private ListViewItem LvCreate(JIndex.File x)
{
var lv = new ListViewItem {
Text = Path.GetFileName(x.Path), Tag = x
};
lv.SubItems.Add(new ListViewItem.ListViewSubItem {
Text = Directory.GetParent(x.Path).Name
});
lv.SubItems.Add(new ListViewItem.ListViewSubItem {
Text = x.Path
});
lv.SubItems.Add(new ListViewItem.ListViewSubItem {
Text = x.Id.ToString()
});
return(lv);
}
public List <JIndex.Grouping> GetAllWords(JIndex.File x) //, string phraseToFind = "")
{
List <JIndex.Grouping> L = new List <JIndex.Grouping>();
try
{
using (PdfDocument document = PdfDocument.Open(x.Path))
{
Dictionary <string, ulong> D = new Dictionary <string, ulong>();
double perc = (100.0 * _CurrentSize) / _TotalSize;
var tsEL = DateTime.Now - _DateStart;
double secs = tsEL.TotalSeconds;
int estimated;
_Velocity = _CurrentSize / secs;
estimated = (int)(_TotalSize / _Velocity);
tsEL = TimeSpan.FromSeconds((int)secs);
var tsES = TimeSpan.FromSeconds(estimated);
var P = (int)(perc * 0.45 + 5);
JProgressSet(P, $"[{perc.ToString("F1", CultureInfo.InvariantCulture)}%] EL {tsEL} ES {tsES}"
+ $" Processing file {x.Id} of {_Index.Files.Count}: {Path.GetFileName(x.Path)} ");
DateTime wc = DateTime.Now;
try
{
foreach (Page page in document.GetPages())
{
// string pageText = page.Text;
foreach (Word word in page.GetWords())
{
if (D.ContainsKey(word.Text))
{
D[word.Text]++;
}
else
{
D[word.Text] = 0;
}
// TODO: queue words to find phrases - move it outside this code.
if (DateTime.Now > wc)
{
perc = (100.0 * _CurrentSize) / _TotalSize;
tsEL = DateTime.Now - _DateStart;
secs = tsEL.TotalSeconds;
_Velocity = _CurrentSize / secs;
estimated = (int)(_TotalSize / _Velocity);
tsEL = TimeSpan.FromSeconds((int)secs);
tsES = TimeSpan.FromSeconds(estimated);
P = (int)(perc * 0.45 + 5);
JProgressSet(P, $"[{perc.ToString("F1", CultureInfo.InvariantCulture)}%] EL {tsEL} ES {tsES}"
+ $" Processing file {x.Id} of {_Index.Files.Count}: {Path.GetFileName(x.Path)} ");
wc = DateTime.Now.AddMilliseconds(200);
}
}
}
}
catch { }
_CurrentSize += x.Size;
// JProgressSet(P, $"Collecting file words: {x.Id} Words: {D.Count} {Path.GetFileName(x.Path)}");
foreach (var kv in D)
{
L.Add(new JIndex.Grouping {
GId = _gid++, Count = kv.Value, Id = x.Id, Word = kv.Key
});
}
D.Clear();
}
}
catch
{ }
return(L);
}
