Update utils/helpers.py

utils/helpers.py

@@ -14,6 +14,9 @@ from langchain.text_splitter import (
     RecursiveCharacterTextSplitter,
     SentenceTransformersTokenTextSplitter,
 )
+import matplotlib.pyplot as plt
+import matplotlib.patches as patches
+import streamlit as st
 import chromadb
 from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction
 
@@ -230,3 +233,41 @@ def draw_boxes(image: Any, predictions: List[Dict[str, Any]]) -> Any:
 
     # Return the annotated image
     return image
+
+
+def draw_bounding_boxes_for_textract(image, json_data):
+    # Load the image
+    # image: image = Image.open(image)
+    fig, ax = plt.subplots(1)
+    ax.imshow(image)
+    
+    # Parse the JSON data
+    data = json.loads(json_data)
+    
+    # Check if 'body' key exists and load its content
+    if 'body' in data:
+        blocks = json.loads(data['body'])
+    else:
+        st.error('Invalid JSON data.')
+        return
+    
+    # Iterate through the elements to find bounding boxes
+    for item in blocks:
+        if 'BlockType' in item and item['BlockType'] in ['LINE', 'WORD']:
+            bbox = item['Geometry']['BoundingBox']
+            # Extract coordinates and dimensions
+            left, top, width, height = bbox['Left'], bbox['Top'], bbox['Width'], bbox['Height']
+            # Convert to image coordinates (may need adjustment based on your image dimensions)
+            rect = patches.Rectangle((left * image.width, top * image.height), width * image.width, height * image.height, linewidth=1, edgecolor='r', facecolor='none')
+            ax.add_patch(rect)
+    
+    # Hide axes
+    ax.axis('off')
+    # Save the annotated image to a temporary file
+    temp_file = "temp_annotated_image.png"
+    plt.savefig(temp_file, bbox_inches='tight', pad_inches=0)
+    plt.close()
+    
+    # Display the annotated image in Streamlit
+    st.image(temp_file)
+