def parseftml(fnameorstr, feats=None):
"""parse an FTML document into a list ftmlstrings
Args:
root : ElementTree Element object representing FTML document
Returns:
list of ftmlstring objects, each representing one <test>, in document order
Within <string> elements it removes <em> markup and converts backslash-u notation to Unicode characters.
<testgroup> divisions are ignored and tests from all <testgroups> are collected together.
"""
strs = []
if os.path.exists(fnameorstr):
root = et.parse(fnameorstr)
else:
root = et.fromstring(fnameorstr)
for test in root.findall('.//test'):
s = "".join(test.find('string').itertext())
s = re.sub(r'\\u([a-fA-F0-9]{4,6})',
lambda m: chr(int(m.group(1), 16)), s)
stylename = test.get('stylename', None)
if stylename is None:
lfeats = None
lang = None
else:
style = root.find(f'./head/styles/style[@name="{stylename}"]')
lfeats = style.get('feats', None)
if lfeats is not None:
lfeats = dict(parseFeat(t.strip()) for t in lfeats.split(','))
if feats is not None:
lfeats.update(feats)
elif feats is not None:
lfeats = feats
else:
lfeats = None
lang = style.get('lang', None)
rtl = test.get('rtl', "").lower() in ("true", "1")
for w in s.split():
s = UserString(w)
s.feats = lfeats
s.lang = lang
s.rtl = rtl
strs.append(s)
return strs
