private void SearchResultsListItemControl_DataContextChanged(object sender, DependencyPropertyChangedEventArgs e)
{
TextContext.Inlines.Clear();
// Check that we know what this DataContext is
PDFSearchResult search_result = DataContext as PDFSearchResult;
if (null == search_result)
{
return;
}
string current_context_sentence = search_result.context_sentence;
string current_context_sentence_lower = current_context_sentence.ToLower();
while (!String.IsNullOrEmpty(current_context_sentence))
{
// Look for the next keyword
int next_k = -1;
int next_keyword_pos = Int32.MaxValue;
for (int k = 0; k < search_result.keywords.Length; ++k)
{
int pos = current_context_sentence_lower.IndexOf(search_result.keywords[k]);
if (-1 < pos && pos < next_keyword_pos)
{
next_k = k;
next_keyword_pos = pos;
}
}
// If we have a keyword
if (-1 != next_k)
{
// Add a span of everything up to the keyword
{
Run run = new Run(current_context_sentence.Substring(0, next_keyword_pos));
TextContext.Inlines.Add(run);
}
{
Run run = new Run(current_context_sentence.Substring(next_keyword_pos, search_result.keywords[next_k].Length));
run.FontWeight = FontWeights.Bold;
run.Foreground = TextSearchBrushes.Instance.GetBrushPair(next_k).BorderBrush;
TextContext.Inlines.Add(run);
}
current_context_sentence = current_context_sentence.Substring(next_keyword_pos + search_result.keywords[next_k].Length);
current_context_sentence_lower = current_context_sentence.ToLower();
}
else
{
Run run = new Run(current_context_sentence);
TextContext.Inlines.Add(run);
current_context_sentence = "";
current_context_sentence_lower = "";
}
}
}
void ListSearchResults_SelectionChanged(object sender, SelectionChangedEventArgs e)
{
PDFSearchResult search_result = ListSearchResults.SelectedItem as PDFSearchResult;
if (null != SearchSelectionChanged)
{
SearchSelectionChanged(search_result);
}
}
void ListSearchResults_MouseUp(object sender, MouseButtonEventArgs e)
{
PDFSearchResult search_result = ListSearchResults.SelectedItem as PDFSearchResult;
if (null != SearchClicked)
{
SearchClicked(search_result);
}
}
public static List <PDFSearchResult> SearchPage(PDFDocument pdf_document, int page, string terms, MatchDelegate match)
{
// Tidy up the keywords
if (null == terms)
{
terms = "";
}
string[] keywords = GenerateIndividualKeywords(terms);
List <PDFSearchResult> search_results = new List <PDFSearchResult>();
WordList words = new WordList();
var SPLITTER_WHITESPACE = new char[] { ' ', '\n', '\r', '\t' };
// Add the comments
{
if (1 == page && !String.IsNullOrEmpty(pdf_document.Comments))
{
var splits = pdf_document.Comments.Split(SPLITTER_WHITESPACE, StringSplitOptions.RemoveEmptyEntries);
foreach (var split in splits)
{
words.Add(new Word {
Text = split
});
}
}
}
// Add the annotations
{
foreach (var pdf_annotation in pdf_document.GetAnnotations())
{
if (page == pdf_annotation.Page)
{
if (!String.IsNullOrEmpty(pdf_annotation.Text))
{
var splits = pdf_annotation.Text.Split(SPLITTER_WHITESPACE, StringSplitOptions.RemoveEmptyEntries);
foreach (var split in splits)
{
words.Add(new Word {
Text = split
});
}
}
}
}
}
// Add the PDF running text
{
WordList words_pdf = pdf_document.GetOCRText(page);
if (null != words_pdf)
{
words.AddRange(words_pdf);
}
}
// Find the text
if (null != words && null != keywords)
{
// Split keywords
string[][] split_keywords = new string[keywords.Length][];
for (int i = 0; i < keywords.Length; ++i)
{
split_keywords[i] = StringTools.Split_NotInDelims(keywords[i].ToLower(), '"', '"', " ").ToArray();
}
for (int w = 0; w < words.Count; ++w)
{
Word first_word = words[w];
string first_word_lower = first_word.Text.ToLower();
for (int i = 0; i < split_keywords.Length; ++i)
{
string[] split_keyword = split_keywords[i];
// Ignore spurious empty keyword sets
if (1 > split_keyword.Length)
{
continue;
}
// Don't process single keywords that are too short
if (2 > split_keyword[0].Length)
{
continue;
}
// Process the first word - if it doesn't match we are done here
if (!match(first_word_lower, split_keyword[0]))
{
continue;
}
// If there are more words we have to get a little crafty and check the remaining words
bool follows_match = true;
for (int j = 0; j < split_keyword.Length; ++j)
{
if (w + j < words.Count)
{
Word follow_word = words[w + j];
string follow_word_lower = follow_word.Text.ToLower();
if (!match(follow_word_lower, split_keyword[j]))
{
follows_match = false;
break;
}
}
else
{
follows_match = false;
break;
}
}
// If the remaining words dont match, bail
if (!follows_match)
{
continue;
}
// If we get here, the word (any any follow words) match
{
PDFSearchResult search_result = new PDFSearchResult();
search_results.Add(search_result);
// Store the page
search_result.keywords = keywords;
search_result.page = page;
// Get the words associated with this result
{
search_result.keyword_index = i;
search_result.words = new Word[split_keyword.Length];
for (int j = 0; j < split_keyword.Length; ++j)
{
Word follow_word = words[w + j];
search_result.words[j] = follow_word;
}
}
// Create the context sentence
{
int MIN_CONTEXT_SIZE = 3;
int MAX_CONTEXT_SIZE = 10;
bool ellipsis_start = false;
bool ellipsis_end = false;
int w_start = w;
while (w_start > 0)
{
// Stop at a preceding sentence
if (ContainsASentenceTerminator(words[w_start - 1].Text))
{
if (w - w_start >= MIN_CONTEXT_SIZE)
{
break;
}
}
// Stop if we are going too far
if (w - w_start > MAX_CONTEXT_SIZE)
{
ellipsis_start = true;
break;
}
--w_start;
}
int w_end = w;
while (w_end < words.Count)
{
// Stop at the end of a sentence
if (ContainsASentenceTerminator(words[w_end].Text))
{
if (w_end - w >= MIN_CONTEXT_SIZE)
{
break;
}
}
// Stop if we are going too far
if (w_end - w > MAX_CONTEXT_SIZE)
{
ellipsis_end = true;
break;
}
if (w_end + 1 == words.Count)
{
break;
}
++w_end;
}
StringBuilder sb = new StringBuilder();
sb.AppendFormat("p{0}: ", page);
if (ellipsis_start)
{
sb.Append("...");
}
for (int w_current = w_start; w_current <= w_end; ++w_current)
{
sb.Append(words[w_current].Text);
sb.Append(" ");
}
if (ellipsis_end)
{
sb.Append("...");
}
search_result.context_sentence = sb.ToString();
}
}
}
}
}
return(search_results);
}
private void ListSearchResults_SelectionChanged(object sender, SelectionChangedEventArgs e)
{
PDFSearchResult search_result = ListSearchResults.SelectedItem as PDFSearchResult;
SearchSelectionChanged?.Invoke(search_result);
}
private void ListSearchResults_MouseUp(object sender, MouseButtonEventArgs e)
{
PDFSearchResult search_result = ListSearchResults.SelectedItem as PDFSearchResult;
SearchClicked?.Invoke(search_result);
}