/// <summary>
/// Разбить на слова.
/// </summary>
/// <param name="src">Исходная строка.</param>
/// <returns>Разбитая на слова строка.</returns>
public IEnumerable<string> Split(string src)
{
var ws = new Windows.Data.Text.WordsSegmenter("ru-RU");
var tokens = ws.GetTokens(src ?? "");
foreach (var token in tokens)
{
yield return token.Text;
}
}
/// <summary>
/// This is the click handler for the 'Word Segments' button.
///
/// When this button is activated, the Text Segmentation API will calculate
/// the word segment from the given input string and character index for that string,
/// and return the WordSegment object that contains the index within its text bounds.
/// Segment breaking behavior is based off of the language-tag input, which defines
/// which language rules to use.
///
/// </summary>
/// <param name="sender">The object that raised the event.</param>
/// <param name="e">Event data that describes the click action on the button.</param>
private void WordSegmentButton_Click(object sender, RoutedEventArgs e)
{
Button b = sender as Button;
if (b != null)
{
// Initialize and obtain input values
StringBuilder notifyText = new StringBuilder();
// Obtain the input string value, check for non-emptiness
String inputStringText = inputStringBox.Text;
if (String.IsNullOrEmpty(inputStringText))
{
notifyText = new StringBuilder("Cannot compute word segments: input string is empty.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
// Obtain the language tag value, check for non-emptiness
// Ex. Valid Values:
// "en-US" (English (United States))
// "fr-FR" (French (France))
// "de-DE" (German (Germany))
// "ja-JP" (Japanese (Japan))
// "ar-SA" (Arabic (Saudi Arabia))
// "zh-CN" (China (PRC))
String languageTagText = languageTagBox.Text;
if (String.IsNullOrEmpty(languageTagText))
{
notifyText.AppendLine("Language tag input is empty ... using generic-language segmentation rules.");
languageTagText = "und"; // This is used for non language-specific locales. 'und' is short for 'undetermined'.
}
else
{
if (!Windows.Globalization.Language.IsWellFormed(languageTagText))
{
notifyText = new StringBuilder("Language tag is not well formed: \"" + languageTagText + "\"");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Obtain the input Index
String inputIndexString = indexBox.Text;
uint inputIndex = 0;
if (String.IsNullOrEmpty(inputIndexString))
{
notifyText.AppendLine("No input index provided ... using first segment reference (index = 0) as default.");
}
else
{
try
{
inputIndex = Convert.ToUInt32(indexBox.Text);
}
catch (FormatException)
{
notifyText = new StringBuilder("Invalid index supplied.\n\nPlease check that this value is valid, and non-negative.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
catch (OverflowException)
{
notifyText = new StringBuilder("Invalid index supplied: Negative-valued index.\n\nPlease check that this value is valid, and non-negative.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
if ((inputIndex < 0) || (inputIndex >= inputStringText.Length))
{
notifyText = new StringBuilder("Invalid index supplied ... cannot use a negative index, or an index that is out of bounds of the input string.\n\nPlease re-check the index value, and try again.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Notify that we are going to calculate word segment
notifyText.AppendLine("\nFinding the word segment for the given index ...\n");
notifyText.AppendLine("Input: \"" + inputStringText + "\"");
notifyText.AppendLine("Language Tag: \"" + languageTagText + "\"");
notifyText.AppendLine("Index: " + inputIndex + "\n");
// Construct the WordsSegmenter instance
var segmenter = new Windows.Data.Text.WordsSegmenter(languageTagText);
// Obtain the token segment
var tokenSegment = segmenter.GetTokenAt(inputStringText, inputIndex);
notifyText.AppendLine("Indexed segment: \"" + tokenSegment.Text + "\"");
// Set output box text to the contents of the StringBuilder instance
rootPage.NotifyUser(notifyText.ToString(), NotifyType.StatusMessage);
}
}
/// <summary>
/// This is the click handler for the 'Word Segments' button.
///
/// When this button is activated, the Text Segmentation API will calculate
/// the word segment from the given input string and character index for that string,
/// and return the WordSegment object that contains the index within its text bounds.
/// Segment breaking behavior is based off of the language-tag input, which defines
/// which language rules to use.
///
/// </summary>
/// <param name="sender">The object that raised the event.</param>
/// <param name="e">Event data that describes the click action on the button.</param>
private void WordSegmentButton_Click(object sender, RoutedEventArgs e)
{
// Initialize and obtain input values
StringBuilder notifyText = new StringBuilder();
// Obtain the input string value, check for non-emptiness
String inputStringText = inputStringBox.Text;
if (String.IsNullOrEmpty(inputStringText))
{
notifyText = new StringBuilder("Cannot compute word segments: input string is empty.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
// Obtain the language tag value, check for non-emptiness
// Ex. Valid Values:
// "en-US" (English (United States))
// "fr-FR" (French (France))
// "de-DE" (German (Germany))
// "ja-JP" (Japanese (Japan))
// "ar-SA" (Arabic (Saudi Arabia))
// "zh-CN" (China (PRC))
String languageTagText = languageTagBox.Text;
if (String.IsNullOrEmpty(languageTagText))
{
notifyText.AppendLine("Language tag input is empty ... using generic-language segmentation rules.");
languageTagText = "und"; // This is used for non language-specific locales. 'und' is short for 'undetermined'.
}
else
{
if (!Windows.Globalization.Language.IsWellFormed(languageTagText))
{
notifyText = new StringBuilder("Language tag is not well formed: \"" + languageTagText + "\"");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Obtain the input Index
String inputIndexString = indexBox.Text;
uint inputIndex = 0;
if (String.IsNullOrEmpty(inputIndexString))
{
notifyText.AppendLine("No input index provided ... using first segment reference (index = 0) as default.");
}
else
{
try
{
inputIndex = Convert.ToUInt32(indexBox.Text);
}
catch (FormatException)
{
notifyText = new StringBuilder("Invalid index supplied.\n\nPlease check that this value is valid, and non-negative.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
catch (OverflowException)
{
notifyText = new StringBuilder("Invalid index supplied: Negative-valued index.\n\nPlease check that this value is valid, and non-negative.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
if ((inputIndex < 0) || (inputIndex >= inputStringText.Length))
{
notifyText = new StringBuilder("Invalid index supplied ... cannot use a negative index, or an index that is out of bounds of the input string.\n\nPlease re-check the index value, and try again.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Notify that we are going to calculate word segment
notifyText.AppendLine("\nFinding the word segment for the given index ...\n");
notifyText.AppendLine("Input: \"" + inputStringText + "\"");
notifyText.AppendLine("Language Tag: \"" + languageTagText + "\"");
notifyText.AppendLine("Index: " + inputIndex + "\n");
// Construct the WordsSegmenter instance
var segmenter = new Windows.Data.Text.WordsSegmenter(languageTagText);
// Obtain the token segment
var tokenSegment = segmenter.GetTokenAt(inputStringText, inputIndex);
notifyText.AppendLine("Indexed segment: \"" + tokenSegment.Text + "\"");
// Set output box text to the contents of the StringBuilder instance
rootPage.NotifyUser(notifyText.ToString(), NotifyType.StatusMessage);
}
/// <summary>
/// This is the click handler for the 'Word Segments' button.
///
/// When this button is activated, the Text Segmentation API will calculate
/// the word segments from the given input string and language tag, and then
/// print out the word and alternate-form segments for that language.
///
/// </summary>
/// <param name="sender">The object that raised the event.</param>
/// <param name="e">Event data that describes the click action on the button.</param>
private void WordSegmentsButton_Click(object sender, RoutedEventArgs e)
{
// Initialize and obtain input values
StringBuilder notifyText = new StringBuilder();
// Obtain the input string value, check for non-emptiness
String inputStringText = inputStringBox.Text;
if (String.IsNullOrEmpty(inputStringText))
{
notifyText = new StringBuilder("Cannot compute word segments: input string is empty.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
// Obtain the language tag value, check for non-emptiness
// Ex. Valid Values:
// "en-US" (English (United States))
// "fr-FR" (French (France))
// "de-DE" (German (Germany))
// "ja-JP" (Japanese (Japan))
// "ar-SA" (Arabic (Saudi Arabia))
// "zh-CN" (China (PRC))
String languageTagText = languageTagBox.Text;
if (String.IsNullOrEmpty(languageTagText))
{
notifyText.AppendLine("Language tag input is empty ... using generic-language segmentation rules.");
languageTagText = "und"; // This is used for non language-specific locales. 'und' is short for 'undetermined'.
}
else
{
if (!Windows.Globalization.Language.IsWellFormed(languageTagText))
{
notifyText = new StringBuilder("Language tag is not well formed: \"" + languageTagText + "\"");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Notify that we are going to calculate word segments
notifyText.AppendLine("\nCalculating word segments ...\n");
notifyText.AppendLine("Input: \"" + inputStringText + "\"");
notifyText.AppendLine("Language Tag: \"" + languageTagText + "\"");
// Construct the WordsSegmenter instance
var segmenter = new Windows.Data.Text.WordsSegmenter(languageTagText);
var wordSegments = segmenter.GetTokens(inputStringText);
// Iterate over the tokenized segments
foreach (var wordSegment in wordSegments)
{
notifyText.AppendLine("\tWord---->\"" + wordSegment.Text + "\"");
foreach (var alternateForm in wordSegment.AlternateForms)
{
notifyText.AppendLine("\t\tAlternate---->\"" + alternateForm.AlternateText + "\"");
}
}
// send notifyText to the output pane
rootPage.NotifyUser(notifyText.ToString(), NotifyType.StatusMessage);
}
/// <summary>
/// This is the click handler for the 'Word Segments' button.
///
/// When this button is activated, the Text Segmentation API will calculate
/// the word segments from the given input string and language tag, and then
/// print out the word and alternate-form segments for that language.
///
/// </summary>
/// <param name="sender">The object that raised the event.</param>
/// <param name="e">Event data that describes the click action on the button.</param>
private void WordSegmentsButton_Click(object sender, RoutedEventArgs e)
{
// Initialize and obtain input values
StringBuilder notifyText = new StringBuilder();
// Obtain the input string value, check for non-emptiness
String inputStringText = inputStringBox.Text;
if (String.IsNullOrEmpty(inputStringText))
{
notifyText = new StringBuilder("Cannot compute word segments: input string is empty.");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
// Obtain the language tag value, check for non-emptiness
// Ex. Valid Values:
// "en-US" (English (United States))
// "fr-FR" (French (France))
// "de-DE" (German (Germany))
// "ja-JP" (Japanese (Japan))
// "ar-SA" (Arabic (Saudi Arabia))
// "zh-CN" (China (PRC))
String languageTagText = languageTagBox.Text;
if (String.IsNullOrEmpty(languageTagText))
{
notifyText.AppendLine("Language tag input is empty ... using generic-language segmentation rules.");
languageTagText = "und"; // This is used for non language-specific locales. 'und' is short for 'undetermined'.
}
else
{
if (!Windows.Globalization.Language.IsWellFormed(languageTagText))
{
notifyText = new StringBuilder("Language tag is not well formed: \"" + languageTagText + "\"");
rootPage.NotifyUser(notifyText.ToString(), NotifyType.ErrorMessage);
return;
}
}
// Notify that we are going to calculate word segments
notifyText.AppendLine("\nCalculating word segments ...\n");
notifyText.AppendLine("Input: \"" + inputStringText + "\"");
notifyText.AppendLine("Language Tag: \"" + languageTagText + "\"");
// Construct the WordsSegmenter instance
var segmenter = new Windows.Data.Text.WordsSegmenter(languageTagText);
var wordSegments = segmenter.GetTokens(inputStringText);
// Iterate over the tokenized segments
foreach (var wordSegment in wordSegments)
{
notifyText.AppendLine("\tWord---->\"" + wordSegment.Text + "\"");
foreach (var alternateForm in wordSegment.AlternateForms)
{
notifyText.AppendLine("\t\tAlternate---->\"" + alternateForm.AlternateText + "\"");
}
}
// send notifyText to the output pane
rootPage.NotifyUser(notifyText.ToString(), NotifyType.StatusMessage);
}
