Fix custom backend functionality

camelot/backends/image_conversion.py

@@ -1,5 +1,6 @@
 """Classes and functions for the ImageConversionBackend backends."""
 
+from typing import Any
 from typing import Dict
 from typing import List
 from typing import Type
@@ -22,7 +23,7 @@ class ImageConversionError(ValueError): # noqa D101
 class ImageConversionBackend:
  """Classes the ImageConversionBackend backend."""
 
- def __init__(self, backend: str = "poppler", use_fallback: bool = True) -> None:
+ def __init__(self, backend: Any = "poppler", use_fallback: bool = True) -> None:
  """Initialize the conversion backend .
 
  Parameters
@@ -37,15 +38,70 @@ def __init__(self, backend: str = "poppler", use_fallback: bool = True) -> None:
  ValueError
  Raise an error if the backend is not supported.
  """
- if backend not in BACKENDS.keys():
- raise ValueError(f"Image conversion backend {backend!r} not supported")
-
- self.backend: str = backend
+ self.backend: ConversionBackend = self.get_backend(backend)
  self.use_fallback: bool = use_fallback
  self.fallbacks: List[str] = list(
- filter(lambda x: x != backend, BACKENDS.keys())
+ filter(lambda x: isinstance(backend, str) and x != backend, BACKENDS.keys())
  )
 
+ def get_backend(self, backend):
+ """Retrieve the specified backend for processing.
+
+ This method checks if the provided backend is a string representing
+ a known backend or an object implementing a 'convert' method. It
+ returns an instance of the backend if valid.
+
+ Parameters
+ ----------
+ backend : str or object
+ The backend to retrieve. This can be:
+ - A string ('poppler' or 'ghostscript') corresponding to a pre-defined backend.
+ - An object that must implement a 'convert' method.
+
+ Returns
+ -------
+ object
+ An instance of the specified backend.
+
+ Raises
+ ------
+ NotImplementedError
+ If the backend is a string that is not recognized or if it is an
+ object that does not implement the 'convert' method.
+
+ Examples
+ --------
+ >> backend_instance = get_backend('poppler')
+ >> backend_instance = get_backend(my_custom_backend)
+
+ Notes
+ -----
+ The valid backends are defined in the BACKENDS dictionary. The
+ method verifies the backend type and raises exceptions for
+ unsupported backends.
+ """
+
+ def implements_convert():
+ methods = [
+ method for method in dir(backend) if method.startswith("__") is False
+ ]
+ return "convert" in methods
+
+ if isinstance(backend, str):
+ if backend not in BACKENDS.keys():
+ raise NotImplementedError(
+ f"Unknown backend {backend!r} specified. Please use either 'poppler' or 'ghostscript'."
+ )
+
+ return BACKENDS[backend]()
+ else:
+ if not implements_convert():
+ raise NotImplementedError(
+ f"{backend!r} must implement a 'convert' method"
+ )
+
+ return backend
+
  def convert(self, pdf_path: str, png_path: str) -> None:
  """Convert PDF to png_path.
 
@@ -64,8 +120,7 @@ def convert(self, pdf_path: str, png_path: str) -> None:
  [description]
  """
  try:
- converter = BACKENDS[self.backend]()
- converter.convert(pdf_path, png_path)
+ self.backend.convert(pdf_path, png_path)
  except Exception as f:
  if self.use_fallback:
  for fallback in self.fallbacks:

camelot/parsers/lattice.py

@@ -6,7 +6,6 @@
 from typing import Any
 
 from ..backends import ImageConversionBackend
-from ..backends.image_conversion import BACKENDS
 from ..image_processing import adaptive_threshold
 from ..image_processing import find_contours
 from ..image_processing import find_joints
@@ -119,34 +118,10 @@ def __init__(
  self.iterations = iterations
  self.resolution = resolution
  self.use_fallback = use_fallback
- self.backend = Lattice._get_backend(backend)
  self.icb = ImageConversionBackend(use_fallback=use_fallback, backend=backend)
  self.image_path = None
  self.pdf_image = None
 
- @staticmethod
- def _get_backend(backend):
- def implements_convert():
- methods = [
- method for method in dir(backend) if method.startswith("__") is False
- ]
- return "convert" in methods
-
- if isinstance(backend, str):
- if backend not in BACKENDS.keys():
- raise NotImplementedError(
- f"Unknown backend {backend!r} specified. Please use either 'poppler' or 'ghostscript'."
- )
-
- return BACKENDS[backend]()
- else:
- if not implements_convert():
- raise NotImplementedError(
- f"{backend!r} must implement a 'convert' method"
- )
-
- return backend
-
  @staticmethod
  def _shift_index(
  table: Any, r_idx: int, c_idx: int, direction: str

tests/test_common.py

@@ -5,6 +5,7 @@
 from pandas.testing import assert_frame_equal
 
 import camelot
+from camelot.backends.ghostscript_backend import GhostscriptBackend
 from camelot.core import Table
 from camelot.core import TableList
 from camelot.io import PDFHandler
@@ -52,6 +53,15 @@ def test_repr_ghostscript(testdir):
  assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
 
 
+@skip_on_windows
+def test_repr_ghostscript_custom_backend(testdir):
+ filename = os.path.join(testdir, "foo.pdf")
+ tables = camelot.read_pdf(filename, backend=GhostscriptBackend())
+ assert repr(tables) == "<TableList n=1>"
+ assert repr(tables[0]) == "<Table shape=(7, 7)>"
+ assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
+
+
 @skip_pdftopng
 def test_url_poppler():
  url = "https://pypdf-table-extraction.readthedocs.io/en/latest/_static/pdf/foo.pdf"
@@ -70,6 +80,15 @@ def test_url_ghostscript(testdir):
  assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
 
 
+@skip_on_windows
+def test_url_ghostscript_custom_backend(testdir):
+ url = "https://pypdf-table-extraction.readthedocs.io/en/latest/_static/pdf/foo.pdf"
+ tables = camelot.read_pdf(url, backend=GhostscriptBackend())
+ assert repr(tables) == "<TableList n=1>"
+ assert repr(tables[0]) == "<Table shape=(7, 7)>"
+ assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
+
+
 @skip_pdftopng
 def test_pages_poppler():
  url = "https://pypdf-table-extraction.readthedocs.io/en/latest/_static/pdf/foo.pdf"
@@ -112,6 +131,30 @@ def test_pages_ghostscript():
  assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
 
 
+@skip_on_windows
+def test_pages_ghostscript_custom_backend():
+ url = "https://pypdf-table-extraction.readthedocs.io/en/latest/_static/pdf/foo.pdf"
+ custom_backend = GhostscriptBackend()
+ tables = camelot.read_pdf(url, backend=custom_backend, use_fallback=False)
+ assert repr(tables) == "<TableList n=1>"
+ assert repr(tables[0]) == "<Table shape=(7, 7)>"
+ assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
+
+ tables = camelot.read_pdf(
+ url, pages="1-end", backend=custom_backend, use_fallback=False
+ )
+ assert repr(tables) == "<TableList n=1>"
+ assert repr(tables[0]) == "<Table shape=(7, 7)>"
+ assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
+
+ tables = camelot.read_pdf(
+ url, pages="all", backend=custom_backend, use_fallback=False
+ )
+ assert repr(tables) == "<TableList n=1>"
+ assert repr(tables[0]) == "<Table shape=(7, 7)>"
+ assert repr(tables[0].cells[0][0]) == "<Cell x1=120 y1=218 x2=165 y2=234>"
+
+
 def test_table_order():
  def _make_table(page, order):
  t = Table([], [])

tests/test_image_conversion_backend.py

@@ -33,7 +33,7 @@ def convert(self, pdf_path, png_path):
 def test_poppler_backend_error_when_no_use_fallback(patch_backends):
  backend = ImageConversionBackend(backend="poppler", use_fallback=False)
 
- message = "Image conversion failed with image conversion backend 'poppler'"
+ message = r"Image conversion failed with image conversion backend.+Poppler"
  with pytest.raises(ValueError, match=message):
  backend.convert("foo", "bar")
 
@@ -57,3 +57,8 @@ def test_ghostscript_backend_error_when_use_fallback(monkeypatch):
  message = "Image conversion failed with image conversion backend 'poppler'\n error: Image conversion failed"
  with pytest.raises(ValueError, match=message):
  backend.convert("foo", "bar")
+
+
+@pytest.mark.xfail
+def test_invalid_backend():
+ ImageConversionBackend(backend="invalid_backend")