void adjustWords1()
{
string plainText = extractPlainText(tempPdfFilePath);
char[] lineSeparatingChars = { '\n' };
char[] wordSeparatingChars = { ' ' };
string[] lines = plainText.Split(lineSeparatingChars, StringSplitOptions.RemoveEmptyEntries);
List <string> wordCombinationList = new List <string>();
foreach (string line in lines)
{
string[] words = line.Split(wordSeparatingChars, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < words.Length; i++)
{
string wordCombination = "";
for (int j = i; j < words.Length; j++)
{
wordCombination += (i == j ? "" : " ") + words[j];
wordCombinationList.Add(wordCombination);
}
}
}
for (int i = 0; i < completedWordList.Count; i++)
{
double bestSimilarity = 0;
int bestSimilarWordId = 0;
string agent1 = Regex.Replace(completedWordList[i].word, @"[^0-9a-zA-Z$@]+", "");
for (int j = 0; j < wordCombinationList.Count; j++)
{
string agent2 = Regex.Replace(wordCombinationList[j], @"[^0-9a-zA-Z$@]+", "");
double similarity = ContractNote.getSimilarity(agent1, agent2);
if (bestSimilarity < similarity)
{
bestSimilarity = similarity;
bestSimilarWordId = j;
}
}
if (bestSimilarity > 0.8)
{
completedWordList[i].word = wordCombinationList[bestSimilarWordId];
}
}
}
void adjustWords2()
{
string plainText = extractPlainText(tempPdfFilePath);
char[] lineSeparatingChars = { '\n' };
char[] wordSeparatingChars = { ' ' };
string[] lines = plainText.Split(lineSeparatingChars, StringSplitOptions.RemoveEmptyEntries);
List <string> wordCombinationList = new List <string>();
foreach (string line in lines)
{
string[] words = line.Split(wordSeparatingChars, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < words.Length; i++)
{
string wordCombination = "";
for (int j = i; j < words.Length; j++)
{
wordCombination += (i == j ? "" : " ") + words[j];
wordCombinationList.Add(wordCombination);
}
}
}
try
{
using (PdfReader reader = new PdfReader(tempPdfFilePath))
{
var pageRectangle = reader.GetPageSize(1);
float pageWidth = pageRectangle.Width;
float pageHeight = pageRectangle.Height;
float sizeRate = 2754f / 595;
for (int i = 0; i < completedWordList.Count; i++)
{
double bestSimilarity = 0;
string word = completedWordList[i].word;
float w = completedWordList[i].w / sizeRate;
float h = completedWordList[i].h / sizeRate;
float x = completedWordList[i].x / sizeRate;
float y = pageHeight - completedWordList[i].y / sizeRate - h;
RenderFilter[] filter = { new RegionTextRenderFilter(new System.util.RectangleJ(x, y, w, h)) };
ITextExtractionStrategy strategy = new FilteredTextRenderListener(new LocationTextExtractionStrategy(), filter);
string text = PdfTextExtractor.GetTextFromPage(reader, 1, strategy);
string[] wordSeparatingStrings = { " " };
string[] words = text.Split(wordSeparatingStrings, StringSplitOptions.RemoveEmptyEntries);
if (words.Length > 0)
{
bestSimilarity = ContractNote.getSimilarity(word, words[0]);
if (bestSimilarity > 0.8)
{
completedWordList[i].word = words[0];
}
}
int bestSimilarWordId = 0;
string agent1 = Regex.Replace(word, @"[^0-9a-zA-Z$@]+", "");
for (int j = 0; j < wordCombinationList.Count; j++)
{
string agent2 = Regex.Replace(wordCombinationList[j], @"[^0-9a-zA-Z$@]+", "");
double similarity = ContractNote.getSimilarity(agent1, agent2);
if (bestSimilarity < similarity)
{
bestSimilarity = similarity;
bestSimilarWordId = j;
}
}
if (bestSimilarity > 0.8)
{
completedWordList[i].word = wordCombinationList[bestSimilarWordId];
}
}
}
}
catch { }
}