Add pypdfium2 rendering backend (experimental patch)

Author: mara004

camelot/backends/image_conversion.py

@@ -2,12 +2,13 @@
 
 from .poppler_backend import PopplerBackend
 from .ghostscript_backend import GhostscriptBackend
+from .pdfium_backend import PdfiumBackend
 
-BACKENDS = {"poppler": PopplerBackend, "ghostscript": GhostscriptBackend}
+BACKENDS = {"poppler": PopplerBackend, "ghostscript": GhostscriptBackend, "pdfium": PdfiumBackend}
 
 
 class ImageConversionBackend(object):
-    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")
 

camelot/backends/pdfium_backend.py

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+
+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
+        image = doc[0].render(scale=resolution/72).to_pil()
+        image.save(png_path)

setup.py

@@ -24,7 +24,13 @@
     "tabulate>=0.8.9",
 ]
 
-base_requires = ["ghostscript>=0.7", "opencv-python>=3.4.2.17", "pdftopng>=0.2.3"]
+base_requires = [
+    "opencv-python>=3.4.2.17",
+    "pypdfium2>=4,<5",
+    "pillow",
+    "ghostscript>=0.7",  # deprecate?
+    "pdftopng>=0.2.3",  # deprecate?
+]
 
 plot_requires = [
     "matplotlib>=2.2.3",

tests/test_common.py

@@ -59,6 +59,14 @@ def test_password():
     assert_frame_equal(df, tables[0].df)
 
 
+def test_repr_pdfium():
+    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>"
+
+
 def test_repr_poppler():
     filename = os.path.join(testdir, "foo.pdf")
     tables = camelot.read_pdf(filename, backend="poppler")
@@ -76,6 +84,14 @@ def test_repr_ghostscript():
     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>"
+
+
 def test_url_poppler():
     url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"
     tables = camelot.read_pdf(url, backend="poppler")
@@ -93,6 +109,24 @@ def test_url_ghostscript():
     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>"
+
+
 def test_pages_poppler():
     url = "https://camelot-py.readthedocs.io/en/master/_static/pdf/foo.pdf"
     tables = camelot.read_pdf(url, backend="poppler")

tests/test_image_conversion_backend.py

@@ -29,7 +29,7 @@ def test_poppler_backend_error_when_no_use_fallback(monkeypatch):
     monkeypatch.setattr(
         "camelot.backends.image_conversion.BACKENDS", BACKENDS, raising=True
     )
-    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):
@@ -44,7 +44,7 @@ def test_ghostscript_backend_when_use_fallback(monkeypatch):
     monkeypatch.setattr(
         "camelot.backends.image_conversion.BACKENDS", BACKENDS, raising=True
     )
-    backend = ImageConversionBackend()
+    backend = ImageConversionBackend(backend="poppler")
     backend.convert("foo", "bar")
 
 
@@ -53,7 +53,7 @@ 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):