Parse full sentence when recognition is complete

ButterscotchV · Jan 30, 2024 · 61b180b · 61b180b
1 parent 04df123
commit 61b180b
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 9 deletions.
diff --git a/ButterSTT/SpeechToTextHandler.cs b/ButterSTT/SpeechToTextHandler.cs
@@ -147,9 +147,20 @@ private void OnAprilTokens(AprilResultKind result, AprilToken[] tokens)
                     ? EnglishCapitalization.Capitalize(aprilOutput.ToString().Trim())
                     : "";
 
-            messageQueue.CurParagraph = EnglishTextParser.ParseParagraph(aprilOutputString);
             if (result == AprilResultKind.FinalRecognition)
+            {
+                messageQueue.CurParagraph = EnglishTextParser.ParseParagraph(
+                    aprilOutputString,
+                    wordRegex: EnglishTextParser.WordKeepUrl()
+                );
                 messageQueue.FinishCurrentParagraph();
+            }
+            else
+            {
+                messageQueue.CurParagraph = EnglishTextParser.ParseParagraph(aprilOutputString);
+            }
+
+            Console.WriteLine(messageQueue.GetCurrentMessage());
 
             try
             {

diff --git a/ButterSTT/TextProcessing/EnglishCapitalization.cs b/ButterSTT/TextProcessing/EnglishCapitalization.cs
@@ -4,9 +4,9 @@ namespace ButterSTT.TextProcessing
 {
     public static partial class EnglishCapitalization
     {
-        public static string Capitalize(string message)
+        public static string Capitalize(string message, Regex? regex = null)
         {
-            return BasicCapitals().Replace(message.ToLower(), c => c.Value.ToUpper());
+            return (regex ?? BasicCapitals()).Replace(message.ToLower(), c => c.Value.ToUpper());
         }
 
         // Capitalizes starts of sentences and standalone "I"s, must be run on a lowercase string

diff --git a/ButterSTT/TextProcessing/EnglishTextParser.cs b/ButterSTT/TextProcessing/EnglishTextParser.cs
@@ -5,20 +5,31 @@ namespace ButterSTT.TextProcessing
 {
     public static partial class EnglishTextParser
     {
-        public static Paragraph ParseParagraph(string text)
+        public static Paragraph ParseParagraph(
+            string text,
+            Regex? regex = null,
+            Regex? wordRegex = null,
+            bool addSpaces = true
+        )
         {
-            Sentence[] sentences = SentenceKeepUrl()
+            Sentence[] sentences = (regex ?? SentenceKeepUrl())
                 .Matches(text)
-                .Select(m => ParseSentence(m.Value))
+                .Select(m => ParseSentence(m.Value, regex: wordRegex, addSpaces: addSpaces))
                 .ToArray();
             return new Paragraph(sentences);
         }
 
-        public static Sentence ParseSentence(string text)
+        public static Sentence ParseSentence(
+            string text,
+            Regex? regex = null,
+            bool addSpaces = true
+        )
         {
-            Word[] words = WordOnlyCompleteKeepUrl()
+            Word[] words = (regex ?? WordOnlyCompleteKeepUrl())
                 .Matches(text)
-                .Select(m => new Word(m.Value.EndsWith(' ') ? m.Value : m.Value + " "))
+                .Select(m => new Word(
+                    addSpaces && !m.Value.EndsWith(' ') ? m.Value + " " : m.Value
+                ))
                 .ToArray();
             return new Sentence(words);
         }