解析搜狗音乐的音乐下载地址

//
//  SougouMusicParser.h
//
//  Created by scott.8an@gmail.com on 12-3-13.
//  Copyright (c) 2012年 littleworn llc. All rights reserved.
//

#import <Foundation/Foundation.h>
#import "ASIHTTPRequest.h"
#import "ASINetworkQueue.h"
#import "TFHpple.h"
#import "XPathQuery.h"
#import "TFHppleElement.h"
#import "DataType.h"

//页面地址
#define kRequestURL(s_songName,i_pageNumber) \
        [NSString stringWithFormat:@"http://mp3.sogou.com/music.so?pf=mp3&query=%@&page=%i&w=02009900&dr=1",s_songName,i_pageNumber]

@interface SogouMusicParser : NSObject

+ (SogouMusicParser*)shareInstance;

/**
 发送成功通知的数组结构:
 [
    {
        "song"="save me",
        "artist"="Queen",
        "size"="4.38M",
        "firstChoose_url"="http://ziyuan1.myi.cn/film/YINYUE/fir/firT1563.mp3",
        "other_link"= [
            "http://mul1.tximg.cn/music/group/bbs/mp3/7/090528/1243462409315.mp3",
            "http://api.ning.com/files/SRIkTa2lzwqL1jt26257Yzj3TCFZ6jAJD2HaYhYB*N469JhRrd01xDttiHqjJD7K9WkWr69u6LnXChFfA2Wid0jWUfiIB33m/f.i.r._.mp3"
        ]
    },...
 ]
 **/
- (void)runToGetSongInfoFromSogouWithKeyword:(NSString*)kw pageNumber:(NSInteger)pgNum;
@end

//
//  SougouMusicParser.m
//
//  Created by scott.8an@gmail.com on 12-3-13.
//  Copyright (c) 2012年 littleworn llc. All rights reserved.
//

#import "SogouMusicParser.h"

NSString *const SogouMusicParseSuccessNotification = @"SogouMusicParseSuccessNotification";
NSString *const SogouMusicParseFailedNotification = @"SogouMusicParseFailedNotification";

static SogouMusicParser *parser_ = nil;

@interface SogouMusicParser (Private)
//获得跳转地址的urls
/**
 [
 "http://mp3.sogou.com/down.so?gid=11950F9C68D7EDE7&globalId=1f940cf0056326c8&query=%CE%D2%C",
 "http://mp3.sogou.com/down.so?gid=11950F9C68D7EDE7&globalId=1f940cf0056326c8&query=%CE%D2%C",
 ...
 ]
 **/
- (NSArray*)getJumpToDownloadPageURLsWithKeyword:(NSString*)kw pageNumber:(int)pgNum;

/**
 返回字典结构:
 {
 "song"="save me",
 "artist"="Queen",
 "size"="4.38M",
 "firstChoose_url"="http://ziyuan1.myi.cn/film/YINYUE/fir/firT1563.mp3",
 "other_link"= [
 "http://mul1.tximg.cn/music/group/bbs/mp3/7/090528/1243462409315.mp3",
 "http://api.ning.com/files/SRIkTa2lzwqL1jt26257Yzj3TCFZ6jAJD2HaYhYB*N469JhRrd01xDttiHqjJD7K9WkWr69u6LnXChFfA2Wid0jWUfiIB33m/f.i.r._.mp3"
 ]
 }
 **/
- (NSDictionary*)getMusicInfoByJumpToUrl:(NSString*)jumpToUrl;
@end


@implementation SogouMusicParser

+ (SogouMusicParser*)shareInstance{
    if (!parser_) {
        parser_ = [[self alloc] init];
    }
    return parser_;
}

- (NSArray*)getJumpToDownloadPageURLsWithKeyword:(NSString*)kw pageNumber:(int)pgNum{
    if (kw && [kw length]) {
        //初始化需要返回的数组
        NSMutableArray *urlsArr = [NSMutableArray arrayWithCapacity:0];

        //去掉搜索条件两端空格
        NSString *pureKW = [kw stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]];

        //用 + 号连接需要要搜索的绑定条件
        NSArray *kwArr = [pureKW componentsSeparatedByString:@" "];
        NSMutableString *neededKw = [NSMutableString stringWithCapacity:0];
        if (kwArr && [kwArr count]) {
            for (NSString *str in kwArr) {
                [neededKw appendFormat:@"%@+",str];
            }

            //去掉最后一个 + 号
            [neededKw deleteCharactersInRange:NSMakeRange([neededKw length]-1, 1)];
        }else{
            neededKw = (NSMutableString*)kw;
        }
        //对url编码
        NSString *requestURLStr = kRequestURL(neededKw,pgNum);
        NSString *requestURLStrEnc = [requestURLStr stringByAddingPercentEscapesUsingEncoding:CFStringConvertEncodingToNSStringEncoding(kCFStringEncodingGB_18030_2000)];

        //请求页面地址
        ASIHTTPRequest *request = [ASIHTTPRequest requestWithURL:[NSURL URLWithString:requestURLStrEnc]];
        [request startSynchronous];

        //获得返回数据
        //NSString *responseStr = [request responseString];
        NSData  *responseData = [request responseData];
        //NSLog(@"%@",responseStr);

        if (responseData) {
            TFHpple *help = [TFHpple hppleWithHTMLData:responseData];
            if (help) {
                //用xpath解析获得歌曲的跳转地址
                //获得tbody节点
                NSString *xPath = @"//*[@id='songlist']";
                TFHppleElement *tableNode = [help peekAtSearchWithXPathQuery:xPath];
                TFHppleElement *tBodyNode = tableNode.firstChild;
                if (tBodyNode) {
                    NSMutableArray *childrenNodes = (NSMutableArray*)tBodyNode.children;
                    if (childrenNodes && [childrenNodes count]) {
                        //移除第一个node
                        [childrenNodes removeObjectAtIndex:0];
                        for (TFHppleElement *element in childrenNodes) {
                            NSArray *tdNodeArr = element.children;
                            if (tdNodeArr && [tdNodeArr count]) {
                                for (TFHppleElement *tdEle in tdNodeArr) {
                                    TFHppleElement *aNode = tdEle.firstChild;
                                    if (aNode && [[aNode objectForKey:@"action"] isEqualToString:@"down"]) {
                                        /**
                                         window.open('/down.so?gid=1A6F5F0BA3CDD9C8&globalId=137c9db3dc0bf7bf&query=%B2%BB%C3%F0%B5%C4%B0%AE&tgid=629801e642a2eadc&pf=mp3&s=%CC%B7%D3%BD%F7%EB&t=%B2%BB%C3%F0%B5%C4%B0%AE&size=4195343&ac=0&c','
                                         ','width=431,height=495,scrollbars=no');
                                         uigsPB('consume=music_down&music_down=28');return(false);
                                         */
                                        NSString *urlThatMixed = [aNode objectForKey:@"onclick"];
                                        NSArray *urlParts = [urlThatMixed componentsSeparatedByString:@"'"];
                                        //取第二个作为uri
                                        NSString *uri = [urlParts objectAtIndex:1];

                                        if (uri && [uri length]) {
                                            NSString *url = [NSString stringWithFormat:@"http://mp3.sogou.com%@",uri];
                                            //NSLog(@"组合新的跳转地址：http://mp3.sogou.com%@",uri);

                                            [urlsArr addObject:url];
                                        }
                                    }
                                }
                            }
                        }
                        if ([urlsArr count]) {
                            return urlsArr;
                        }
                    }
                }
            }
        }else{
            //再次尝试
            [self getJumpToDownloadPageURLsWithKeyword:kw pageNumber:pgNum];
        }
    }
    return nil;
}

- (NSDictionary*)getMusicInfoByJumpToUrl:(NSString*)jumpToUrl{
    if (jumpToUrl && [jumpToUrl length]) {
        //获得下载页面,不需要编码，因为获得的数据已经是编码过了的
        ASIHTTPRequest *request = [ASIHTTPRequest requestWithURL:[NSURL URLWithString:jumpToUrl]];
        [request startSynchronous];

        //获得返回界面
        NSString *responseStr = [request responseString];
        NSData *responseData = [request responseData];
      // NSLog(@"返回下载界面的网页信息:%@",responseStr);

        if (responseData) {
            NSMutableDictionary *dictionary = [NSMutableDictionary dictionaryWithCapacity:0];

            //解析页面
            TFHpple *help = [TFHpple hppleWithHTMLData:responseData];

            /**
             1.查找歌曲的下载地址
             **/

            //href="..+.mp3"
            NSString *pattern = @"href=\"..+.mp3\"";
            NSRegularExpression *exp = [NSRegularExpression regularExpressionWithPattern:pattern
                                                                                 options:NSRegularExpressionCaseInsensitive
                                                                                   error:nil];
            NSArray *metchedArr = [exp matchesInString:responseStr options:NSMatchingReportProgress range:NSMakeRange(0, [responseStr length])];

            if (metchedArr && [metchedArr count]) {
                //用数组保存其他连接
                NSMutableArray *otherLinksArr = [NSMutableArray arrayWithCapacity:0];
                for (NSTextCheckingResult *result in metchedArr) {
                    NSRange range = result.range;
                    NSString *urlWithHREF = [responseStr substringWithRange:range];
                    if (urlWithHREF && [urlWithHREF length]) {
                        NSArray *urlParts = [urlWithHREF componentsSeparatedByString:@"\""];
                        if (urlParts && [urlParts count]>2) {
                            NSString *musicURL = [urlParts objectAtIndex:1];

                            //把第一个获得的地址存到首选里去
                            NSString *songURL = [dictionary objectForKey:@"firstChoose_url"];
                            if (!songURL || [songURL length]<1) {
                                [dictionary setObject:musicURL forKey:@"firstChoose_url"];
                            }else{
                                //去重
                                if (![musicURL isEqualToString:songURL]) {
                                    //用数组保存其他链接
                                    [otherLinksArr addObject:musicURL];
                                }
                            }
                        }
                    }
                }

                //保存其他链接到字典
                [dictionary setObject:otherLinksArr forKey:@"other_link"];
            }

            //如果下载地址解析失败，一下内容都没有必要存在
            NSString *firstChoose_url = [dictionary objectForKey:@"firstChoose_url"];
            if (firstChoose_url && [firstChoose_url length]) {
                /**
                 2.查找歌曲的名称
                 **/
                NSString *xPath = @"//*[@class=\"info1\"]";
                TFHppleElement *info1Node = [help peekAtSearchWithXPathQuery:xPath];
                TFHppleElement *aNode = info1Node.firstChild;
                NSString *songName = [aNode objectForKey:@"title"];
                if (songName && [songName length]) {
                    [dictionary setObject:songName forKey:@"song"];
                }else{
                    [dictionary setObject:@"Unknown" forKey:@"song"];
                }


                /**
                 3.查找艺术家
                 **/
                NSArray *info1Children = info1Node.children;
                if (info1Children && [info1Children count]) {
                    BOOL isArtistTag = NO;
                    for (TFHppleElement *node in info1Children) {
                        if ([node.tagName isEqualToString:@"a"] && isArtistTag) {
                            NSString *artist = [node objectForKey:@"title"];
                            if (artist && [artist length]) {
                                [dictionary setObject:artist forKey:@"artist"];
                                isArtistTag =NO;
                            }
                        }
                        isArtistTag = YES;
                    }
                }
                NSString *hasArtist = [dictionary objectForKey:@"artist"];
                if (!hasArtist || [hasArtist length]<1) {
                    [dictionary setObject:@"Unknown" forKey:@"artist"];
                }

                /**
                 4.查找歌曲的大小
                 **/
                xPath = @"//*[@class=\"info2\"]";
                TFHppleElement *info2Node = [help peekAtSearchWithXPathQuery:xPath];
                NSArray *children = info2Node.children;
                if (children && [children count]) {
                    for (TFHppleElement *node in children) {
                        if ([node.tagName isEqualToString:@"strong"]) {
                            NSString *sizeDesc = node.content;
                            if (sizeDesc && [sizeDesc length]) {
                                [dictionary setObject:sizeDesc forKey:@"size"];
                            }
                        }
                    }
                }
                NSString *hasSize = [dictionary objectForKey:@"size"];
                if (!hasSize || [hasSize length]<1) {
                    [dictionary setObject:@"Unknown" forKey:@"size"];
                }

                //NSLog(@"***************歌曲信息:%@",dictionary);

                //返回字典
                return dictionary;
            }
            return nil;
        }else{
            //重新请求，直到获取成功
            [self getMusicInfoByJumpToUrl:jumpToUrl];
        }
    }
    return nil;
}

- (void)runToGetSongInfoFromSogouWithKeyword:(NSString*)kw pageNumber:(NSInteger)pgNum{
    if (kw && [kw length]) {
        NSMutableArray *songsInfoArr = [NSMutableArray arrayWithCapacity:0];

        NSArray *songsURLsArr = [self getJumpToDownloadPageURLsWithKeyword:kw pageNumber:pgNum];

        //如果第一次请求失败，再次请求
        if (!songsInfoArr || [songsInfoArr count]<1) {
            songsURLsArr = [self getJumpToDownloadPageURLsWithKeyword:kw pageNumber:pgNum];
        }
        //如果第二次请求失败，再次请求
        if (!songsInfoArr || [songsInfoArr count]<1) {
            songsURLsArr = [self getJumpToDownloadPageURLsWithKeyword:kw pageNumber:pgNum];
        }

        if (songsURLsArr && [songsURLsArr count]) {
            for (NSString *url in songsURLsArr) {
                NSDictionary *songInfo = [self getMusicInfoByJumpToUrl:url];
                if (songInfo && [songInfo count]) {
                    [songsInfoArr addObject:songInfo];
                }
            }
        }

        if ([songsInfoArr count]) {
            [[NSNotificationCenter defaultCenter] postNotificationName:SogouMusicParseSuccessNotification
                                                                object:nil userInfo:[NSDictionary dictionaryWithObject:songsInfoArr forKey:@"songsInfo"]];
        }else{
            [[NSNotificationCenter defaultCenter] postNotificationName:SogouMusicParseFailedNotification
                                                                object:nil userInfo:[NSDictionary dictionaryWithObject:@"This page does not exist" 
                                                                                                                forKey:@"msg"]];
        }
    }else{
        [[NSNotificationCenter defaultCenter] postNotificationName:SogouMusicParseFailedNotification
                                                            object:nil userInfo:[NSDictionary dictionaryWithObject:@"Check what you've been input" 
                                                                                                            forKey:@"msg"]];
    }
}
@end