def __init__(self, form, msd=[], v_insts=[]):
(self.form, self.msd) = (form.split('+'), msd)
self.scount = 0
r = ''
#print self.form
for f in self.form:
#print (f)
if f.isdigit():
r += '(.+)'
else:
r += f
self.scount += len(f)
self.regex = r
#print (r)
#print '\n'
self.cregex = re.compile(self.regex)
# vars
collect_vars = defaultdict(set)
for vs in v_insts:
#print vs
for (i, v) in vs:
collect_vars[i].add(v)
self.v_regex = []
for (_, ss) in collect_vars.iteritems():
self.v_regex.append(
re.compile(genregex.genregex(ss, pvalue=0.05).pyregex()))
def paradigms_to_foma(paradigms, grammarname, pval):
"""Converts iterable of paradigms to foma-script (as a string)."""
parvars = {}
rstring = u''
defstring = u''
substring = u''
par_is_constrained = {}
alphabet = paradigms_to_alphabet(paradigms)
alphabet = {'"' + a + '"' for a in alphabet}
alphstring = 'def Alph ' + u'|'.join(alphabet) + ';\n'
for paradigm in paradigms:
# if paradigm.count < 3 and grammarname == 'Gunconstrained':
# continue
par_is_constrained[paradigm.name] = False
parstrings = []
for formnumber, form in enumerate(paradigm.forms):
tagstrings = map(
lambda (feature, value): u'"' + feature + u'"' + u' = ' + u'"'
+ value + u'"', form.msd)
parstring = u''
for idx, (is_var, slot) in enumerate(paradigm.slots):
if is_var:
parvarname = nospace(paradigm.name) + '=var' + str(idx)
if parvarname not in parvars:
r = genregex.genregex(slot, pvalue=pval, length=False)
parvars[parvarname] = True
if r.fomaregex() != '?+':
par_is_constrained[paradigm.name] = True
defstring += 'def ' + parvarname + ' ' + r.fomaregex(
).replace('?', 'Alph') + ';\n'
parstring += ' [' + parvarname + '] '
else:
thisslot = escape_fixed_string(slot[formnumber])
baseformslot = escape_fixed_string(slot[0])
parstring += u' [' + thisslot + u':' + baseformslot + u'] '
parstring += u'0:["[" ' + u' " " '.join(tagstrings) + u' "]"]'
parstrings.append(parstring)
#if grammarname != 'Gcodnstrained' or par_is_constrained[paradigm.name]:
rstring += u'def ' + nospace(
paradigm.name) + u'|\n'.join(parstrings) + u';\n'
#parnames = [nospace(paradigm.name) for paradigm in paradigms if ' ' not in paradigm.name]
parnames = []
for paradigm in paradigms:
#if ' ' not in paradigm.name and (grammarname != 'Gconstrdained' or par_is_constrained[paradigm.name]):
parnames.append(nospace(paradigm.name))
rstring += u'def ' + grammarname + u' ' + u' | '.join(parnames) + u';'
return alphstring + defstring + rstring
def __init__(self, form, msd=[], v_insts=[]):
(self.form,self.msd) = (form.split('+'), msd)
self.scount = 0
r = ''
for f in self.form:
if f.isdigit():
r += '(.+)'
else:
r += f
self.scount += len(f)
self.regex = r
self.cregex = re.compile(self.regex)
# vars
collect_vars = defaultdict(set)
for vs in v_insts:
for (i,v) in vs:
collect_vars[i].add(v)
self.v_regex = []
for (_,ss) in collect_vars.iteritems():
self.v_regex.append(re.compile(genregex.genregex(ss,pvalue=0.05).pyregex()))
