NSDictionary *RSEntitiesDictionary(void) {
static NSDictionary *entitiesDictionary = nil;
static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^{
entitiesDictionary =
@{@"#034": @"\"",
@"#038": @"&",
@"#38": @"&",
@"#039": @"'",
@"#145": @"",
@"#146": @"",
@"#147": @"“",
@"#148": @"”",
@"#149": @"•",
@"#150": @"-",
@"#151": @"—",
@"#153": @"™",
@"#160": RSXMLStringWithValue(160),
@"#161": @"¡",
@"#162": @"¢",
@"#163": @"£",
@"#164": @"?",
@"#165": @"¥",
@"#166": @"?",
@"#167": @"§",
@"#168": @"¨",
@"#169": @"©",
@"#170": @"©",
@"#171": @"«",
@"#172": @"¬",
@"#173": @"¬",
@"#174": @"®",
@"#175": @"¯",
@"#176": @"°",
@"#177": @"±",
@"#178": @" ",
@"#179": @" ",
@"#180": @"´",
@"#181": @"µ",
@"#182": @"µ",
@"#183": @"·",
@"#184": @"¸",
@"#185": @" ",
@"#186": @"º",
@"#187": @"»",
@"#188": @"1/4",
@"#189": @"1/2",
@"#190": @"1/2",
@"#191": @"¿",
@"#192": @"À",
@"#193": @"Á",
@"#194": @"Â",
@"#195": @"Ã",
@"#196": @"Ä",
@"#197": @"Å",
@"#198": @"Æ",
@"#199": @"Ç",
@"#200": @"È",
@"#201": @"É",
@"#202": @"Ê",
@"#203": @"Ë",
@"#204": @"Ì",
@"#205": @"Í",
@"#206": @"Î",
@"#207": @"Ï",
@"#208": @"?",
@"#209": @"Ñ",
@"#210": @"Ò",
@"#211": @"Ó",
@"#212": @"Ô",
@"#213": @"Õ",
@"#214": @"Ö",
@"#215": @"x",
@"#216": @"Ø",
@"#217": @"Ù",
@"#218": @"Ú",
@"#219": @"Û",
@"#220": @"Ü",
@"#221": @"Y",
@"#222": @"?",
@"#223": @"ß",
@"#224": @"à",
@"#225": @"á",
@"#226": @"â",
@"#227": @"ã",
@"#228": @"ä",
@"#229": @"å",
@"#230": @"æ",
@"#231": @"ç",
@"#232": @"è",
@"#233": @"é",
@"#234": @"ê",
@"#235": @"ë",
@"#236": @"ì",
@"#237": @"í",
@"#238": @"î",
@"#239": @"ï",
@"#240": @"?",
@"#241": @"ñ",
@"#242": @"ò",
@"#243": @"ó",
@"#244": @"ô",
@"#245": @"õ",
@"#246": @"ö",
@"#247": @"÷",
@"#248": @"ø",
@"#249": @"ù",
@"#250": @"ú",
@"#251": @"û",
@"#252": @"ü",
@"#253": @"y",
@"#254": @"?",
@"#255": @"ÿ",
@"#32": @" ",
@"#34": @"\"",
@"#39": @"",
@"#8194": @" ",
@"#8195": @" ",
@"#8211": @"-",
@"#8212": @"—",
@"#8216": @"",
@"#8217": @"",
@"#8220": @"“",
@"#8221": @"”",
@"#8230": @"…",
@"#8617": RSXMLStringWithValue(8617),
@"AElig": @"Æ",
@"Aacute": @"Á",
@"Acirc": @"Â",
@"Agrave": @"À",
@"Aring": @"Å",
@"Atilde": @"Ã",
@"Auml": @"Ä",
@"Ccedil": @"Ç",
@"Dstrok": @"?",
@"ETH": @"?",
@"Eacute": @"É",
@"Ecirc": @"Ê",
@"Egrave": @"È",
@"Euml": @"Ë",
@"Iacute": @"Í",
@"Icirc": @"Î",
@"Igrave": @"Ì",
@"Iuml": @"Ï",
@"Ntilde": @"Ñ",
@"Oacute": @"Ó",
@"Ocirc": @"Ô",
@"Ograve": @"Ò",
@"Oslash": @"Ø",
@"Otilde": @"Õ",
@"Ouml": @"Ö",
@"Pi": @"Π",
@"THORN": @"?",
@"Uacute": @"Ú",
@"Ucirc": @"Û",
@"Ugrave": @"Ù",
@"Uuml": @"Ü",
@"Yacute": @"Y",
@"aacute": @"á",
@"acirc": @"â",
@"acute": @"´",
@"aelig": @"æ",
@"agrave": @"à",
@"amp": @"&",
@"apos": @"'",
@"aring": @"å",
@"atilde": @"ã",
@"auml": @"ä",
@"brkbar": @"?",
@"brvbar": @"?",
@"ccedil": @"ç",
@"cedil": @"¸",
@"cent": @"¢",
@"copy": @"©",
@"curren": @"?",
@"deg": @"°",
@"die": @"?",
@"divide": @"÷",
@"eacute": @"é",
@"ecirc": @"ê",
@"egrave": @"è",
@"eth": @"?",
@"euml": @"ë",
@"euro": @"€",
@"frac12": @"1/2",
@"frac14": @"1/4",
@"frac34": @"3/4",
@"gt": @">",
@"hearts": @"♥",
@"hellip": @"…",
@"iacute": @"í",
@"icirc": @"î",
@"iexcl": @"¡",
@"igrave": @"ì",
@"iquest": @"¿",
@"iuml": @"ï",
@"laquo": @"«",
@"ldquo": @"“",
@"lsquo": @"",
@"lt": @"<",
@"macr": @"¯",
@"mdash": @"—",
@"micro": @"µ",
@"middot": @"·",
@"ndash": @"-",
@"not": @"¬",
@"ntilde": @"ñ",
@"oacute": @"ó",
@"ocirc": @"ô",
@"ograve": @"ò",
@"ordf": @"ª",
@"ordm": @"º",
@"oslash": @"ø",
@"otilde": @"õ",
@"ouml": @"ö",
@"para": @"¶",
@"pi": @"π",
@"plusmn": @"±",
@"pound": @"£",
@"quot": @"\"",
@"raquo": @"»",
@"rdquo": @"”",
@"reg": @"®",
@"rsquo": @"",
@"sect": @"§",
@"shy": @" ",
@"sup1": @" ",
@"sup2": @" ",
@"sup3": @" ",
@"szlig": @"ß",
@"thorn": @"?",
@"times": @"x",
@"trade": @"™",
@"uacute": @"ú",
@"ucirc": @"û",
@"ugrave": @"ù",
@"uml": @"¨",
@"uuml": @"ü",
@"yacute": @"y",
@"yen": @"¥",
@"yuml": @"ÿ",
@"infin": @"∞",
@"nbsp": RSXMLStringWithValue(160),
@"#x21A9": RSXMLStringWithValue(8617),
@"#xFE0E": RSXMLStringWithValue(65038),
@"#x2019": RSXMLStringWithValue(8217),
@"#x2026": RSXMLStringWithValue(8230),
@"#x201C": RSXMLStringWithValue(8220),
@"#x201D": RSXMLStringWithValue(8221),
@"#x2014": RSXMLStringWithValue(8212)};
});
return entitiesDictionary;
}

View File

@ -0,0 +1,25 @@
//
// RSDateParser.h
// RSXML
//
// Created by Brent Simmons on 3/25/15.
// Copyright (c) 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
/*Common web dates -- RFC 822 and 8601 -- are handled here:
the formats you find in JSON and XML feeds.
Any of these may return nil. They may also return garbage, given bad input.*/
NSDate *RSDateWithString(NSString *dateString);
/*If you're using a SAX parser, you have the bytes and don't need to convert to a string first.
It's faster and uses less memory.
(Assumes bytes are UTF-8 or ASCII. If you're using the libxml SAX parser, this will work.)*/
NSDate *RSDateWithBytes(const char *bytes, NSUInteger numberOfBytes);

View File

@ -0,0 +1,435 @@
//
// RSDateParser.m
// RSXML
//
// Created by Brent Simmons on 3/25/15.
// Copyright (c) 2015 Ranchero Software, LLC. All rights reserved.
//
#import <time.h>
#import "RSDateParser.h"
typedef struct {
const char *abbreviation;
const NSInteger offsetHours;
const NSInteger offsetMinutes;
} RSTimeZoneAbbreviationAndOffset;
#define kNumberOfTimeZones 96
static const RSTimeZoneAbbreviationAndOffset timeZoneTable[kNumberOfTimeZones] = {
{"GMT", 0, 0}, //Most common at top, for performance
{"PDT", -7, 0}, {"PST", -8, 0}, {"EST", -5, 0}, {"EDT", -4, 0},
{"MDT", -6, 0}, {"MST", -7, 0}, {"CST", -6, 0}, {"CDT", -5, 0},
{"ACT", -8, 0}, {"AFT", 4, 30}, {"AMT", 4, 0}, {"ART", -3, 0},
{"AST", 3, 0}, {"AZT", 4, 0}, {"BIT", -12, 0}, {"BDT", 8, 0},
{"ACST", 9, 30}, {"AEST", 10, 0}, {"AKST", -9, 0}, {"AMST", 5, 0},
{"AWST", 8, 0}, {"AZOST", -1, 0}, {"BIOT", 6, 0}, {"BRT", -3, 0},
{"BST", 6, 0}, {"BTT", 6, 0}, {"CAT", 2, 0}, {"CCT", 6, 30},
{"CET", 1, 0}, {"CEST", 2, 0}, {"CHAST", 12, 45}, {"ChST", 10, 0},
{"CIST", -8, 0}, {"CKT", -10, 0}, {"CLT", -4, 0}, {"CLST", -3, 0},
{"COT", -5, 0}, {"COST", -4, 0}, {"CVT", -1, 0}, {"CXT", 7, 0},
{"EAST", -6, 0}, {"EAT", 3, 0}, {"ECT", -4, 0}, {"EEST", 3, 0},
{"EET", 2, 0}, {"FJT", 12, 0}, {"FKST", -4, 0}, {"GALT", -6, 0},
{"GET", 4, 0}, {"GFT", -3, 0}, {"GILT", 7, 0}, {"GIT", -9, 0},
{"GST", -2, 0}, {"GYT", -4, 0}, {"HAST", -10, 0}, {"HKT", 8, 0},
{"HMT", 5, 0}, {"IRKT", 8, 0}, {"IRST", 3, 30}, {"IST", 2, 0},
{"JST", 9, 0}, {"KRAT", 7, 0}, {"KST", 9, 0}, {"LHST", 10, 30},
{"LINT", 14, 0}, {"MAGT", 11, 0}, {"MIT", -9, 30}, {"MSK", 3, 0},
{"MUT", 4, 0}, {"NDT", -2, 30}, {"NFT", 11, 30}, {"NPT", 5, 45},
{"NT", -3, 30}, {"OMST", 6, 0}, {"PETT", 12, 0}, {"PHOT", 13, 0},
{"PKT", 5, 0}, {"RET", 4, 0}, {"SAMT", 4, 0}, {"SAST", 2, 0},
{"SBT", 11, 0}, {"SCT", 4, 0}, {"SLT", 5, 30}, {"SST", 8, 0},
{"TAHT", -10, 0}, {"THA", 7, 0}, {"UYT", -3, 0}, {"UYST", -2, 0},
{"VET", -4, 30}, {"VLAT", 10, 0}, {"WAT", 1, 0}, {"WET", 0, 0},
{"WEST", 1, 0}, {"YAKT", 9, 0}, {"YEKT", 5, 0}
}; /*See http://en.wikipedia.org/wiki/List_of_time_zone_abbreviations for list*/
#pragma mark - Parser
enum {
RSJanuary = 1,
RSFebruary,
RSMarch,
RSApril,
RSMay,
RSJune,
RSJuly,
RSAugust,
RSSeptember,
RSOctober,
RSNovember,
RSDecember
};
static NSInteger nextMonthValue(const char *bytes, NSUInteger numberOfBytes, NSUInteger startingIndex, NSUInteger *finalIndex) {
/*Months are 1-based -- January is 1, Dec is 12.
Lots of short-circuits here. Not strict. GIGO.*/
NSUInteger i;// = startingIndex;
NSUInteger numberOfAlphaCharactersFound = 0;
char monthCharacters[3] = {0, 0, 0};
for (i = startingIndex; i < numberOfBytes; i++) {
*finalIndex = i;
char character = bytes[i];
BOOL isAlphaCharacter = (BOOL)isalpha(character);
if (!isAlphaCharacter && numberOfAlphaCharactersFound < 1)
continue;
if (!isAlphaCharacter && numberOfAlphaCharactersFound > 0)
break;
numberOfAlphaCharactersFound++;
if (numberOfAlphaCharactersFound == 1) {
if (character == 'F' || character == 'f')
return RSFebruary;
if (character == 'S' || character == 's')
return RSSeptember;
if (character == 'O' || character == 'o')
return RSOctober;
if (character == 'N' || character == 'n')
return RSNovember;
if (character == 'D' || character == 'd')
return RSDecember;
}
monthCharacters[numberOfAlphaCharactersFound - 1] = character;
if (numberOfAlphaCharactersFound >=3)
break;
}
if (numberOfAlphaCharactersFound < 2)
return NSNotFound;
if (monthCharacters[0] == 'J' || monthCharacters[0] == 'j') { //Jan, Jun, Jul
if (monthCharacters[1] == 'a' || monthCharacters[i] == 'A')
return RSJanuary;
if (monthCharacters[1] == 'u' || monthCharacters[1] == 'U') {
if (monthCharacters[2] == 'n' || monthCharacters[2] == 'N')
return RSJune;
return RSJuly;
}
return RSJanuary;
}
if (monthCharacters[0] == 'M' || monthCharacters[0] == 'm') { //March, May
if (monthCharacters[2] == 'y' || monthCharacters[2] == 'Y')
return RSMay;
return RSMarch;
}
if (monthCharacters[0] == 'A' || monthCharacters[0] == 'a') { //April, August
if (monthCharacters[1] == 'u' || monthCharacters[1] == 'U')
return RSAugust;
return RSApril;
}
return RSJanuary; //should never get here
}
static NSInteger nextNumericValue(const char *bytes, NSUInteger numberOfBytes, NSUInteger startingIndex, NSUInteger maximumNumberOfDigits, NSUInteger *finalIndex) {
/*maximumNumberOfDigits has a maximum limit of 4 (for time zone offsets and years).
*finalIndex will be the index of the last character looked at.*/
if (maximumNumberOfDigits > 4)
maximumNumberOfDigits = 4;
NSUInteger i = 0;
NSUInteger numberOfDigitsFound = 0;
NSInteger digits[4] = {0, 0, 0, 0};
for (i = startingIndex; i < numberOfBytes; i++) {
*finalIndex = i;
BOOL isDigit = (BOOL)isdigit(bytes[i]);
if (!isDigit && numberOfDigitsFound < 1)
continue;
if (!isDigit && numberOfDigitsFound > 0)
break;
digits[numberOfDigitsFound] = bytes[i] - 48; // '0' is 48
numberOfDigitsFound++;
if (numberOfDigitsFound >= maximumNumberOfDigits)
break;
}
if (numberOfDigitsFound < 1)
return NSNotFound;
if (numberOfDigitsFound == 1)
return digits[0];
if (numberOfDigitsFound == 2)
return (digits[0] * 10) + digits[1];
if (numberOfDigitsFound == 3)
return (digits[0] * 100) + (digits[1] * 10) + digits[2];
return (digits[0] * 1000) + (digits[1] * 100) + (digits[2] * 10) + digits[3];
}
static BOOL hasAtLeastOneAlphaCharacter(const char *s) {
NSUInteger length = strlen(s);
NSUInteger i = 0;
for (i = 0; i < length; i++) {
if (isalpha(s[i]))
return YES;
}
return NO;
}
#pragma mark - Time Zones and offsets
static NSInteger offsetInSecondsForTimeZoneAbbreviation(const char *abbreviation) {
/*Linear search should be fine. It's a C array, and short (under 100 items).
Most common time zones are at the beginning of the array. (We can tweak this as needed.)*/
NSUInteger i;
for (i = 0; i < kNumberOfTimeZones; i++) {
RSTimeZoneAbbreviationAndOffset zone = timeZoneTable[i];
if (strcmp(abbreviation, zone.abbreviation) == 0) {
if (zone.offsetHours < 0)
return (zone.offsetHours * 60 * 60) - (zone.offsetMinutes * 60);
return (zone.offsetHours * 60 * 60) + (zone.offsetMinutes * 60);
}
}
return 0;
}
static NSInteger offsetInSecondsForOffsetCharacters(const char *timeZoneCharacters) {
BOOL isPlus = timeZoneCharacters[0] == '+';
NSUInteger finalIndex = 0;
NSInteger hours = nextNumericValue(timeZoneCharacters, strlen(timeZoneCharacters), 0, 2, &finalIndex);
NSInteger minutes = nextNumericValue(timeZoneCharacters, strlen(timeZoneCharacters), finalIndex + 1, 2, &finalIndex);
if (hours == NSNotFound)
hours = 0;
if (minutes == NSNotFound)
minutes = 0;
if (hours == 0 && minutes == 0)
return 0;
NSInteger seconds = (hours * 60 * 60) + (minutes * 60);
if (!isPlus)
seconds = 0 - seconds;
return seconds;
}
static const char *rs_GMT = "GMT";
static const char *rs_UTC = "UTC";
static NSInteger parsedTimeZoneOffset(const char *bytes, NSUInteger numberOfBytes, NSUInteger startingIndex) {
/*Examples: GMT Z +0000 -0000 +07:00 -0700 PDT EST
Parse into char[5] -- drop any colon characters. If numeric, calculate seconds from GMT.
If alpha, special-case GMT and Z, otherwise look up in time zone list to get offset.*/
char timeZoneCharacters[6] = {0, 0, 0, 0, 0, 0}; //nil-terminated last character
NSUInteger i = 0;
NSUInteger numberOfCharactersFound = 0;
for (i = startingIndex; i < numberOfBytes; i++) {
char ch = bytes[i];
if (ch == ':' || ch == ' ')
continue;
if (isdigit(ch) || isalpha(ch) || ch == '+' || ch == '-') {
numberOfCharactersFound++;
timeZoneCharacters[numberOfCharactersFound - 1] = ch;
}
if (numberOfCharactersFound >= 5)
break;
}
if (numberOfCharactersFound < 1 || timeZoneCharacters[0] == 'Z' || timeZoneCharacters[0] == 'z')
return 0;
if (strcasestr(timeZoneCharacters, rs_GMT) != nil || strcasestr(timeZoneCharacters, rs_UTC))
return 0;
if (hasAtLeastOneAlphaCharacter(timeZoneCharacters))
return offsetInSecondsForTimeZoneAbbreviation(timeZoneCharacters);
return offsetInSecondsForOffsetCharacters(timeZoneCharacters);
}
#pragma mark - Date Creation
static NSDate *dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(NSInteger year, NSInteger month, NSInteger day, NSInteger hour, NSInteger minute, NSInteger second, NSInteger milliseconds, NSInteger timeZoneOffset) {
struct tm timeInfo;
timeInfo.tm_sec = (int)second;
timeInfo.tm_min = (int)minute;
timeInfo.tm_hour = (int)hour;
timeInfo.tm_mday = (int)day;
timeInfo.tm_mon = (int)(month - 1); //It's 1-based coming in
timeInfo.tm_year = (int)(year - 1900); //see time.h -- it's years since 1900
timeInfo.tm_wday = -1;
timeInfo.tm_yday = -1;
timeInfo.tm_isdst = -1;
timeInfo.tm_gmtoff = 0;//[timeZone secondsFromGMT];
timeInfo.tm_zone = nil;
NSTimeInterval rawTime = (NSTimeInterval)(timegm(&timeInfo) - timeZoneOffset); //timegm instead of mktime (which uses local time zone)
if (rawTime == (time_t)ULONG_MAX) {
/*NSCalendar is super-amazingly-slow (which is partly why RSDateParser exists), so this is used only when the date is far enough in the future (19 January 2038 03:14:08Z on 32-bit systems) that timegm fails. If profiling says that this is a performance issue, then you've got a weird app that needs to work with dates far in the future.*/
NSDateComponents *dateComponents = [NSDateComponents new];
dateComponents.timeZone = [NSTimeZone timeZoneForSecondsFromGMT:timeZoneOffset];
dateComponents.year = year;
dateComponents.month = month;
dateComponents.day = day;
dateComponents.hour = hour;
dateComponents.minute = minute;
dateComponents.second = second + (milliseconds / 1000);
return [[NSCalendar autoupdatingCurrentCalendar] dateFromComponents:dateComponents];
}
if (milliseconds > 0) {
rawTime += ((float)milliseconds / 1000.0f);
}
return [NSDate dateWithTimeIntervalSince1970:rawTime];
}
#pragma mark - Standard Formats
static NSDate *RSParsePubDateWithBytes(const char *bytes, NSUInteger numberOfBytes) {
/*@"EEE',' dd MMM yyyy HH':'mm':'ss ZZZ"
@"EEE, dd MMM yyyy HH:mm:ss zzz"
@"dd MMM yyyy HH:mm zzz"
@"dd MMM yyyy HH:mm ZZZ"
@"EEE, dd MMM yyyy"
@"EEE, dd MMM yyyy HH:mm zzz"
etc.*/
NSUInteger finalIndex = 0;
NSInteger day = 1;
NSInteger month = RSJanuary;
NSInteger year = 1970;
NSInteger hour = 0;
NSInteger minute = 0;
NSInteger second = 0;
NSInteger timeZoneOffset = 0;
day = nextNumericValue(bytes, numberOfBytes, 0, 2, &finalIndex);
if (day < 1 || day == NSNotFound)
day = 1;
month = nextMonthValue(bytes, numberOfBytes, finalIndex + 1, &finalIndex);
year = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 4, &finalIndex);
hour = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
if (hour == NSNotFound)
hour = 0;
minute = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
if (minute == NSNotFound)
minute = 0;
NSUInteger currentIndex = finalIndex + 1;
BOOL hasSeconds = (currentIndex < numberOfBytes) && (bytes[currentIndex] == ':');
if (hasSeconds)
second = nextNumericValue(bytes, numberOfBytes, currentIndex, 2, &finalIndex);
currentIndex = finalIndex + 1;
BOOL hasTimeZone = (currentIndex < numberOfBytes) && (bytes[currentIndex] == ' ');
if (hasTimeZone)
timeZoneOffset = parsedTimeZoneOffset(bytes, numberOfBytes, currentIndex);
return dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(year, month, day, hour, minute, second, 0, timeZoneOffset);
}
static NSDate *RSParseW3CWithBytes(const char *bytes, NSUInteger numberOfBytes) {
/*@"yyyy'-'MM'-'dd'T'HH':'mm':'ss"
@"yyyy-MM-dd'T'HH:mm:sszzz"
@"yyyy-MM-dd'T'HH:mm:ss'.'SSSzzz"
etc.*/
NSUInteger finalIndex = 0;
NSInteger day = 1;
NSInteger month = RSJanuary;
NSInteger year = 1970;
NSInteger hour = 0;
NSInteger minute = 0;
NSInteger second = 0;
NSInteger milliseconds = 0;
NSInteger timeZoneOffset = 0;
year = nextNumericValue(bytes, numberOfBytes, 0, 4, &finalIndex);
month = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
day = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
hour = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
minute = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
second = nextNumericValue(bytes, numberOfBytes, finalIndex + 1, 2, &finalIndex);
NSUInteger currentIndex = finalIndex + 1;
BOOL hasMilliseconds = (currentIndex < numberOfBytes) && (bytes[currentIndex] == '.');
if (hasMilliseconds) {
milliseconds = nextNumericValue(bytes, numberOfBytes, currentIndex, 3, &finalIndex);
currentIndex = finalIndex + 1;
}
timeZoneOffset = parsedTimeZoneOffset(bytes, numberOfBytes, currentIndex);
return dateWithYearMonthDayHourMinuteSecondAndTimeZoneOffset(year, month, day, hour, minute, second, milliseconds, timeZoneOffset);
}
static BOOL dateIsPubDate(const char *bytes, NSUInteger numberOfBytes) {
NSUInteger i = 0;
for (i = 0; i < numberOfBytes; i++) {
if (bytes[i] == ' ' || bytes[i] == ',')
return YES;
}
return NO;
}
static BOOL numberOfBytesIsOutsideReasonableRange(NSUInteger numberOfBytes) {
return numberOfBytes < 6 || numberOfBytes > 50;
}
#pragma mark - API
NSDate *RSDateWithBytes(const char *bytes, NSUInteger numberOfBytes) {
if (numberOfBytesIsOutsideReasonableRange(numberOfBytes))
return nil;
if (dateIsPubDate(bytes, numberOfBytes))
return RSParsePubDateWithBytes(bytes, numberOfBytes);
return RSParseW3CWithBytes(bytes, numberOfBytes);
}
NSDate *RSDateWithString(NSString *dateString) {
const char *utf8String = [dateString UTF8String];
return RSDateWithBytes(utf8String, strlen(utf8String));
}

View File

@ -0,0 +1,31 @@
//
// RSXMLInternal.h
// RSXML
//
// Created by Brent Simmons on 12/26/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
NS_ASSUME_NONNULL_BEGIN
BOOL RSXMLIsEmpty(id _Nullable obj);
BOOL RSXMLStringIsEmpty(NSString * _Nullable s);
@interface NSString (RSXMLInternal)
- (NSString *)rsxml_md5HashString;
@end
@interface NSDictionary (RSXMLInternal)
- (nullable id)rsxml_objectForCaseInsensitiveKey:(NSString *)key;
@end
NS_ASSUME_NONNULL_END

View File

@ -0,0 +1,83 @@
//
// RSXMLInternal.m
// RSXML
//
// Created by Brent Simmons on 12/26/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import <CommonCrypto/CommonDigest.h>
#import "RSXMLInternal.h"
static BOOL RSXMLIsNil(id obj) {
return obj == nil || obj == [NSNull null];
}
BOOL RSXMLIsEmpty(id obj) {
if (RSXMLIsNil(obj)) {
return YES;
}
if ([obj respondsToSelector:@selector(count)]) {
return [obj count] < 1;
}
if ([obj respondsToSelector:@selector(length)]) {
return [obj length] < 1;
}
return NO; /*Shouldn't get here very often.*/
}
BOOL RSXMLStringIsEmpty(NSString *s) {
return RSXMLIsNil(s) || s.length < 1;
}
@implementation NSString (RSXMLInternal)
- (NSData *)rsxml_md5Hash {
NSData *data = [self dataUsingEncoding:NSUTF8StringEncoding];
unsigned char hash[CC_MD5_DIGEST_LENGTH];
CC_MD5(data.bytes, (CC_LONG)data.length, hash);
return [NSData dataWithBytes:(const void *)hash length:CC_MD5_DIGEST_LENGTH];
}
- (NSString *)rsxml_md5HashString {
NSData *md5Data = [self rsxml_md5Hash];
const Byte *bytes = md5Data.bytes;
return [NSString stringWithFormat:@"%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x", bytes[0], bytes[1], bytes[2], bytes[3], bytes[4], bytes[5], bytes[6], bytes[7], bytes[8], bytes[9], bytes[10], bytes[11], bytes[12], bytes[13], bytes[14], bytes[15]];
}
@end
@implementation NSDictionary (RSXMLInternal)
- (nullable id)rsxml_objectForCaseInsensitiveKey:(NSString *)key {
id obj = self[key];
if (obj) {
return obj;
}
for (NSString *oneKey in self.allKeys) {
if ([oneKey isKindOfClass:[NSString class]] && [key caseInsensitiveCompare:oneKey] == NSOrderedSame) {
return self[oneKey];
}
}
return nil;
}
@end

View File

@ -0,0 +1,13 @@
//
// RSAtomParser.h
// RSXML
//
// Created by Brent Simmons on 1/15/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import "FeedParser.h"
@interface RSAtomParser : NSObject <FeedParser>
@end

View File

@ -0,0 +1,604 @@
//
// RSAtomParser.m
// RSXML
//
// Created by Brent Simmons on 1/15/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import <libxml/xmlstring.h>
#import "RSAtomParser.h"
#import "RSSAXParser.h"
#import "FeedParser.h"
#import "RSParsedFeed.h"
#import "RSParsedArticle.h"
#import "RSXMLData.h"
#import "NSString+RSXML.h"
#import "RSDateParser.h"
@interface RSAtomParser () <RSSAXParserDelegate>
@property (nonatomic) NSData *feedData;
@property (nonatomic) NSString *urlString;
@property (nonatomic) BOOL endFeedFound;
@property (nonatomic) BOOL parsingXHTML;
@property (nonatomic) BOOL parsingSource;
@property (nonatomic) BOOL parsingArticle;
@property (nonatomic) BOOL parsingAuthor;
@property (nonatomic) NSMutableArray *attributesStack;
@property (nonatomic, readonly) NSDictionary *currentAttributes;
@property (nonatomic) NSMutableString *xhtmlString;
@property (nonatomic) NSString *link;
@property (nonatomic) NSString *title;
@property (nonatomic) NSMutableArray *articles;
@property (nonatomic) NSDate *dateParsed;
@property (nonatomic) RSSAXParser *parser;
@property (nonatomic, readonly) RSParsedArticle *currentArticle;
@property (nonatomic, readonly) NSDate *currentDate;
@end
@implementation RSAtomParser
#pragma mark - Class Methods
+ (BOOL)canParseFeed:(RSXMLData *)xmlData {
// Checking for '<feed' and '<entry' within first n characters should do it.
@autoreleasepool {
NSData *feedData = xmlData.data;
NSString *s = [[NSString alloc] initWithBytesNoCopy:(void *)feedData.bytes length:feedData.length encoding:NSUTF8StringEncoding freeWhenDone:NO];
if (!s) {
s = [[NSString alloc] initWithData:feedData encoding:NSUTF8StringEncoding];
}
if (!s) {
s = [[NSString alloc] initWithData:feedData encoding:NSUnicodeStringEncoding];
}
if (!s) {
return NO;
}
static const NSInteger numberOfCharactersToSearch = 4096;
NSRange rangeToSearch = NSMakeRange(0, numberOfCharactersToSearch);
if (s.length < numberOfCharactersToSearch) {
rangeToSearch.length = s.length;
}
NSRange feedRange = [s rangeOfString:@"<feed" options:NSLiteralSearch range:rangeToSearch];
NSRange entryRange = [s rangeOfString:@"<entry" options:NSLiteralSearch range:rangeToSearch];
if (feedRange.length < 1 || entryRange.length < 1) {
return NO;
}
if (feedRange.location > entryRange.location) {
return NO; // Wrong order.
}
}
return YES;
}
#pragma mark - Init
- (instancetype)initWithXMLData:(RSXMLData *)xmlData {
self = [super init];
if (!self) {
return nil;
}
_feedData = xmlData.data;
_urlString = xmlData.urlString;
_parser = [[RSSAXParser alloc] initWithDelegate:self];
_attributesStack = [NSMutableArray new];
_articles = [NSMutableArray new];
return self;
}
#pragma mark - API
- (RSParsedFeed *)parseFeed:(NSError **)error {
[self parse];
RSParsedFeed *parsedFeed = [[RSParsedFeed alloc] initWithURLString:self.urlString title:self.title link:self.link articles:self.articles];
return parsedFeed;
}
#pragma mark - Constants
static NSString *kTypeKey = @"type";
static NSString *kXHTMLType = @"xhtml";
static NSString *kRelKey = @"rel";
static NSString *kAlternateValue = @"alternate";
static NSString *kHrefKey = @"href";
static NSString *kXMLKey = @"xml";
static NSString *kBaseKey = @"base";
static NSString *kLangKey = @"lang";
static NSString *kXMLBaseKey = @"xml:base";
static NSString *kXMLLangKey = @"xml:lang";
static NSString *kTextHTMLValue = @"text/html";
static NSString *kRelatedValue = @"related";
static NSString *kShortURLValue = @"shorturl";
static NSString *kHTMLValue = @"html";
static NSString *kEnValue = @"en";
static NSString *kTextValue = @"text";
static NSString *kSelfValue = @"self";
static const char *kID = "id";
static const NSInteger kIDLength = 3;
static const char *kTitle = "title";
static const NSInteger kTitleLength = 6;
static const char *kContent = "content";
static const NSInteger kContentLength = 8;
static const char *kSummary = "summary";
static const NSInteger kSummaryLength = 8;
static const char *kLink = "link";
static const NSInteger kLinkLength = 5;
static const char *kPublished = "published";
static const NSInteger kPublishedLength = 10;
static const char *kUpdated = "updated";
static const NSInteger kUpdatedLength = 8;
static const char *kAuthor = "author";
static const NSInteger kAuthorLength = 7;
static const char *kEntry = "entry";
static const NSInteger kEntryLength = 6;
static const char *kSource = "source";
static const NSInteger kSourceLength = 7;
static const char *kFeed = "feed";
static const NSInteger kFeedLength = 5;
static const char *kType = "type";
static const NSInteger kTypeLength = 5;
static const char *kRel = "rel";
static const NSInteger kRelLength = 4;
static const char *kAlternate = "alternate";
static const NSInteger kAlternateLength = 10;
static const char *kHref = "href";
static const NSInteger kHrefLength = 5;
static const char *kXML = "xml";
static const NSInteger kXMLLength = 4;
static const char *kBase = "base";
static const NSInteger kBaseLength = 5;
static const char *kLang = "lang";
static const NSInteger kLangLength = 5;
static const char *kTextHTML = "text/html";
static const NSInteger kTextHTMLLength = 10;
static const char *kRelated = "related";
static const NSInteger kRelatedLength = 8;
static const char *kShortURL = "shorturl";
static const NSInteger kShortURLLength = 9;
static const char *kHTML = "html";
static const NSInteger kHTMLLength = 5;
static const char *kEn = "en";
static const NSInteger kEnLength = 3;
static const char *kText = "text";
static const NSInteger kTextLength = 5;
static const char *kSelf = "self";
static const NSInteger kSelfLength = 5;
#pragma mark - Parsing
- (void)parse {
self.dateParsed = [NSDate date];
@autoreleasepool {
[self.parser parseData:self.feedData];
[self.parser finishParsing];
}
// Optimization: make articles do calculations on this background thread.
[self.articles makeObjectsPerformSelector:@selector(calculateArticleID)];
}
- (void)addArticle {
RSParsedArticle *article = [[RSParsedArticle alloc] initWithFeedURL:self.urlString];
article.dateParsed = self.dateParsed;
[self.articles addObject:article];
}
- (RSParsedArticle *)currentArticle {
return self.articles.lastObject;
}
- (NSDictionary *)currentAttributes {
return self.attributesStack.lastObject;
}
- (NSDate *)currentDate {
return RSDateWithBytes(self.parser.currentCharacters.bytes, self.parser.currentCharacters.length);
}
- (void)addFeedLink {
if (self.link && self.link.length > 0) {
return;
}
NSString *related = self.currentAttributes[kRelKey];
if (related == kAlternateValue) {
self.link = self.currentAttributes[kHrefKey];
}
}
- (void)addFeedTitle {
if (self.title.length < 1) {
self.title = self.parser.currentStringWithTrimmedWhitespace;
}
}
- (void)addLink {
NSString *urlString = self.currentAttributes[kHrefKey];
if (urlString.length < 1) {
return;
}
NSString *rel = self.currentAttributes[kRelKey];
if (rel.length < 1) {
rel = kAlternateValue;
}
if (rel == kAlternateValue) {
if (!self.currentArticle.link) {
self.currentArticle.link = urlString;
}
}
else if (rel == kRelatedValue) {
if (!self.currentArticle.permalink) {
self.currentArticle.permalink = urlString;
}
}
}
- (void)addContent {
self.currentArticle.body = [self currentStringWithHTMLEntitiesDecoded];
}
- (void)addSummary {
if (!self.currentArticle.body) {
self.currentArticle.body = [self currentStringWithHTMLEntitiesDecoded];
}
}
- (NSString *)currentStringWithHTMLEntitiesDecoded {
return [self.parser.currentStringWithTrimmedWhitespace rs_stringByDecodingHTMLEntities];
}
- (void)addArticleElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix {
if (prefix) {
return;
}
if (RSSAXEqualTags(localName, kID, kIDLength)) {
self.currentArticle.guid = self.parser.currentStringWithTrimmedWhitespace;
}
else if (RSSAXEqualTags(localName, kTitle, kTitleLength)) {
self.currentArticle.title = [self currentStringWithHTMLEntitiesDecoded];
}
else if (RSSAXEqualTags(localName, kContent, kContentLength)) {
[self addContent];
}
else if (RSSAXEqualTags(localName, kSummary, kSummaryLength)) {
[self addSummary];
}
else if (RSSAXEqualTags(localName, kLink, kLinkLength)) {
[self addLink];
}
else if (RSSAXEqualTags(localName, kPublished, kPublishedLength)) {
self.currentArticle.datePublished = self.currentDate;
}
else if (RSSAXEqualTags(localName, kUpdated, kUpdatedLength)) {
self.currentArticle.dateModified = self.currentDate;
}
}
- (void)addXHTMLTag:(const xmlChar *)localName {
if (!localName) {
return;
}
[self.xhtmlString appendString:@"<"];
[self.xhtmlString appendString:[NSString stringWithUTF8String:(const char *)localName]];
if (self.currentAttributes.count < 1) {
[self.xhtmlString appendString:@">"];
return;
}
for (NSString *oneKey in self.currentAttributes) {
[self.xhtmlString appendString:@" "];
NSString *oneValue = self.currentAttributes[oneKey];
[self.xhtmlString appendString:oneKey];
[self.xhtmlString appendString:@"=\""];
oneValue = [oneValue stringByReplacingOccurrencesOfString:@"\"" withString:@"&quot;"];
[self.xhtmlString appendString:oneValue];
[self.xhtmlString appendString:@"\""];
}
[self.xhtmlString appendString:@">"];
}
#pragma mark - RSSAXParserDelegate
- (void)saxParser:(RSSAXParser *)SAXParser XMLStartElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri numberOfNamespaces:(NSInteger)numberOfNamespaces namespaces:(const xmlChar **)namespaces numberOfAttributes:(NSInteger)numberOfAttributes numberDefaulted:(int)numberDefaulted attributes:(const xmlChar **)attributes {
if (self.endFeedFound) {
return;
}
NSDictionary *xmlAttributes = [self.parser attributesDictionary:attributes numberOfAttributes:numberOfAttributes];
if (!xmlAttributes) {
xmlAttributes = [NSDictionary dictionary];
}
[self.attributesStack addObject:xmlAttributes];
if (self.parsingXHTML) {
[self addXHTMLTag:localName];
return;
}
if (RSSAXEqualTags(localName, kEntry, kEntryLength)) {
self.parsingArticle = YES;
[self addArticle];
return;
}
if (RSSAXEqualTags(localName, kAuthor, kAuthorLength)) {
self.parsingAuthor = YES;
return;
}
if (RSSAXEqualTags(localName, kSource, kSourceLength)) {
self.parsingSource = YES;
return;
}
BOOL isContentTag = RSSAXEqualTags(localName, kContent, kContentLength);
BOOL isSummaryTag = RSSAXEqualTags(localName, kSummary, kSummaryLength);
if (self.parsingArticle && (isContentTag || isSummaryTag)) {
NSString *contentType = xmlAttributes[kTypeKey];
if ([contentType isEqualToString:kXHTMLType]) {
self.parsingXHTML = YES;
self.xhtmlString = [NSMutableString stringWithString:@""];
return;
}
}
if (!self.parsingArticle && RSSAXEqualTags(localName, kLink, kLinkLength)) {
[self addFeedLink];
return;
}
[self.parser beginStoringCharacters];
}
- (void)saxParser:(RSSAXParser *)SAXParser XMLEndElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri {
if (RSSAXEqualTags(localName, kFeed, kFeedLength)) {
self.endFeedFound = YES;
return;
}
if (self.endFeedFound) {
return;
}
if (self.parsingXHTML) {
BOOL isContentTag = RSSAXEqualTags(localName, kContent, kContentLength);
BOOL isSummaryTag = RSSAXEqualTags(localName, kSummary, kSummaryLength);
if (self.parsingArticle && (isContentTag || isSummaryTag)) {
if (isContentTag) {
self.currentArticle.body = [self.xhtmlString copy];
}
else if (isSummaryTag) {
if (self.currentArticle.body.length < 1) {
self.currentArticle.body = [self.xhtmlString copy];
}
}
}
if (isContentTag || isSummaryTag) {
self.parsingXHTML = NO;
}
[self.xhtmlString appendString:@"</"];
[self.xhtmlString appendString:[NSString stringWithUTF8String:(const char *)localName]];
[self.xhtmlString appendString:@">"];
}
else if (RSSAXEqualTags(localName, kAuthor, kAuthorLength)) {
self.parsingAuthor = NO;
}
else if (RSSAXEqualTags(localName, kEntry, kEntryLength)) {
self.parsingArticle = NO;
}
else if (self.parsingArticle && !self.parsingSource) {
[self addArticleElement:localName prefix:prefix];
}
else if (RSSAXEqualTags(localName, kSource, kSourceLength)) {
self.parsingSource = NO;
}
else if (!self.parsingArticle && !self.parsingSource && RSSAXEqualTags(localName, kTitle, kTitleLength)) {
[self addFeedTitle];
}
[self.attributesStack removeLastObject];
}
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForName:(const xmlChar *)name prefix:(const xmlChar *)prefix {
if (prefix && RSSAXEqualTags(prefix, kXML, kXMLLength)) {
if (RSSAXEqualTags(name, kBase, kBaseLength)) {
return kXMLBaseKey;
}
if (RSSAXEqualTags(name, kLang, kLangLength)) {
return kXMLLangKey;
}
}
if (prefix) {
return nil;
}
if (RSSAXEqualTags(name, kRel, kRelLength)) {
return kRelKey;
}
if (RSSAXEqualTags(name, kType, kTypeLength)) {
return kTypeKey;
}
if (RSSAXEqualTags(name, kHref, kHrefLength)) {
return kHrefKey;
}
if (RSSAXEqualTags(name, kAlternate, kAlternateLength)) {
return kAlternateValue;
}
return nil;
}
static BOOL equalBytes(const void *bytes1, const void *bytes2, NSUInteger length) {
return memcmp(bytes1, bytes2, length) == 0;
}
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForValue:(const void *)bytes length:(NSUInteger)length {
static const NSUInteger alternateLength = kAlternateLength - 1;
static const NSUInteger textHTMLLength = kTextHTMLLength - 1;
static const NSUInteger relatedLength = kRelatedLength - 1;
static const NSUInteger shortURLLength = kShortURLLength - 1;
static const NSUInteger htmlLength = kHTMLLength - 1;
static const NSUInteger enLength = kEnLength - 1;
static const NSUInteger textLength = kTextLength - 1;
static const NSUInteger selfLength = kSelfLength - 1;
if (length == alternateLength && equalBytes(bytes, kAlternate, alternateLength)) {
return kAlternateValue;
}
if (length == textHTMLLength && equalBytes(bytes, kTextHTML, textHTMLLength)) {
return kTextHTMLValue;
}
if (length == relatedLength && equalBytes(bytes, kRelated, relatedLength)) {
return kRelatedValue;
}
if (length == shortURLLength && equalBytes(bytes, kShortURL, shortURLLength)) {
return kShortURLValue;
}
if (length == htmlLength && equalBytes(bytes, kHTML, htmlLength)) {
return kHTMLValue;
}
if (length == enLength && equalBytes(bytes, kEn, enLength)) {
return kEnValue;
}
if (length == textLength && equalBytes(bytes, kText, textLength)) {
return kTextValue;
}
if (length == selfLength && equalBytes(bytes, kSelf, selfLength)) {
return kSelfValue;
}
return nil;
}
- (void)saxParser:(RSSAXParser *)SAXParser XMLCharactersFound:(const unsigned char *)characters length:(NSUInteger)length {
if (self.parsingXHTML) {
[self.xhtmlString appendString:[[NSString alloc] initWithBytesNoCopy:(void *)characters length:length encoding:NSUTF8StringEncoding freeWhenDone:NO]];
}
}
@end

View File

@ -0,0 +1,13 @@
//
// RSRSSParser.h
// RSXML
//
// Created by Brent Simmons on 1/6/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import "FeedParser.h"
@interface RSRSSParser : NSObject <FeedParser>
@end

View File

@ -0,0 +1,469 @@
//
// RSRSSParser.m
// RSXML
//
// Created by Brent Simmons on 1/6/15.
// Copyright (c) 2015 Ranchero Software LLC. All rights reserved.
//
#import <libxml/xmlstring.h>
#import "RSRSSParser.h"
#import "RSSAXParser.h"
#import "RSParsedFeed.h"
#import "RSParsedArticle.h"
#import "RSXMLData.h"
#import "RSXMLInternal.h"
#import "NSString+RSXML.h"
#import "RSDateParser.h"
@interface RSRSSParser () <RSSAXParserDelegate>
@property (nonatomic) NSData *feedData;
@property (nonatomic) NSString *urlString;
@property (nonatomic) NSDictionary *currentAttributes;
@property (nonatomic) RSSAXParser *parser;
@property (nonatomic) NSMutableArray *articles;
@property (nonatomic) BOOL parsingArticle;
@property (nonatomic, readonly) RSParsedArticle *currentArticle;
@property (nonatomic) BOOL parsingChannelImage;
@property (nonatomic, readonly) NSDate *currentDate;
@property (nonatomic) BOOL endRSSFound;
@property (nonatomic) NSString *link;
@property (nonatomic) NSString *title;
@property (nonatomic) NSDate *dateParsed;
@end
@implementation RSRSSParser
#pragma mark - Class Methods
+ (BOOL)canParseFeed:(RSXMLData *)xmlData {
// Checking for '<rss' and '<channel>' within first n characters should do it.
// TODO: handle RSS 1.0
@autoreleasepool {
NSData *feedData = xmlData.data;
NSString *s = [[NSString alloc] initWithBytesNoCopy:(void *)feedData.bytes length:feedData.length encoding:NSUTF8StringEncoding freeWhenDone:NO];
if (!s) {
s = [[NSString alloc] initWithData:feedData encoding:NSUTF8StringEncoding];
}
if (!s) {
s = [[NSString alloc] initWithData:feedData encoding:NSUnicodeStringEncoding];
}
if (!s) {
return NO;
}
static const NSInteger numberOfCharactersToSearch = 4096;
NSRange rangeToSearch = NSMakeRange(0, numberOfCharactersToSearch);
if (s.length < numberOfCharactersToSearch) {
rangeToSearch.length = s.length;
}
NSRange rssRange = [s rangeOfString:@"<rss" options:NSLiteralSearch range:rangeToSearch];
NSRange channelRange = [s rangeOfString:@"<channel>" options:NSLiteralSearch range:rangeToSearch];
if (rssRange.length < 1 || channelRange.length < 1) {
return NO;
}
if (rssRange.location > channelRange.location) {
return NO; // Wrong order.
}
}
return YES;
}
#pragma mark - Init
- (instancetype)initWithXMLData:(RSXMLData *)xmlData {
self = [super init];
if (!self) {
return nil;
}
_feedData = xmlData.data;
_urlString = xmlData.urlString;
_parser = [[RSSAXParser alloc] initWithDelegate:self];
_articles = [NSMutableArray new];
return self;
}
#pragma mark - API
- (RSParsedFeed *)parseFeed:(NSError **)error {
[self parse];
RSParsedFeed *parsedFeed = [[RSParsedFeed alloc] initWithURLString:self.urlString title:self.title link:self.link articles:self.articles];
return parsedFeed;
}
#pragma mark - Constants
static NSString *kIsPermaLinkKey = @"isPermaLink";
static NSString *kURLKey = @"url";
static NSString *kLengthKey = @"length";
static NSString *kTypeKey = @"type";
static NSString *kFalseValue = @"false";
static NSString *kTrueValue = @"true";
static NSString *kContentEncodedKey = @"content:encoded";
static NSString *kDCDateKey = @"dc:date";
static NSString *kDCCreatorKey = @"dc:creator";
static NSString *kRDFAboutKey = @"rdf:about";
static const char *kItem = "item";
static const NSInteger kItemLength = 5;
static const char *kImage = "image";
static const NSInteger kImageLength = 6;
static const char *kLink = "link";
static const NSInteger kLinkLength = 5;
static const char *kTitle = "title";
static const NSInteger kTitleLength = 6;
static const char *kDC = "dc";
static const NSInteger kDCLength = 3;
static const char *kCreator = "creator";
static const NSInteger kCreatorLength = 8;
static const char *kDate = "date";
static const NSInteger kDateLength = 5;
static const char *kContent = "content";
static const NSInteger kContentLength = 8;
static const char *kEncoded = "encoded";
static const NSInteger kEncodedLength = 8;
static const char *kGuid = "guid";
static const NSInteger kGuidLength = 5;
static const char *kPubDate = "pubDate";
static const NSInteger kPubDateLength = 8;
static const char *kAuthor = "author";
static const NSInteger kAuthorLength = 7;
static const char *kDescription = "description";
static const NSInteger kDescriptionLength = 12;
static const char *kRSS = "rss";
static const NSInteger kRSSLength = 4;
static const char *kURL = "url";
static const NSInteger kURLLength = 4;
static const char *kLength = "length";
static const NSInteger kLengthLength = 7;
static const char *kType = "type";
static const NSInteger kTypeLength = 5;
static const char *kIsPermaLink = "isPermaLink";
static const NSInteger kIsPermaLinkLength = 12;
static const char *kRDF = "rdf";
static const NSInteger kRDFlength = 4;
static const char *kAbout = "about";
static const NSInteger kAboutLength = 6;
static const char *kFalse = "false";
static const NSInteger kFalseLength = 6;
static const char *kTrue = "true";
static const NSInteger kTrueLength = 5;
#pragma mark - Parsing
- (void)parse {
self.dateParsed = [NSDate date];
@autoreleasepool {
[self.parser parseData:self.feedData];
[self.parser finishParsing];
}
// Optimization: make articles do calculations on this background thread.
[self.articles makeObjectsPerformSelector:@selector(calculateArticleID)];
}
- (void)addArticle {
RSParsedArticle *article = [[RSParsedArticle alloc] initWithFeedURL:self.urlString];
article.dateParsed = self.dateParsed;
[self.articles addObject:article];
}
- (RSParsedArticle *)currentArticle {
return self.articles.lastObject;
}
- (void)addFeedElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix {
if (prefix != NULL) {
return;
}
if (RSSAXEqualTags(localName, kLink, kLinkLength)) {
if (!self.link) {
self.link = self.parser.currentStringWithTrimmedWhitespace;
}
}
else if (RSSAXEqualTags(localName, kTitle, kTitleLength)) {
self.title = self.parser.currentStringWithTrimmedWhitespace;
}
}
- (void)addDCElement:(const xmlChar *)localName {
if (RSSAXEqualTags(localName, kCreator, kCreatorLength)) {
self.currentArticle.author = self.parser.currentStringWithTrimmedWhitespace;
}
else if (RSSAXEqualTags(localName, kDate, kDateLength)) {
self.currentArticle.datePublished = self.currentDate;
}
}
- (void)addGuid {
self.currentArticle.guid = self.parser.currentStringWithTrimmedWhitespace;
NSString *isPermaLinkValue = [self.currentAttributes rsxml_objectForCaseInsensitiveKey:@"ispermalink"];
if (!isPermaLinkValue || ![isPermaLinkValue isEqualToString:@"false"]) {
self.currentArticle.permalink = [self urlString:self.currentArticle.guid];
}
}
- (NSString *)urlString:(NSString *)s {
/*Resolve against home page URL (if available) or feed URL.*/
if ([[s lowercaseString] hasPrefix:@"http"]) {
return s;
}
if (!self.link) {
//TODO: get feed URL and use that to resolve URL.*/
return s;
}
NSURL *baseURL = [NSURL URLWithString:self.link];
if (!baseURL) {
return s;
}
NSURL *resolvedURL = [NSURL URLWithString:s relativeToURL:baseURL];
if (resolvedURL.absoluteString) {
return resolvedURL.absoluteString;
}
return s;
}
- (NSString *)currentStringWithHTMLEntitiesDecoded {
return [self.parser.currentStringWithTrimmedWhitespace rs_stringByDecodingHTMLEntities];
}
- (void)addArticleElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix {
if (RSSAXEqualTags(prefix, kDC, kDCLength)) {
[self addDCElement:localName];
return;
}
if (RSSAXEqualTags(prefix, kContent, kContentLength) && RSSAXEqualTags(localName, kEncoded, kEncodedLength)) {
self.currentArticle.body = [self currentStringWithHTMLEntitiesDecoded];
return;
}
if (prefix != NULL) {
return;
}
if (RSSAXEqualTags(localName, kGuid, kGuidLength)) {
[self addGuid];
}
else if (RSSAXEqualTags(localName, kPubDate, kPubDateLength)) {
self.currentArticle.datePublished = self.currentDate;
}
else if (RSSAXEqualTags(localName, kAuthor, kAuthorLength)) {
self.currentArticle.author = self.parser.currentStringWithTrimmedWhitespace;
}
else if (RSSAXEqualTags(localName, kLink, kLinkLength)) {
self.currentArticle.link = [self urlString:self.parser.currentStringWithTrimmedWhitespace];
}
else if (RSSAXEqualTags(localName, kDescription, kDescriptionLength)) {
if (!self.currentArticle.body) {
self.currentArticle.body = [self currentStringWithHTMLEntitiesDecoded];
}
}
else if (RSSAXEqualTags(localName, kTitle, kTitleLength)) {
self.currentArticle.title = [self currentStringWithHTMLEntitiesDecoded];
}
}
- (NSDate *)currentDate {
return RSDateWithBytes(self.parser.currentCharacters.bytes, self.parser.currentCharacters.length);
}
#pragma mark - RSSAXParserDelegate
- (void)saxParser:(RSSAXParser *)SAXParser XMLStartElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri numberOfNamespaces:(NSInteger)numberOfNamespaces namespaces:(const xmlChar **)namespaces numberOfAttributes:(NSInteger)numberOfAttributes numberDefaulted:(int)numberDefaulted attributes:(const xmlChar **)attributes {
if (self.endRSSFound) {
return;
}
NSDictionary *xmlAttributes = nil;
if (RSSAXEqualTags(localName, kItem, kItemLength) || RSSAXEqualTags(localName, kGuid, kGuidLength)) {
xmlAttributes = [self.parser attributesDictionary:attributes numberOfAttributes:numberOfAttributes];
}
if (self.currentAttributes != xmlAttributes) {
self.currentAttributes = xmlAttributes;
}
if (!prefix && RSSAXEqualTags(localName, kItem, kItemLength)) {
[self addArticle];
self.parsingArticle = YES;
if (xmlAttributes && xmlAttributes[kRDFAboutKey]) { /*RSS 1.0 guid*/
self.currentArticle.guid = xmlAttributes[kRDFAboutKey];
self.currentArticle.permalink = self.currentArticle.guid;
}
}
else if (!prefix && RSSAXEqualTags(localName, kImage, kImageLength)) {
self.parsingChannelImage = YES;
}
if (!self.parsingChannelImage) {
[self.parser beginStoringCharacters];
}
}
- (void)saxParser:(RSSAXParser *)SAXParser XMLEndElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri {
if (self.endRSSFound) {
return;
}
if (RSSAXEqualTags(localName, kRSS, kRSSLength)) {
self.endRSSFound = YES;
}
else if (RSSAXEqualTags(localName, kImage, kImageLength)) {
self.parsingChannelImage = NO;
}
else if (RSSAXEqualTags(localName, kItem, kItemLength)) {
self.parsingArticle = NO;
}
else if (self.parsingArticle) {
[self addArticleElement:localName prefix:prefix];
}
else if (!self.parsingChannelImage) {
[self addFeedElement:localName prefix:prefix];
}
}
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForName:(const xmlChar *)name prefix:(const xmlChar *)prefix {
if (RSSAXEqualTags(prefix, kRDF, kRDFlength)) {
if (RSSAXEqualTags(name, kAbout, kAboutLength)) {
return kRDFAboutKey;
}
return nil;
}
if (prefix) {
return nil;
}
if (RSSAXEqualTags(name, kIsPermaLink, kIsPermaLinkLength)) {
return kIsPermaLinkKey;
}
if (RSSAXEqualTags(name, kURL, kURLLength)) {
return kURLKey;
}
if (RSSAXEqualTags(name, kLength, kLengthLength)) {
return kLengthKey;
}
if (RSSAXEqualTags(name, kType, kTypeLength)) {
return kTypeKey;
}
return nil;
}
static BOOL equalBytes(const void *bytes1, const void *bytes2, NSUInteger length) {
return memcmp(bytes1, bytes2, length) == 0;
}
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForValue:(const void *)bytes length:(NSUInteger)length {
static const NSUInteger falseLength = kFalseLength - 1;
static const NSUInteger trueLength = kTrueLength - 1;
if (length == falseLength && equalBytes(bytes, kFalse, falseLength)) {
return kFalseValue;
}
if (length == trueLength && equalBytes(bytes, kTrue, trueLength)) {
return kTrueValue;
}
return nil;
}
@end

View File

@ -0,0 +1,36 @@
//
// RSOPMLAttributes.h
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
// OPML allows for arbitrary attributes.
// These are the common attributes in OPML files used as RSS subscription lists.
extern NSString *OPMLTextKey; //text
extern NSString *OPMLTitleKey; //title
extern NSString *OPMLDescriptionKey; //description
extern NSString *OPMLTypeKey; //type
extern NSString *OPMLVersionKey; //version
extern NSString *OPMLHMTLURLKey; //htmlUrl
extern NSString *OPMLXMLURLKey; //xmlUrl
@interface NSDictionary (RSOPMLAttributes)
// A frequent error in OPML files is to mess up the capitalization,
// so these do a case-insensitive lookup.
@property (nonatomic, readonly) NSString *opml_text;
@property (nonatomic, readonly) NSString *opml_title;
@property (nonatomic, readonly) NSString *opml_description;
@property (nonatomic, readonly) NSString *opml_type;
@property (nonatomic, readonly) NSString *opml_version;
@property (nonatomic, readonly) NSString *opml_htmlUrl;
@property (nonatomic, readonly) NSString *opml_xmlUrl;
@end

View File

@ -0,0 +1,66 @@
//
// RSOPMLAttributes.m
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSOPMLAttributes.h"
#import "RSXMLInternal.h"
NSString *OPMLTextKey = @"text";
NSString *OPMLTitleKey = @"title";
NSString *OPMLDescriptionKey = @"description";
NSString *OPMLTypeKey = @"type";
NSString *OPMLVersionKey = @"version";
NSString *OPMLHMTLURLKey = @"htmlUrl";
NSString *OPMLXMLURLKey = @"xmlUrl";
@implementation NSDictionary (RSOPMLAttributes)
- (NSString *)opml_text {
return [self rsxml_objectForCaseInsensitiveKey:OPMLTextKey];
}
- (NSString *)opml_title {
return [self rsxml_objectForCaseInsensitiveKey:OPMLTitleKey];
}
- (NSString *)opml_description {
return [self rsxml_objectForCaseInsensitiveKey:OPMLDescriptionKey];
}
- (NSString *)opml_type {
return [self rsxml_objectForCaseInsensitiveKey:OPMLTypeKey];
}
- (NSString *)opml_version {
return [self rsxml_objectForCaseInsensitiveKey:OPMLVersionKey];
}
- (NSString *)opml_htmlUrl {
return [self rsxml_objectForCaseInsensitiveKey:OPMLHMTLURLKey];
}
- (NSString *)opml_xmlUrl {
return [self rsxml_objectForCaseInsensitiveKey:OPMLXMLURLKey];
}
@end

View File

@ -0,0 +1,17 @@
//
// RSOPMLDocument.h
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
#import "RSOPMLItem.h"
@interface RSOPMLDocument : RSOPMLItem
@property (nonatomic) NSString *title;
@end

View File

@ -0,0 +1,13 @@
//
// RSOPMLDocument.m
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSOPMLDocument.h"
@implementation RSOPMLDocument
@end

View File

@ -0,0 +1,23 @@
//
// RSOPMLFeedSpecifier.h
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@interface RSOPMLFeedSpecifier : NSObject
- (instancetype)initWithTitle:(NSString *)title feedDescription:(NSString *)feedDescription homePageURL:(NSString *)homePageURL feedURL:(NSString *)feedURL;
@property (nonatomic, readonly) NSString *title;
@property (nonatomic, readonly) NSString *feedDescription;
@property (nonatomic, readonly) NSString *homePageURL;
@property (nonatomic, readonly) NSString *feedURL;
@end

View File

@ -0,0 +1,50 @@
//
// RSOPMLFeedSpecifier.m
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSOPMLFeedSpecifier.h"
#import "RSXMLInternal.h"
@implementation RSOPMLFeedSpecifier
- (instancetype)initWithTitle:(NSString *)title feedDescription:(NSString *)feedDescription homePageURL:(NSString *)homePageURL feedURL:(NSString *)feedURL {
NSParameterAssert(!RSXMLIsEmpty(feedURL));
self = [super init];
if (!self) {
return nil;
}
if (RSXMLIsEmpty(title)) {
_title = nil;
}
else {
_title = title;
}
if (RSXMLIsEmpty(feedDescription)) {
_feedDescription = nil;
}
else {
_feedDescription = feedDescription;
}
if (RSXMLIsEmpty(homePageURL)) {
_homePageURL = nil;
}
else {
_homePageURL = homePageURL;
}
_feedURL = feedURL;
return self;
}
@end

View File

@ -0,0 +1,26 @@
//
// RSOPMLItem.h
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSOPMLFeedSpecifier;
@interface RSOPMLItem : NSObject
@property (nonatomic) NSDictionary *attributes;
@property (nonatomic) NSArray *children;
- (void)addChild:(RSOPMLItem *)child;
@property (nonatomic, readonly) RSOPMLFeedSpecifier *OPMLFeedSpecifier; //May be nil.
@property (nonatomic, readonly) NSString *titleFromAttributes; //May be nil.
@property (nonatomic, readonly) BOOL isFolder;
@end

View File

@ -0,0 +1,86 @@
//
// RSOPMLItem.m
// RSXML
//
// Created by Brent Simmons on 2/28/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import "RSOPMLItem.h"
#import "RSOPMLAttributes.h"
#import "RSOPMLFeedSpecifier.h"
#import "RSXMLInternal.h"
@interface RSOPMLItem ()
@property (nonatomic) NSMutableArray *mutableChildren;
@end
@implementation RSOPMLItem
@synthesize children = _children;
@synthesize OPMLFeedSpecifier = _OPMLFeedSpecifier;
- (NSArray *)children {
return [self.mutableChildren copy];
}
- (void)setChildren:(NSArray *)children {
_children = children;
self.mutableChildren = [_children mutableCopy];
}
- (void)addChild:(RSOPMLItem *)child {
if (!self.mutableChildren) {
self.mutableChildren = [NSMutableArray new];
}
[self.mutableChildren addObject:child];
}
- (RSOPMLFeedSpecifier *)OPMLFeedSpecifier {
if (_OPMLFeedSpecifier) {
return _OPMLFeedSpecifier;
}
NSString *feedURL = self.attributes.opml_xmlUrl;
if (RSXMLIsEmpty(feedURL)) {
return nil;
}
_OPMLFeedSpecifier = [[RSOPMLFeedSpecifier alloc] initWithTitle:self.attributes.opml_title feedDescription:self.attributes.opml_description homePageURL:self.attributes.opml_htmlUrl feedURL:feedURL];
return _OPMLFeedSpecifier;
}
- (NSString *)titleFromAttributes {
NSString *title = self.attributes.opml_title;
if (title) {
return title;
}
title = self.attributes.opml_text;
if (title) {
return title;
}
return nil;
}
- (BOOL)isFolder {
return self.mutableChildren.count > 0;
}
@end

View File

@ -0,0 +1,29 @@
//
// RSOPMLParser.h
// RSXML
//
// Created by Brent Simmons on 7/12/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
@class RSXMLData;
@class RSOPMLDocument;
typedef void (^RSParsedOPMLBlock)(RSOPMLDocument *OPMLDocument, NSError *error);
void RSParseOPML(RSXMLData *xmlData, RSParsedOPMLBlock callback); //async; calls back on main thread.
@interface RSOPMLParser: NSObject
- (instancetype)initWithXMLData:(RSXMLData *)xmlData;
@property (nonatomic, readonly) RSOPMLDocument *OPMLDocument;
@property (nonatomic, readonly) NSError *error;
@end

View File

@ -0,0 +1,297 @@
//
// RSOPMLParser.m
// RSXML
//
// Created by Brent Simmons on 7/12/15.
// Copyright © 2015 Ranchero Software, LLC. All rights reserved.
//
#import "RSOPMLParser.h"
#import <libxml/xmlstring.h>
#import "RSXMLData.h"
#import "RSSAXParser.h"
#import "RSOPMLItem.h"
#import "RSOPMLDocument.h"
#import "RSOPMLAttributes.h"
#import "RSXMLError.h"
void RSParseOPML(RSXMLData *xmlData, RSParsedOPMLBlock callback) {
NSCParameterAssert(xmlData);
NSCParameterAssert(callback);
dispatch_async(dispatch_get_global_queue(QOS_CLASS_DEFAULT, 0), ^{
@autoreleasepool {
RSOPMLParser *parser = [[RSOPMLParser alloc] initWithXMLData:xmlData];
RSOPMLDocument *document = parser.OPMLDocument;
NSError *error = parser.error;
dispatch_async(dispatch_get_main_queue(), ^{
callback(document, error);
});
}
});
}
@interface RSOPMLParser () <RSSAXParserDelegate>
@property (nonatomic, readwrite) RSOPMLDocument *OPMLDocument;
@property (nonatomic, readwrite) NSError *error;
@property (nonatomic) NSMutableArray *itemStack;
@end
@implementation RSOPMLParser
#pragma mark - Init
- (instancetype)initWithXMLData:(RSXMLData *)XMLData {
self = [super init];
if (!self) {
return nil;
}
[self parse:XMLData];
return self;
}
#pragma mark - Private
- (void)parse:(RSXMLData *)XMLData {
@autoreleasepool {
if (![self canParseData:XMLData.data]) {
NSString *filename = nil;
NSURL *url = [NSURL URLWithString:XMLData.urlString];
if (url && url.isFileURL) {
filename = url.path.lastPathComponent;
}
if ([XMLData.urlString hasPrefix:@"http"]) {
filename = XMLData.urlString;
}
if (!filename) {
filename = XMLData.urlString;
}
self.error = RSOPMLWrongFormatError(filename);
return;
}
RSSAXParser *parser = [[RSSAXParser alloc] initWithDelegate:self];
self.itemStack = [NSMutableArray new];
self.OPMLDocument = [RSOPMLDocument new];
[self pushItem:self.OPMLDocument];
[parser parseData:XMLData.data];
[parser finishParsing];
}
}
- (BOOL)canParseData:(NSData *)d {
// Check for <opml and <outline near the top.
@autoreleasepool {
NSString *s = [[NSString alloc] initWithBytesNoCopy:(void *)d.bytes length:d.length encoding:NSUTF8StringEncoding freeWhenDone:NO];
if (!s) {
NSDictionary *options = @{NSStringEncodingDetectionSuggestedEncodingsKey : @[@(NSUTF8StringEncoding)]};
(void)[NSString stringEncodingForData:d encodingOptions:options convertedString:&s usedLossyConversion:nil];
}
if (!s) {
return NO;
}
static const NSInteger numberOfCharactersToSearch = 4096;
NSRange rangeToSearch = NSMakeRange(0, numberOfCharactersToSearch);
if (s.length < numberOfCharactersToSearch) {
rangeToSearch.length = s.length;
}
NSRange opmlRange = [s rangeOfString:@"<opml" options:NSCaseInsensitiveSearch range:rangeToSearch];
if (opmlRange.length < 1) {
return NO;
}
NSRange outlineRange = [s rangeOfString:@"<outline" options:NSLiteralSearch range:rangeToSearch];
if (outlineRange.length < 1) {
return NO;
}
if (outlineRange.location < opmlRange.location) {
return NO;
}
}
return YES;
}
- (void)pushItem:(RSOPMLItem *)item {
[self.itemStack addObject:item];
}
- (void)popItem {
NSAssert(self.itemStack.count > 0, nil);
/*If itemStack is empty, bad things are happening.
But we still shouldn't crash in production.*/
if (self.itemStack.count > 0) {
[self.itemStack removeLastObject];
}
}
- (RSOPMLItem *)currentItem {
return self.itemStack.lastObject;
}
#pragma mark - RSSAXParserDelegate
static const char *kOutline = "outline";
static const char kOutlineLength = 8;
- (void)saxParser:(RSSAXParser *)SAXParser XMLStartElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri numberOfNamespaces:(NSInteger)numberOfNamespaces namespaces:(const xmlChar **)namespaces numberOfAttributes:(NSInteger)numberOfAttributes numberDefaulted:(int)numberDefaulted attributes:(const xmlChar **)attributes {
if (!RSSAXEqualTags(localName, kOutline, kOutlineLength)) {
return;
}
RSOPMLItem *item = [RSOPMLItem new];
item.attributes = [SAXParser attributesDictionary:attributes numberOfAttributes:numberOfAttributes];
[[self currentItem] addChild:item];
[self pushItem:item];
}
- (void)saxParser:(RSSAXParser *)SAXParser XMLEndElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri {
if (RSSAXEqualTags(localName, kOutline, kOutlineLength)) {
[self popItem];
}
}
static const char *kText = "text";
static const NSInteger kTextLength = 5;
static const char *kTitle = "title";
static const NSInteger kTitleLength = 6;
static const char *kDescription = "description";
static const NSInteger kDescriptionLength = 12;
static const char *kType = "type";
static const NSInteger kTypeLength = 5;
static const char *kVersion = "version";
static const NSInteger kVersionLength = 8;
static const char *kHTMLURL = "htmlUrl";
static const NSInteger kHTMLURLLength = 8;
static const char *kXMLURL = "xmlUrl";
static const NSInteger kXMLURLLength = 7;
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForName:(const xmlChar *)name prefix:(const xmlChar *)prefix {
if (prefix) {
return nil;
}
size_t nameLength = strlen((const char *)name);
if (nameLength == kTextLength - 1) {
if (RSSAXEqualTags(name, kText, kTextLength)) {
return OPMLTextKey;
}
if (RSSAXEqualTags(name, kType, kTypeLength)) {
return OPMLTypeKey;
}
}
else if (nameLength == kTitleLength - 1) {
if (RSSAXEqualTags(name, kTitle, kTitleLength)) {
return OPMLTitleKey;
}
}
else if (nameLength == kXMLURLLength - 1) {
if (RSSAXEqualTags(name, kXMLURL, kXMLURLLength)) {
return OPMLXMLURLKey;
}
}
else if (nameLength == kVersionLength - 1) {
if (RSSAXEqualTags(name, kVersion, kVersionLength)) {
return OPMLVersionKey;
}
if (RSSAXEqualTags(name, kHTMLURL, kHTMLURLLength)) {
return OPMLHMTLURLKey;
}
}
else if (nameLength == kDescriptionLength - 1) {
if (RSSAXEqualTags(name, kDescription, kDescriptionLength)) {
return OPMLDescriptionKey;
}
}
return nil;
}
static const char *kRSSUppercase = "RSS";
static const char *kRSSLowercase = "rss";
static const NSUInteger kRSSLength = 3;
static NSString *RSSUppercaseValue = @"RSS";
static NSString *RSSLowercaseValue = @"rss";
static NSString *emptyString = @"";
static BOOL equalBytes(const void *bytes1, const void *bytes2, NSUInteger length) {
return memcmp(bytes1, bytes2, length) == 0;
}
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForValue:(const void *)bytes length:(NSUInteger)length {
if (length < 1) {
return emptyString;
}
if (length == kRSSLength) {
if (equalBytes(bytes, kRSSUppercase, kRSSLength)) {
return RSSUppercaseValue;
}
else if (equalBytes(bytes, kRSSLowercase, kRSSLength)) {
return RSSLowercaseValue;
}
}
return nil;
}
@end

View File

@ -0,0 +1,68 @@
//
// RSSAXParser.h
// RSXML
//
// Created by Brent Simmons on 3/25/15.
// Copyright (c) 2015 Ranchero Software, LLC. All rights reserved.
//
@import Foundation;
/*Thread-safe, not re-entrant.
Calls to the delegate will happen on the same thread where the parser runs.
This is a low-level streaming XML parser, a thin wrapper for libxml2's SAX parser. It doesn't do much Foundation-ifying quite on purpose -- because the goal is performance and low memory use.
This class is not meant to be sub-classed. Use the delegate methods.
*/
@class RSSAXParser;
@protocol RSSAXParserDelegate <NSObject>
@optional
- (void)saxParser:(RSSAXParser *)SAXParser XMLStartElement:(const unsigned char *)localName prefix:(const unsigned char *)prefix uri:(const unsigned char *)uri numberOfNamespaces:(NSInteger)numberOfNamespaces namespaces:(const unsigned char **)namespaces numberOfAttributes:(NSInteger)numberOfAttributes numberDefaulted:(int)numberDefaulted attributes:(const unsigned char **)attributes;
- (void)saxParser:(RSSAXParser *)SAXParser XMLEndElement:(const unsigned char *)localName prefix:(const unsigned char *)prefix uri:(const unsigned char *)uri;
- (void)saxParser:(RSSAXParser *)SAXParser XMLCharactersFound:(const unsigned char *)characters length:(NSUInteger)length;
- (void)saxParserDidReachEndOfDocument:(RSSAXParser *)SAXParser; /*If canceled, may not get called (but might).*/
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForName:(const unsigned char *)name prefix:(const unsigned char *)prefix; /*Okay to return nil. Prefix may be nil.*/
- (NSString *)saxParser:(RSSAXParser *)SAXParser internedStringForValue:(const void *)bytes length:(NSUInteger)length;
@end
void RSSAXInitLibXMLParser(void); // Needed by RSSAXHTMLParser.
/*For use by delegate.*/
BOOL RSSAXEqualTags(const unsigned char *localName, const char *tag, NSInteger tagLength);
@interface RSSAXParser : NSObject
- (instancetype)initWithDelegate:(id<RSSAXParserDelegate>)delegate;
- (void)parseData:(NSData *)data;
- (void)parseBytes:(const void *)bytes numberOfBytes:(NSUInteger)numberOfBytes;
- (void)finishParsing;
- (void)cancel;
@property (nonatomic, strong, readonly) NSData *currentCharacters; /*nil if not storing characters. UTF-8 encoded.*/
@property (nonatomic, strong, readonly) NSString *currentString; /*Convenience to get string version of currentCharacters.*/
@property (nonatomic, strong, readonly) NSString *currentStringWithTrimmedWhitespace;
- (void)beginStoringCharacters; /*Delegate can call from XMLStartElement. Characters will be available in XMLEndElement as currentCharacters property. Storing characters is stopped after each XMLEndElement.*/
/*Delegate can call from within XMLStartElement. Returns nil if numberOfAttributes < 1.*/
- (NSDictionary *)attributesDictionary:(const unsigned char **)attributes numberOfAttributes:(NSInteger)numberOfAttributes;
@end

View File

@ -0,0 +1,346 @@
//
// RSSAXParser.m
// RSXML
//
// Created by Brent Simmons on 3/25/15.
// Copyright (c) 2015 Ranchero Software, LLC. All rights reserved.
//
#import <libxml/tree.h>
#import <libxml/xmlstring.h>
#import "RSSAXParser.h"
#import "RSXMLInternal.h"
@interface RSSAXParser ()
@property (nonatomic, weak) id<RSSAXParserDelegate> delegate;
@property (nonatomic, assign) xmlParserCtxtPtr context;
@property (nonatomic, assign) BOOL storingCharacters;
@property (nonatomic) NSMutableData *characters;
@property (nonatomic) BOOL delegateRespondsToInternedStringMethod;
@property (nonatomic) BOOL delegateRespondsToInternedStringForValueMethod;
@property (nonatomic) BOOL delegateRespondsToStartElementMethod;
@property (nonatomic) BOOL delegateRespondsToEndElementMethod;
@property (nonatomic) BOOL delegateRespondsToCharactersFoundMethod;
@property (nonatomic) BOOL delegateRespondsToEndOfDocumentMethod;
@end
@implementation RSSAXParser
+ (void)initialize {
RSSAXInitLibXMLParser();
}
#pragma mark - Init
- (instancetype)initWithDelegate:(id<RSSAXParserDelegate>)delegate {
self = [super init];
if (self == nil)
return nil;
_delegate = delegate;
if ([_delegate respondsToSelector:@selector(saxParser:internedStringForName:prefix:)]) {
_delegateRespondsToInternedStringMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:internedStringForValue:length:)]) {
_delegateRespondsToInternedStringForValueMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:XMLStartElement:prefix:uri:numberOfNamespaces:namespaces:numberOfAttributes:numberDefaulted:attributes:)]) {
_delegateRespondsToStartElementMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:XMLEndElement:prefix:uri:)]) {
_delegateRespondsToEndElementMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParser:XMLCharactersFound:length:)]) {
_delegateRespondsToCharactersFoundMethod = YES;
}
if ([_delegate respondsToSelector:@selector(saxParserDidReachEndOfDocument:)]) {
_delegateRespondsToEndOfDocumentMethod = YES;
}
return self;
}
#pragma mark - Dealloc
- (void)dealloc {
if (_context != nil) {
xmlFreeParserCtxt(_context);
_context = nil;
}
_delegate = nil;
}
#pragma mark - API
static xmlSAXHandler saxHandlerStruct;
- (void)parseData:(NSData *)data {
[self parseBytes:data.bytes numberOfBytes:data.length];
}
- (void)parseBytes:(const void *)bytes numberOfBytes:(NSUInteger)numberOfBytes {
if (self.context == nil) {
self.context = xmlCreatePushParserCtxt(&saxHandlerStruct, (__bridge void *)self, nil, 0, nil);
xmlCtxtUseOptions(self.context, XML_PARSE_RECOVER | XML_PARSE_NOENT);
}
@autoreleasepool {
xmlParseChunk(self.context, (const char *)bytes, (int)numberOfBytes, 0);
}
}
- (void)finishParsing {
NSAssert(self.context != nil, nil);
if (self.context == nil)
return;
@autoreleasepool {
xmlParseChunk(self.context, nil, 0, 1);
xmlFreeParserCtxt(self.context);
self.context = nil;
self.characters = nil;
}
}
- (void)cancel {
@autoreleasepool {
xmlStopParser(self.context);
}
}
- (void)beginStoringCharacters {
self.storingCharacters = YES;
self.characters = [NSMutableData new];
}
- (void)endStoringCharacters {
self.storingCharacters = NO;
self.characters = nil;
}
- (NSData *)currentCharacters {
if (!self.storingCharacters) {
return nil;
}
return self.characters;
}
- (NSString *)currentString {
NSData *d = self.currentCharacters;
if (RSXMLIsEmpty(d)) {
return nil;
}
return [[NSString alloc] initWithData:d encoding:NSUTF8StringEncoding];
}
- (NSString *)currentStringWithTrimmedWhitespace {
return [self.currentString stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]];
}
#pragma mark - Attributes Dictionary
- (NSDictionary *)attributesDictionary:(const xmlChar **)attributes numberOfAttributes:(NSInteger)numberOfAttributes {
if (numberOfAttributes < 1 || !attributes) {
return nil;
}
NSMutableDictionary *d = [NSMutableDictionary new];
@autoreleasepool {
NSInteger i = 0, j = 0;
for (i = 0, j = 0; i < numberOfAttributes; i++, j+=5) {
NSUInteger lenValue = (NSUInteger)(attributes[j + 4] - attributes[j + 3]);
NSString *value = nil;
if (self.delegateRespondsToInternedStringForValueMethod) {
value = [self.delegate saxParser:self internedStringForValue:(const void *)attributes[j + 3] length:lenValue];
}
if (!value) {
value = [[NSString alloc] initWithBytes:(const void *)attributes[j + 3] length:lenValue encoding:NSUTF8StringEncoding];
}
NSString *attributeName = nil;
if (self.delegateRespondsToInternedStringMethod) {
attributeName = [self.delegate saxParser:self internedStringForName:(const xmlChar *)attributes[j] prefix:(const xmlChar *)attributes[j + 1]];
}
if (!attributeName) {
attributeName = [NSString stringWithUTF8String:(const char *)attributes[j]];
if (attributes[j + 1]) {
NSString *attributePrefix = [NSString stringWithUTF8String:(const char *)attributes[j + 1]];
attributeName = [NSString stringWithFormat:@"%@:%@", attributePrefix, attributeName];
}
}
if (value && attributeName) {
d[attributeName] = value;
}
}
}
return d;
}
#pragma mark - Equal Tags
BOOL RSSAXEqualTags(const xmlChar *localName, const char *tag, NSInteger tagLength) {
if (!localName) {
return NO;
}
return !strncmp((const char *)localName, tag, (size_t)tagLength);
}
#pragma mark - Callbacks
- (void)xmlEndDocument {
@autoreleasepool {
if (self.delegateRespondsToEndOfDocumentMethod) {
[self.delegate saxParserDidReachEndOfDocument:self];
}
[self endStoringCharacters];
}
}
- (void)xmlCharactersFound:(const xmlChar *)ch length:(NSUInteger)length {
@autoreleasepool {
if (self.storingCharacters) {
[self.characters appendBytes:(const void *)ch length:length];
}
if (self.delegateRespondsToCharactersFoundMethod) {
[self.delegate saxParser:self XMLCharactersFound:ch length:length];
}
}
}
- (void)xmlStartElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri numberOfNamespaces:(int)numberOfNamespaces namespaces:(const xmlChar **)namespaces numberOfAttributes:(int)numberOfAttributes numberDefaulted:(int)numberDefaulted attributes:(const xmlChar **)attributes {
@autoreleasepool {
if (self.delegateRespondsToStartElementMethod) {
[self.delegate saxParser:self XMLStartElement:localName prefix:prefix uri:uri numberOfNamespaces:numberOfNamespaces namespaces:namespaces numberOfAttributes:numberOfAttributes numberDefaulted:numberDefaulted attributes:attributes];
}
}
}
- (void)xmlEndElement:(const xmlChar *)localName prefix:(const xmlChar *)prefix uri:(const xmlChar *)uri {
@autoreleasepool {
if (self.delegateRespondsToEndElementMethod) {
[self.delegate saxParser:self XMLEndElement:localName prefix:prefix uri:uri];
}
[self endStoringCharacters];
}
}
@end
static void startElementSAX(void *context, const xmlChar *localname, const xmlChar *prefix, const xmlChar *URI, int nb_namespaces, const xmlChar **namespaces, int nb_attributes, int nb_defaulted, const xmlChar **attributes) {
[(__bridge RSSAXParser *)context xmlStartElement:localname prefix:prefix uri:URI numberOfNamespaces:nb_namespaces namespaces:namespaces numberOfAttributes:nb_attributes numberDefaulted:nb_defaulted attributes:attributes];
}
static void endElementSAX(void *context, const xmlChar *localname, const xmlChar *prefix, const xmlChar *URI) {
[(__bridge RSSAXParser *)context xmlEndElement:localname prefix:prefix uri:URI];
}
static void charactersFoundSAX(void *context, const xmlChar *ch, int len) {
[(__bridge RSSAXParser *)context xmlCharactersFound:ch length:(NSUInteger)len];
}
static void endDocumentSAX(void *context) {
[(__bridge RSSAXParser *)context xmlEndDocument];
}
static xmlSAXHandler saxHandlerStruct = {
nil, /* internalSubset */
nil, /* isStandalone */
nil, /* hasInternalSubset */
nil, /* hasExternalSubset */
nil, /* resolveEntity */
nil, /* getEntity */
nil, /* entityDecl */
nil, /* notationDecl */
nil, /* attributeDecl */
nil, /* elementDecl */
nil, /* unparsedEntityDecl */
nil, /* setDocumentLocator */
nil, /* startDocument */
endDocumentSAX, /* endDocument */
nil, /* startElement*/
nil, /* endElement */
nil, /* reference */
charactersFoundSAX, /* characters */
nil, /* ignorableWhitespace */
nil, /* processingInstruction */
nil, /* comment */
nil, /* warning */
nil, /* error */
nil, /* fatalError //: unused error() get all the errors */
nil, /* getParameterEntity */
nil, /* cdataBlock */
nil, /* externalSubset */
XML_SAX2_MAGIC,
nil,
startElementSAX, /* startElementNs */
endElementSAX, /* endElementNs */
nil /* serror */
};
void RSSAXInitLibXMLParser(void) {
static dispatch_once_t onceToken;
dispatch_once(&onceToken, ^{
xmlInitParser();
});
}