add: check for the text being small so it can't go below 0, also file isn't image.png but the current timestamp

main.py

@@ -1,5 +1,5 @@
-import io
-import base64
+import os
+import time
 from typing import Dict, Any
 
 import numpy as np
@@ -39,14 +39,16 @@ def extract_text_from_regions(
 
 
 def predict(image: np.ndarray, target_lang: str):
+    timestamp = str(int(time.time() * 1000000))  # Generate a unique timestamp
+    temp_filename = f"image_{timestamp}.png"
 
     image = Image.fromarray(image)
-    image.save("image.png")
+    image.save(temp_filename)
 
     try:
         np_image = np.array(image)
 
-        results = predict_bounding_boxes(object_detection_model, "image.png")
+        results = predict_bounding_boxes(object_detection_model, temp_filename)
         extract_text_from_regions(np_image, target_lang, results)
 
         return np_image
@@ -54,3 +56,11 @@ def predict(image: np.ndarray, target_lang: str):
     except Exception as e:
         print(f"Error: {str(e)}")
         return None
+
+    finally:
+        # Clean up the temporary file
+        if os.path.exists(temp_filename):
+            try:
+                os.remove(temp_filename)
+            except OSError as e:
+                print(f"Warning: Could not remove temporary file {temp_filename}: {e}")

utils/write_text_on_image.py

@@ -64,6 +64,8 @@ def add_text(image: np.ndarray, text: str, contour: np.ndarray):
     line_height = 16
     font_size = 14
     wrapping_ratio = 0.075
+    min_font_size = 6  # Minimum font size to prevent going to 0
+    max_iterations = 20  # Prevent infinite loops
 
     wrap_width = max(1, int(w * wrapping_ratio))
     wrapped_text = textwrap.fill(text, width=wrap_width, break_long_words=True)
@@ -73,22 +75,40 @@ def add_text(image: np.ndarray, text: str, contour: np.ndarray):
     lines = wrapped_text.split("\n")
     total_text_height = (len(lines)) * line_height
 
-    while total_text_height > h:
-        line_height -= 2
-        font_size -= 2
-        wrapping_ratio += 0.025
+    iterations = 0
+    while (
+        total_text_height > h
+        and font_size > min_font_size
+        and iterations < max_iterations
+    ):
+        line_height = max(line_height - 2, min_font_size)
+        font_size = max(font_size - 2, min_font_size)
+        wrapping_ratio = min(wrapping_ratio + 0.025, 0.5)
         wrap_width = max(1, int(w * wrapping_ratio))
         wrapped_text = textwrap.fill(text, width=wrap_width, break_long_words=True)
         font = ImageFont.truetype(font_path, size=font_size)
         lines = wrapped_text.split("\n")
         total_text_height = (len(lines)) * line_height
+        iterations += 1
+
+    # If text still doesn't fit after all adjustments, truncate it
+    if total_text_height > h:
+        max_lines = max(1, h // line_height)
+        lines = lines[:max_lines]
+        if len(lines) < len(wrapped_text.split("\n")):
+            # Add ellipsis to last line if text was truncated
+            if lines:
+                lines[-1] = lines[-1][: max(0, len(lines[-1]) - 3)] + "..."
 
     # Vertical centering
-    text_y = y + (h - total_text_height) // 2
+    actual_text_height = len(lines) * line_height
+    text_y = y + max(0, (h - actual_text_height) // 2)
 
     for line in lines:
         text_length = draw.textlength(line, font=font)
-        text_x = x + (w - text_length) // 2
+        text_x = x + max(
+            0, (w - text_length) // 2
+        )  # Ensure x coordinate is not negative
         draw.text((text_x, text_y), line, font=font, fill=(0, 0, 0))
         text_y += line_height