def read_config(fname):
rdr = None
try:
rdr = LineReader(fname)
for s in rdr.lines():
# Ignore comments
ix = s.find("#")
if ix >= 0:
s = s[0:ix]
s = s.strip()
if not s:
# Blank line: ignore
continue
if s[0] == "[" and s[-1] == "]":
# Handle section name, if found (it is redundant in this case)
section = s[1:-1].strip().lower()
if section != "meanings":
raise ConfigError("Unknown section name '{0}'".format(section))
continue
Meanings.add_entry(s)
except ConfigError as e:
# Add file name and line number information to the exception
# if it's not already there
if rdr is not None:
e.set_pos(rdr.fname(), rdr.line())
raise e
def add_entry(s):
""" Handle additional word meanings in the settings section """
# Format: stofn ordmynd ordfl fl (default ob) beyging (default -)
a = s.split()
if len(a) < 2 or len(a) > 5:
raise ConfigError("Meaning should have two to five arguments, {0} given".format(len(a)))
stofn = None
fl = None
beyging = None
if len(a) == 2:
# Short format: only ordmynd and ordfl
ordmynd = a[0]
ordfl = a[1]
else:
# Full format: at least three arguments, stofn ordmynd ordfl
stofn = a[0]
ordmynd = a[1]
ordfl = a[2]
fl = a[3] if len(a) >= 4 else None
beyging = a[4] if len(a) >= 5 else None
if len(a) == 2 and "-" in ordmynd:
# Creating new meanings by prefixing existing ones
Meanings.add_composite(ordmynd, ordfl)
else:
Meanings.add(stofn, ordmynd, ordfl, fl, beyging)
def add_entry(s):
""" Handle additional word meanings in the settings section """
# Format: stofn ordmynd ordfl fl (default ob) beyging (default -)
a = s.split()
if len(a) < 2 or len(a) > 5:
raise ConfigError(
"Meaning should have two to five arguments, {0} given".format(len(a))
)
stofn = None
fl = None
beyging = None
utg = None
if 3 <= len(a) <= 4 and re.match(r"\d+$", a[-1]):
# An id number (utg) is being given in the last argument
utg = int(a[-1])
# Cut it off the end and proceed
a = a[:-1]
if len(a) <= 3:
# Short format: ordmynd ordfl [fl]
ordmynd = a[0]
ordfl = a[1]
if len(a) == 3:
fl = a[2]
else:
# Full format: at least four arguments, stofn ordmynd ordfl fl
if utg is not None:
raise ConfigError(
"An id number (utg) can't be specified with a full meaning"
)
stofn = a[0]
ordmynd = a[1]
ordfl = a[2]
fl = a[3]
beyging = a[4] if len(a) >= 5 else None
if utg is not None and not "-" in ordmynd:
raise ConfigError(
"An id number (utg) should only be specified for a composite word"
)
if len(a) <= 3 and "-" in ordmynd:
# Creating new meanings by prefixing existing ones
Meanings.add_composite(ordmynd, ordfl, fl, utg=utg)
else:
assert utg is None
Meanings.add(stofn, ordmynd, ordfl, fl, beyging)
def add_composite(stofn, ordfl):
""" Add composite word forms by putting a prefix on existing BIN word forms.
Called from the config file handler. """
assert stofn is not None
assert ordfl is not None
a = stofn.split("-")
if len(a) != 2:
raise ConfigError("Composite word meaning must contain a single hyphen")
prefix = a[0]
stem = a[1]
m = Meanings.forms(stem)
if m:
for w in m:
if w.ordfl == ordfl:
t = (prefix + w.stofn, -1, ordfl, w.fl, prefix + w.ordmynd, w.beyging)
Meanings.DICT[prefix + w.ordmynd].append(t)
Meanings.ROOT[prefix + w.stofn].append(t)
def add_composite(stofn, ordfl, fl, *, utg=None):
""" Add composite word forms by putting a prefix on existing BIN word forms.
Called from the config file handler. """
assert stofn is not None
assert ordfl is not None
# Handle cases like 'Suður-Ameríku-ríki' correctly
a = stofn.rsplit("-", maxsplit=1)
assert len(a) == 2
stem = a[1]
if stem[0].isupper():
# Uppercase stem (such as 'Norður-Makedónía'): keep the hyphen
prefix = a[0] + "-"
else:
prefix = a[0]
m = Meanings.forms(stem)
if m:
last_utg = None
for w in m:
if w.ordfl == ordfl and (utg is None or utg == w.utg):
# Check for ambiguity between different lemmas
if last_utg is None:
last_utg = w.utg
elif w.utg != last_utg:
# We are encountering two different ids (utg):
# this probably needs to be disambiguated
raise ConfigError(
"Ambiguous word stem: {0}/{1}".format(w.stofn, w.ordfl)
)
# Matches the requested category and also the
# id number (utg), if given: create a new entry in the
# Meanings dict, having id -1
t = (
prefix + w.stofn,
-1,
ordfl,
fl or w.fl,
prefix + w.ordmynd,
w.beyging,
)
Meanings.DICT[prefix + w.ordmynd].append(t)
Meanings.ROOT[prefix + w.stofn].append(t)
