def warning_on_break_types_that_arent_recognised(self):
break_element = xml_element("w:br", {"w:type": "unknownBreakType"}, [])
result = _read_document_xml_element(break_element)
expected_warning = results.warning(
"Unsupported break type: unknownBreakType")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)
def when_imagedata_element_has_no_relationship_id_then_it_is_ignored_with_warning(self):
imagedata_element = xml_element("v:imagedata")
result = _read_document_xml_element(imagedata_element)
expected_warning = results.warning("A v:imagedata element without a relationship ID was ignored")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)
def warning_if_non_cell_in_table_row(self):
element = xml_element("w:tbl", {}, [w_tr(xml_element("w:p"))])
result = _read_document_xml_element(element)
expected_warning = results.warning(
"unexpected non-cell element in table row, cell merging may be incorrect"
)
assert_equal([expected_warning], result.messages)
def warning_if_unsupported_image_type(self, mocks):
drawing_element = _create_inline_image(
blip=_embedded_blip("rId5"),
description="It's a hat",
)
relationships = Relationships([
_image_relationship("rId5", "media/hat.emf"),
])
docx_file = mocks.mock()
funk.allows(docx_file).open("word/media/hat.emf").returns(io.BytesIO(self.IMAGE_BYTES))
content_types = mocks.mock()
funk.allows(content_types).find_content_type("word/media/hat.emf").returns("image/x-emf")
result = _read_document_xml_element(
drawing_element,
content_types=content_types,
relationships=relationships,
docx_file=docx_file,
)
assert_equal("image/x-emf", result.value.content_type)
expected_warning = results.warning("Image of type image/x-emf is unlikely to display in web browsers")
assert_equal([expected_warning], result.messages)
def when_imagedata_element_has_no_relationship_id_then_it_is_ignored_with_warning(self):
imagedata_element = xml_element("v:imagedata")
result = _read_document_xml_element(imagedata_element)
expected_warning = results.warning("A v:imagedata element without a relationship ID was ignored")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)
def warning_if_unsupported_image_type(self, context):
drawing_element = _create_inline_image(
blip=_embedded_blip("rId5"),
description="It's a hat",
)
image_bytes = b"Not an image at all!"
relationships = Relationships({
"rId5": Relationship(target="media/hat.emf")
})
docx_file = context.mock()
funk.allows(docx_file).open("word/media/hat.emf").returns(io.BytesIO(image_bytes))
content_types = context.mock()
funk.allows(content_types).find_content_type("word/media/hat.emf").returns("image/x-emf")
result = _read_document_xml_element(
drawing_element,
content_types=content_types,
relationships=relationships,
docx_file=docx_file,
)
assert_equal("image/x-emf", result.value[0].content_type)
expected_warning = results.warning("Image of type image/x-emf is unlikely to display in web browsers")
assert_equal([expected_warning], result.messages)
def warning_is_emitted_if_paragraph_style_is_unrecognised():
result = convert_document_element_to_html(
documents.paragraph(style_name="TipsParagraph", children=[
_run_with_text("Tip")
]),
)
assert_equal([results.warning("Unrecognised paragraph style: TipsParagraph")], result.messages)
def warning_if_unsupported_image_type(self, context):
drawing_element = _create_inline_image(
relationship_id="rId5",
description="It's a hat",
)
image_bytes = b"Not an image at all!"
relationships = Relationships(
{"rId5": Relationship(target="media/hat.emf")})
docx_file = context.mock()
funk.allows(docx_file).open("word/media/hat.emf").returns(
io.BytesIO(image_bytes))
content_types = context.mock()
funk.allows(content_types).find_content_type(
"word/media/hat.emf").returns("image/x-emf")
result = read_document_xml_element(
drawing_element,
content_types=content_types,
relationships=relationships,
docx_file=docx_file,
)
assert_equal("image/x-emf", result.value[0].content_type)
expected_warning = results.warning(
"Image of type image/x-emf is unlikely to display in web browsers")
assert_equal([expected_warning], result.messages)
def warn_if_images_stored_outside_of_document_are_specified_when_passing_fileobj_without_name():
fileobj = io.BytesIO()
with open(test_path("external-picture.docx"), "rb") as source_fileobj:
shutil.copyfileobj(source_fileobj, fileobj)
result = mammoth.convert_to_html(fileobj=fileobj)
assert_equal("", result.value)
assert_equal([results.warning("could not find external image 'tiny-picture.png', fileobj has no name")], result.messages)
def warning_if_style_mapping_is_not_understood():
style_map = """
!!!!
p => h1"""
with open(test_path("single-paragraph.docx"), "rb") as fileobj:
result = mammoth.convert_to_html(fileobj=fileobj, style_map=style_map)
assert_equal("<h1>Walking on imported air</h1>", result.value)
warning = "Did not understand this style mapping, so ignored it: !!!!"
assert_equal([results.warning(warning)], result.messages)
def warning_if_style_mapping_is_not_understood():
style_map = """
!!!!
p => h1"""
with open(test_path("single-paragraph.docx"), "rb") as fileobj:
result = mammoth.convert_to_html(fileobj=fileobj, style_map=style_map)
assert_equal("<h1>Walking on imported air</h1>", result.value)
warning = "Did not understand this style mapping, so ignored it: !!!!"
assert_equal([results.warning(warning)], result.messages)
def warning_is_emitted_if_paragraph_style_is_unrecognised():
result = convert_document_element_to_html(
documents.paragraph(style_id="Heading1",
style_name="Heading 1",
children=[_run_with_text("Tip")]), )
assert_equal([
results.warning(
"Unrecognised paragraph style: Heading 1 (Style ID: Heading1)")
], result.messages)
def warning_is_emitted_when_table_style_cannot_be_found(self):
style_xml = xml_element("w:tblStyle", {"w:val": "TableNormal"})
properties_xml = xml_element("w:tblPr", {}, [style_xml])
table_xml = xml_element("w:tbl", {}, [properties_xml])
result = _read_document_xml_element(table_xml, styles=Styles.EMPTY)
table = result.value
assert_equal("TableNormal", table.style_id)
assert_equal(None, table.style_name)
assert_equal([results.warning("Table style with ID TableNormal was referenced but not defined in the document")], result.messages)
def warning_is_emitted_when_paragraph_style_cannot_be_found(self):
style_xml = xml_element("w:pStyle", {"w:val": "Heading1"})
properties_xml = xml_element("w:pPr", {}, [style_xml])
paragraph_xml = xml_element("w:p", {}, [properties_xml])
result = _read_document_xml_element(paragraph_xml, styles=Styles.EMPTY)
paragraph = result.value
assert_equal("Heading1", paragraph.style_id)
assert_equal(None, paragraph.style_name)
assert_equal([results.warning("Paragraph style with ID Heading1 was referenced but not defined in the document")], result.messages)
def warning_is_emitted_when_table_style_cannot_be_found(self):
style_xml = xml_element("w:tblStyle", {"w:val": "TableNormal"})
properties_xml = xml_element("w:tblPr", {}, [style_xml])
table_xml = xml_element("w:tbl", {}, [properties_xml])
result = _read_document_xml_element(table_xml, styles=Styles.EMPTY)
table = result.value
assert_equal("TableNormal", table.style_id)
assert_equal(None, table.style_name)
assert_equal([results.warning("Table style with ID TableNormal was referenced but not defined in the document")], result.messages)
def warning_is_emitted_when_paragraph_style_cannot_be_found(self):
style_xml = xml_element("w:pStyle", {"w:val": "Heading1"})
properties_xml = xml_element("w:pPr", {}, [style_xml])
paragraph_xml = xml_element("w:p", {}, [properties_xml])
result = _read_document_xml_element(paragraph_xml, styles=Styles.EMPTY)
paragraph = result.value
assert_equal("Heading1", paragraph.style_id)
assert_equal(None, paragraph.style_name)
assert_equal([results.warning("Paragraph style with ID Heading1 was referenced but not defined in the document")], result.messages)
def warning_is_emitted_when_run_style_cannot_be_found(self):
style_xml = xml_element("w:rStyle", {"w:val": "Heading1Char"})
properties_xml = xml_element("w:rPr", {}, [style_xml])
run_xml = xml_element("w:r", {}, [properties_xml])
result = _read_document_xml_element(run_xml, styles=Styles.EMPTY)
run = result.value
assert_equal("Heading1Char", run.style_id)
assert_equal(None, run.style_name)
assert_equal([results.warning("Run style with ID Heading1Char was referenced but not defined in the document")], result.messages)
def warning_is_emitted_when_run_style_cannot_be_found(self):
style_xml = xml_element("w:rStyle", {"w:val": "Heading1Char"})
properties_xml = xml_element("w:rPr", {}, [style_xml])
run_xml = xml_element("w:r", {}, [properties_xml])
result = _read_document_xml_element(run_xml, styles=Styles.EMPTY)
run = result.value
assert_equal("Heading1Char", run.style_id)
assert_equal(None, run.style_name)
assert_equal([results.warning("Run style with ID Heading1Char was referenced but not defined in the document")], result.messages)
def warn_if_images_stored_outside_of_document_are_specified_when_passing_fileobj_without_name(
):
fileobj = io.BytesIO()
with open(test_path("external-picture.docx"), "rb") as source_fileobj:
shutil.copyfileobj(source_fileobj, fileobj)
result = mammoth.convert_to_html(fileobj=fileobj)
assert_equal("", result.value)
assert_equal([
results.warning(
"could not find external image 'tiny-picture.png', fileobj has no name"
)
], result.messages)
def warning_if_non_cell_in_table_row(self):
element = xml_element("w:tbl", {}, [w_tr(xml_element("w:p"))])
result = _read_document_xml_element(element)
expected_warning = results.warning("unexpected non-cell element in table row, cell merging may be incorrect")
assert_equal([expected_warning], result.messages)
def unrecognised_elements_emit_warning(self):
element = xml_element("w:huh", {}, [])
result = _read_document_xml_element(element)
expected_warning = results.warning("An unrecognised element was ignored: w:huh")
assert_equal([expected_warning], result.messages)
def warning_on_breaks_that_arent_line_breaks(self):
break_element = xml_element("w:br", {"w:type": "page"}, [])
result = _read_document_xml_element(break_element)
expected_warning = results.warning("Unsupported break type: page")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)
def unrecognised_elements_emit_warning():
element = xml_element("w:huh", {}, [])
result = _read_document_xml_element(element)
expected_warning = results.warning("An unrecognised element was ignored: w:huh")
assert_equal([expected_warning], result.messages)
def warning_on_break_types_that_arent_recognised(self):
break_element = xml_element("w:br", {"w:type": "unknownBreakType"}, [])
result = _read_document_xml_element(break_element)
expected_warning = results.warning("Unsupported break type: unknownBreakType")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)
def warning_on_breaks_that_arent_line_breaks(self):
break_element = xml_element("w:br", {"w:type": "page"}, [])
result = _read_document_xml_element(break_element)
expected_warning = results.warning("Unsupported break type: page")
assert_equal([expected_warning], result.messages)
assert_equal(None, result.value)