def test_read_form_416():
url = (
"https://www.fda.gov/downloads/AboutFDA/ReportsManualsForms/Forms/UCM074728.pdf"
)
reader = PdfReader(BytesIO(get_pdf_from_url(url, name="issue_416.pdf")))
fields = reader.get_form_text_fields()
assert len(fields) > 0
def test_get_form(src, expected, expected_get_fields):
"""Check if we can read out form data."""
src = os.path.join(RESOURCE_ROOT, src)
reader = PdfReader(src)
fields = reader.get_form_text_fields()
assert fields == expected
with open("tmp-fields-report.txt", "w") as f:
fields = reader.get_fields(fileobj=f)
assert fields == expected_get_fields
if fields:
for field in fields.values():
# Just access the attributes
[
field.field_type,
field.parent,
field.kids,
field.name,
field.alternate_name,
field.mapping_name,
field.flags,
field.value,
field.default_value,
field.additional_actions,
]
# cleanup
os.remove("tmp-fields-report.txt")