/// <summary>Initializes the base class with the given word, then computes
/// ushort representations of the metaphone keys computed by the
/// base class</summary>
///
/// <param name="word">Word for which to compute metaphone keys</param>
public ShortDoubleMetaphone(String word)
: base(word)
{
m_primaryShortKey = ShortDoubleMetaphone.metaphoneKeyToShort(this.PrimaryKey);
if (this.AlternateKey != null)
{
m_alternateShortKey = ShortDoubleMetaphone.metaphoneKeyToShort(this.AlternateKey);
}
else
{
m_alternateShortKey = METAPHONE_INVALID_KEY;
}
}
/// <summary>Sets a new current word, computing the string and ushort representations
/// of the metaphone keys of the given word.
///
/// Note that this uses the new modifier, which hides the base class
/// computeKeys. The base class's computeKeys is then explicitly
/// called as part of the function body. It is important to note that
/// this is NOT equivalent to overriding a virtual function, in that
/// polymorphism is not provided. In this case, polymorphism is of no
/// value, while the potential efficiency gained by not using virtual
/// methods is quite valuable.</summary>
///
/// <param name="word">New current word for which to compute metaphone keys</param>
new public void computeKeys(String word)
{
base.computeKeys(word);
m_primaryShortKey = ShortDoubleMetaphone.metaphoneKeyToShort(this.PrimaryKey);
if (this.AlternateKey != null)
{
m_alternateShortKey = ShortDoubleMetaphone.metaphoneKeyToShort(this.AlternateKey);
}
else
{
m_alternateShortKey = METAPHONE_INVALID_KEY;
}
}
private void btn_ProperNouns_Click(object sender, RibbonControlEventArgs e)
{
Word.Document doc = Globals.ThisAddIn.Application.ActiveDocument;
HashSet <string> wordlist = new HashSet <string>();
foreach (Word.Range rng in TextHelpers.GetText(doc))
{
string txt = rng.Text;
Word.Style style = rng.get_Style();
if (style != null)
{
Regex re_heading = new Regex(@"(?i)(heading|title|date|toc)");
Match m = re_heading.Match(style.NameLocal);
if (m.Success)
{
continue;
}
}
HashSet <string> propers = new HashSet <string>();
propers = TextHelpers.ProperNouns(txt);
wordlist.UnionWith(propers);
}
//Produce the groupings
HashSet <string> capped = TextHelpers.KeepCaps(wordlist);
//DoubleMetaphone
Dictionary <ushort, List <string> > mpgroups = new Dictionary <ushort, List <string> >();
//Dictionary<string, List<string>> mpgroups = new Dictionary<string, List<string>>();
ShortDoubleMetaphone sdm = new ShortDoubleMetaphone();
//HashSet<string> tested = new HashSet<string>();
foreach (string word in capped)
{
/*
* if (tested.Contains(word))
* {
* continue;
* }
* else
* {
* tested.Add(word);
* }
*/
sdm.computeKeys(word);
ushort pri = sdm.PrimaryShortKey;
ushort alt = sdm.AlternateShortKey;
if (mpgroups.ContainsKey(pri))
{
mpgroups[pri].Add(word);
}
else
{
List <string> node = new List <string>();
node.Add(word);
mpgroups[pri] = node;
}
if (mpgroups.ContainsKey(alt))
{
mpgroups[alt].Add(word);
}
else
{
List <string> node = new List <string>();
node.Add(word);
mpgroups[alt] = node;
}
}
//Edit Distance
List <string> dtested = new List <string>();
int mindist;
int.TryParse(Properties.Settings.Default.mindist, out mindist);
if (mindist == 0)
{
mindist = 2;
}
Dictionary <string, List <string> > distgroups = new Dictionary <string, List <string> >();
foreach (string word1 in capped)
{
if (dtested.Contains(word1))
{
continue;
}
else
{
dtested.Add(word1);
}
if (word1.Length <= mindist)
{
continue;
}
foreach (string word2 in capped)
{
if (word2.Length <= mindist)
{
continue;
}
int dist = TextHelpers.EditDistance(word1, word2);
//int percent = (int)Math.Round((dist / word1.Length) * 100.0);
if ((dist > 0) && (dist <= mindist))
//if ((dist > 0) && (percent <= distpercent))
{
dtested.Add(word2);
if (distgroups.ContainsKey(word1))
{
distgroups[word1].Add(word2);
}
else
{
List <string> node = new List <string>();
node.Add(word2);
distgroups[word1] = node;
}
}
}
}
//Create new document
Word.Document newdoc = Globals.ThisAddIn.Application.Documents.Add();
Word.View view = Globals.ThisAddIn.Application.ActiveWindow.View;
view.DisplayPageBoundaries = false;
Word.Paragraph pgraph;
//Intro text
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Heading 1"]);
pgraph.Range.Text = "Proper Noun Checker\n";
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Normal"]);
pgraph.Range.Text = "This tool only looks at words that start with a capital letter. It then uses phonetic comparison and edit distance to find other proper nouns that are similar. Words in all caps (acronyms) are not included.\n";
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.Range.Text = "The system tries to ignore words at the beginning of sentences and in headers. This means some errors may go unseen, so use multiple tools!\n";
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.Range.Text = "Most of what you see here are false positives! That's unavoidable. But it still catches certain otherwise-hard-to-find misspellings.\n";
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.Range.InsertBreak(Word.WdBreakType.wdSectionBreakContinuous);
Word.Section sec = newdoc.Sections[2];
sec.PageSetup.TextColumns.SetCount(2);
sec.PageSetup.TextColumns.LineBetween = -1;
//Distance
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Heading 2"]);
//pgraph.KeepWithNext = 0;
pgraph.Range.Text = "Edit Distance (" + mindist + ")\n";
foreach (string key in distgroups.Keys)
{
List <string> group = distgroups[key];
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Normal"]);
pgraph.Range.Text = key + ", " + string.Join(", ", group) + "\n";
}
pgraph.Range.InsertBreak(Word.WdBreakType.wdPageBreak);
//pgraph = newdoc.Content.Paragraphs.Add();
//pgraph.Range.InsertBreak(Word.WdBreakType.wdSectionBreakContinuous);
//Word.InlineShape line = pgraph.Range.InlineShapes.AddHorizontalLineStandard();
//line.Height = 2;
//line.Fill.Solid();
//line.HorizontalLineFormat.NoShade = true;
//line.Fill.ForeColor.RGB = System.Drawing.ColorTranslator.ToOle(System.Drawing.Color.Black);
//line.HorizontalLineFormat.PercentWidth = 90;
//line.HorizontalLineFormat.Alignment = WdHorizontalLineAlignment.wdHorizontalLineAlignCenter;
//sec = newdoc.Sections[3];
//sec.PageSetup.TextColumns.SetCount(2);
//sec.PageSetup.TextColumns.LineBetween = -1;
//Metaphone
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Heading 2"]);
//pgraph.KeepWithNext = 0;
pgraph.Range.Text = "Phonetic Comparisons\n";
foreach (ushort key in mpgroups.Keys)
{
if (key == 65535)
{
continue;
}
List <string> group = mpgroups[key];
if (group.Count > 1)
{
pgraph = newdoc.Content.Paragraphs.Add();
pgraph.set_Style(newdoc.Styles["Normal"]);
pgraph.Range.Text = string.Join(", ", group) + "\n";
}
}
//pgraph = newdoc.Content.Paragraphs.Add();
//pgraph.Range.InsertBreak(Word.WdBreakType.wdSectionBreakContinuous);
newdoc.GrammarChecked = true;
}