/// <summary>
/// // key: url, value: 保存地址
/// </summary>
/// <param name="downloadItems"></param>
private void DownadFiles(string albumTitle, Dictionary <string, string> downloadItems, int paralleCount = 4)
{
var totalCount = downloadItems.Values?.Count;
if (totalCount == 0)
{
return;
}
System.Console.WriteLine($"开始下载专辑{albumTitle}, 共{totalCount}个音频");
Stopwatch timer = Stopwatch.StartNew();
var options = new ParallelOptions {
MaxDegreeOfParallelism = paralleCount
};
Parallel.ForEach(downloadItems, options, downloadItem =>
{
var url = downloadItem.Key;
var savePath = downloadItem.Value;
try
{
DownloadSingleFile(url, savePath);
}
catch (Exception ex)
{
HtmlParseLogger.Error($"下载{Path.GetFileName(savePath)}出错({url}), Exception: {ex.Message}");
}
Console.WriteLine($"下载完成, 剩余: {--totalCount}个");
});
timer.Stop();
System.Console.WriteLine($"专辑{albumTitle}下载完成, 用时{timer.Elapsed.TotalSeconds}s, 共{totalCount}个音频");
}
/// <summary>
/// 部分页面没有MP3, 只有MP4视频, 需要检测MP4实际地址 (如： 乌盆记-1 言菊朋 http://www.bavc.com.cn/w10279097.htm?page=1)
/// 找到播放器id部分, 根据id推测出下载地址
/// player id: <div id="piv_d69fff2eae3766ef7aff664b0ab2b61d_d"></div>
/// 下载地址模板: http://mpv.videocc.net/d69fff2eae/{0}/{1}_1.mp4
/// </summary>
///
/// 以 piv_d69fff2eae3766ef7aff664b0ab2b61d_d 为例
/// id部分只要两个_之间的字符 d69fff2eae3766ef7aff664b0ab2b61d
/// mp4地址模板中:
/// {0} 为 "d69fff2eae3766ef7aff664b0ab2b61d" 最后一个字母d
/// {1} 为 ”d69fff2eae3766ef7aff664b0ab2b61d“
/// 则实际MP4地址为: https://mpv.videocc.net/d69fff2eae/d/d69fff2eae3766ef7aff664b0ab2b61d_1.mp4
public static string GuessMp4DownloadUrl(string title, string mp3InfoUrl)
{
// 复制Program.LoadMp3Info
var homePageNode = HtmlCacheParser.LoadHtmlNode(mp3InfoUrl);
// <div id="plv_d69fff2eae176f68eb79b5e0575cc75b_d"></div>
var main2 = homePageNode.SelectSingleNode("/html/body/table[4]/tbody/tr");
var targetNode = main2.SelectNodes(".//div").FirstOrDefault(x => x.Attributes["id"] != null && x.Attributes["id"].Value.StartsWith("plv_"));
if (targetNode == null)
{
HtmlParseLogger.Error($"mp4 解析错误 {title}({mp3InfoUrl})页面信息不存在");
}
var mp4UrlTemplate = "https://mpv.videocc.net/d69fff2eae/{0}/{1}_1.mp4";
var idVal = targetNode.Attributes["id"].Value;
var first_ = idVal.IndexOf("_");
var last_ = idVal.LastIndexOf("_");
if (first_ > -1 && last_ > -1 && first_ < last_)
{
//"plv_d69fff2eae176f68eb79b5e0575cc75b_d"
// d69fff2eae176f68eb79b5e0575cc75b
var mp4ItemKey = idVal.Substring(first_ + 1, last_ - first_ - 1);
return(string.Format(mp4UrlTemplate, mp4ItemKey[mp4ItemKey.Length - 1], mp4ItemKey));
}
return(string.Empty);
}
public void DownloadMp4Items(List <Mp4Item> mp4Items, bool deleteIfExist = false)
{
var mp4DownloadDIr = System.IO.Path.Combine(
StaticVariables.GetDownloadDir(),
StaticVariables.MP4_FOLDER_NAME);
if (Directory.Exists(mp4DownloadDIr))
{
Directory.CreateDirectory(mp4DownloadDIr);
}
foreach (var categoryInfo in mp4Items.GroupBy(x => x.CategoryName))
{
var categoryName = categoryInfo.Key;
foreach (var albumInfo in categoryInfo.GroupBy(x => x.AblumName))
{
// albumInfo
var albumName = albumInfo.Key;
var albumMp4Items = albumInfo.ToArray();
if (albumMp4Items.Length == 0)
{
continue;
}
// key: url, value: 保存地址
Dictionary <string, string> downloadItems = new Dictionary <string, string>();
foreach (var item in albumMp4Items)
{
if (string.IsNullOrEmpty(item.Mp4DownloadUrl))
{
HtmlParseLogger.Error($"{item.Mp3FileName} 没有下载地址，跳过");
continue;
}
var dir = Path.Combine(mp4DownloadDIr,
categoryName,
albumName);
if (!Directory.Exists(dir))
{
Directory.CreateDirectory(dir);
}
var mp3Name = Path.Combine(dir, $"{item.Mp3FileName}.mp4");
downloadItems.Add(item.Mp4DownloadUrl, mp3Name);
}
DownadFiles(albumName, downloadItems);
}
//OperaHtmlParser.Mp4Item
}
}
// 下载专辑
private void DownloadMediaItem(string mp3DownloadDir, string mp4DownloadDir, MediaItem mediaItem)
{
if (!Directory.Exists(mp3DownloadDir))
{
Directory.CreateDirectory(mp3DownloadDir);
}
if (!Directory.Exists(mp4DownloadDir))
{
Directory.CreateDirectory(mp4DownloadDir);
}
System.Console.WriteLine($"下载专辑: {mediaItem.Title}...");
// key: url, value: 文件保存地址
Dictionary <string, string> downloadItems = new Dictionary <string, string>();
// 下载封面图片 "_cover.jpg" 放到MP3文件夹中
var coverImgName = Path.Combine(mp3DownloadDir, StaticVariables.ALBUM_COVER_NAME);
downloadItems.Add(mediaItem.ImageUrl, coverImgName);
// 准备mp3/mp4下载地址
foreach (var mp3Info in mediaItem.Mp3Items)
{
var saveFilePath = string.Empty;
var downloadUrl = string.Empty;
if (!string.IsNullOrEmpty(mp3Info.Mp3DownloadUrl))
{
downloadUrl = mp3Info.Mp3DownloadUrl;
saveFilePath = Path.Combine(mp3DownloadDir, $"{mp3Info.Title}.mp3");
}
else if (!string.IsNullOrEmpty(mp3Info.Mp4DownloadUrl))
{
downloadUrl = mp3Info.Mp4DownloadUrl;
saveFilePath = Path.Combine(mp4DownloadDir, $"{mp3Info.Title}.mp4");
}
else
{
HtmlParseLogger.Error($"{mediaItem.Title} 没有下载地址，跳过");
continue;
}
downloadItems.Add(downloadUrl, saveFilePath);
}
// 开始下载
var totalCount = mediaItem.Mp3Items?.Count;
System.Console.WriteLine($"开始下载专辑{mediaItem.Title}, 共{totalCount}个音频");
DownadFiles(mediaItem.Title, downloadItems);
}
/// url为某个具体MP3播放页面
/// 从url中读取MP3相关信息
static void LoadMp3Info(Mp3Info mp3Info)
{
if (mp3Info == null || string.IsNullOrEmpty(mp3Info.Url))
{
throw new ArgumentException(nameof(mp3Info));
}
// 播放页面左侧有导航栏的情况（少数是这样）
// 无导航： http://www.bavc.com.cn/w10276740.htm?page=1
// 有导航： http://www.bavc.com.cn/w10276738.htm?page=1
var homePageNode = HtmlCacheParser.LoadHtmlNode(mp3Info.Url);
var tdNodes = homePageNode.SelectNodes("/html/body/table[4]/tbody/tr/td");
if (tdNodes == null || tdNodes.Count == 0)
{
HtmlParseLogger.Error($"{mp3Info.Title}({mp3Info.Url})页面信息不存在");
}
// 音频页面有导航栏xpath不同
bool hasSidebar = tdNodes.Count > 1;
HtmlNode mainNode = null;
if (!hasSidebar)
{
mainNode = homePageNode.SelectSingleNode("/html/body/table[4]/tbody/tr/td/table[4]");
}
else
{
mainNode = homePageNode.SelectSingleNode("/html/body/table[4]/tbody/tr/td[3]/center/table[last()]");
}
if (mainNode == null)
{
HtmlParseLogger.Error($"{mp3Info.Title}({mp3Info.Url})页面信息不存在");
}
// 标题(列表页已经获取了) /html/body/table[4]/tbody/tr/td/table[4]/tbody/tr[2]/td/table/tbody/tr[1]/td/p[1]/span
// 当前读取方式会显示‘Your browser does not support’，<a>写出了地址
mp3Info.Mp3DownloadUrl = mainNode.SelectSingleNode(".//table[1]//table[1]//a")?.Attributes["href"]?.Value;
// 如果没有MP3, 检测MP4
if (string.IsNullOrEmpty(mp3Info.Mp3DownloadUrl))
{
mp3Info.Mp4DownloadUrl = GuessMp4DownloadUrl(mp3Info.Title, mp3Info.Url);
}
// 歌词标题和歌词在一个td里, 标题带有<strong>
// /html/body/table[4]/tbody/tr/td/table[4]/tbody/tr[2]/td/table/tbody/tr[5]/td/strong
// /html/body/table[4]/tbody/tr/td/table[4]/tbody/tr[2]/td/table/tbody/tr[4]/td
// 歌词部分有两种可能
HtmlNode lyricNode = mainNode.SelectSingleNode("./tbody/tr[last()-1]/td/table/tbody/tr[last()]");
if (lyricNode == null)
{
lyricNode = mainNode.SelectSingleNode("./tbody/tr[last()-2]/td/table/tbody/tr[last()]");
}
if (lyricNode == null)
{
HtmlParseLogger.Error($"{mp3Info.Title}({mp3Info.Url})歌词信息不存在, 跳过");
}
// 大部分标题在<p><strong>{Title}</strong></p>中
// 目前只发现13. 《梨园名票唱腔选》第三集 "06.武家坡-2 夏山楼主 高亭"没有<strong>, 标题<p>{Title}</p>中
var lyricTitle = (lyricNode?.SelectSingleNode(".//strong") ??
lyricNode?.SelectSingleNode(".//p"))?.InnerText;
var lyric = lyricNode?.InnerText;
// .Replace(" ", "")
lyric = lyric?.Replace("\r\n", "");
// 歌词去掉标题部分
if (!string.IsNullOrEmpty(lyricTitle) &&
!string.IsNullOrEmpty(lyric))
{
var titleIndex = lyric.IndexOf(lyricTitle);
// 标题出现在开头
if (titleIndex == 0 && lyric.Length > lyricTitle.Length)
{
lyric = lyric.Substring(lyricTitle.Length);
}
}
mp3Info.Lyric = lyric;
mp3Info.LyricTitle = lyricTitle;
}
