Clean up

CHANGELOG.md

@@ -15,10 +15,18 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
-- Updated `LLMTextProcessor` and `TTSService` to normalize text input by
-  splitting into individual characters before aggregation. This ensures proper
-  sentence boundary detection when LLMs return multiple sentences in a single
-  chunk (e.g., Google Gemini).
+- Text Aggregation Improvements:
+
+  - Updated `LLMTextProcessor` and `TTSService` to normalize text input by
+    splitting into individual characters before aggregation. This ensures proper
+    sentence boundary detection when LLMs return multiple sentences in a single
+    chunk (e.g., Google Gemini).
+  - All text aggregators now properly support character-by-character streaming
+    input.
+  - Refactored text aggregators to use inheritance: `SkipTagsAggregator` and
+    `PatternPairAggregator` now inherit from `SimpleTextAggregator`, reusing
+    its lookahead-based sentence detection logic via
+    `_check_sentence_with_lookahead()`.
 
 - Updated `AICFilter` to use Quail STT as the default model
   (`AICModelType.QUAIL_STT`). Quail STT is optimized for human-to-machine
@@ -58,6 +66,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   voice-ui-kit's conversational panel rending of the LLM output after a
   function call.
 
+- Fixed a bug in `PatternPairAggregator` where pattern handlers could be called
+  multiple times for patterns with `MatchAction.KEEP` or `MatchAction.AGGREGATE`
+  actions.
+
 ## [0.0.96] - 2025-11-26 🦃 "Happy Thanksgiving!" 🦃
 
 ### Added

src/pipecat/utils/text/simple_text_aggregator.py

@@ -60,13 +60,19 @@ class SimpleTextAggregator(BaseTextAggregator):
         # Add new text to buffer
         self._text += text
 
-        # Delegate to sentence detection logic
         return await self._check_sentence_with_lookahead(text)
 
     async def _check_sentence_with_lookahead(self, text: str) -> Optional[Aggregation]:
         """Check for sentence boundaries using lookahead logic.
 
         This method implements the core sentence detection logic with lookahead.
+        When sentence-ending punctuation is detected, it waits for the next
+        non-whitespace character before calling NLTK. This disambiguates cases
+        like "$29." (not a sentence) vs "$29. Next" (sentence ends at period).
+        Whitespace alone is not meaningful lookahead since it appears in both
+        cases. Instead, the first non-whitespace character after the punctuation
+        is used to confirm the sentence boundary.
+
         Subclasses can call this via super() to reuse the lookahead behavior
         while adding their own logic (e.g., tag handling, pattern matching).