OCR'ing navigate page separately from preview page.

Author: samuelclay

handlers/navigation_handler.py

@@ -321,10 +321,7 @@ def navigate(
                     if progress and progress.get("time_left"):
                         # Try to cycle to get page/location information
                         page_text = progress.get("reading_time_indicator", "")
-                        percentage = progress.get("percentage", "")
-                        updated_progress = cycle_page_indicator_if_needed(
-                            self.automator.driver, page_text, f"{percentage}%" if percentage else None
-                        )
+                        updated_progress = cycle_page_indicator_if_needed(self.automator.driver, page_text)
                         if updated_progress:
                             screenshot_data["progress"] = updated_progress
 
@@ -390,6 +387,13 @@ def navigate(
         if not success:
             return {"error": "Navigation failed"}, 500
 
+        # After navigation, capture page info at the user's actual position (not preview position)
+        # This is only needed when we're doing a preview, to get the correct page number
+        navigation_page_info = None
+        if success and preview_count != 0:
+            # Extract just the page indicator at the navigation position
+            _, navigation_page_info, _ = self._extract_screenshot_for_ocr(f"nav_pos")
+
         # If preview was requested, handle it after navigating
         preview_ocr_text = None
         preview_page_info = None
@@ -404,10 +408,10 @@ def navigate(
                 )
 
             if preview_success and preview_ocr_text:
-                # Use the page info from the preview if available, otherwise get current page data
-                if preview_page_info:
-                    # Use the page info extracted from OCR of the preview page
-                    progress = preview_page_info
+                # Use the page info from the navigation position, NOT the preview position
+                if navigation_page_info:
+                    # Use the page info extracted at navigation position
+                    progress = navigation_page_info
                 else:
                     # Fall back to getting current page data after navigation
                     progress = self.automator.state_machine.reader_handler.get_reading_progress(
@@ -464,10 +468,7 @@ def navigate(
                 if progress and progress.get("time_left"):
                     # Try to cycle to get page/location information
                     page_text = progress.get("reading_time_indicator", "")
-                    percentage = progress.get("percentage", "")
-                    updated_progress = cycle_page_indicator_if_needed(
-                        self.automator.driver, page_text, f"{percentage}%" if percentage else None
-                    )
+                    updated_progress = cycle_page_indicator_if_needed(self.automator.driver, page_text)
                     if updated_progress:
                         screenshot_data["progress"] = updated_progress
 
@@ -795,14 +796,11 @@ def _extract_screenshot_for_ocr(self, prefix: str) -> Tuple[Optional[str], Optio
 
                 # Extract page information from OCR results
                 page_indicator_text = ocr_results.get("page_indicator_text")
-                percentage_text = ocr_results.get("percentage_text")
 
                 # Parse the page indicators and handle time-based indicators
-                if page_indicator_text or percentage_text:
+                if page_indicator_text:
                     # Use the cycle function which will tap if needed for time-based indicators
-                    page_info = cycle_page_indicator_if_needed(
-                        self.automator.driver, page_indicator_text, percentage_text
-                    )
+                    page_info = cycle_page_indicator_if_needed(self.automator.driver, page_indicator_text)
                 else:
                     page_info = None
 

server/utils/ocr_utils.py

@@ -335,14 +335,14 @@ def process_ocr(image_content, clean_ui_elements=True) -> Tuple[Optional[str], O
         return None, combined_error
 
 
-def extract_page_indicator_regions(image_bytes):
-    """Extract the page indicator and percentage regions from a screenshot.
+def extract_page_indicator_region(image_bytes):
+    """Extract the page indicator region from a screenshot.
 
     Args:
         image_bytes: The screenshot as bytes
 
     Returns:
-        tuple: (page_indicator_bytes, percentage_bytes) - Cropped regions as bytes
+        bytes: page_indicator_bytes - Cropped region as bytes
     """
     from io import BytesIO
 
@@ -353,7 +353,7 @@ def extract_page_indicator_regions(image_bytes):
         img = Image.open(BytesIO(image_bytes))
         width, height = img.size
 
-        # Define crop regions based on proportions
+        # Define crop region based on proportions
         # Bottom-left for page/location indicator
         # The page number is in the bottom 6% of screen (bottom 80px of 1400px)
         page_indicator_box = (
@@ -363,46 +363,31 @@ def extract_page_indicator_regions(image_bytes):
             height,  # Bottom edge
         )
 
-        # Bottom-right for percentage
-        percentage_box = (
-            int(width * 0.5),  # Start at 50% from left
-            int(height * 0.94),  # Start at 94% from top (bottom 6%)
-            width,  # Right edge
-            height,  # Bottom edge
-        )
-
-        # Crop the regions
+        # Crop the region
         page_indicator_img = img.crop(page_indicator_box)
-        percentage_img = img.crop(percentage_box)
 
         # Convert back to bytes
         page_indicator_bytes = BytesIO()
         page_indicator_img.save(page_indicator_bytes, format="PNG")
         page_indicator_bytes = page_indicator_bytes.getvalue()
 
-        percentage_bytes = BytesIO()
-        percentage_img.save(percentage_bytes, format="PNG")
-        percentage_bytes = percentage_bytes.getvalue()
-
         logger.debug(f"Cropped page indicator region: {page_indicator_box}")
-        logger.debug(f"Cropped percentage region: {percentage_box}")
 
-        return page_indicator_bytes, percentage_bytes
+        return page_indicator_bytes
 
     except Exception as e:
-        logger.error(f"Error extracting page indicator regions: {e}", exc_info=True)
-        return None, None
+        logger.error(f"Error extracting page indicator region: {e}", exc_info=True)
+        return None
 
 
-def parse_page_indicators(page_indicator_text, percentage_text):
-    """Parse page indicator and percentage text to extract structured progress data.
+def parse_page_indicators(page_indicator_text):
+    """Parse page indicator text to extract structured progress data.
 
     Args:
         page_indicator_text: OCR text from page indicator region (e.g., "Page 123 of 456", "8 mins left in chapter")
-        percentage_text: OCR text from percentage region (e.g., "87%")
 
     Returns:
-        dict: Progress information with current_page/location, total_pages/locations, percentage, and/or time_left
+        dict: Progress information with current_page/location, total_pages/locations, and/or time_left
     """
     progress = {}
 
@@ -435,17 +420,6 @@ def parse_page_indicators(page_indicator_text, percentage_text):
             progress["current_page"] = None
             progress["total_pages"] = None
 
-    # Parse percentage
-    if percentage_text:
-        percentage_match = re.search(r"(\d+)%", percentage_text)
-        if percentage_match:
-            progress["percentage"] = int(percentage_match.group(1))
-            logger.info(f"Extracted percentage: {progress['percentage']}%")
-        else:
-            progress["percentage"] = None
-    else:
-        progress["percentage"] = None
-
     return progress
 
 
@@ -456,13 +430,13 @@ def process_screenshot_with_regions(image_bytes):
         image_bytes: The screenshot as bytes
 
     Returns:
-        dict: Contains 'main_text', 'page_indicator_text', 'percentage_text', and any errors
+        dict: Contains 'main_text', 'page_indicator_text', and any errors
     """
     from io import BytesIO
 
     from PIL import Image
 
-    result = {"main_text": None, "page_indicator_text": None, "percentage_text": None, "errors": []}
+    result = {"main_text": None, "page_indicator_text": None, "errors": []}
 
     try:
         # Load the image once
@@ -489,8 +463,8 @@ def process_screenshot_with_regions(image_bytes):
         elif main_error:
             result["errors"].append(f"Main text OCR error: {main_error}")
 
-        # Extract page indicator regions
-        page_indicator_bytes, percentage_bytes = extract_page_indicator_regions(image_bytes)
+        # Extract page indicator region
+        page_indicator_bytes = extract_page_indicator_region(image_bytes)
 
         # OCR page indicator
         if page_indicator_bytes:
@@ -503,19 +477,6 @@ def process_screenshot_with_regions(image_bytes):
             elif page_error:
                 result["errors"].append(f"Page indicator OCR error: {page_error}")
 
-        # OCR percentage
-        if percentage_bytes:
-            percent_text, percent_error = KindleOCR.process_ocr(percentage_bytes, clean_ui_elements=False)
-            if percent_text:
-                # Clean up the text - remove any extra whitespace
-                percent_text = percent_text.strip()
-                result["percentage_text"] = percent_text
-                logger.info(f"OCR: Percentage extracted: '{percent_text}'")
-            elif percent_error:
-                result["errors"].append(f"Percentage OCR error: {percent_error}")
-        else:
-            logger.warning("OCR: No percentage bytes extracted")
-
         return result
 
     except Exception as e:
@@ -626,19 +587,18 @@ def is_ocr_requested(default=False):
     return perform_ocr
 
 
-def cycle_page_indicator_if_needed(driver, page_indicator_text, percentage_text=None):
+def cycle_page_indicator_if_needed(driver, page_indicator_text):
     """If time-based indicator is detected, tap to cycle through formats to get page/location.
 
     Args:
         driver: The Appium driver instance
         page_indicator_text: The OCR'd text from the page indicator region
-        percentage_text: Optional percentage text
 
     Returns:
         dict: Updated progress information with page/location data if found
     """
     # First parse what we have
-    progress = parse_page_indicators(page_indicator_text, percentage_text)
+    progress = parse_page_indicators(page_indicator_text)
 
     # Check if we got a time-based indicator or "Learning reading speed" instead of page/location
     if (
@@ -710,7 +670,6 @@ def process_screenshot_response(screenshot_id, screenshot_path, use_base64=False
 
             ocr_text = ocr_results.get("main_text")
             page_indicator_text = ocr_results.get("page_indicator_text")
-            percentage_text = ocr_results.get("percentage_text")
             errors = ocr_results.get("errors", [])
 
             if ocr_text:
@@ -720,14 +679,12 @@ def process_screenshot_response(screenshot_id, screenshot_path, use_base64=False
                 logger.info(f"OCR text extracted successfully, length: {len(ocr_text)} characters")
 
                 # Log what we got from the page regions
-                logger.info(
-                    f"Page indicator text: '{page_indicator_text}', Percentage text: '{percentage_text}'"
-                )
+                logger.info(f"Page indicator text: '{page_indicator_text}'")
 
                 # Parse and add page progress information if extracted
                 # Note: We can't use cycle_page_indicator_if_needed here because we don't have access to the driver
                 # The cycling should be handled by the calling code that has access to the driver
-                progress = parse_page_indicators(page_indicator_text, percentage_text)
+                progress = parse_page_indicators(page_indicator_text)
 
                 # Log the parsed progress
                 logger.info(f"Parsed progress: {progress}")