def process(self, path, metadata):
ext = os.path.splitext(path)[1]
if ext not in self.supported_extensions:
return path, metadata
basename = os.path.basename(path)
src_path = os.path.join(copy_to_secure_location(path), basename)
src_dir = os.path.dirname(src_path)
remove_file_dir(path)
# Remove <SDFIELD> tags if any
cleaned_html = rename_sdfield_tags(
open(src_path, 'rb').read().decode('utf-8'))
with open(src_path, 'wb') as fd:
fd.write(cleaned_html.encode('utf-8'))
error_file = os.path.join(src_dir, 'tidy-errors')
cmd = 'tidy -asxhtml -clean -indent -modify -utf8 -f %s %s' % (
error_file, src_path)
os.system(cmd)
os.unlink(error_file)
return src_path, metadata
def process(self, path, metadata):
ext = os.path.splitext(path)[1]
if ext not in self.supported_extensions:
return path, metadata
basename = os.path.basename(path)
src_path = os.path.join(
copy_to_secure_location(path), basename)
src_dir = os.path.dirname(src_path)
remove_file_dir(path)
# Remove <SDFIELD> tags if any
cleaned_html = rename_sdfield_tags(
open(src_path, 'rb').read().decode('utf-8'))
with open(src_path, 'wb') as fd:
fd.write(cleaned_html.encode('utf-8'))
error_file = os.path.join(src_dir, 'tidy-errors')
cmd = 'tidy -asxhtml -clean -indent -modify -utf8 -f %s %s' % (
error_file, src_path)
os.system(cmd)
os.unlink(error_file)
return src_path, metadata
def test_rename_sdfield_tags_nested(self):
html_input = '<p>Blah<sdfield>12<span>b</span></sdfield></p>'
result = rename_sdfield_tags(html_input)
expected = '<p>Blah<span class="sdfield">12<span>b</span></span></p>'
assert result == expected
def test_rename_sdfield_tags_empty(self):
html_input = '<p>Blah</p>'
result = rename_sdfield_tags(html_input)
expected = '<p>Blah</p>'
assert result == expected
def test_rename_sdfield_tags_uppercase(self):
html_input = '<P>Blah<SDFIELD TYPE="PAGE">8</SDFIELD></P>'
result = rename_sdfield_tags(html_input)
expected = '<P>Blah<span class="sdfield" TYPE="PAGE">8</span></P>'
assert result == expected
def test_rename_sdfield_tags(self):
html_input = '<p>Blah<sdfield type="PAGE">8</sdfield></p>'
result = rename_sdfield_tags(html_input)
expected = '<p>Blah<span class="sdfield" type="PAGE">8</span></p>'
assert result == expected
