private async Task _savePageAsync()
{
var book = await BookHelper.OpenAsync(WebBrowser.Source);
if (book != null)
{
Toast.ShowInfo($"下载成功！共下载了 {book.Count} 章");
return;
}
book = new Book()
{
Name = WebBrowser.DocumentTitle,
IsLocal = false,
Count = 1,
Url = WebBrowser.Source.AbsoluteUri
};
var chapter = new BookChapter()
{
Name = book.Name,
Url = WebBrowser.Source.ToString(),
};
chapter.Content = BookHelper.HtmlToText(await GetBody());
SqlHelper.Conn.Open();
book.Save();
chapter.BookId = book.Id;
chapter.Save();
SqlHelper.Conn.Close();
Toast.ShowInfo("页面保存成功！");
}
/// <summary>
/// 直接保存不返回
/// </summary>
/// <param name="lines"></param>
/// <param name="book"></param>
public static void GetBookChapter(IList <string> lines, Book book)
{
var index = _getChapterTitle(lines);
int start;
var position = 1;
BookChapter chapter;
while (index >= 0)
{
start = index;
index = _getChapterTitle(index + 1, lines);
chapter = new BookChapter()
{
Name = lines[start],
Position = position,
Content = _getChapterContent(lines, start, index - 1),
BookId = book.Id
};
chapter.Save();
position++;
}
book.Count = position - 1;
book.Save();
}
/// <summary>
/// 直接保存不返回
/// </summary>
/// <param name="file"></param>
/// <param name="book"></param>
/// <returns></returns>
public static async Task GetBookChapterAsync(StorageFile file, Book book)
{
var lines = new List <string>(); // 已获取的行
string line; // 当前行
var maxScore = 0; // 最高分数
var maxLine = -1; //最高分数对应的行号
var maxCount = 0;
var count = 0; // 当前字数
var lineLength = 0; // 当前行的长度
var index = -1; // 当前行在数组中的序号
var score = 0; // 当前行分数
var lastMaxLine = -1;
var position = 1; // 章节总数
BookChapter chapter; // 章节
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
using (var inputStream = await file.OpenReadAsync())
using (var classicStream = inputStream.AsStreamForRead())
using (var streamReader = new StreamReader(classicStream, StorageHelper.GetEncoding(classicStream, Encoding.GetEncoding("gbk"))))
{
while (streamReader.Peek() >= 0)
{
line = streamReader.ReadLine();
lines.Add(line);
index++;
if (string.IsNullOrWhiteSpace(line))
{
score += 50;
continue;
}
if (lastMaxLine < 0)
{
lastMaxLine = index;
score = 0;
continue;
}
lineLength = line.Length;
count += lineLength;
if (lineLength > 50)
{
score = 0;
continue;
}
if (Pattern.IsMatch(line))
{
// 正则匹配
var match = Pattern.Match(line);
var a = string.IsNullOrEmpty(match.Groups[2].Value);
var b = string.IsNullOrEmpty(match.Groups[3].Value);
if (!a && !b)
{
// 保存当前已搜索的章节
chapter = new BookChapter()
{
Name = lines[lastMaxLine],
Position = position,
Content = _getChapterContent(lines, 0, index - 1),
BookId = book.Id
};
chapter.Save();
position++;
lines.RemoveRange(0, index);
// 初始化
lastMaxLine = 0;
index = 0;
maxScore = 0;
count = 0;
score = 0;
continue;
}
if (count < 500)
{
score = 0;
continue;
}
if (!b)
{
maxScore = score + 100;
maxLine = index;
maxCount = count;
}
else if (!a)
{
maxScore = score + 50;
maxLine = index;
maxCount = count;
}
else if (lineLength < 10)
{
maxScore = score + 100;
maxLine = index;
maxCount = count;
}
score = 0;
continue;
}
if (count < 500)
{
score = 0;
continue;
}
score += (300 - Math.Abs(4000 - count) / 100 + 50 - Math.Abs(10 - lineLength) * 5);
if (line.IndexOf("“") >= 0 || line.IndexOf("‘") >= 0 || line.IndexOf("：") >= 0)
{
score -= 20;
}
if (score > maxScore)
{
maxScore = score;
maxLine = index;
score = 0;
maxCount = count;
}
//Log.Info($"i: {index} l: {lineLength} c:{count} s: {score}");
if (count >= 8000)
{
// 保存当前已搜索的章节 存在问题，下一章过长
chapter = new BookChapter()
{
Name = lines[lastMaxLine],
Position = position,
Content = _getChapterContent(lines, 0, maxLine - 1),
BookId = book.Id
};
chapter.Save();
position++;
lines.RemoveRange(0, maxLine);
// 初始化
lastMaxLine = 0;
index -= maxLine;
maxScore = 0;
count -= maxCount;
continue;
}
score = 0;
}
}
// 最后保存
chapter = new BookChapter()
{
Name = lines[lastMaxLine],
Position = position,
Content = _getChapterContent(lines, 0, lines.Count - 1),
BookId = book.Id
};
chapter.Save();
book.Count = position;
book.Save();
// 原始方法
//var content = await StorageHelper.GetFileTextAsync(file);
//GetBookChapter(GetLines(content), book);
}