private List <TextToken> GetSearchBeforeContext(BookTokenIterator bookTokenIterator, int startTokenId, int count = 8)
{
var result = new List <TextToken>();
var tokenId = startTokenId;
while (--tokenId >= 0 && result.Count < count)
{
bookTokenIterator.MoveTo(tokenId);
bookTokenIterator.MoveNext();
if (bookTokenIterator.Current is NewPageToken)
{
break;
}
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
result.Insert(0, textToken);
}
bookTokenIterator.MoveTo(startTokenId);
bookTokenIterator.MoveNext();
return(result);
}
public string GetText(BookModel book, int tokenOffset, int wordsCount, out int lastTokenId)
{
lastTokenId = -1;
var result = new List <string>();
using (var tokenIterator = new BookTokenIterator(book.GetTokensPath(), TokensTool.GetTokens(book.BookID)))
{
int words = 0;
tokenIterator.MoveTo(tokenOffset);
while (tokenIterator.MoveNext() && words < wordsCount)
{
if (tokenIterator.Current is NewPageToken && result.Count > 0)
{
break;
}
var textToken = tokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
lastTokenId = textToken.ID;
result.Add(textToken.Text);
words++;
}
}
return(string.Join(" ", result));
}
public IEnumerable<TokenBlockBase> GetLines(BookTokenIterator bookTokens, string lastText, int firstTokenID,
int stopTokenID = -1, string stopText = null)
{
_firstTokenID = firstTokenID;
_tree = bookTokens.BuildTree(_firstTokenID);
_lastOpenTag = _tree.Peek();
_fontSize = GetCurrentFontSize();
_separator = false;
bool firstText = true;
_marginLeft = _marginRight = 0.0;
foreach (TagOpenToken openTagToken in _tree.Reverse())
EnterMargin(openTagToken.TextProperties);
if (string.IsNullOrEmpty(stopText) && stopTokenID > 0)
--stopTokenID;
while (bookTokens.MoveNext())
{
foreach (TokenBlockBase baseTokenLine in OutputLines(false))
yield return baseTokenLine;
if (!Append(bookTokens, lastText, stopTokenID, stopText, ref firstText))
break;
}
foreach (TokenBlockBase baseTokenLine in OutputLines(true))
yield return baseTokenLine;
}
private List <TextToken> GetSearchAfterContext(BookTokenIterator bookTokenIterator, int endTokenId, int count = 8)
{
var result = new List <TextToken>();
var tokenId = endTokenId;
while (++tokenId < bookTokenIterator.Count && result.Count < count)
{
bookTokenIterator.MoveTo(tokenId);
bookTokenIterator.MoveNext();
if (bookTokenIterator.Current is NewPageToken)
{
break;
}
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
result.Add(textToken);
}
bookTokenIterator.MoveTo(endTokenId);
bookTokenIterator.MoveNext();
return(result);
}
private List <BookSearchResult> SearchOneWord(BookTokenIterator bookTokenIterator, string query, int count)
{
var result = new List <BookSearchResult>();
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
if (textToken.Text.IndexOf(query, StringComparison.InvariantCultureIgnoreCase) >= 0)
{
var previousContext = GetSearchBeforeContext(bookTokenIterator, textToken.ID);
var afterContext = GetSearchAfterContext(bookTokenIterator, textToken.ID);
result.Add(new BookSearchResult
{
PreviousContext = previousContext,
SearchResult = new List <TextToken> {
textToken
},
NextContext = afterContext
});
if (result.Count >= count)
{
break;
}
}
}
return(result);
}
private List <BookSearchResult> SearchGroupWords(BookTokenIterator bookTokenIterator, List <string> query, int count)
{
var result = new List <BookSearchResult>();
var firstWordQuery = query[0];
var lastWordQuery = query.Last();
TextToken firstWordToken;
while ((firstWordToken = FindFirstWord(bookTokenIterator, firstWordQuery)) != null)
{
var resultSequence = new List <TextToken>();
resultSequence.Add(firstWordToken);
bool findNextSequence = false;
for (int i = 1; i < query.Count - 1; i++)
{
TextToken intermediateToken;
if (CheckIntermediateWord(bookTokenIterator, query[i], out intermediateToken))
{
resultSequence.Add(intermediateToken);
}
else
{
findNextSequence = true;
break;
}
}
if (findNextSequence)
{
continue;
}
TextToken lastToken;
if (CheckLastWord(bookTokenIterator, lastWordQuery, out lastToken))
{
resultSequence.Add(lastToken);
var previousContext = GetSearchBeforeContext(bookTokenIterator, firstWordToken.ID);
var afterContext = GetSearchAfterContext(bookTokenIterator, lastToken.ID);
result.Add(new BookSearchResult
{
PreviousContext = previousContext,
SearchResult = resultSequence,
NextContext = afterContext
});
if (result.Count >= count)
{
break;
}
}
}
return(result);
}
public void Init()
{
if (_bookTokenIterator != null && _book != null)
{
int tokenId = _bookTokenIterator.Current.ID;
_bookTokenIterator.Dispose();
_bookTokenIterator = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID));
_bookTokenIterator.MoveTo(tokenId);
_bookTokenIterator.MoveNext();
}
}
public void Init()
{
if (_bookTokenIterator != null && _book != null)
{
int tokenId = _bookTokenIterator.Current.ID;
_bookTokenIterator.Dispose();
_bookTokenIterator = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID));
_bookTokenIterator.MoveTo(tokenId);
_bookTokenIterator.MoveNext();
}
}
public Task <PageInfo> GetPageAsync(int tokenID, string startText)
{
return(Task.Factory.StartNew(
delegate
{
using (var tokens = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID)))
{
PageInfo page = BookFactory
.GetBookParser(_book.Type, tokens, _fontSize, _pageSize, _images)
.GetPage(tokenID, startText);
return page;
}
}));
}
public Task<PageInfo> GetPageAsync(int tokenID, string startText)
{
return Task.Factory.StartNew(
delegate
{
using (var tokens = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID)))
{
PageInfo page = BookFactory
.GetBookParser(_book.Type, tokens, _fontSize, _pageSize, _images)
.GetPage(tokenID, startText);
return page;
}
});
}
public static IBookBuilder GetBookParser(string bookType, BookTokenIterator bookTokens, int fontSize, Size pageSize, IEnumerable<BookImage> images)
{
var headerSizes = new ReadOnlyCollection<double>(new List<double> {24, 32, 42});
IFontHelper activeFontHelper = GetActiveFontMetrics(AppSettings.Default.FontSettings.FontFamily.Source);
switch (bookType)
{
case "fb2":
case "txt":
case "epub":
case "html":
return new BookBuilder(bookTokens, images, headerSizes, activeFontHelper, pageSize, fontSize , AppSettings.Default.Hyphenation, AppSettings.Default.UseCSSFontSize);
default:
throw new NotSupportedException("Book type '" + bookType + "' is not supported!");
}
}
public Task<List<BookSearchResult>> Search(BookModel book, string query, int count)
{
if (string.IsNullOrEmpty(query) || book == null)
return Task<List<BookSearchResult>>.Factory.StartNew(() => new List<BookSearchResult>());
_bookTokenIterator?.Dispose();
_book = book;
_bookTokenIterator = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID));
_query = PrepareQuery(query);
return Task<List<BookSearchResult>>
.Factory.StartNew(() => Load(_bookTokenIterator, _query, count));
}
private TextToken FindFirstWord(BookTokenIterator bookTokenIterator, string query)
{
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
if (textToken.Text.EndsWith(query, StringComparison.InvariantCultureIgnoreCase))
{
return(textToken);
}
}
return(null);
}
public void Init()
{
if (_bookTokenIterator == null || _book == null) return;
var tokenId = 0;
try
{
tokenId = _bookTokenIterator.Current.ID;
}
catch (Exception)
{
// ignored
}
_bookTokenIterator.Dispose();
_bookTokenIterator = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID));
_bookTokenIterator.MoveTo(tokenId);
_bookTokenIterator.MoveNext();
}
public static IBookBuilder GetBookParser(string bookType, BookTokenIterator bookTokens, int fontSize, Size pageSize, IEnumerable <BookImage> images)
{
var headerSizes = new ReadOnlyCollection <double>(new List <double> {
24, 32, 42
});
IFontHelper activeFontHelper = GetActiveFontMetrics(AppSettings.Default.FontSettings.FontFamily.Source);
switch (bookType)
{
case "fb2":
case "txt":
case "epub":
case "html":
return(new BookBuilder(bookTokens, images, headerSizes, activeFontHelper, pageSize, fontSize, AppSettings.Default.Hyphenation, AppSettings.Default.UseCSSFontSize));
default:
throw new NotSupportedException("Book type '" + bookType + "' is not supported!");
}
}
public BookBuilder(
BookTokenIterator bookTokens,
IEnumerable<BookImage> images,
IList<double> headerSizes,
IFontHelper helper,
Size pageSize,
double textSize,
bool hyphenation,
bool useCssFontSize)
{
_bookTokens = bookTokens;
_images = images;
_headerSizes = headerSizes;
_helper = helper;
_pageSize = pageSize;
_textSize = textSize;
_hyphenation = hyphenation;
_useCssFontSize = useCssFontSize;
}
private bool CheckLastWord(BookTokenIterator bookTokenIterator, string query, out TextToken result)
{
result = null;
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
{
continue;
}
if (textToken.Text.StartsWith(query, StringComparison.InvariantCultureIgnoreCase))
{
result = textToken;
return(true);
}
return(false);
}
return(false);
}
public Task <List <BookSearchResult> > Search(BookModel book, string query, int count)
{
if (string.IsNullOrEmpty(query) || book == null)
{
return(Task <List <BookSearchResult> > .Factory.StartNew(() => new List <BookSearchResult>()));
}
if (_bookTokenIterator != null)
{
_bookTokenIterator.Dispose();
}
_book = book;
_bookTokenIterator = new BookTokenIterator(_book.GetTokensPath(), TokensTool.GetTokens(_book.BookID));
_query = PrepareQuery(query);
return(Task <List <BookSearchResult> >
.Factory.StartNew(() => Load(_bookTokenIterator, _query, count)));
}
public string GetLastParagraphByToken(BookModel book, int tokenOffset, out string pointer)
{
var result = new List<string>();
using (var tokenIterator = new BookTokenIterator(book.GetTokensPath(), TokensTool.GetTokens(book.BookID)))
{
var token = FindNewToken(tokenOffset, tokenIterator);
pointer = token.Pointer;
var tokenId = token.ID;
tokenIterator.MoveTo(tokenId);
while (tokenIterator.MoveNext())
{
if (tokenIterator.Current is TagOpenToken && result.Count > 0)
break;
var textToken = tokenIterator.Current as TextToken;
if (textToken == null)
continue;
result.Add(textToken.Text);
}
}
return string.Join(" ", result);
}
private List<BookSearchResult> Load(BookTokenIterator bookTokenIterator, List<string> query, int count)
{
var result = new List<BookSearchResult>();
try
{
if (query.Count == 1)
{
result = SearchOneWord(bookTokenIterator, query[0], count);
}
if (query.Count > 1)
{
result = SearchGroupWords(bookTokenIterator, query, count);
}
}
catch (TokenIteratorUnableMoveNextException tokenExp)
{
throw new SearchInBookInterruptedException("Book tokenizer exception has occured", tokenExp);
}
return result;
}
private List <BookSearchResult> Load(BookTokenIterator bookTokenIterator, List <string> query, int count)
{
var result = new List <BookSearchResult>();
try
{
if (query.Count == 1)
{
result = SearchOneWord(bookTokenIterator, query[0], count);
}
if (query.Count > 1)
{
result = SearchGroupWords(bookTokenIterator, query, count);
}
}
catch (TokenIteratorUnableMoveNextException tokenExp)
{
throw new SearchInBookInterruptedException("Book tokenizer exception has occured", tokenExp);
}
return(result);
}
public string GetText(BookModel book, int tokenOffset, int wordsCount, out int lastTokenId)
{
lastTokenId = -1;
var result = new List<string>();
using (var tokenIterator = new BookTokenIterator(book.GetTokensPath(), TokensTool.GetTokens(book.BookID)))
{
int words = 0;
tokenIterator.MoveTo(tokenOffset);
while (tokenIterator.MoveNext() && words < wordsCount)
{
if(tokenIterator.Current is NewPageToken && result.Count > 0)
break;
var textToken = tokenIterator.Current as TextToken;
if(textToken == null)
continue;
lastTokenId = textToken.ID;
result.Add(textToken.Text);
words++;
}
}
return string.Join(" ", result);
}
private bool CheckLastWord(BookTokenIterator bookTokenIterator, string query, out TextToken result)
{
result = null;
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
if (textToken.Text.StartsWith(query, StringComparison.InvariantCultureIgnoreCase))
{
result = textToken;
return true;
}
return false;
}
return false;
}
private TextToken FindFirstWord(BookTokenIterator bookTokenIterator, string query)
{
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
if (textToken.Text.EndsWith(query))
{
return textToken;
}
}
return null;
}
private bool Append(BookTokenIterator bookTokens, string lastText, int stopTokenID, string stopText, ref bool firstText)
{
TokenBase token = bookTokens.Current;
var pageBreakToken = token as NewPageToken;
if (pageBreakToken != null)
AppendToLine(pageBreakToken);
var imageToken = token as PictureToken;
if (imageToken != null)
AppendToLine(imageToken);
var openTagToken = token as TagOpenToken;
if (openTagToken != null)
AppendToLine(openTagToken);
var closeTagToken = token as TagCloseToken;
if (closeTagToken != null)
AppendToLine(closeTagToken);
var textSeparatorToken = token as WhitespaceToken;
if (textSeparatorToken != null)
AppendSeparator();
var textToken = token as TextToken;
if (textToken != null && AppendTextToken(textToken, lastText, stopTokenID, stopText, ref firstText) ||
stopTokenID >= 0 && token.ID >= stopTokenID)
return false;
return true;
}
private bool CheckIntermediateWord(BookTokenIterator bookTokenIterator, string query, out TextToken result)
{
result = null;
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
if (textToken.Text.Equals(query))
{
result = textToken;
return true;
}
return false;
}
return false;
}
public BookSearch(BookModel book)
{
_book = book;
_bookTokenIterator = new BookTokenIterator(book.GetTokensPath(), TokensTool.GetTokens(book.BookID));
}
private List<BookSearchResult> SearchOneWord(BookTokenIterator bookTokenIterator, string query, int count)
{
var result = new List<BookSearchResult>();
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
if (textToken.Text.IndexOf(query, StringComparison.InvariantCultureIgnoreCase) >= 0)
{
var previousContext = GetSearchBeforeContext(bookTokenIterator, textToken.ID);
var afterContext = GetSearchAfterContext(bookTokenIterator, textToken.ID);
result.Add(new BookSearchResult
{
PreviousContext = previousContext,
SearchResult = new List<TextToken>{textToken},
NextContext = afterContext
});
if (result.Count >= count)
break;
}
}
return result;
}
private List<BookSearchResult> SearchGroupWords(BookTokenIterator bookTokenIterator, List<string> query, int count)
{
var result = new List<BookSearchResult>();
var firstWordQuery = query[0];
var lastWordQuery = query.Last();
TextToken firstWordToken;
while ((firstWordToken = FindFirstWord(bookTokenIterator, firstWordQuery)) != null)
{
var resultSequence = new List<TextToken>();
resultSequence.Add(firstWordToken);
bool findNextSequence = false;
for (int i = 1; i < query.Count - 1; i++)
{
TextToken intermediateToken;
if (CheckIntermediateWord(bookTokenIterator, query[i], out intermediateToken))
{
resultSequence.Add(intermediateToken);
}
else
{
findNextSequence = true;
break;
}
}
if (findNextSequence)
continue;
TextToken lastToken;
if (CheckLastWord(bookTokenIterator, lastWordQuery, out lastToken))
{
resultSequence.Add(lastToken);
var previousContext = GetSearchBeforeContext(bookTokenIterator, firstWordToken.ID);
var afterContext = GetSearchAfterContext(bookTokenIterator, lastToken.ID);
result.Add(new BookSearchResult
{
PreviousContext = previousContext,
SearchResult = resultSequence,
NextContext = afterContext
});
if (result.Count >= count)
break;
}
}
return result;
}
private TextToken FindFirstWord(BookTokenIterator bookTokenIterator, string query)
{
while (bookTokenIterator.MoveNext())
{
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
if (textToken.Text.EndsWith(query, StringComparison.InvariantCultureIgnoreCase))
{
return textToken;
}
}
return null;
}
private static TagOpenToken FindNewToken(int tokenOffset, BookTokenIterator tokenIterator)
{
List<TagOpenToken> tokens = new List<TagOpenToken>();
var offset = 40;
var searchToken = tokenOffset - offset;
var idx = searchToken;
if (searchToken < 0) searchToken = 0;
tokenIterator.MoveTo(searchToken);
while (tokenIterator.MoveNext() && idx < tokenOffset)
{
if (!(tokenIterator.Current is TagOpenToken)) { idx++; continue;}
var tagToken = tokenIterator.Current as TagOpenToken;
if (tagToken.Name.Contains("p"))
{
tokens.Add(tagToken);
}
idx++;
}
var result = tokens.Count > 0 ? tokens.Last() : FindNewToken(searchToken, tokenIterator);
return result;
}
private List<TextToken> GetSearchBeforeContext(BookTokenIterator bookTokenIterator, int startTokenId, int count = 8)
{
var result = new List<TextToken>();
var tokenId = startTokenId;
while (--tokenId >= 0 && result.Count < count)
{
bookTokenIterator.MoveTo(tokenId);
bookTokenIterator.MoveNext();
if (bookTokenIterator.Current is NewPageToken)
break;
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
result.Insert(0, textToken);
}
bookTokenIterator.MoveTo(startTokenId);
bookTokenIterator.MoveNext();
return result;
}
private List<TextToken> GetSearchAfterContext(BookTokenIterator bookTokenIterator, int endTokenId, int count = 8)
{
var result = new List<TextToken>();
var tokenId = endTokenId;
while (++tokenId < bookTokenIterator.Count && result.Count < count)
{
bookTokenIterator.MoveTo(tokenId);
bookTokenIterator.MoveNext();
if (bookTokenIterator.Current is NewPageToken)
break;
var textToken = bookTokenIterator.Current as TextToken;
if (textToken == null)
continue;
result.Add(textToken);
}
bookTokenIterator.MoveTo(endTokenId);
bookTokenIterator.MoveNext();
return result;
}
