Added support for embedding indexed images (#443)

* added support for palette images

* added test images palette

* removed generate=true

* removed link=None

* fixed previous tests

* removed Trailing whitespace

* testing image filters

* managed "PA" images

* small fix

* generating pdf with pillow 9.1.1

* unsupported image filter error test

* allowing embedding of indexed PNG images

* formatting

* added PIL image

* added test for p/pa images

* porting transparency in the pdf of P images

* fixed typos

* calling get_img_info

* removed trailing whitespace

* fixed image info

* fixed pdfs

Co-authored-by: Lucas Cimon <925560+Lucas-C@users.noreply.github.com>

CHANGELOG.md

@@ -25,6 +25,7 @@ This can also be enabled programmatically with `warnings.simplefilter('default',
 - allowing correctly parsing of SVG files with CSS styling (`style="..."` attribute), thanks to @RedShy
 - [`FPDF.star`](https://pyfpdf.github.io/fpdf2/Shapes.html#regular-star): new method added to draw regular stars, thanks to @digidigital and @RedShy
 - [`FPDF.ink_annotation`](https://pyfpdf.github.io/fpdf2/Annotations.html#ink-annotations): new method added to add path annotations
+- allowing embedding of indexed PNG images without converting them to RGB colorspace, thanks to @RedShy
 - allowing to change appearance of [highlight annotations](https://pyfpdf.github.io/fpdf2/fpdf/fpdf.html#fpdf.fpdf.FPDF.highlight) by specifying a [`TextMarkupType`](https://pyfpdf.github.io/fpdf2/fpdf/enums.html#fpdf.enums.TextMarkupType)
 - documentation on how to control objects transparency: [link to docs](https://pyfpdf.github.io/fpdf2/Transparency.html)
 - documentation on how to create tables and charts using [pandas](https://pandas.pydata.org/) DataFrames: [link to docs](https://pyfpdf.github.io/fpdf2/Maths.html), thanks to @iwayankurniawan

fpdf/fpdf.py

@@ -4000,9 +4000,14 @@ def _putimage(self, info):
         self._out(f"/Height {info['h']}")
 
         if info["cs"] == "Indexed":
+            palette_ref = (
+                pdf_ref(self.n + 2)
+                if self.allow_images_transparency and "smask" in info
+                else pdf_ref(self.n + 1)
+            )
             self._out(
                 f"/ColorSpace [/Indexed /DeviceRGB "
-                f"{len(info['pal']) // 3 - 1} {pdf_ref(self.n + 1)}]"
+                f"{len(info['pal']) // 3 - 1} {palette_ref}]"
             )
         else:
             self._out(f"/ColorSpace /{info['cs']}")
@@ -4044,11 +4049,10 @@ def _putimage(self, info):
         # Palette
         if info["cs"] == "Indexed":
             self._newobj()
-            filter, pal = (
-                ("/Filter /FlateDecode ", zlib.compress(info["pal"]))
-                if self.compress
-                else ("", info["pal"])
-            )
+            if self.compress:
+                filter, pal = ("/Filter /FlateDecode ", zlib.compress(info["pal"]))
+            else:
+                filter, pal = ("", info["pal"])
             self._out(f"<<{filter}/Length {len(pal)}>>")
             self._out(pdf_stream(pal))
             self._out("endobj")

fpdf/image_parsing.py

@@ -56,15 +56,23 @@ def get_img_info(img, image_filter="AUTO", dims=None):
     """
     if Image is None:
         raise EnvironmentError("Pillow not available - fpdf2 cannot insert images")
+
     if not isinstance(img, Image.Image):
         img = Image.open(img)
+
     if dims:
         img = img.resize(dims, resample=RESAMPLE)
+
     if image_filter == "AUTO":
         # Very simple logic for now:
         image_filter = "DCTDecode" if img.format == "JPEG" else "FlateDecode"
-    if img.mode not in ("L", "LA", "RGB", "RGBA"):
+
+    if img.mode in ("P", "PA") and image_filter != "FlateDecode":
+        img = img.convert("RGBA")
+
+    if img.mode not in ("L", "LA", "RGB", "RGBA", "P", "PA"):
         img = img.convert("RGBA")
+
     w, h = img.size
     info = {}
     if img.mode == "L":
@@ -79,6 +87,30 @@ def get_img_info(img, image_filter="AUTO", dims=None):
             "JPXDecode",
         ):
             info["smask"] = _to_data(img, image_filter, select_slice=alpha_channel)
+    elif img.mode == "P":
+        dpn, bpc, colspace = 1, 8, "Indexed"
+        info["data"] = _to_data(img, image_filter)
+        info["pal"] = img.palette.palette
+
+        # check if the P image has transparency
+        if img.info.get("transparency", None) is not None and image_filter not in (
+            "DCTDecode",
+            "JPXDecode",
+        ):
+            # convert to RGBA to get the alpha channel for creating the smask
+            info["smask"] = _to_data(
+                img.convert("RGBA"), image_filter, select_slice=slice(3, None, 4)
+            )
+    elif img.mode == "PA":
+        dpn, bpc, colspace = 1, 8, "Indexed"
+        info["pal"] = img.palette.palette
+        alpha_channel = slice(1, None, 2)
+        info["data"] = _to_data(img, image_filter, remove_slice=alpha_channel)
+        if _has_alpha(img, alpha_channel) and image_filter not in (
+            "DCTDecode",
+            "JPXDecode",
+        ):
+            info["smask"] = _to_data(img, image_filter, select_slice=alpha_channel)
     elif img.mode == "RGB":
         dpn, bpc, colspace = 3, 8, "DeviceRGB"
         info["data"] = _to_data(img, image_filter)
@@ -102,7 +134,6 @@ def get_img_info(img, image_filter="AUTO", dims=None):
             "bpc": bpc,
             "f": image_filter,
             "dp": dp,
-            "pal": "",
             "trns": "",
         }
     )
@@ -113,18 +144,23 @@ def get_img_info(img, image_filter="AUTO", dims=None):
 def _to_data(img, image_filter, **kwargs):
     if image_filter == "FlateDecode":
         return _to_zdata(img, **kwargs)
+
     if img.mode == "LA":
         img = img.convert("L")
+
     if img.mode == "RGBA":
         img = img.convert("RGB")
+
     if image_filter == "DCTDecode":
         compressed_bytes = BytesIO()
         img.save(compressed_bytes, format="JPEG")
         return compressed_bytes.getvalue()
+
     if image_filter == "JPXDecode":
         compressed_bytes = BytesIO()
         img.save(compressed_bytes, format="JPEG2000")
         return compressed_bytes.getvalue()
+
     raise FPDFException(f'Unsupported image filter: "{image_filter}"')
 
 

test/errors/test_FPDF_errors.py

@@ -4,6 +4,8 @@
 import fpdf
 import pytest
 from fpdf.errors import FPDFException, FPDFUnicodeEncodingException
+from fpdf.image_parsing import get_img_info
+from PIL import Image
 
 HERE = Path(__file__).resolve().parent
 
@@ -107,6 +109,13 @@ def test_repeated_calls_to_output(tmp_path):
     assert_pdf_equal(pdf, HERE / "repeated_calls_to_output.pdf", tmp_path)
 
 
+def test_unsupported_image_filter_error():
+    image_filter = "N/A"
+    with pytest.raises(FPDFException) as error:
+        get_img_info(img=Image.open(HERE / "flowers.png"), image_filter=image_filter)
+    assert str(error.value) == f'Unsupported image filter: "{image_filter}"'
+
+
 def test_incorrent_number_of_pages_toc():
     pdf = fpdf.FPDF()
     pdf.add_page()