refactoring in preparationfor alpha channel info

image_prediction/image_extractor/extractors/parsable.py

@@ -1,11 +1,11 @@
 import io
 from functools import partial, lru_cache
-from itertools import chain, starmap, filterfalse
+from itertools import chain, starmap, filterfalse, repeat
 from operator import itemgetter, truth
 
 import fitz
 from PIL import Image
-from funcy import rcompose, compose, curry, merge
+from funcy import rcompose, compose, curry, merge, zipdict
 from tqdm import tqdm
 
 from image_prediction.image_extractor.extractor import ImageExtractor, ImageMetadataPair
@@ -44,6 +44,7 @@ class ParsablePDFImageExtractor(ImageExtractor):
         images = get_images_on_page(self.doc, page)
         metadata = get_metadata_for_images_on_page(page)
         get_image_infos.cache_clear()
+        load_image_handle_from_xref.cache_clear()
 
         image_metadata_pairs = starmap(
             ImageMetadataPair, filter(compose(all, curry(map)(truth)), zip(images, metadata))
@@ -63,7 +64,7 @@ def extract_pages(doc, page_range):
 def get_images_on_page(doc, page: fitz.Page):
     image_infos = get_image_infos(page)
     xrefs = map(itemgetter("xref"), image_infos)
-    images = map(partial(load_image_from_xref, doc), xrefs)
+    images = map(partial(xref_to_image, doc), xrefs)
 
     return images
 
@@ -76,6 +77,11 @@ def get_metadata_for_images_on_page(page: fitz.Page):
     metadata = validate_size_and_passthrough(metadata)
     metadata = map(partial(merge, get_page_metadata(page)), metadata)
 
+    # xrefs = map(itemgetter("xref"), image_infos)
+    # alpha = map(has_alpha_channel, xrefs)
+    # alpha = zipdict(repeat(Info.ALPHA), alpha)
+    # metadata = starmap(merge, zip(alpha, metadata))
+
     yield from metadata
 
 
@@ -87,8 +93,39 @@ def validate_size_and_passthrough(metadata):
     yield from map(validate_box_size, metadata)
 
 
-def load_image_from_xref(doc, xref):
+# def load_image_from_xref(doc, xref):
-    maybe_image = doc.extract_image(xref)
+#
+#     maybe_image = doc.extract_image(xref)
+#     if maybe_image:
+#         smask = doc.extract_image(maybe_image["smask"])
+#         pix1 = fitz.Pixmap(maybe_image)  # (1) pixmap of image w/o alpha
+#         mask = fitz.Pixmap(doc.extract_image(smask)["image"])  # (2) mask pixmap
+#         pix = fitz.Pixmap(pix1, mask)  # (3) copy of pix1, image mask added
+#         im = Image.open(io.BytesIO(pix.tobytes()))
+#     else:
+#         im = None
+#
+#     import IPython
+#     IPython.embed()
+#     return im
+
+
+@lru_cache(maxsize=None)
+def load_image_handle_from_xref(doc, xref):
+    return doc.extract_image(xref)
+
+
+def has_alpha_channel(doc, xref):
+    maybe_image = load_image_handle_from_xref(doc, xref)
+    if maybe_image:
+        maybe_smask = doc.extract_image(maybe_image["smask"])
+        return maybe_smask is not None
+    else:
+        return False
+
+
+def xref_to_image(doc, xref):
+    maybe_image = load_image_handle_from_xref(doc, xref)
     return Image.open(io.BytesIO(maybe_image["image"])) if maybe_image else None
 
 
@@ -98,6 +135,10 @@ def get_image_infos(page: fitz.Page):
 
 
 def get_image_metadata(image_info):
+
+    # import IPython
+    # IPython.embed()
+    # smask = doc.extract_image(maybe_image["smask"])
     x1, y1, x2, y2 = map(rounder, image_info["bbox"])
 
     width = abs(x2 - x1)

image_prediction/info.py

@@ -11,3 +11,4 @@ class Info(Enum):
     X2 = "x2"
     Y1 = "y1"
     Y2 = "y2"
+    # ALPHA = "alpha"

scripts/run_pipeline.py

@@ -35,7 +35,7 @@ def process_pdf(pipeline, pdf_path, page_range=None):
 
 
 def main(args):
-    pipeline = load_pipeline(verbose=True, tolerance=3)
+    pipeline = load_pipeline(verbose=False, tolerance=3)
 
     if os.path.isfile(args.input):
         pdf_paths = [args.input]

test/conftest.py

@@ -194,12 +194,16 @@ def input_size(request):
 def array_to_image(array):
     assert np.all(array <= 1)
     assert np.all(array >= 0)
-    return Image.fromarray(np.uint8(array * 255), mode="RGB")
 
+    if array.shape[-1] == 3:
+        mode = "RGB"
+    elif array.shape[-1] == 4:
+        mode = "RGBA"
+    else:
+        raise ValueError(f"Unexpected number of channels {array.shape[-1]}. Expected 3 or 4.")
 
-@pytest.fixture(params=[{"width": 10, "height": 15, "depth": 3}, {"width": 150, "height": 100, "depth": 3}])
+    # noinspection PyTypeChecker
-def input_size(request):
+    return Image.fromarray(np.uint8(array * 255), mode=mode)
-    return itemgetter("width", "height", "depth")(request.param)
 
 
 @pytest.fixture

test/unit_tests/image_extractor_test.py

@@ -17,7 +17,11 @@ def test_image_extractor_mock(image_extractor, images):
 
 
 @pytest.mark.parametrize("extractor_type", ["parsable_pdf", "default"])
-@pytest.mark.parametrize("input_size", [{"depth": 3, "width": 170, "height": 220}], indirect=["input_size"])
+@pytest.mark.parametrize(
+    "input_size",
+    [{"depth": 3, "width": 170, "height": 220}, {"depth": 3, "width": 170, "height": 220}],
+    indirect=["input_size"],
+)
 def test_parsable_pdf_image_extractor(image_extractor, pdf, images, metadata, input_size):
     images_extracted, metadata_extracted = map(list, extract_images_from_pdf(pdf, image_extractor))
     assert np.allclose(images_to_batch_tensor(images_extracted), images_to_batch_tensor(images))