public static WordOnPage Combine(WordOnPage wop1, WordOnPage wop2, string IndexAs)
{
WordOnPage wop = new WordOnPage();
wop.CombinedWords = new List <string>();
wop.CombinedWords.Add(wop1.Word);
foreach (string str in wop1.CombinedWords)
{
wop.CombinedWords.Add(str);
}
return(wop);
}
public async Task <bool> ProcessIt(Stream DocumentToLoad, DocumentTypes docType, bool DictionaryMatch, bool ExcludeNumberWords)
{
var ret = true;
await Task.Run(async() =>
{
try
{
int PageNum = 0;
if (docType == DocumentTypes.PDF)
{
PagesInBook = new List <BookPage>();
using (PdfDocument pdfDoc = new PdfDocument(new PdfReader(DocumentToLoad)))
{
int maxPages = pdfDoc.GetNumberOfPages();
PageNum = pdfDoc.GetNumberOfPages();
StringBuilder text = new StringBuilder();
for (int x = 1; x <= pdfDoc.GetNumberOfPages(); x++)
{
FireUpdateEvent(EventUpdateTypes.ParsingFromPDF, "Page: " + x.ToString(), maxPages, x, Warnings.Count());
PagesInBook.Add(new BookPage(PdfTextExtractor.GetTextFromPage(pdfDoc.GetPage(x))));
}
}
}
List <WordOnPage> index = new List <WordOnPage>();
int pageCount = 0;
int pageTotal = PagesInBook.Count();
// Loop through all of the pages
foreach (BookPage bp in PagesInBook)
{
pageCount++;
FireUpdateEvent(EventUpdateTypes.Indexing, "Page: " + pageCount.ToString(), pageTotal, pageCount, Warnings.Count());
// Loop through all of the Words on the page
foreach (string word in bp.Words)
{
bool foundInIndex = false;
bool doit = true;
if (ContainsNumber(word) && ExcludeNumberWords)
{
doit = false;
}
if (doit)
{
// Loop through our existing Index
foreach (WordOnPage wop in index.Where(x => x.Word.ToLower().Trim() == word.ToLower().Trim()))
{
// the Word is found in the index
foundInIndex = true;
bool pageAlreadyMarked = false;
// Loop through the pages this word is found on
foreach (PageClass page in wop.Page.Where(p => p.Page == pageCount))
{
pageAlreadyMarked = true;
break;
}
// If this page was not marked for this word
if (!pageAlreadyMarked)
{
// add the page mark
wop.Page.Add(new PageClass(pageCount));
}
break;
}
// If the word was not found in the index
if (!foundInIndex)
{
WordOnPage nwop = new WordOnPage();
nwop.Word = word.Trim();
//nwop.WordType = GetWordType(nwop.Word);
nwop.Page = new List <PageClass>();
nwop.Page.Add(new PageClass(pageCount));
index.Add(nwop);
}
}
}
}
RawWordList = index.OrderBy(p => p.Word).ToList();
int c = 0;
if (DictionaryMatch)
{
foreach (WordOnPage wop in RawWordList)
{
c++;
FireUpdateEvent(EventUpdateTypes.DictionarySearch, wop.Word, RawWordList.Count, c, Warnings.Count());
bool found = false;
foreach (DictionaryWord dic in DictionaryWords.Where(x => x.Word.ToLower().Trim() == wop.Word.ToLower().Trim()))
{
wop.WordType = dic.WordType;
found = true;
break;
}
if (!found)
{
wop.WordType = "Unknown";
}
}
}
List <WordOnPage> finalList = new List <WordOnPage>();
StringBuilder stb = new StringBuilder();
foreach (WordOnPage wop in RawWordList)
{
//if (wop.)
//{
// finalList.Add(wop);
// //StringBuilder sbpages = new StringBuilder();
// //foreach (int page in wop.Page)
// //{
// // sbpages.Append(page.ToString() + ", ");
// //}
// //stb.AppendLine(wop.Word + " - " + sbpages.ToString());
//}
}
// Serialize(ProjectFileToCreate, this);
}
catch (Exception ex)
{
Debug.WriteLine("ERROR:" + ex.Message);
FireUpdateEvent(EventUpdateTypes.Error, ex.Message, 0, 0, 0);
}
});
return(ret);
}
