ADD: output monochrome image support

docs/reference.md

@@ -26,6 +26,7 @@ convert_from_path(
     grayscale=False,
     size=None,
     paths_only=False,
+    mono=False,
 )
 
 convert_from_bytes(
@@ -47,6 +48,7 @@ convert_from_bytes(
     grayscale=False,
     size=None,
     paths_only=False,
+    mono=False,
 )
 ```
 

pdf2image/parsers.py

@@ -7,6 +7,22 @@
 from PIL import Image
 
 
+def parse_buffer_to_pbm(data):
+    """Parse PBM file bytes to Pillow Image"""
+
+    images = []
+
+    index = 0
+
+    while index < len(data):
+        code, size = tuple(data[index : index + 40].split(b"\n")[0:2]) # no rgb section in pbm
+        size_x, size_y = tuple(size.split(b" "))
+        file_size = len(code) + len(size) + 2 + int(size_x) * int(size_y) * 3
+        images.append(Image.open(BytesIO(data[index : index + file_size])))
+        index += file_size
+
+    return images
+
 def parse_buffer_to_ppm(data):
     """Parse PPM file bytes to Pillow Image"""
 

pdf2image/pdf2image.py

@@ -16,6 +16,7 @@
 from .generators import uuid_generator, counter_generator, ThreadSafeGenerator
 
 from .parsers import (
+    parse_buffer_to_pbm,
     parse_buffer_to_pgm,
     parse_buffer_to_ppm,
     parse_buffer_to_jpeg,
@@ -55,6 +56,7 @@ def convert_from_path(
     paths_only=False,
     use_pdftocairo=False,
     timeout=None,
+    mono=False,
 ):
     """
         Description: Convert PDF to Image will throw whenever one of the condition is reached
@@ -98,7 +100,7 @@ def convert_from_path(
 
     # We start by getting the output format, the buffer processing function and if we need pdftocairo
     parsed_fmt, final_extension, parse_buffer_func, use_pdfcairo_format = _parse_format(
-        fmt, grayscale
+        fmt, grayscale, mono
     )
 
     # We use pdftocairo is the format requires it OR we need a transparent output
@@ -170,6 +172,7 @@ def convert_from_path(
             single_file,
             grayscale,
             size,
+            mono
         )
 
         if use_pdfcairo:
@@ -236,6 +239,7 @@ def convert_from_bytes(
     paths_only=False,
     use_pdftocairo=False,
     timeout=None,
+    mono=False,
 ):
     """
         Description: Convert PDF to Image will throw whenever one of the condition is reached
@@ -288,6 +292,7 @@ def convert_from_bytes(
                 paths_only=paths_only,
                 use_pdftocairo=use_pdftocairo,
                 timeout=timeout,
+                mono=mono
             )
     finally:
         os.close(fh)
@@ -308,6 +313,7 @@ def _build_command(
     single_file,
     grayscale,
     size,
+    mono
 ):
     if use_cropbox:
         args.append("-cropbox")
@@ -321,7 +327,7 @@ def _build_command(
     if last_page is not None:
         args.extend(["-l", str(last_page)])
 
-    if fmt not in ["pgm", "ppm"]:
+    if fmt not in ["pgm", "ppm", "pbm"]:
         args.append("-" + fmt)
 
     if fmt in ["jpeg", "jpg"] and jpegopt:
@@ -339,6 +345,9 @@ def _build_command(
     if grayscale:
         args.append("-gray")
 
+    if mono:
+        args.append("-mono")
+
     if size is None:
         pass
     elif isinstance(size, tuple) and len(size) == 2:
@@ -360,7 +369,7 @@ def _build_command(
     return args
 
 
-def _parse_format(fmt, grayscale=False):
+def _parse_format(fmt, grayscale=False, mono=False):
     fmt = fmt.lower()
     if fmt[0] == ".":
         fmt = fmt[1:]
@@ -372,6 +381,8 @@ def _parse_format(fmt, grayscale=False):
         return "tiff", "tif", None, True
     if fmt == "ppm" and grayscale:
         return "pgm", "pgm", parse_buffer_to_pgm, False
+    if fmt == "ppm" and mono:
+        return "pbm", "pbm", parse_buffer_to_pbm, False
     # Unable to parse the format so we'll use the default
     return "ppm", "ppm", parse_buffer_to_ppm, False
 

tests.py

@@ -1228,6 +1228,52 @@ def test_conversion_to_grayscale_from_path_using_dir(self):
             )
         )
 
+    ## Test mono option
+
+    @profile
+    @unittest.skipIf(not POPPLER_INSTALLED, "Poppler is not installed!")
+    def test_conversion_to_mono_from_bytes(self):
+        start_time = time.time()
+        with open("./tests/test_14.pdf", "rb") as pdf_file:
+            images_from_bytes = convert_from_bytes(pdf_file.read(), mono=True)
+            self.assertTrue(images_from_bytes[0].mode == "1")
+        print(
+            "test_conversion_to_mono_from_bytes_14: {} sec".format(
+                (time.time() - start_time) / 14.0
+            )
+        )
+
+    @profile
+    @unittest.skipIf(not POPPLER_INSTALLED, "Poppler is not installed!")
+    def test_conversion_to_mono_from_path(self):
+        start_time = time.time()
+        images_from_path = convert_from_path("./tests/test_14.pdf", mono=True)
+        print("images_from_path 2 ", images_from_path)
+        self.assertTrue(images_from_path[0].mode == "1")
+        [im.close() for im in images_from_path]
+        print(
+            "test_conversion_to_mono_from_path_14: {} sec".format(
+                (time.time() - start_time) / 14.0
+            )
+        )
+
+    @profile
+    @unittest.skipIf(not POPPLER_INSTALLED, "Poppler is not installed!")
+    def test_conversion_to_mono_from_path_using_dir(self):
+        start_time = time.time()
+        with TemporaryDirectory() as path:
+            images_from_path = convert_from_path(
+                "./tests/test_14.pdf", output_folder=path, mono=True
+            )
+            print("images_from_path 1", images_from_path)
+            self.assertTrue(images_from_path[0].mode == "1")
+            [im.close() for im in images_from_path]
+        print(
+            "test_conversion_to_mono_from_path_using_dir_14: {} sec".format(
+                (time.time() - start_time) / 14.0
+            )
+        )
+
     ## Test pathlib support
 
     @profile
@@ -1642,7 +1688,7 @@ def test_pdfinfo_functions_same_number_of_parameters(self):
             len(signature(pdfinfo_from_bytes).parameters),
         )
         print("test_pdfinfo_functions_same_number_of_parameters: {} sec".format(time.time() - start_time))
-    
+
     @unittest.skipIf(not POPPLER_INSTALLED, "Poppler is not installed!")
     def test_timeout_pdfinfo_from_path_241(self):
         start_time = time.time()