public BookResult WriteCommentToFile(Book book, string newComment)
{
BookResult bookResult = fileOperation.WriteCommentToFile(book, newComment);
return(bookResult);
}
public void ReadTxtFile(string filePathName, Language language, BookCategory category, int skipLines, string connectionString, ref Dictionary <string, WordStatistics> wordsRepo)
{
var sw = new Stopwatch();
var fi = new FileInfo(filePathName);
if (fi.Extension != ".txt")
{
return;
}
sw.Start();
var lines = File.ReadLines(filePathName);
var cntdict = new Dictionary <string, Int64>();
Int64 wc = 0;
Int64 tc = 0;
Int64 linenum = 0;
// 定义enumerable 避免可能的对IEnumerable的遍历，这是ReSharper提示的优化项。
var enumerable = lines as IList <string> ?? lines.ToList();
foreach (var line in enumerable)
{
++linenum;
if (linenum <= skipLines) // todo: this is temp for downloaded book
{
continue;
}
var cline = GetLetters(line, language);
foreach (var letter in cline)
{
if (cntdict.ContainsKey(letter))
{
cntdict[letter] += 1;
}
else
{
cntdict[letter] = 1;
++wc;
}
++tc;
}
}
if (wc < 50)
{
Console.WriteLine(
$"{TotalBook}: Parse total words: {tc}, different words: {wc}, less than 50!!!!! {fi.Name.Replace(".txt", "")}");
return;
}
List <KeyValuePair <string, Int64> > myList = cntdict.ToList();
myList.Sort(
(pair1, pair2) => pair2.Value.CompareTo(pair1.Value)
);
var top1020 = string.Join("", myList.Skip(10).Take(10).Select(p => p.Key).ToArray());
bool findbook = true;
try
{
var context = Program.GetBookContext(connectionString);
// 添加跟踪信息
var booktrack = new BookTrack
{
Top1020 = top1020,
OriginalPath = filePathName,
BookInfo = string.Join('\n', enumerable.Skip(skipLines).Take(5))
};
context.Add(booktrack);
context.SaveChanges();
var book = context.Books.SingleOrDefault(p => p.TopIndexWords == top1020); // Include(p => p.Language).Include(p => p.BookCategory).
if (book != null)
{
Console.WriteLine($"book already exists: {fi.Name}, tracked in BookTrack and not calculate statistics.");
return;
}
if (book == null)
{
findbook = false;
book = new Book
{
LanguageId = language.LanguageId,
BookName = fi.Name.Replace(".txt", ""),
LastDateTime = DateTime.UtcNow,
BookInfo = string.Join('\n', enumerable.Skip(skipLines).Take(5))
};
}
book.PrevWordCount = book.LastWordCount;
book.PrevTotalCount = book.LastTotalCount;
book.LastWordCount = wc;
book.LastTotalCount = tc;
if (findbook)
{
context.Update(book);
}
else
{
book.TopIndexWords = top1020;
// 2018年4月17日 EF Core Add function returns negative id，why?
context.Add(book);
}
context.SaveChanges();
MapBookCategory map = new MapBookCategory
{
BookId = book.BookId,
BookCategoryId = category.BookCategoryId
};
context.Add(map);
context.SaveChanges();
var br = new BookResult
{
Book = book,
BookId = findbook ? book.BookId : 0,
LanguageId = language.LanguageId,
BookTrack = booktrack.BookTrackId,
ResultDateTime = DateTime.UtcNow,
WordCount = wc,
TotalCount = tc,
Top10 = string.Join("", myList.Take(10).Select(p => p.Key).ToArray()),
Top1020 = top1020,
Top50 = string.Join("", myList.Take(50).Select(p => p.Key).ToArray())
};
context.Add(br);
context.SaveChanges();
var newwords = new List <WordStatistics>();
var wrs = new List <WordResult>();
foreach (var dic in cntdict)
{
wrs.Add(new WordResult
{
Book = book,
BookId = book.BookId,
WordCount = dic.Value,
WordLetter = dic.Key
});
if (!wordsRepo.ContainsKey(dic.Key))
{
wordsRepo[dic.Key] = new WordStatistics
{
WordUnicode = dic.Key,
TotalBook = 1,
TotalWords = wc,
TotalOccur = dic.Value,
MaxRatio = (double)dic.Value / (double)wc,
MaxWords = wc,
MaxOccur = dic.Value,
BookCategory = category,
BookCategoryId = category.BookCategoryId,
FirstBookId = book.BookId,
WordLength = dic.Key.Length
};
}
else
{
var wordsta = wordsRepo[dic.Key];
wordsta.TotalBook += 1;
wordsta.MaxOccur = Math.Max(wordsta.MaxOccur, dic.Value);
wordsta.MaxWords = Math.Max(wordsta.MaxWords, wc);
wordsta.MaxRatio = Math.Max(wordsta.MaxRatio, (double)dic.Value / (double)wc);
wordsta.TotalOccur += dic.Value;
wordsta.TotalWords += wc;
//if (findbook) todo 这里的计算有误！
//{
// wordsta.TotalOccur -= ?? not book.PrevWordCount;
// wordsta.TotalWords -= ?? not book.PrevTotalCount;
//}
#if false
#else
wordsRepo[dic.Key] = wordsta;
#endif
}
}
#if false
context.AddRange(wrs);
context.UpdateRange(wordsrepo.Values);
context.SaveChanges();
context.AddRange(newwords);
context.SaveChanges();
#else
#endif
sw.Stop();
++TotalBook;
TotalWords += wc;
TotalTime += sw.ElapsedMilliseconds;
Console.WriteLine(
$"{TotalBook}: Parse finished, total words: {tc}, different words: {wc}, average elapsed milliseconds: {(double) TotalTime / (double) TotalBook}, average words: {(double) TotalWords / (double) TotalBook}. {book.BookName}");
Int64 i = 0;
foreach (var count in myList)
{
if (i++ > 10)
{
break;
}
Console.WriteLine($"{i}: {count.Key}: {count.Value}");
}
}
catch (Exception e)
{
// must create new context since old context has dirty data
var context = Program.GetBookContext(connectionString);
context?.Add(new BookException
{
Top1020 = top1020,
ErrorMsg = e.ToString(),
BookInfo = fi.FullName
});
context?.SaveChanges();
Console.WriteLine($"{TotalBook}: Parse failed: {e}");
}
}
public BookResult incrementLikeCount(Book book, User user)
{
BookResult bookResult = fileOperation.ReadTheBookFromFileChangeTheLikeCount(book, user);
return(bookResult);
}
public IActionResult Result(BookResult result)
{
return(View(result));
}