private static void WriteBadDateSpaces(string str)
{
(var badDates, var goodDates) = TxtReader.FindBadDateSpaces(str);
Console.WriteLine();
Console.WriteLine("Number of bad date descriptions {0} out of {1}", badDates.Count, goodDates.Count + badDates.Count);
foreach (Match match in badDates)
{
Console.WriteLine(match.ToString());
}
Console.WriteLine();
}
private static void WriteAllQuotes(string str)
{
(MatchCollection matches, int wordCount, int wholeStringWordCount) = TxtReader.CheckQuotes(str);
Console.WriteLine();
Console.WriteLine("Number of quotes: {0} -- {1}%", matches.Count, (wordCount * 100) / wholeStringWordCount);
int i = 0;
foreach (Match match in matches)
{
i++;
Console.WriteLine("Quote {0}: {1}", i, match);
}
Console.WriteLine();
}
public static void Main(string[] args)
{
List <string> workSentences;
string[] workWords;
string wholeWorkString;
if (File.Exists(@"WorkFile.txt"))
{
(workSentences, workWords, wholeWorkString) = TxtReader.LoadSentences(@"WorkFile.txt", numberOfWords);
var startupPath = @"source\";
if (Directory.Exists(startupPath))
{
// This path is a directory
ProcessDirectory(startupPath, workSentences);
}
else
{
Console.WriteLine("{0} is not a valid path.", startupPath);
}
WriteWordsCountWithPercents(workWords);
WriteBadDateSpaces(wholeWorkString);
WriteAllQuotes(wholeWorkString);
StringBuilder builder = new StringBuilder();
foreach (var word in workWords)
{
builder.Append(word + " ");
}
HtmlGenerator generator = new HtmlGenerator();
generator.AddText(builder.ToString());
generator.AddMaches(matchedSentences, searchGroups);
var html = generator.GenerateHtml();
using (System.IO.StreamWriter file = new System.IO.StreamWriter(@"Output.html", false))
{
file.Write(html);
}
}
else
{
Console.WriteLine("WorkFile.txt not found");
}
Console.ReadKey();
}
public static void ProcessDirectory(string targetDirectory, List <string> sourceSentences)
{
string[] fileEntries = Directory.GetFiles(targetDirectory, "*.txt");
string[] pdfEntries = Directory.GetFiles(targetDirectory, "*.pdf");
foreach (string pdfName in pdfEntries)
{
if (fileEntries.All(x => x.Replace(".txt", "") != pdfName.Replace(".pdf", "")))
{
TxtReader.ExtractTextFromPdfToTxt(pdfName, targetDirectory);
}
}
fileEntries = Directory.GetFiles(targetDirectory, "*.txt");
Console.WriteLine("Number of loaded sources: " + fileEntries.Length);
Console.WriteLine();
foreach (var fileName in fileEntries)
{
ProcessFile(fileName, sourceSentences);
}
}
public static void ProcessFile(string path, List <string> sourceSentences)
{
// load source
(List <string> sentences, string[] wordArray, string str) = TxtReader.LoadSentences(path, numberOfWords);
//print filename
Console.WriteLine(path);
Console.WriteLine();
//find all sentences
var matches = sourceSentences.Intersect(sentences);
matches = TxtReader.CombineFollowingSencences(matches.ToList(), numberOfWords);
WriteFileMatches(matches.ToList());
foreach (var match in matches)
{
matchedSentences.Add(match);
}
}