def process_page_section(index, page, section, verbose):
pagetitle = unicode(page.title())
subpagetitle = re.sub("^.*:", "", pagetitle)
def pagemsg(txt):
msg("Page %s %s: %s" % (index, pagetitle, txt))
def expand_text(tempcall):
return blib.expand_text(tempcall, pagetitle, pagemsg, verbose)
if not page.exists():
pagemsg("WARNING: Page doesn't exist, skipping")
return None
parsed = blib.parse_text(section)
noun_table_templates = []
noun_old_templates = []
for t in parsed.filter_templates():
if unicode(t.name) == "ru-decl-noun-see":
pagemsg("Found ru-decl-noun-see, skipping")
return None
for t in parsed.filter_templates():
if unicode(t.name) == "ru-noun-table":
noun_table_templates.append(t)
if unicode(t.name) == "ru-noun-old":
noun_old_templates.append(t)
if len(noun_table_templates) > 1:
pagemsg("WARNING: Found multiple ru-noun-table templates, skipping")
return None
if len(noun_old_templates) > 1:
pagemsg("WARNING: Found multiple ru-noun-old templates, skipping")
return None
if len(noun_table_templates) < 1:
if noun_old_templates:
pagemsg("WARNING: No ru-noun-table templates but found ru-noun-old template(s): %s" %
", ".join(unicode(x) for x in noun_old_templates))
return unicode(parsed), 0, 0, 0, []
for t in parsed.filter_templates():
if unicode(t.name) in ["ru-noun+", "ru-proper noun+"]:
pagemsg("Found ru-noun+ or ru-proper noun+, skipping")
return None
headword_templates = []
for t in parsed.filter_templates():
if unicode(t.name) in ["ru-noun", "ru-proper noun"]:
headword_templates.append(t)
if len(headword_templates) > 1:
pagemsg("WARNING: Found multiple headword templates, skipping")
return None
if len(headword_templates) < 1:
return unicode(parsed), 0, 0, 0, []
noun_table_template = noun_table_templates[0]
noun_old_template = noun_old_templates[0] if len(noun_old_templates) == 1 else None
headword_template = headword_templates[0]
frobbed_manual_translit = []
decl_templates = [x for x in [noun_table_template, noun_old_template] if x]
if verbose:
pagemsg("Found headword template: %s" % unicode(headword_template))
pagemsg("Found decl template: %s" % unicode(noun_table_template))
if noun_old_template:
pagemsg("Found old decl template: %s" % unicode(noun_old_template))
# Retrieve headword translit and maybe transfer to decl
headword_tr = getparam(headword_template, "tr")
if headword_tr:
if verbose:
pagemsg("Found headword manual translit tr=%s" % headword_tr)
if "," in headword_tr:
pagemsg("WARNING: Comma in headword manual translit, skipping: %s" %
headword_tr)
return None
# Punt if multi-arg-set, can't handle yet
for decl_template in decl_templates:
for param in decl_template.params:
if not param.showkey:
val = unicode(param.value)
if val == "or":
pagemsg("WARNING: Manual translit and multi-decl templates, can't handle, skipping: %s" % unicode(decl_template))
return None
if val == "-" or val == "_" or val.startswith("join:"):
pagemsg("WARNING: Manual translit and multi-word templates, can't handle, skipping: %s" % unicode(decl_template))
return None
for i in xrange(2, 10):
if getparam(headword_template, "tr%s" % i):
pagemsg("WARNING: Headword template has translit param tr%s, can't handle, skipping: %s" % (
i, unicode(headword_template)))
return None
if runoun.arg1_is_stress(getparam(decl_template, "1")):
lemma_arg = "2"
else:
lemma_arg = "1"
lemmaval = getparam(decl_template, lemma_arg)
if not lemmaval:
lemmaval = subpagetitle
if "//" in lemmaval:
m = re.search("^(.*?)//(.*)$", lemmaval)
if m.group(2) != headword_tr:
pagemsg("WARNING: Found existing manual translit in decl template %s, but doesn't match headword translit %s; skipping" % (
lemmaval, headword_tr))
return None
else:
pagemsg("Already found manual translit in decl template %s" %
lemmaval)
else:
lemmaval += "//" + headword_tr
orig_decl_template = unicode(decl_template)
decl_template.add(lemma_arg, lemmaval)
pagemsg("Replacing decl %s with %s" % (orig_decl_template,
unicode(decl_template)))
frobbed_manual_translit = [headword_tr]
genders = blib.fetch_param_chain(headword_template, "2", "g")
bian_replaced = 0
# Change a=bi in decl to a=ia or a=ai, depending on order of anim/inan in
# headword template
for decl_template in decl_templates:
if getparam(decl_template, "a") in ["b", "bi", "bian", "both"]:
saw_in = -1
saw_an = -1
for i,g in enumerate(genders):
if re.search(r"\bin\b", g) and saw_in < 0:
saw_in = i
if re.search(r"\ban\b", g) and saw_an < 0:
saw_an = i
if saw_in >= 0 and saw_an >= 0:
orig_decl_template = unicode(decl_template)
if saw_in < saw_an:
pagemsg("Replacing a=bi with a=ia in decl template")
decl_template.add("a", "ia")
bian_replaced = 1
else:
pagemsg("Replacing a=bi with a=ai in decl template")
decl_template.add("a", "ai")
bian_replaced = 1
pagemsg("Replacing decl %s with %s" % (orig_decl_template,
unicode(decl_template)))
generate_template = re.sub(r"^\{\{ru-noun-table", "{{ru-generate-noun-args",
unicode(noun_table_template))
generate_result = expand_text(generate_template)
if not generate_result:
pagemsg("WARNING: Error generating noun args, skipping")
return None
args = ru.split_generate_args(generate_result)
genders = runoun.check_old_noun_headword_forms(headword_template, args,
subpagetitle, pagemsg)
if genders == None:
return None
new_params = []
for param in noun_table_template.params:
new_params.append((param.name, param.value))
orig_headword_template = unicode(headword_template)
params_to_preserve = runoun.fix_old_headword_params(headword_template,
new_params, genders, pagemsg)
if params_to_preserve == None:
return None
if unicode(headword_template.name) == "ru-proper noun":
# If proper noun and n is both then we need to add n=both because
# proper noun+ defaults to n=sg
if args["n"] == "b" and not getparam(headword_template, "n"):
pagemsg("Adding n=both to headword tempate")
headword_template.add("n", "both")
# Correspondingly, if n is sg then we can usually remove n=sg;
# but we need to check that the number is actually sg with n=sg
# removed because of the possibility of plurale tantum lemmas
if args["n"] == "s":
generate_template_with_ndef = generate_template.replace("}}", "|ndef=sg}}")
generate_template_with_ndef = re.sub(r"\|n=s[^=|{}]*", "",
generate_template_with_ndef)
generate_result = expand_text(generate_template_with_ndef)
if not generate_result:
pagemsg("WARNING: Error generating noun args, skipping")
return None
ndef_args = ru.split_generate_args(generate_result)
if ndef_args["n"] == "s":
existing_n = getparam(headword_template, "n")
if existing_n and not re.search(r"^s", existing_n):
pagemsg("WARNING: Something wrong: Found n=%s, not singular" %
existing_n)
else:
pagemsg("Removing n=sg from headword tempate")
rmparam(headword_template, "n")
else:
pagemsg("WARNING: Unable to remove n= from headword template because n=%s" %
ndef_args["n"])
headword_template.params.extend(params_to_preserve)
ru_noun_changed = 0
ru_proper_noun_changed = 0
if unicode(headword_template.name) == "ru-noun":
headword_template.name = "ru-noun+"
ru_noun_changed = 1
else:
headword_template.name = "ru-proper noun+"
ru_proper_noun_changed = 1
pagemsg("Replacing headword %s with %s" % (orig_headword_template, unicode(headword_template)))
return unicode(parsed), ru_noun_changed, ru_proper_noun_changed, bian_replaced, frobbed_manual_translit
def process_page(page, index, parsed):
global args
pagetitle = unicode(page.title())
subpagetitle = re.sub("^.*:", "", pagetitle)
def pagemsg(txt):
msg("Page %s %s: %s" % (index, pagetitle, txt))
pagemsg("Processing")
if ":" in pagetitle:
pagemsg("WARNING: Colon in page title, skipping")
return
def expand_text(tempcall):
return blib.expand_text(tempcall, pagetitle, pagemsg, args.verbose)
origtext = page.text
parsed = blib.parse_text(origtext)
# Find the declension arguments for LEMMA and inflected form INFL,
# the WORDINDth word in the expression. Return value is a tuple of
# four items: a list of (NAME, VALUE) tuples for the arguments, whether
# the word is an adjective, the value of n= (if given), and the value
# of a= (if given).
def find_decl_args(lemma, infl, wordind):
declpage = pywikibot.Page(site, lemma)
if rulib.remove_accents(infl) == lemma:
wordlink = "[[%s]]" % infl
else:
wordlink = "[[%s|%s]]" % (lemma, infl)
if not declpage.exists():
if lemma in is_short_adj or re.search(u"(ий|ый|ой)$", lemma):
pagemsg(
"WARNING: Page doesn't exist, assuming word #%s adjectival: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return [("1", wordlink), ("2", "+")], True, None, None
else:
pagemsg(
"WARNING: Page doesn't exist, can't locate decl for word #%s, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
parsed = blib.parse_text(declpage.text)
decl_templates = []
headword_templates = []
decl_z_templates = []
for t in parsed.filter_templates():
tname = unicode(t.name)
if tname in ["ru-noun-table", "ru-decl-adj"]:
pagemsg("find_decl_args: Found decl template: %s" % unicode(t))
decl_templates.append(t)
if tname in ["ru-noun", "ru-proper noun"]:
pagemsg("find_decl_args: Found headword template: %s" %
unicode(t))
headword_templates.append(t)
if tname in ["ru-decl-noun-z"]:
pagemsg("find_decl_args: Found z-decl template: %s" %
unicode(t))
decl_z_templates.append(t)
if not decl_templates:
if decl_z_templates:
# {{ru-decl-noun-z|звезда́|f-in|d|ё}}
# {{ru-decl-noun-z|ёж|m-inan|b}}
if len(decl_z_templates) > 1:
pagemsg(
"WARNING: Multiple decl-z templates during decl lookup for word #%s, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
else:
decl_z_template = decl_z_templates[0]
headword_template = None
pagemsg("find_decl_args: Using z-decl template: %s" %
unicode(decl_z_template))
if len(headword_templates) == 0:
pagemsg(
"WARNING: find_decl_args: No headword templates for use with z-decl template conversion during decl lookup for word #%s: lemma=%s, infl=%s, zdecl=%s"
% (wordind, lemma, infl, unicode(decl_z_template)))
elif len(headword_templates) > 1:
pagemsg(
"WARNING: find_decl_args: Multiple headword templates for use with z-decl template conversion during decl lookup for word #%s, ignoring: lemma=%s, infl=%s, zdecl=%s"
% (wordind, lemma, infl, unicode(decl_z_template)))
else:
headword_template = headword_templates[0]
pagemsg(
"find_decl_args: For word #%s, lemma=%s, infl=%s, using headword template %s for use with z-decl template %s"
%
(wordind, lemma, infl, unicode(headword_template),
unicode(decl_z_template)))
decl_template = runounlib.convert_zdecl_to_ru_noun_table(
decl_z_template,
subpagetitle,
pagemsg,
headword_template=headword_template)
decl_templates = [decl_template]
elif "[[Category:Russian indeclinable nouns]]" in declpage.text or [
x for x in headword_templates if getparam(x, "3") == "-"
]:
return [("1", wordlink), ("2", "$")], False, None, None
else:
pagemsg(
"WARNING: No decl template during decl lookup for word #%s, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
if len(decl_templates) == 1:
decl_template = decl_templates[0]
else:
# Multiple decl templates
for t in decl_templates:
if unicode(t.name) == "ru-decl-adj" and re.search(
u"(ий|ый|ой)$", lemma):
pagemsg(
"WARNING: Multiple decl templates during decl lookup for word #%s, assuming adjectival: lemma=%s, infl=%s"
% (wordind, lemma, infl))
decl_template = t
break
else:
if lemma in use_given_decl:
overriding_decl = use_given_decl[lemma]
pagemsg(
"WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, using overriding declension %s: lemma=%s, infl=%s"
% (wordind, overriding_decl, lemma, infl))
decl_template = blib.parse_text(
overriding_decl).filter_templates()[0]
elif pagetitle in use_given_page_decl:
overriding_decl = use_given_page_decl[pagetitle].get(
lemma, None)
if not overriding_decl:
pagemsg(
"WARNING: Missing entry for ambiguous-decl lemma for word #%s, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return
else:
pagemsg(
"WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, using overriding declension %s: lemma=%s, infl=%s"
% (wordind, overriding_decl, lemma, infl))
decl_template = blib.parse_text(
overriding_decl).filter_templates()[0]
else:
pagemsg(
"WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
pagemsg("find_decl_args: Using decl template: %s" %
unicode(decl_template))
if unicode(decl_template.name) == "ru-decl-adj":
if re.search(ur"\bь\b", getparam(decl_template, "2"), re.U):
return [("1", wordlink), ("2", u"+ь")], True, None, None
else:
return [("1", wordlink), ("2", "+")], True, None, None
# ru-noun-table
assert unicode(decl_template.name) == "ru-noun-table"
# Split out the arg sets in the declension and check the
# lemma of each one, taking care to handle cases where there is no lemma
# (it would default to the page name).
highest_numbered_param = 0
for p in decl_template.params:
pname = unicode(p.name)
if re.search("^[0-9]+$", pname):
highest_numbered_param = max(highest_numbered_param,
int(pname))
# Now gather the numbered arguments into arg sets. Code taken from
# ru-noun.lua.
offset = 0
arg_sets = []
arg_set = []
for i in xrange(1, highest_numbered_param + 2):
end_arg_set = False
val = getparam(decl_template, str(i))
if i == highest_numbered_param + 1:
end_arg_set = True
elif val == "_" or val == "-" or re.search("^join:", val):
pagemsg(
"WARNING: Found multiword decl during decl lookup for word #%s, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
elif val == "or":
end_arg_set = True
if end_arg_set:
arg_sets.append(arg_set)
arg_set = []
offset = i
else:
arg_set.append(val)
canon_infl = rulib.remove_accents(infl).lower()
canon_lemma = lemma.lower()
ispl = False
need_sc1 = False
found_gender = None
if canon_infl != canon_lemma:
for sgend, plend, gender, is_sc1 in pl_data:
if sgend:
check_sgend = sgend
else:
check_sgend = consonant_re
if re.search(check_sgend + "$",
canon_lemma) and canon_infl == re.sub(
sgend + "$", plend, canon_lemma):
ispl = True
found_gender = gender
need_sc1 = is_sc1
break
else:
pagemsg(
"WARNING: For word#%s, inflection not same as lemma, not recognized as plural, can't handle, skipping: lemma=%s, infl=%s"
% (wordind, lemma, infl))
return None
# Substitute the wordlink for any lemmas in the declension.
# If plural, also add gender and verify special case (1) as necessary.
# Concatenate all the numbered params, substituting the wordlink into
# the lemma as necessary.
numbered_params = []
for arg_set in arg_sets:
lemma_arg = 0
if len(arg_set) > 0 and runounlib.arg1_is_stress(arg_set[0]):
lemma_arg = 1
if len(arg_set) <= lemma_arg:
arg_set.append("")
arglemma = arg_set[lemma_arg]
manualtr = ""
if "//" in arglemma:
arglemma, manualtr = re.search("^(.*?)(//.*?)$",
arglemma).groups()
if (not arglemma or arglemma.lower() == infl.lower()
or rulib.is_monosyllabic(infl)
and rulib.remove_accents(arglemma).lower()
== rulib.remove_accents(infl).lower() or ispl and
rulib.remove_accents(arglemma).lower() == lemma.lower()):
arg_set[lemma_arg] = wordlink + manualtr
else:
pagemsg(
"WARNING: Can't sub word link %s into decl lemma %s%s" %
(wordlink, arg_set[lemma_arg], ispl and ", skipping"
or ""))
if ispl:
return None
if ispl:
# Add the gender
if len(arg_set) <= lemma_arg + 1:
arg_set.append("")
declarg = arg_set[lemma_arg + 1]
# First, sub in gender
m = re.search("(3f|[mfn])", declarg)
if found_gender == "mf":
if not m:
pagemsg(
u"WARNING: For singular in -ь and plural in -и, need gender in singular and don't have it, word #%s, skipping: lemma=%s, infl=%s"
% (wordinfl, lemma, infl))
return None
decl_gender = m.group(1)
if decl_gender == "n":
pagemsg(
u"WARNING: For singular in -ь and plural in -и, can't have neuter gender for word #%s, skipping: lemma=%s, infl=%s"
% (wordinfl, lemma, infl))
return None
elif decl_gender in ["m", "3f"]:
pagemsg(
u"Singular in -ь and plural in -и, already found gender %s in decl for word #%s, taking no action: lemma=%s, infl=%s"
% (decl_gender, wordind, lemma, infl))
else:
assert gender == "f"
pagemsg(
u"Singular in -ь and plural in -и, replacing f with 3f so singular will be recognized for word #%s: lemma=%s, infl=%s"
% (wordind, lemma, infl))
declarg = re.sub("f", "3f", declarg, 1)
else:
if m:
decl_gender = m.group(1)
if decl_gender == found_gender:
pagemsg(
"Already found gender %s in decl for word #%s, taking no action: lemma=%s, infl=%s"
% (found_gender, wordind, lemma, infl))
else:
pagemsg(
"WARNING: Found wrong gender %s in decl for word #%s, forcibly replacing with lemma-form-derived gender %s: lemma=%s, infl=%s"
% (decl_gender, wordind, found_gender, lemma,
infl))
declarg = re.sub("(3f|[mfn])", found_gender,
declarg, 1)
else:
pagemsg(
"No gender in decl for word #%s, adding gender %s: lemma=%s, infl=%s"
% (wordind, found_gender, lemma, infl))
declarg = found_gender + declarg
# Now check special case 1
if need_sc1 != ("(1)" in declarg):
if need_sc1:
pagemsg(
"WARNING: Irregular plural calls for special case (1), but not present in decl arg for word #%s, skipping: declarg=%s, lemma=%s, infl=%s"
% (wordind, declarg, lemma, infl))
return None
else:
pagemsg(
"WARNING: Special case (1) present in decl arg but plural for word #%s is regular, skipping: declarg=%s, lemma=%s, infl=%s"
% (wordind, declarg, lemma, infl))
return None
arg_set[lemma_arg + 1] = declarg
if numbered_params:
numbered_params.append("or")
numbered_params.extend(arg_set)
# Now gather all params, including named ones.
params = []
params.extend(
(str(i + 1), val)
for i, val in zip(xrange(len(numbered_params)), numbered_params))
num = None
anim = None
for p in decl_template.params:
pname = unicode(p.name)
val = unicode(p.value)
if pname == "a":
anim = val
elif pname == "n":
num = val
elif pname == "notes":
params.append((pname, val))
elif pname == "title":
pagemsg(
"WARNING: Found explicit title= for word #%s, ignoring: lemma=%s, infl=%s, title=%s"
% (wordind, lemma, infl, val))
elif re.search("^[0-9]+$", pname):
pass
else:
keepparam = True
if pname == "loc":
if pagetitle in keep_locative:
pagemsg(
"Keeping locative for word #%s because page in keep_locative: loc=%s, lemma=%s, infl=%s"
% (wordind, val, lemma, infl))
else:
pagemsg(
"WARNING: Discarding locative for word #%s: loc=%s, lemma=%s, infl=%s"
% (wordind, val, lemma, infl))
keepparam = False
if pname == "par":
pagemsg(
"WARNING: Discarding partitive for word #%s: par=%s, lemma=%s, infl=%s"
% (wordind, val, lemma, infl))
keepparam = False
if pname == "voc":
pagemsg(
"WARNING: Discarding vocative for word #%s: voc=%s, lemma=%s, infl=%s"
% (wordind, val, lemma, infl))
keepparam = False
if keepparam:
if pname == "loc" and re.search(ur"^(на|в)\b", val, re.U):
pagemsg(
u"WARNING: на or в found in loc= for word #%s, may not work in multi-word lemma: loc=%s, lemma=%s, infl=%s"
% (wordind, val, lemma, infl))
pname += str(wordind)
params.append((pname, val))
def process_page(index, page, save, verbose):
pagetitle = unicode(page.title())
subpagetitle = re.sub("^.*:", "", pagetitle)
def pagemsg(txt):
msg("Page %s %s: %s" % (index, pagetitle, txt))
pagemsg("Processing")
if ":" in pagetitle:
pagemsg("WARNING: Colon in page title, skipping")
return
def expand_text(tempcall):
return blib.expand_text(tempcall, pagetitle, pagemsg, verbose)
origtext = page.text
parsed = blib.parse_text(origtext)
# Find the declension arguments for LEMMA and inflected form INFL,
# the WORDINDth word in the expression. Return value is a tuple of
# four items: a list of (NAME, VALUE) tuples for the arguments, whether
# the word is an adjective, the value of n= (if given), and the value
# of a= (if given).
def find_decl_args(lemma, infl, wordind):
declpage = pywikibot.Page(site, lemma)
if ru.remove_accents(infl) == lemma:
wordlink = "[[%s]]" % infl
else:
wordlink = "[[%s|%s]]" % (lemma, infl)
if not declpage.exists():
if lemma in is_short_adj or re.search(u"(ий|ый|ой)$", lemma):
pagemsg("WARNING: Page doesn't exist, assuming word #%s adjectival: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return [("1", wordlink), ("2", "+")], True, None, None
else:
pagemsg("WARNING: Page doesn't exist, can't locate decl for word #%s, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
parsed = blib.parse_text(declpage.text)
decl_templates = []
headword_templates = []
decl_z_templates = []
for t in parsed.filter_templates():
tname = unicode(t.name)
if tname in ["ru-noun-table", "ru-decl-adj"]:
pagemsg("find_decl_args: Found decl template: %s" % unicode(t))
decl_templates.append(t)
if tname in ["ru-noun", "ru-proper noun"]:
pagemsg("find_decl_args: Found headword template: %s" % unicode(t))
headword_templates.append(t)
if tname in ["ru-decl-noun-z"]:
pagemsg("find_decl_args: Found z-decl template: %s" % unicode(t))
decl_z_templates.append(t)
if not decl_templates:
if decl_z_templates:
# {{ru-decl-noun-z|звезда́|f-in|d|ё}}
# {{ru-decl-noun-z|ёж|m-inan|b}}
if len(decl_z_templates) > 1:
pagemsg("WARNING: Multiple decl-z templates during decl lookup for word #%s, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
else:
decl_z_template = decl_z_templates[0]
headword_template = None
pagemsg("find_decl_args: Using z-decl template: %s" %
unicode(decl_z_template))
if len(headword_templates) == 0:
pagemsg("WARNING: find_decl_args: No headword templates for use with z-decl template conversion during decl lookup for word #%s: lemma=%s, infl=%s, zdecl=%s" %
(wordind, lemma, infl, unicode(decl_z_template)))
elif len(headword_templates) > 1:
pagemsg("WARNING: find_decl_args: Multiple headword templates for use with z-decl template conversion during decl lookup for word #%s, ignoring: lemma=%s, infl=%s, zdecl=%s" %
(wordind, lemma, infl, unicode(decl_z_template)))
else:
headword_template = headword_templates[0]
pagemsg("find_decl_args: For word #%s, lemma=%s, infl=%s, using headword template %s for use with z-decl template %s" %
(wordind, lemma, infl, unicode(headword_template),
unicode(decl_z_template)))
decl_template = runoun.convert_zdecl_to_ru_noun_table(decl_z_template,
subpagetitle, pagemsg, headword_template=headword_template)
decl_templates = [decl_template]
elif "[[Category:Russian indeclinable nouns]]" in declpage.text or [
x for x in headword_templates if getparam(x, "3") == "-"]:
return [("1", wordlink), ("2", "$")], False, None, None
else:
pagemsg("WARNING: No decl template during decl lookup for word #%s, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
if len(decl_templates) == 1:
decl_template = decl_templates[0]
else:
# Multiple decl templates
for t in decl_templates:
if unicode(t.name) == "ru-decl-adj" and re.search(u"(ий|ый|ой)$", lemma):
pagemsg("WARNING: Multiple decl templates during decl lookup for word #%s, assuming adjectival: lemma=%s, infl=%s" %
(wordind, lemma, infl))
decl_template = t
break
else:
if lemma in use_given_decl:
overriding_decl = use_given_decl[lemma]
pagemsg("WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, using overriding declension %s: lemma=%s, infl=%s" %
(wordind, overriding_decl, lemma, infl))
decl_template = blib.parse_text(overriding_decl).filter_templates()[0]
elif pagetitle in use_given_page_decl:
overriding_decl = use_given_page_decl[pagetitle].get(lemma, None)
if not overriding_decl:
pagemsg("WARNING: Missing entry for ambiguous-decl lemma for word #%s, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return
else:
pagemsg("WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, using overriding declension %s: lemma=%s, infl=%s" %
(wordind, overriding_decl, lemma, infl))
decl_template = blib.parse_text(overriding_decl).filter_templates()[0]
else:
pagemsg("WARNING: Multiple decl templates during decl lookup for word #%s and not adjectival, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
pagemsg("find_decl_args: Using decl template: %s" % unicode(decl_template))
if unicode(decl_template.name) == "ru-decl-adj":
if re.search(ur"\bь\b", getparam(decl_template, "2"), re.U):
return [("1", wordlink), ("2", u"+ь")], True, None, None
else:
return [("1", wordlink), ("2", "+")], True, None, None
# ru-noun-table
assert unicode(decl_template.name) == "ru-noun-table"
# Split out the arg sets in the declension and check the
# lemma of each one, taking care to handle cases where there is no lemma
# (it would default to the page name).
highest_numbered_param = 0
for p in decl_template.params:
pname = unicode(p.name)
if re.search("^[0-9]+$", pname):
highest_numbered_param = max(highest_numbered_param, int(pname))
# Now gather the numbered arguments into arg sets. Code taken from
# ru-noun.lua.
offset = 0
arg_sets = []
arg_set = []
for i in xrange(1, highest_numbered_param + 2):
end_arg_set = False
val = getparam(decl_template, str(i))
if i == highest_numbered_param + 1:
end_arg_set = True
elif val == "_" or val == "-" or re.search("^join:", val):
pagemsg("WARNING: Found multiword decl during decl lookup for word #%s, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
elif val == "or":
end_arg_set = True
if end_arg_set:
arg_sets.append(arg_set)
arg_set = []
offset = i
else:
arg_set.append(val)
canon_infl = ru.remove_accents(infl).lower()
canon_lemma = lemma.lower()
ispl = False
need_sc1 = False
found_gender = None
if canon_infl != canon_lemma:
for sgend, plend, gender, is_sc1 in pl_data:
if sgend:
check_sgend = sgend
else:
check_sgend = consonant_re
if re.search(check_sgend + "$", canon_lemma) and canon_infl == re.sub(sgend + "$", plend, canon_lemma):
ispl = True
found_gender = gender
need_sc1 = is_sc1
break
else:
pagemsg("WARNING: For word#%s, inflection not same as lemma, not recognized as plural, can't handle, skipping: lemma=%s, infl=%s" %
(wordind, lemma, infl))
return None
# Substitute the wordlink for any lemmas in the declension.
# If plural, also add gender and verify special case (1) as necessary.
# Concatenate all the numbered params, substituting the wordlink into
# the lemma as necessary.
numbered_params = []
for arg_set in arg_sets:
lemma_arg = 0
if len(arg_set) > 0 and runoun.arg1_is_stress(arg_set[0]):
lemma_arg = 1
if len(arg_set) <= lemma_arg:
arg_set.append("")
arglemma = arg_set[lemma_arg]
manualtr = ""
if "//" in arglemma:
arglemma, manualtr = re.search("^(.*?)(//.*?)$", arglemma).groups()
if (not arglemma or arglemma.lower() == infl.lower() or
ru.is_monosyllabic(infl) and ru.remove_accents(arglemma).lower() ==
ru.remove_accents(infl).lower() or
ispl and ru.remove_accents(arglemma).lower() == lemma.lower()
):
arg_set[lemma_arg] = wordlink + manualtr
else:
pagemsg("WARNING: Can't sub word link %s into decl lemma %s%s" % (
wordlink, arg_set[lemma_arg], ispl and ", skipping" or ""))
if ispl:
return None
if ispl:
# Add the gender
if len(arg_set) <= lemma_arg + 1:
arg_set.append("")
declarg = arg_set[lemma_arg + 1]
# First, sub in gender
m = re.search("(3f|[mfn])", declarg)
if found_gender == "mf":
if not m:
pagemsg(u"WARNING: For singular in -ь and plural in -и, need gender in singular and don't have it, word #%s, skipping: lemma=%s, infl=%s" %
(wordinfl, lemma, infl))
return None
decl_gender = m.group(1)
if decl_gender == "n":
pagemsg(u"WARNING: For singular in -ь and plural in -и, can't have neuter gender for word #%s, skipping: lemma=%s, infl=%s" %
(wordinfl, lemma, infl))
return None
elif decl_gender in ["m", "3f"]:
pagemsg(u"Singular in -ь and plural in -и, already found gender %s in decl for word #%s, taking no action: lemma=%s, infl=%s" %
(decl_gender, wordind, lemma, infl))
else:
assert gender == "f"
pagemsg(u"Singular in -ь and plural in -и, replacing f with 3f so singular will be recognized for word #%s: lemma=%s, infl=%s" %
(wordind, lemma, infl))
declarg = re.sub("f", "3f", declarg, 1)
else:
if m:
decl_gender = m.group(1)
if decl_gender == found_gender:
pagemsg("Already found gender %s in decl for word #%s, taking no action: lemma=%s, infl=%s" %
(found_gender, wordind, lemma, infl))
else:
pagemsg("WARNING: Found wrong gender %s in decl for word #%s, forcibly replacing with lemma-form-derived gender %s: lemma=%s, infl=%s" %
(decl_gender, wordind, found_gender, lemma, infl))
declarg = re.sub("(3f|[mfn])", found_gender, declarg, 1)
else:
pagemsg("No gender in decl for word #%s, adding gender %s: lemma=%s, infl=%s" %
(wordind, found_gender, lemma, infl))
declarg = found_gender + declarg
# Now check special case 1
if need_sc1 != ("(1)" in declarg):
if need_sc1:
pagemsg("WARNING: Irregular plural calls for special case (1), but not present in decl arg for word #%s, skipping: declarg=%s, lemma=%s, infl=%s" % (
wordind, declarg, lemma, infl))
return None
else:
pagemsg("WARNING: Special case (1) present in decl arg but plural for word #%s is regular, skipping: declarg=%s, lemma=%s, infl=%s" % (
wordind, declarg, lemma, infl))
return None
arg_set[lemma_arg + 1] = declarg
if numbered_params:
numbered_params.append("or")
numbered_params.extend(arg_set)
# Now gather all params, including named ones.
params = []
params.extend((str(i+1), val) for i, val in zip(xrange(len(numbered_params)), numbered_params))
num = None
anim = None
for p in decl_template.params:
pname = unicode(p.name)
val = unicode(p.value)
if pname == "a":
anim = val
elif pname == "n":
num = val
elif pname == "notes":
params.append((pname, val))
elif pname == "title":
pagemsg("WARNING: Found explicit title= for word #%s, ignoring: lemma=%s, infl=%s, title=%s" %
(wordind, lemma, infl, val))
elif re.search("^[0-9]+$", pname):
pass
else:
keepparam = True
if pname == "loc":
if pagetitle in keep_locative:
pagemsg("Keeping locative for word #%s because page in keep_locative: loc=%s, lemma=%s, infl=%s" % (
wordind, val, lemma, infl))
else:
pagemsg("WARNING: Discarding locative for word #%s: loc=%s, lemma=%s, infl=%s" % (
wordind, val, lemma, infl))
keepparam = False
if pname == "par":
pagemsg("WARNING: Discarding partitive for word #%s: par=%s, lemma=%s, infl=%s" % (
wordind, val, lemma, infl))
keepparam = False
if pname == "voc":
pagemsg("WARNING: Discarding vocative for word #%s: voc=%s, lemma=%s, infl=%s" % (
wordind, val, lemma, infl))
keepparam = False
if keepparam:
if pname == "loc" and re.search(ur"^(на|в)\b", val, re.U):
pagemsg(u"WARNING: на or в found in loc= for word #%s, may not work in multi-word lemma: loc=%s, lemma=%s, infl=%s" %
(wordind, val, lemma, infl))
pname += str(wordind)
params.append((pname, val))
def process_page_section(index, page, section, verbose):
pagetitle = unicode(page.title())
subpagetitle = re.sub("^.*:", "", pagetitle)
def pagemsg(txt):
msg("Page %s %s: %s" % (index, pagetitle, txt))
def expand_text(tempcall):
return blib.expand_text(tempcall, pagetitle, pagemsg, verbose)
if not page.exists():
pagemsg("WARNING: Page doesn't exist, skipping")
return None
parsed = blib.parse_text(section)
noun_table_templates = []
noun_old_templates = []
for t in parsed.filter_templates():
if unicode(t.name) == "ru-decl-noun-see":
pagemsg("Found ru-decl-noun-see, skipping")
return None
for t in parsed.filter_templates():
if unicode(t.name) == "ru-noun-table":
noun_table_templates.append(t)
if unicode(t.name) == "ru-noun-old":
noun_old_templates.append(t)
if len(noun_table_templates) > 1:
pagemsg("WARNING: Found multiple ru-noun-table templates, skipping")
return None
if len(noun_old_templates) > 1:
pagemsg("WARNING: Found multiple ru-noun-old templates, skipping")
return None
if not noun_table_templates and not noun_old_templates:
return unicode(parsed), 0, 0, 0, []
for t in parsed.filter_templates():
if unicode(t.name) in ["ru-noun+", "ru-proper noun+"]:
pagemsg("Found ru-noun+ or ru-proper noun+, skipping")
return None
headword_templates = []
for t in parsed.filter_templates():
if unicode(t.name) in ["ru-noun", "ru-proper noun"]:
headword_templates.append(t)
if len(headword_templates) > 1:
pagemsg("WARNING: Found multiple headword templates, skipping")
return None
if len(headword_templates) < 1:
return unicode(parsed), 0, 0, 0, []
noun_table_template = noun_table_templates[0] if len(
noun_table_templates) == 1 else None
noun_old_template = noun_old_templates[0] if len(
noun_old_templates) == 1 else None
if noun_old_template and not noun_table_template:
noun_table_template = noun_old_template
noun_old_template = None
headword_template = headword_templates[0]
frobbed_manual_translit = []
decl_templates = [x for x in [noun_table_template, noun_old_template] if x]
if verbose:
pagemsg("Found headword template: %s" % unicode(headword_template))
pagemsg("Found decl template: %s" % unicode(noun_table_template))
if noun_old_template:
pagemsg("Found old decl template: %s" % unicode(noun_old_template))
# Retrieve headword translit and maybe transfer to decl
headword_tr = getparam(headword_template, "tr")
if headword_tr:
if verbose:
pagemsg("Found headword manual translit tr=%s" % headword_tr)
if "," in headword_tr:
pagemsg(
"WARNING: Comma in headword manual translit, skipping: %s" %
headword_tr)
return None
# Punt if multi-arg-set, can't handle yet
for decl_template in decl_templates:
for param in decl_template.params:
if not param.showkey:
val = unicode(param.value)
if val == "or":
pagemsg(
"WARNING: Manual translit and multi-decl templates, can't handle, skipping: %s"
% unicode(decl_template))
return None
if val == "-" or val == "_" or val.startswith("join:"):
pagemsg(
"WARNING: Manual translit and multi-word templates, can't handle, skipping: %s"
% unicode(decl_template))
return None
for i in xrange(2, 10):
if getparam(headword_template, "tr%s" % i):
pagemsg(
"WARNING: Headword template has translit param tr%s, can't handle, skipping: %s"
% (i, unicode(headword_template)))
return None
if runounlib.arg1_is_stress(getparam(decl_template, "1")):
lemma_arg = "2"
else:
lemma_arg = "1"
lemmaval = getparam(decl_template, lemma_arg)
if not lemmaval:
lemmaval = subpagetitle
if "//" in lemmaval:
m = re.search("^(.*?)//(.*)$", lemmaval)
if m.group(2) != headword_tr:
pagemsg(
"WARNING: Found existing manual translit in decl template %s, but doesn't match headword translit %s; skipping"
% (lemmaval, headword_tr))
return None
else:
pagemsg(
"Already found manual translit in decl template %s" %
lemmaval)
else:
lemmaval += "//" + headword_tr
orig_decl_template = unicode(decl_template)
decl_template.add(lemma_arg, lemmaval)
pagemsg("Replacing decl %s with %s" %
(orig_decl_template, unicode(decl_template)))
frobbed_manual_translit = [headword_tr]
genders = blib.fetch_param_chain(headword_template, "2", "g")
bian_replaced = 0
# Change a=bi in decl to a=ia or a=ai, depending on order of anim/inan in
# headword template
for decl_template in decl_templates:
if getparam(decl_template, "a") in ["b", "bi", "bian", "both"]:
saw_in = -1
saw_an = -1
for i, g in enumerate(genders):
if re.search(r"\bin\b", g) and saw_in < 0:
saw_in = i
if re.search(r"\ban\b", g) and saw_an < 0:
saw_an = i
if saw_in >= 0 and saw_an >= 0:
orig_decl_template = unicode(decl_template)
if saw_in < saw_an:
pagemsg("Replacing a=bi with a=ia in decl template")
decl_template.add("a", "ia")
bian_replaced = 1
else:
pagemsg("Replacing a=bi with a=ai in decl template")
decl_template.add("a", "ai")
bian_replaced = 1
pagemsg("Replacing decl %s with %s" %
(orig_decl_template, unicode(decl_template)))
generate_template = re.sub(
r"^\{\{ru-noun-old", "{{ru-generate-noun-args|old=1",
re.sub(r"^\{\{ru-noun-table", "{{ru-generate-noun-args",
unicode(noun_table_template)))
generate_result = expand_text(generate_template)
if not generate_result:
pagemsg("WARNING: Error generating noun args, skipping")
return None
args = blib.split_generate_args(generate_result)
genders = runounlib.check_old_noun_headword_forms(headword_template, args,
subpagetitle, pagemsg)
if genders == None:
return None
new_params = []
for param in noun_table_template.params:
new_params.append((param.name, param.value))
orig_headword_template = unicode(headword_template)
params_to_preserve = runounlib.fix_old_headword_params(
headword_template, new_params, genders, pagemsg)
if params_to_preserve == None:
return None
if unicode(headword_template.name) == "ru-proper noun":
# If proper noun and n is both then we need to add n=both because
# proper noun+ defaults to n=sg
if args["n"] == "b" and not getparam(headword_template, "n"):
pagemsg("Adding n=both to headword tempate")
headword_template.add("n", "both")
# Correspondingly, if n is sg then we can usually remove n=sg;
# but we need to check that the number is actually sg with n=sg
# removed because of the possibility of plurale tantum lemmas
if args["n"] == "s":
generate_template_with_ndef = generate_template.replace(
"}}", "|ndef=sg}}")
generate_template_with_ndef = re.sub(r"\|n=s[^=|{}]*", "",
generate_template_with_ndef)
generate_result = expand_text(generate_template_with_ndef)
if not generate_result:
pagemsg("WARNING: Error generating noun args, skipping")
return None
ndef_args = blib.split_generate_args(generate_result)
if ndef_args["n"] == "s":
existing_n = getparam(headword_template, "n")
if existing_n and not re.search(r"^s", existing_n):
pagemsg(
"WARNING: Something wrong: Found n=%s, not singular" %
existing_n)
else:
pagemsg("Removing n=sg from headword tempate")
rmparam(headword_template, "n")
else:
pagemsg(
"WARNING: Unable to remove n= from headword template because n=%s"
% ndef_args["n"])
headword_template.params.extend(params_to_preserve)
ru_noun_changed = 0
ru_proper_noun_changed = 0
if unicode(headword_template.name) == "ru-noun":
headword_template.name = "ru-noun+"
ru_noun_changed = 1
else:
headword_template.name = "ru-proper noun+"
ru_proper_noun_changed = 1
if unicode(noun_table_template).startswith("{{ru-noun-old"):
headword_template.add("old", "1")
pagemsg("Replacing headword %s with %s" %
(orig_headword_template, unicode(headword_template)))
return unicode(
parsed
), ru_noun_changed, ru_proper_noun_changed, bian_replaced, frobbed_manual_translit
