Add pypdfium2 rendering backend

camelot/backends/image_conversion.py

@@ -1,18 +1,24 @@
+from .pdfium_backend import PdfiumBackend
 from .ghostscript_backend import GhostscriptBackend
 from .poppler_backend import PopplerBackend
 
 
-BACKENDS = {"poppler": PopplerBackend, "ghostscript": GhostscriptBackend}
+BACKENDS = {
+    "pdfium": PdfiumBackend,
+    "poppler": PopplerBackend,
+    "ghostscript": GhostscriptBackend,
+}
 
 
 class ImageConversionBackend:
-    def __init__(self, backend="poppler", use_fallback=True):
+    def __init__(self, backend="pdfium", use_fallback=True):
         if backend not in BACKENDS.keys():
             raise ValueError(f"Image conversion backend '{backend}' not supported")
 
         self.backend = backend
         self.use_fallback = use_fallback
-        self.fallbacks = list(filter(lambda x: x != backend, BACKENDS.keys()))
+        self.fallbacks = list(BACKENDS.keys())
+        self.fallbacks.remove(self.backend)
 
     def convert(self, pdf_path, png_path):
         try:

camelot/backends/pdfium_backend.py

@@ -0,0 +1,14 @@
+try:
+    import pypdfium2 as pdfium
+except Exception:
+    pdfium = None
+
+class PdfiumBackend(object):
+    def convert(self, pdf_path, png_path, resolution=300):
+        if not pdfium:
+            raise OSError("pypdfium2 is not installed.")
+        doc = pdfium.PdfDocument(pdf_path)
+        assert len(doc) == 1
+        doc.init_forms()
+        image = doc[0].render(scale=resolution/72).to_pil()
+        image.save(png_path)

noxfile.py

@@ -155,7 +155,7 @@ def mypy(session: Session) -> None:
         session.run("mypy", f"--python-executable={sys.executable}", "noxfile.py")
 
 
-base_requires = ["ghostscript>=0.7", "opencv-python>=3.4.2.17"]
+base_requires = ["ghostscript>=0.7", "opencv-python>=3.4.2.17", "pypdfium2>=4,<5"]
 
 plot_requires = [
     "matplotlib>=2.2.3",

pyproject.toml

@@ -57,7 +57,8 @@ myst-parser = {version = ">=0.16.1"}
 camelot = "camelot.__main__:main"
 
 [tool.poetry.group.base.dependencies]
-ghostscript = "^0.7"
+pypdfium2 = "^4"
+ghostscript = "^0.7"  # remove in favor of pypdfium2?
 opencv-python = "^4.7.0.68"
 
 

tests/test_common.py

@@ -50,6 +50,14 @@ def test_password(testdir):
     assert_frame_equal(df, tables[0].df)
 
 
+def test_repr_pdfium(testdir):
+    filename = os.path.join(testdir, "foo.pdf")
+    tables = camelot.read_pdf(filename, backend="pdfium")
+    assert repr(tables) == "<TableList n=1>"
+    assert repr(tables[0]) == "<Table shape=(7, 7)>"
+    assert repr(tables[0].cells[0][0]) == "<Cell x1=121 y1=218 x2=165 y2=234>"
+
+
 @skip_pdftopng
 def test_repr_poppler(testdir):
     filename = os.path.join(testdir, "foo.pdf")
@@ -68,6 +76,14 @@ def test_repr_ghostscript(testdir):
     assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
 
 
+def test_url_pdfium():
+    url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"
+    tables = camelot.read_pdf(url, backend="pdfium")
+    assert repr(tables) == "<TableList n=1>"
+    assert repr(tables[0]) == "<Table shape=(7, 7)>"
+    assert repr(tables[0].cells[0][0]) == "<Cell x1=121 y1=218 x2=165 y2=234>"
+
+
 @skip_pdftopng
 def test_url_poppler():
     url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"
@@ -86,6 +102,24 @@ def test_url_ghostscript(testdir):
     assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
 
 
+def test_pages_pdfium():
+    url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"
+    tables = camelot.read_pdf(url, backend="pdfium")
+    assert repr(tables) == "<TableList n=1>"
+    assert repr(tables[0]) == "<Table shape=(7, 7)>"
+    assert repr(tables[0].cells[0][0]) == "<Cell x1=121 y1=218 x2=165 y2=234>"
+
+    tables = camelot.read_pdf(url, pages="1-end", backend="pdfium")
+    assert repr(tables) == "<TableList n=1>"
+    assert repr(tables[0]) == "<Table shape=(7, 7)>"
+    assert repr(tables[0].cells[0][0]) == "<Cell x1=121 y1=218 x2=165 y2=234>"
+
+    tables = camelot.read_pdf(url, pages="all", backend="pdfium")
+    assert repr(tables) == "<TableList n=1>"
+    assert repr(tables[0]) == "<Table shape=(7, 7)>"
+    assert repr(tables[0].cells[0][0]) == "<Cell x1=121 y1=218 x2=165 y2=234>"
+
+
 @skip_pdftopng
 def test_pages_poppler():
     url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"

tests/test_image_conversion_backend.py

@@ -2,6 +2,8 @@
 
 from camelot.backends import ImageConversionBackend
 
+# TODO consider adding pdfium backend
+
 
 @pytest.fixture
 def patch_backends(monkeypatch):
@@ -31,15 +33,15 @@ def convert(self, pdf_path, png_path):
 
 
 def test_poppler_backend_error_when_no_use_fallback(patch_backends):
-    backend = ImageConversionBackend(use_fallback=False)
+    backend = ImageConversionBackend(backend="poppler", use_fallback=False)
 
     message = "Image conversion failed with image conversion backend 'poppler'"
     with pytest.raises(ValueError, match=message):
         backend.convert("foo", "bar")
 
 
 def test_ghostscript_backend_when_use_fallback(patch_backends):
-    backend = ImageConversionBackend()
+    backend = ImageConversionBackend(backend="ghostscript")
     backend.convert("foo", "bar")
 
 
@@ -49,7 +51,7 @@ def test_ghostscript_backend_error_when_use_fallback(monkeypatch):
     monkeypatch.setattr(
         "camelot.backends.image_conversion.BACKENDS", backends, raising=True
     )
-    backend = ImageConversionBackend()
+    backend = ImageConversionBackend(backend="poppler")
 
     message = "Image conversion failed with image conversion backend 'ghostscript'"
     with pytest.raises(ValueError, match=message):