adding some pattern recognition for displaying some interesting things for learning activity

app.py

@@ -11,6 +11,8 @@ import requests
 from io import BytesIO
 import cohere
 import os
+import re
+
 
 title = "# Welcome to AyaTonic"
 description = "Learn a New Language With Aya"
@@ -20,6 +22,47 @@ load_dotenv()
 COHERE_API_KEY = os.getenv('CO_API_KEY')
 SEAMLESSM4T = os.getenv('SEAMLESSM4T')
 
+
+# Regular expression patterns for each color
+patterns = {
+    "red": r'<span style="color: red;">(.*?)</span>',
+    "blue": r'<span style="color: blue;">(.*?)</span>',
+    "green": r'<span style="color: green;">(.*?)</span>',
+}
+
+# Dictionaries to hold the matches
+matches = {
+    "red": [],
+    "blue": [],
+    "green": [],
+}
+class TaggedPhraseExtractor:
+    def __init__(self, text=''):
+        self.text = text
+        self.patterns = {}
+
+    def set_text(self, text):
+        """Set the text to search within."""
+        self.text = text
+
+    def add_pattern(self, color, pattern):
+        """Add a new color and its associated pattern."""
+        self.patterns[color] = pattern
+
+    def extract_phrases(self):
+        """Extract phrases for all colors and patterns added."""
+        matches = {color: re.findall(pattern, self.text) for color, pattern in self.patterns.items()}
+        return matches
+
+    def print_phrases(self):
+        """Extract phrases and print them."""
+        matches = self.extract_phrases()
+        for color, phrases in matches.items():
+            print(f"Phrases with color {color}:")
+            for phrase in phrases:
+                print(f"- {phrase}")
+            print()  
+            
 co = cohere.Client(COHERE_API_KEY)
 audio_client = Client(SEAMLESSM4T)