def _replace_ids(can_anns, _from, to, excluded_anns=None, debug=False):
if excluded_anns is None:
excluded_anns = set()
for can_ann in can_anns:
if isinstance(can_ann, Text) and Text not in excluded_anns:
# These annotations can not reference others, pass
pass
elif isinstance(can_ann, Event) and Event not in excluded_anns:
can_ann.args = [(a, to if v in _from else v)
for a, v in can_ann.args]
elif isinstance(can_ann, Modifier) and Modifier not in excluded_anns:
if can_ann.target in _from:
can_ann.target = to
elif isinstance(can_ann, Equiv) and Equiv not in excluded_anns:
if debug:
print >> stderr, 'Will purge from equiv:', _from & set(
can_ann.members), to
new_members = [m for m in can_ann.members if m not in _from]
if new_members != can_ann.members:
can_ann.members = new_members
can_ann.members.append(to)
sort_nicely(can_ann.members)
elif not any(
isinstance(can_ann, c) for c in (
Text,
Event,
Modifier,
Equiv,
)):
assert False, 'unknown canonical annotation type'
def _replace_ids(can_anns, _from, to, excluded_anns=None, debug=False):
if excluded_anns is None:
excluded_anns = set()
for can_ann in can_anns:
if isinstance(can_ann, Text) and Text not in excluded_anns:
# These annotations can not reference others, pass
pass
elif isinstance(can_ann, Event) and Event not in excluded_anns:
can_ann.args = [(a, to if v in _from else v)
for a, v in can_ann.args]
elif isinstance(can_ann, Modifier) and Modifier not in excluded_anns:
if can_ann.target in _from:
can_ann.target = to
elif isinstance(can_ann, Equiv) and Equiv not in excluded_anns:
if debug:
print >> stderr, 'Will purge from equiv:', _from & set(can_ann.members), to
new_members = [m for m in can_ann.members if m not in _from]
if new_members != can_ann.members:
can_ann.members = new_members
can_ann.members.append(to)
sort_nicely(can_ann.members)
elif not any(isinstance(can_ann, c) for c in (Text, Event, Modifier,
Equiv, )):
assert False, 'unknown canonical annotation type'
eq_anns_to_remove = set()
eq_anns = [a for a in can_anns if isinstance(a, Equiv)]
for eq_set in equivalent_texts.itervalues():
#print 'EQ:', eq_set
#print eq_anns
for eq_ann_a, eq_ann_b in combinations(eq_anns, 2):
#print 'A:', eq_ann_a.members
#print 'B:', eq_ann_b.members
# Does the equivalent texts bridge the equivalences?
if ((set(eq_ann_a.members) & eq_set)
and (set(eq_ann_b.members) & eq_set)):
if debug:
print >> stderr, 'Merging:', eq_ann_a, 'and', eq_ann_b
eq_ann_a.members = eq_ann_a.members + [a for a in eq_ann_b.members
if a not in eq_ann_a.members]
sort_nicely(eq_ann_a.members)
eq_ann_b.members = eq_ann_b.members + [a for a in eq_ann_a.members
if a not in eq_ann_b.members]
sort_nicely(eq_ann_b.members)
# We can now safely remove the latter annotation later on
eq_anns_to_remove.add(eq_ann_b)
if debug and eq_anns_to_remove:
print >> stderr, 'Will remove redundant Equivs:', eq_anns_to_remove
can_anns = [a for a in can_anns if a not in eq_anns_to_remove]
# Merge text into unique (texts, types) since we now lack textual bounds
ids_to_remove = set()
for eq_set in equivalent_texts.itervalues():
eqs = [a for a in eq_set]
sort_nicely(eqs)
to_keep = eqs[0]
not_to_keep = set(eq_set - set((to_keep, )))
eq_anns = [a for a in can_anns if isinstance(a, Equiv)]
for eq_set in equivalent_texts.itervalues():
#print 'EQ:', eq_set
#print eq_anns
for eq_ann_a, eq_ann_b in combinations(eq_anns, 2):
#print 'A:', eq_ann_a.members
#print 'B:', eq_ann_b.members
# Does the equivalent texts bridge the equivalences?
if ((set(eq_ann_a.members) & eq_set)
and (set(eq_ann_b.members) & eq_set)):
if debug:
print >> stderr, 'Merging:', eq_ann_a, 'and', eq_ann_b
eq_ann_a.members = eq_ann_a.members + [
a for a in eq_ann_b.members if a not in eq_ann_a.members
]
sort_nicely(eq_ann_a.members)
eq_ann_b.members = eq_ann_b.members + [
a for a in eq_ann_a.members if a not in eq_ann_b.members
]
sort_nicely(eq_ann_b.members)
# We can now safely remove the latter annotation later on
eq_anns_to_remove.add(eq_ann_b)
if debug and eq_anns_to_remove:
print >> stderr, 'Will remove redundant Equivs:', eq_anns_to_remove
can_anns = [a for a in can_anns if a not in eq_anns_to_remove]
# Merge text into unique (texts, types) since we now lack textual bounds
ids_to_remove = set()
for eq_set in equivalent_texts.itervalues():
eqs = [a for a in eq_set]
sort_nicely(eqs)
to_keep = eqs[0]