Implement PNG compression

pdf_segmented/__init__.py

@@ -14,7 +14,7 @@
 #   You should have received a copy of the GNU Affero General Public License
 #   along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
-from .compression import compress_pages
+from .compression import CompressionOptions, compress_pages
 from .input.xcf import xcf_get_pages
 from .output.pdf import pdf_write_pages
 from .segmentation import segment_pages
@@ -31,7 +31,7 @@ def convert_file(
 	output_format: Optional[str] = None,
 	fg_compression: str = 'jbig2',
 	bg_compression: str = 'jpeg',
-	jpeg_quality: Optional[float] = None
+	options: CompressionOptions = CompressionOptions()
 ) -> None:
 	# Create temporary directory
 	tempdir = tempfile.mkdtemp('pdf-segmented')
@@ -63,7 +63,7 @@ def convert_file(
 			segmented_pages=segmented_pages,
 			fg_compression=fg_compression,
 			bg_compression=bg_compression,
-			jpeg_quality=jpeg_quality,
+			options=options,
 			tempdir=tempdir
 		)
 		

pdf_segmented/__main__.py

@@ -15,6 +15,7 @@
 #   along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
 from . import convert_file
+from .compression import CompressionOptions
 
 import argparse
 
@@ -27,8 +28,10 @@ parser.add_argument('input_file')
 parser.add_argument('output_file')
 parser.add_argument('--input-format', choices=['xcf'])
 parser.add_argument('--output-format', choices=['pdf'])
-parser.add_argument('--fg-compression', default='jbig2', choices=['jbig2'])
-parser.add_argument('--bg-compression', default='jpeg', choices=['jpeg'])
+parser.add_argument('--fg-compression', default='jbig2', choices=['jbig2', 'png'])
+parser.add_argument('--bg-compression', default='jpeg', choices=['jpeg', 'jp2', 'png'])
+parser.add_argument('--jp2-lossless', action='store_true')
+parser.add_argument('--jp2-rate', type=float)
 parser.add_argument('--jpeg-quality', type=float)
 
 args = parser.parse_args()
@@ -41,5 +44,9 @@ convert_file(
 	output_format=args.output_format,
 	fg_compression=args.fg_compression,
 	bg_compression=args.bg_compression,
+	options=CompressionOptions(
+		jp2_lossless=args.jp2_lossless,
+		jp2_rate=args.jp2_rate,
 		jpeg_quality=args.jpeg_quality
 	)
+)

pdf_segmented/compression/__init__.py

@@ -22,7 +22,9 @@ class CompressedLayer:
 		pass
 
 from .jbig2 import jbig2_compress_layer
+from .jp2 import jp2_compress_layer
 from .jpeg import jpeg_compress_layer
+from .png import png_compress_layer
 from ..segmentation import SegmentedPage
 
 from PIL import Image
@@ -30,6 +32,12 @@ from PIL import Image
 from dataclasses import dataclass
 from typing import Generator, Iterable, Optional
 
+@dataclass
+class CompressionOptions:
+	jp2_lossless: bool = False
+	jp2_rate: Optional[float] = None
+	jpeg_quality: Optional[float] = None
+
 @dataclass
 class CompressedPage:
 	fg: CompressedLayer
@@ -39,7 +47,7 @@ def compress_pages(
 	segmented_pages: Iterable[SegmentedPage],
 	fg_compression: str,
 	bg_compression: str,
-	jpeg_quality: Optional[float],
+	options: CompressionOptions,
 	tempdir: str
 ) -> Generator[CompressedPage]:
 	
@@ -49,7 +57,7 @@ def compress_pages(
 			segmented_page=segmented_page,
 			fg_compression=fg_compression,
 			bg_compression=bg_compression,
-			jpeg_quality=jpeg_quality,
+			options=options,
 			tempdir=tempdir
 		)
 
@@ -57,7 +65,7 @@ def compress_page(
 	segmented_page: SegmentedPage,
 	fg_compression: str,
 	bg_compression: str,
-	jpeg_quality: Optional[float],
+	options: CompressionOptions,
 	tempdir: str
 ) -> CompressedPage:
 	
@@ -66,13 +74,15 @@ def compress_page(
 		fg=compress_layer(
 			layer=segmented_page.fg,
 			compression=fg_compression,
-			jpeg_quality=jpeg_quality,
+			is_foreground=True,
+			options=options,
 			tempdir=tempdir
 		),
 		bg=compress_layer(
 			layer=segmented_page.bg,
 			compression=bg_compression,
-			jpeg_quality=jpeg_quality,
+			is_foreground=False,
+			options=options,
 			tempdir=tempdir
 		)
 	)
@@ -80,14 +90,19 @@ def compress_page(
 def compress_layer(
 	layer: Image,
 	compression: str,
-	jpeg_quality: Optional[float],
+	is_foreground: bool,
+	options: CompressionOptions,
 	tempdir: str
 ) -> CompressedLayer:
 	
 	# Compress the given layer
 	if compression == 'jbig2':
 		return jbig2_compress_layer(layer=layer, tempdir=tempdir)
+	elif compression == 'jp2':
+		return jp2_compress_layer(layer=layer, jp2_lossless=options.jp2_lossless, jp2_rate=options.jp2_rate)
 	elif compression == 'jpeg':
-		return jpeg_compress_layer(layer=layer, jpeg_quality=jpeg_quality)
+		return jpeg_compress_layer(layer=layer, jpeg_quality=options.jpeg_quality)
+	elif compression == 'png':
+		return png_compress_layer(layer=layer, is_foreground=is_foreground)
 	else:
 		raise NotImplementedError()

pdf_segmented/output/pdf.py

@@ -16,10 +16,12 @@
 
 from ..compression import CompressedLayer, CompressedPage
 from ..compression.jbig2 import JBIG2Layer
+from ..compression.jp2 import JP2Layer
 from ..compression.jpeg import JPEGLayer
+from ..compression.png import PNGLayer
 from ..input import InputPages
 
-from pikepdf import ContentStreamInstruction, Name, Operator, Page, Pdf, Stream, unparse_content_stream
+from pikepdf import ContentStreamInstruction, Dictionary, Name, Operator, Page, Pdf, Stream, unparse_content_stream
 
 from typing import Generator
 
@@ -41,8 +43,8 @@ def pdf_write_pages(
 		
 		# Write each layer to the page
 		content_instructions = []
-		pdf_write_layer(input_pages=input_pages, pdf=pdf, page=page, layer=compressed_page.bg, content_instructions=content_instructions)
-		pdf_write_layer(input_pages=input_pages, pdf=pdf, page=page, layer=compressed_page.fg, content_instructions=content_instructions)
+		pdf_write_layer(input_pages=input_pages, pdf=pdf, page=page, layer=compressed_page.bg, is_foreground=False, content_instructions=content_instructions)
+		pdf_write_layer(input_pages=input_pages, pdf=pdf, page=page, layer=compressed_page.fg, is_foreground=True, content_instructions=content_instructions)
 		
 		# Generate content stream
 		wrapped_instructions = [
@@ -62,6 +64,7 @@ def pdf_write_layer(
 	pdf: Pdf,
 	page: Page,
 	layer: CompressedLayer,
+	is_foreground: bool,
 	content_instructions,
 ) -> None:
 	
@@ -78,6 +81,17 @@ def pdf_write_layer(
 			BitsPerComponent=1,
 			Mask=[1, 1]  # Layer mask
 		)
+	elif isinstance(layer, JP2Layer):
+		pdf_write_image(
+			input_pages=input_pages,
+			pdf=pdf,
+			page=page,
+			value=layer.data,
+			content_instructions=content_instructions,
+			ColorSpace=Name.DeviceRGB,
+			Filter=Name.JPXDecode,
+			BitsPerComponent=8
+		)
 	elif isinstance(layer, JPEGLayer):
 		pdf_write_image(
 			input_pages=input_pages,
@@ -89,6 +103,42 @@ def pdf_write_layer(
 			Filter=Name.DCTDecode,
 			BitsPerComponent=8
 		)
+	elif isinstance(layer, PNGLayer):
+		if is_foreground:
+			# See PDF 1.7 section 7.4.4.3
+			# See also the implementation in img2pdf
+			pdf_write_image(
+				input_pages=input_pages,
+				pdf=pdf,
+				page=page,
+				value=layer.get_flate_data(),
+				content_instructions=content_instructions,
+				ColorSpace=Name.DeviceGray,
+				Filter=Name.FlateDecode,
+				BitsPerComponent=1,
+				Mask=[1, 1],  # Layer mask
+				DecodeParms=Dictionary(
+					Predictor=15,  # PNG prediction (on encoding, PNG optimum) - this is the only allowed value in a PNG file
+					BitsPerComponent=1,  # Default is 8 so must set this here
+					Columns=input_pages.width
+				)
+			)
+		else:
+			pdf_write_image(
+				input_pages=input_pages,
+				pdf=pdf,
+				page=page,
+				value=layer.get_flate_data(),
+				content_instructions=content_instructions,
+				ColorSpace=Name.DeviceRGB,
+				Filter=Name.FlateDecode,
+				BitsPerComponent=8,
+				DecodeParms=Dictionary(
+					Predictor=15,
+					Colors=3,  # Default is 1 so must set this here
+					Columns=input_pages.width
+				)
+			)
 	else:
 		raise NotImplementedError()