corverb for corverb in corverbs
if corverb.startswith("*"))
if isrefl:
refverb = re.sub(u"с[ья]$", "",
verb) + gender_arg
correfverbs = []
for corverb in corverbs:
correfverbs.append(
"%s|g=%s" %
(re.sub(u"с[ья]$", "",
re.sub(r"^\*", "", corverb)),
"impf" if headword_aspect == "pf"
or corverb.startswith("*") else "pf"))
else:
refverb = (re.search(u"и́?$", verb) and verb +
u"сь" or rulib.try_to_stress(verb) +
u"ся") + gender_arg
correfverbs = []
for corverb in corverbs:
impf_override = corverb.startswith("*")
corverb = re.sub(r"^\*", "", corverb)
correfverbs.append("%s|g=%s" % (
(re.search(u"и́?$", corverb)
and corverb + u"сь" or
rulib.try_to_stress(corverb) + u"ся"),
"impf" if headword_aspect == "pf"
or impf_override else "pf"))
if headword_aspect == "pf" or corverb_impf_override:
refverbs = correfverbs + [refverb]
else:
refverbs = [refverb] + correfverbs
def try_to_stress(form):
if "//" in form:
m = re.search("^(.*?)//(.*)$", form)
# FIXME: This should stress the translit as well
return rulib.try_to_stress(m.group(1)) + "//" + m.group(2)
return rulib.try_to_stress(form)
def convert_zdecl_to_ru_noun_table(decl_z_template, subpagetitle, pagemsg,
headword_template=None):
zdecl = unicode(decl_z_template)
zdeclcopy = blib.parse_text(zdecl).filter_templates()[0]
decl_template = blib.parse_text("{{ru-noun-table}}").filter_templates()[0]
# {{ru-decl-noun-z|звезда́|f-in|d|ё}}
# {{ru-decl-noun-z|ёж|m-inan|b}}
def getp(param):
rmparam(zdeclcopy, param)
return getparam(decl_z_template, param).strip()
zlemma = getp("1")
zgender_anim = getp("2")
zstress = getp("3")
zspecial = re.sub(u"ё", u";ё", getp("4"))
m = re.search(r"^([mfn])-(an|in|inan)$", zgender_anim)
if not m:
pagemsg("WARNING: Unable to recognize z-decl gender/anim spec, skipping: %s" %
zgender_anim)
return None
zgender, zanim = m.groups()
if not zlemma:
pagemsg("WARNING: Empty lemma, skipping: %s" % zdecl)
return None
# Remove unnecessary gender
need_gender = (re.search(u"[иы]́?$", zlemma) or
zgender == "n" and re.search(u"[яа]́?$", zlemma) or
zgender == "m" and re.search(u"[яа]́?$", zlemma) and "(1)" in zspecial or
zlemma.endswith(u"ь"))
if not need_gender:
normal_gender = (re.search(u"[оеё]́?$", zlemma) and "n" or
re.search(u"[ая]́?$", zlemma) and "f" or "m")
if normal_gender != zgender:
pagemsg("WARNING: Gender mismatch, normal gender=%s, explicit gender=%s, keeping gender" %
(normal_gender, zgender))
need_gender = True
if need_gender:
pagemsg("Preserving gender in z-decl: %s" % zdecl)
zspecial = zgender + zspecial
else:
pagemsg("Not preserving gender in z-decl: %s" % zdecl)
# Remove unnecessary stress
stressed_lemma = ru.try_to_stress(zlemma)
def check_defstress(defstr, reason):
if defstr == zstress:
pagemsg("Removing stress %s as default because %s: stressed_lemma=%s, template=%s" %
(defstr, reason, stressed_lemma, zdecl))
return defstr
if ru.is_nonsyllabic(stressed_lemma):
default_stress = check_defstress("b", "nonsyllabic lemma")
elif re.search(u"([аяоеыи]́|ё́?)$", stressed_lemma):
default_stress = check_defstress("b", "ending-accented lemma")
# No need for special-casing for ёнок or а́нин, as they are considered
# accent a by ru-decl-noun-z
else:
default_stress = check_defstress("a", "stem-accented lemma")
if default_stress == zstress:
zstress = ""
else:
pagemsg("Not removing stress %s: %s" % (zstress, zdecl))
# Remove unnecessary lemma
if ru.try_to_stress(subpagetitle) == stressed_lemma:
pagemsg(u"Removing lemma %s because identical to subpagetitle %s (modulo monosyllabic stress differences): %s" %
(zlemma, subpagetitle, zdecl))
zlemma = ""
if zstress:
decl_template.add("1", zstress)
offset = 1
else:
offset = 0
decl_template.add(str(1 + offset), zlemma)
decl_template.add(str(2 + offset), zspecial)
if not getparam(decl_template, "3"):
rmparam(decl_template, "3")
if not getparam(decl_template, "2"):
rmparam(decl_template, "2")
if not getparam(decl_template, "1"):
rmparam(decl_template, "1")
headword_anim_spec = headword_template and extract_headword_anim_spec(headword_template)
def anim_mismatch(zdecl_an, allowed_headword_ans):
if headword_anim_spec and headword_anim_spec not in allowed_headword_ans:
pagemsg("WARNING: z-decl anim %s disagrees with headword-derived %s (%s allowed): zdecl=%s, headword=%s" %
(zdecl_an, headword_anim_spec, ",".join(allowed_headword_ans),
zdecl, unicode(headword_template)))
if zanim == "an":
anim_mismatch(zanim, ["an"])
pagemsg("Preserving z-decl -an as a=an: %s" % zdecl)
decl_template.add("a", "an")
elif zanim == "inan":
anim_mismatch(zanim, ["ai", "ia"])
if headword_anim_spec in ["ai", "ia"]:
pagemsg("Converting z-decl -inan to a=%s: %s" %
(headword_anim_spec, zdecl))
decl_template.add("a", headword_anim_spec)
else:
pagemsg("WARNING: Unable to convert z-decl -inan to a=ai or a=ia, preserving as a=bi: zdecl=%s, headword=%s" %
(zdecl, unicode(headword_template or "(no headword)")))
decl_template.add("a", "bi")
else:
assert(zanim == "in")
anim_mismatch(zanim, ["in"])
pagemsg("Dropping z-decl -in as default: %s" % zdecl)
znum = getp("n")
if znum:
if znum == "pl":
pagemsg("WARNING: Found n=pl in z-decl, should convert manually to plural lemma: %s" %
zdecl)
pagemsg("Preserving z-decl n=%s: %s" % (znum, zdecl))
decl_template.add("n", znum)
preserve_params = [
'nom_sg', 'gen_sg', 'dat_sg', 'acc_sg', 'ins_sg', 'prp_sg',
'nom_pl', 'gen_pl', 'dat_pl', 'acc_pl', 'ins_pl', 'prp_pl',
'voc'
]
renamed_params = {'prp_sg':'pre_sg', 'prp_pl':'pre_pl'}
for param in preserve_params:
val = getp(param)
if not val:
continue
newval = fixup_link(val)
newvals = re.split(r"\s*,\s*", newval)
newvals = [re.sub(r"^\[\[([^\[\]|]*)\]\]$", r"\1", x) for x in newvals]
newval= ",".join(newvals)
newparam = renamed_params.get(param, param)
pagemsg("Preserving z-decl override %s=%s%s%s: %s" % (
newparam, newval,
"" if newparam == param else "; renamed from %s" % param,
"" if newval == val else "; canonicalized from %s=%s" % (param, val),
zdecl))
decl_template.add(newparam, newval)
loc = getp("loc")
if loc:
if loc == u"в":
newloc = u"в +"
elif loc == u"на":
newloc = u"на +"
else:
newloc = u"в/на +"
pagemsg("Preserving z-decl locative loc=%s (canonicalized from loc=%s): %s" %
(newloc, loc, zdecl))
decl_template.add("loc", newloc)
par = getp("par")
if par:
newpar="+"
pagemsg("Preserving z-decl partitive par=%s (canonicalized from par=%s): %s" %
(newpar, par, zdecl))
decl_template.add('par', newpar)
notes = getp("note")
if notes:
pagemsg("WARNING: Found z-decl note=<%s>, converting to notes= but probably needs fixing up with footnote symbol and pltail or similar: %s" %
(notes, zdecl))
decl_template.add('notes', notes)
if zdeclcopy.params:
pagemsg("WARNING: Extraneous params in z-decl: %s" % unicode(zdeclcopy))
#pagemsg("Replacing z-decl %s with regular decl %s" %
# (zdecl, unicode(decl_template)))
return decl_template
def convert_zdecl_to_ru_noun_table(decl_z_template,
subpagetitle,
pagemsg,
headword_template=None):
zdecl = unicode(decl_z_template)
zdeclcopy = blib.parse_text(zdecl).filter_templates()[0]
decl_template = blib.parse_text("{{ru-noun-table}}").filter_templates()[0]
# {{ru-decl-noun-z|звезда́|f-in|d|ё}}
# {{ru-decl-noun-z|ёж|m-inan|b}}
def getp(param):
rmparam(zdeclcopy, param)
return getparam(decl_z_template, param).strip()
zlemma = getp("1")
zgender_anim = getp("2")
zstress = getp("3")
zspecial = re.sub(u"ё", u";ё", getp("4"))
m = re.search(r"^([mfn])-(an|in|inan)$", zgender_anim)
if not m:
pagemsg(
"WARNING: Unable to recognize z-decl gender/anim spec, skipping: %s"
% zgender_anim)
return None
zgender, zanim = m.groups()
if not zlemma:
pagemsg("WARNING: Empty lemma, skipping: %s" % zdecl)
return None
# Remove unnecessary gender
need_gender = (re.search(u"[иы]́?$", zlemma)
or zgender == "n" and re.search(u"[яа]́?$", zlemma)
or zgender == "m" and re.search(u"[яа]́?$", zlemma)
and "(1)" in zspecial or zlemma.endswith(u"ь"))
if not need_gender:
normal_gender = (re.search(u"[оеё]́?$", zlemma) and "n"
or re.search(u"[ая]́?$", zlemma) and "f" or "m")
if normal_gender != zgender:
pagemsg(
"WARNING: Gender mismatch, normal gender=%s, explicit gender=%s, keeping gender"
% (normal_gender, zgender))
need_gender = True
if need_gender:
pagemsg("Preserving gender in z-decl: %s" % zdecl)
zspecial = zgender + zspecial
else:
pagemsg("Not preserving gender in z-decl: %s" % zdecl)
# Remove unnecessary stress
stressed_lemma = rulib.try_to_stress(zlemma)
def check_defstress(defstr, reason):
if defstr == zstress:
pagemsg(
"Removing stress %s as default because %s: stressed_lemma=%s, template=%s"
% (defstr, reason, stressed_lemma, zdecl))
return defstr
if rulib.is_nonsyllabic(stressed_lemma):
default_stress = check_defstress("b", "nonsyllabic lemma")
elif re.search(u"([аяоеыи]́|ё́?)$", stressed_lemma):
default_stress = check_defstress("b", "ending-accented lemma")
# No need for special-casing for ёнок or а́нин, as they are considered
# accent a by ru-decl-noun-z
else:
default_stress = check_defstress("a", "stem-accented lemma")
if default_stress == zstress:
zstress = ""
else:
pagemsg("Not removing stress %s: %s" % (zstress, zdecl))
# Remove unnecessary lemma
if rulib.try_to_stress(subpagetitle) == stressed_lemma:
pagemsg(
u"Removing lemma %s because identical to subpagetitle %s (modulo monosyllabic stress differences): %s"
% (zlemma, subpagetitle, zdecl))
zlemma = ""
if zstress:
decl_template.add("1", zstress)
offset = 1
else:
offset = 0
decl_template.add(str(1 + offset), zlemma)
decl_template.add(str(2 + offset), zspecial)
if not getparam(decl_template, "3"):
rmparam(decl_template, "3")
if not getparam(decl_template, "2"):
rmparam(decl_template, "2")
if not getparam(decl_template, "1"):
rmparam(decl_template, "1")
headword_anim_spec = headword_template and extract_headword_anim_spec(
headword_template)
def anim_mismatch(zdecl_an, allowed_headword_ans):
if headword_anim_spec and headword_anim_spec not in allowed_headword_ans:
pagemsg(
"WARNING: z-decl anim %s disagrees with headword-derived %s (%s allowed): zdecl=%s, headword=%s"
%
(zdecl_an, headword_anim_spec, ",".join(allowed_headword_ans),
zdecl, unicode(headword_template)))
if zanim == "an":
anim_mismatch(zanim, ["an"])
pagemsg("Preserving z-decl -an as a=an: %s" % zdecl)
decl_template.add("a", "an")
elif zanim == "inan":
anim_mismatch(zanim, ["ai", "ia"])
if headword_anim_spec in ["ai", "ia"]:
pagemsg("Converting z-decl -inan to a=%s: %s" %
(headword_anim_spec, zdecl))
decl_template.add("a", headword_anim_spec)
else:
pagemsg(
"WARNING: Unable to convert z-decl -inan to a=ai or a=ia, preserving as a=bi: zdecl=%s, headword=%s"
% (zdecl, unicode(headword_template or "(no headword)")))
decl_template.add("a", "bi")
else:
assert (zanim == "in")
anim_mismatch(zanim, ["in"])
pagemsg("Dropping z-decl -in as default: %s" % zdecl)
znum = getp("n")
if znum:
if znum == "pl":
pagemsg(
"WARNING: Found n=pl in z-decl, should convert manually to plural lemma: %s"
% zdecl)
pagemsg("Preserving z-decl n=%s: %s" % (znum, zdecl))
decl_template.add("n", znum)
preserve_params = [
'nom_sg', 'gen_sg', 'dat_sg', 'acc_sg', 'ins_sg', 'prp_sg', 'nom_pl',
'gen_pl', 'dat_pl', 'acc_pl', 'ins_pl', 'prp_pl', 'voc'
]
renamed_params = {'prp_sg': 'pre_sg', 'prp_pl': 'pre_pl'}
for param in preserve_params:
val = getp(param)
if not val:
continue
newval = fixup_link(val)
newvals = re.split(r"\s*,\s*", newval)
newvals = [re.sub(r"^\[\[([^\[\]|]*)\]\]$", r"\1", x) for x in newvals]
newval = ",".join(newvals)
newparam = renamed_params.get(param, param)
pagemsg("Preserving z-decl override %s=%s%s%s: %s" %
(newparam, newval,
"" if newparam == param else "; renamed from %s" % param,
"" if newval == val else "; canonicalized from %s=%s" %
(param, val), zdecl))
decl_template.add(newparam, newval)
loc = getp("loc")
if loc:
if loc == u"в":
newloc = u"в +"
elif loc == u"на":
newloc = u"на +"
else:
newloc = u"в/на +"
pagemsg(
"Preserving z-decl locative loc=%s (canonicalized from loc=%s): %s"
% (newloc, loc, zdecl))
decl_template.add("loc", newloc)
par = getp("par")
if par:
newpar = "+"
pagemsg(
"Preserving z-decl partitive par=%s (canonicalized from par=%s): %s"
% (newpar, par, zdecl))
decl_template.add('par', newpar)
notes = getp("note")
if notes:
pagemsg(
"WARNING: Found z-decl note=<%s>, converting to notes= but probably needs fixing up with footnote symbol and pltail or similar: %s"
% (notes, zdecl))
decl_template.add('notes', notes)
if zdeclcopy.params:
pagemsg("WARNING: Extraneous params in z-decl: %s" %
unicode(zdeclcopy))
#pagemsg("Replacing z-decl %s with regular decl %s" %
# (zdecl, unicode(decl_template)))
return decl_template
def try_to_stress(form):
if "//" in form:
m = re.search("^(.*?)//(.*)$", form)
# FIXME: This should stress the translit as well
return ru.try_to_stress(m.group(1)) + "//" + m.group(2)
return ru.try_to_stress(form)
