From 7ae4b0f55811d63d12b6eb9ac8a534f99e836e00 Mon Sep 17 00:00:00 2001
From: RunasSudo <runassudo@yingtongli.me>
Date: Thu, 8 May 2025 21:24:21 +1000
Subject: [PATCH] Improve performance of reading XCF metadata

---
 pdf_segmented/input/xcf.py | 50 ++++++++++++++------------------------
 1 file changed, 18 insertions(+), 32 deletions(-)

diff --git a/pdf_segmented/input/xcf.py b/pdf_segmented/input/xcf.py
index d7c2fb5..4652586 100644
--- a/pdf_segmented/input/xcf.py
+++ b/pdf_segmented/input/xcf.py
@@ -35,45 +35,31 @@ def xcf_get_pages(input_file: str) -> InputPages:
 	height = None
 	
 	# Read metadata
-	proc = subprocess.run(['magick', 'identify', '-verbose', input_file], capture_output=True, encoding='utf-8', check=True)
-	for line in proc.stdout.splitlines():
-		if line == 'Image:':
-			num_layers += 1
-		elif line.startswith('  Geometry: '):
-			layer_width = float(line[len('  Geometry: '):line.index('x')])
-			layer_height = float(line[line.index('x')+1:line.index('+')])
-			if (width is not None and layer_width != width) or (height is not None and layer_height != height):
-				print('Error: Image with variable-dimension layers is not supported ({}x{} vs {}x{})'.format(layer_width, layer_height, width, height))
-				sys.exit(1)
-			width = layer_width
-			height = layer_height
-		elif line.startswith('  Resolution: '):
-			resolution_x = float(line[len('  Resolution: '):line.index('x')])
-			resolution_y = float(line[line.index('x')+1:])
-			if resolution_x != resolution_y:
-				raise Exception('Unexpected non-square DPI ({}x{})'.format(resolution_x, resolution_y))
-			if dpi is not None and resolution_x != dpi:
-				raise Exception('Unexpected variable DPI image ({} vs {})'.format(resolution_x, dpi))
-			dpi = resolution_x
-		elif line.startswith('  Units: '):
-			if line != '  Units: PixelsPerInch':
-				raise Exception('Unexpected Units (expected PixelsPerInch, got {})'.format(line[len('  Units: '):]))
+	proc = subprocess.run(['magick', 'identify', '-format', '%n;%W;%H;%x;%y;%U|', input_file], capture_output=True, encoding='utf-8', check=True)
+	magick_layers_data = proc.stdout.strip('|').split('|')
 	
-	if num_layers == 0:
-		raise Exception('Unexpected 0 layers')
-	if dpi is None:
-		raise Exception('Unexpected no DPI information')
-	if width is None:
-		raise Exception('Unexpected no width information')
-	if height is None:
-		raise Exception('Unexpected no height information')
+	if any(d != magick_layers_data[0] for d in magick_layers_data[1:]):
+		raise Exception('Different metadata between XCF layers')
+	
+	magick_layer_data = magick_layers_data[0].split(';')
+	num_layers = int(magick_layer_data[0])
+	width = int(magick_layer_data[1])
+	height = int(magick_layer_data[2])
+	resolution_x = float(magick_layer_data[3])
+	resolution_y = float(magick_layer_data[4])
+	resolution_units = magick_layer_data[5]
+	
+	if resolution_units != 'PixelsPerInch':
+		raise Exception('Unexpected resolution units (expected PixelsPerInch, got {})'.format(resolution_units))
+	if resolution_x != resolution_y:
+		raise Exception('Unexpected non-square DPI ({}x{})'.format(resolution_x, resolution_y))
 	
 	return InputPages(
 		file_name=input_file,
 		num_pages=num_layers,
 		width=width,
 		height=height,
-		dpi=dpi,
+		dpi=resolution_x,
 		pages=_do_get_pages(input_file, num_layers)
 	)