NetNewsWire/Frameworks/RSParser/HTML/RSHTMLLinkParser.m

152 lines
3.1 KiB
Mathematica
Raw Normal View History

//
// RSHTMLLinkParser.m
// RSParser
//
// Created by Brent Simmons on 8/7/16.
// Copyright © 2016 Ranchero Software, LLC. All rights reserved.
//
#import <libxml/xmlstring.h>
#import <RSParser/RSHTMLLinkParser.h>
#import <RSParser/RSSAXHTMLParser.h>
#import <RSParser/RSSAXParser.h>
#import <RSParser/RSParserInternal.h>
#import <RSParser/ParserData.h>
@interface RSHTMLLinkParser() <RSSAXHTMLParserDelegate>
@property (nonatomic, readonly) NSMutableArray *links;
2017-06-26 01:32:07 +02:00
@property (nonatomic, readonly) ParserData *parserData;
@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
2017-06-26 01:40:29 +02:00
+ (NSArray *)htmlLinksWithParserData:(ParserData *)parserData {
2017-06-26 01:32:07 +02:00
RSHTMLLinkParser *parser = [[self alloc] initWithParserData:parserData];
return parser.links;
}
#pragma mark - Init
2017-06-26 01:32:07 +02:00
- (instancetype)initWithParserData:(ParserData *)parserData {
2017-06-26 01:32:07 +02:00
NSParameterAssert(parserData.data);
NSParameterAssert(parserData.url);
self = [super init];
if (!self) {
return nil;
}
_links = [NSMutableArray new];
2017-06-26 01:32:07 +02:00
_parserData = parserData;
_dictionaries = [NSMutableArray new];
2017-06-26 01:32:07 +02:00
_baseURL = [NSURL URLWithString:parserData.url];
[self parse];
return self;
}
#pragma mark - Parse
- (void)parse {
RSSAXHTMLParser *parser = [[RSSAXHTMLParser alloc] initWithDelegate:self];
2017-06-26 01:32:07 +02:00
[parser parseData:self.parserData.data];
[parser finishParsing];
}
- (RSHTMLLink *)currentLink {
return self.links.lastObject;
}
static NSString *kHrefKey = @"href";
- (NSString *)urlStringFromDictionary:(NSDictionary *)d {
NSString *href = [d rsparser_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 rsparser_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];
2017-06-26 01:32:07 +02:00
if (!RSParserObjectIsEmpty(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