def process_page(page, index, parsed):
pagetitle = unicode(page.title())
def pagemsg(txt):
msg("Page %s %s: %s" % (index, pagetitle, txt))
notes = []
pagemsg("Processing")
head = None
for t in parsed.filter_templates():
origt = unicode(t)
tn = tname(t)
if tn == "be-adj":
head = getparam(t, "1")
if getparam(t, "head2"):
pagemsg("WARNING: Can't handle head2= yet in {{rfinfl}}")
head = None
elif tn == "rfinfl" and getparam(t, "1") == "be" and getparam(
t, "2") == "adjective":
if not head:
pagemsg("WARNING: Found {{rfinfl}} and don't have head: %s" %
origt)
elif " " in head:
pagemsg(
"Found {{rfinfl}} but head %s has space in it, skipping" %
head)
else:
rmparam(t, "2")
t.add("1", head)
blib.set_template_name(t, "be-adecl")
notes.append("add Belarusian adjective declension for %s" %
head)
elif tn == "be-decl-adj":
word = getparam(t, "1") + getparam(t, "2")
if belib.needs_accents(word):
pagemsg("WARNING: Word %s needs accent: %s" % (word, origt))
continue
t.add("1", word)
rmparam(t, "2")
blib.set_template_name(t, "be-adecl")
notes.append("convert {{be-decl-adj}} to {{be-adecl}}")
elif tn == "be-adj-table":
notesval = getparam(t, "notes")
if notesval:
notesval = re.sub("^: ", "", notesval)
if not re.search(r"\.\s*$", notesval):
notesval = re.sub(r"(\s*)$", r".\1", notesval)
t.add("footnote",
notesval,
before="notes",
preserve_spacing=False)
rmparam(t, "notes")
blib.set_template_name(t, "be-adecl-manual")
notes.append("convert {{be-adj-table}} to {{be-adecl-manual}}")
if origt != unicode(t):
pagemsg("Replaced %s with %s" % (origt, unicode(t)))
return unicode(parsed), notes
def process_adj_headt(t):
origt = unicode(t)
def getp(param):
return getparam(t, param)
tr = getp("tr")
head = getp("head")
if getp("1"):
pagemsg("WARNING: Has 1=%s: %s" % (getp("1"), origt))
return
must_continue = False
for param in t.params:
pn = pname(param)
if pn not in ["head", "tr"]:
pagemsg("WARNING: Unrecognized param %s=%s, skipping: %s" %
(pn, unicode(param.value), origt))
must_continue = True
break
if must_continue:
return False
del t.params[:]
if not head:
head = pagetitle
if belib.needs_accents(head):
pagemsg("WARNING: Head %s missing accents: %s" % (head, origt))
t.add("1", head)
if tr:
t.add("tr", tr)
if origt != unicode(t):
notes.append("fix up {{be-adj}} to use new param convention")
pagemsg("Replaced %s with %s" % (origt, unicode(t)))
return True
def check_multi_stressed(maxparam):
for i in xrange(1, maxparam + 1):
val = getparam(t, str(i))
vals = re.split(r",\s*", val)
for v in vals:
if be.is_multi_stressed(v):
pagemsg(
"WARNING: Param %s=%s has multiple stresses: %s" %
((str(i), val, unicode(t))))
if be.needs_accents(v):
pagemsg("WARNING: Param %s=%s has missing stress: %s" %
((str(i), val, unicode(t))))
def split_form(form):
forms = re.split(r",\s*", form.strip())
forms = [re.sub(r"^\[\[([^\[\]]*)\]\]$", r"\1", f) for f in forms]
forms = [belib.add_accent_to_o(f) for f in forms]
for f in forms:
if "[[" in f:
pagemsg("WARNING: Link in form %s: headword=%s, decl=%s" %
(f, origt, origdeclt))
if belib.needs_accents(f):
pagemsg("WARNING: Form %s missing accents: headword=%s, decl=%s" %
(f, origt, origdeclt))
forms = [f for f in forms if f != "-"]
return forms
def process_verb_headt(t):
origt = unicode(t)
def getp(param):
return getparam(t, param)
tr = getp("tr")
if getp("2"):
head = getp("1")
g = getp("2")
else:
head = getp("head")
g = getp("1") or getp("a")
pf = blib.fetch_param_chain(t, "pf", "pf")
impf = blib.fetch_param_chain(t, "impf", "impf")
must_continue = False
for param in t.params:
pn = pname(param)
if pn not in ["head", "tr", "1", "a", "2", "pf", "pf2", "pf3",
"impf", "impf2", "impf3"]:
pagemsg("WARNING: Unrecognized param %s=%s, skipping: %s" %
(pn, unicode(param.value), origt))
must_continue = True
break
if must_continue:
return False
del t.params[:]
if not head:
head = pagetitle
if belib.needs_accents(head):
pagemsg("WARNING: Head %s missing accents: %s" % (head, origt))
if not g:
pagemsg("WARNING: No aspect in verb headword: %s" % origt)
g = "?"
t.add("1", head)
if tr:
t.add("tr", tr)
t.add("2", g)
blib.set_param_chain(t, pf, "pf", "pf")
blib.set_param_chain(t, impf, "impf", "impf")
if origt != unicode(t):
notes.append("fix up {{be-verb}} to use new param convention")
pagemsg("Replaced %s with %s" % (origt, unicode(t)))
return True