def section_action(self, page, lang, section_content):
super(ReplaceOldEmptyMorphoRu, self).section_action(page, lang, section_content)
title = page.title
if lang != 'ru':
return
m = re.search(u'(=== *Морфологические и синтаксические свойства *=== *(.*?)\n)=',
section_content, re.UNICODE | re.DOTALL)
if not m:
return
old_body = m.group(2)
p = re.compile(
u"""(\{\{
(?P<title>морфо\s*) # заголовок
\|(?P<params>[^{}]+?(\{\{[^{}]+?\}\})?[^{}]+?) # параметры
\}\})""",
flags=re.UNICODE + re.DOTALL + re.VERBOSE)
parts = p.findall(old_body)
new_body = old_body
for part in parts:
old_params = part[2]
params = old_params
params = re.sub(u'(префд|прист|суфф-?|корень|инф|соед|оконч-?|интер|суффд|частица|постфикс)\d*=', '', params)
params = re.sub(u'\|', '', params).strip()
if not params:
# append_dict_list(self.old_empty_morpho, title, part[0])
new_body = new_body.replace(part[0], u'{{морфо-ru|}}')
# params = part[2].strip().split('|')
# for value in params:
# value = value.strip()
# if '=' in value:
# name, value = value.split('=', 1)
# if '-' in value:
# append_dict_list(self.morpho_wrong_hyphens, title, part[0])
# break
if old_body != new_body:
# print '=' * 100
# print old_body
# print '-' * 100
# print new_body
# print '=' * 100
old_section_content = section_content
new_section_content = section_content.replace(old_body, new_body)
try:
old_content = get_wiki_page_content(title)
except NoPage:
send_wiki_mail(subject="pywikibot.exceptions.NoPage: " + title,
message=title)
return
except IsRedirectPage:
send_wiki_mail(
subject="pywikibot.exceptions.IsRedirectPage: " + title,
message=title)
return
new_content = old_content.replace(old_section_content, new_section_content)
if new_content != old_content:
desc = u"Замена пустого {{морфо}} на пустой {{морфо-ru}}"
save_wiki_page(title, new_content, desc)
def handle(self, *args, **options):
if not self.name:
raise NotImplementedError()
append_file('files/timing', u'%s @ %s' % (self.current_datetime, self.name))
print '@', self.current_datetime, '-', "Starting wikt command:", \
self.name
is_started = os.path.exists(self.state_filename)
if is_started:
print '# Command already started'
if not self.skip_mail:
send_wiki_mail(
subject="Wikt command already started: " + self.name,
message=open(self.state_filename).read()
)
return
save_file(self.state_filename,
"Started at " + self.current_datetime)
self.wikt_handle(*args, **options)
os.remove(self.state_filename)
append_file('files/timing', u'%s . %s' % (self.current_datetime, self.name))
def check_for_start(self):
if not self.last_action_filename:
raise NotImplementedError()
last_action = datetime.now() # default value for email
delta = 'None'
if os.path.exists(self.state_filename):
content = load_file(self.state_filename)
last_action = datetime.strptime(content, self.date_format)
delta = datetime.now() - last_action
if delta < timedelta(minutes=3) and not self.skip_mail:
print "Don't start yet: " + self.last_action_filename
send_wiki_mail(
subject="Don't start yet: " + self.last_action_filename,
message="%s // %s" % (last_action, delta)
)
raise Exception("Don't start yet!!")
print "Successfully started: " + self.last_action_filename
send_wiki_mail(
subject="Successfully started: " + self.last_action_filename,
message="%s // %s" % (last_action, delta)
)
append_file('files/timing',
u'%s > %s' % (self.current_datetime_Kyiv,
self.last_action_filename))
def section_action(self, page, lang, section_content):
super(ReplaceOldMorphoRu, self).section_action(page, lang, section_content)
title = page.title
if lang != 'ru':
return
m = re.search(u'(=== *Морфологические и синтаксические свойства *=== *(.*?)\n)=',
section_content, re.UNICODE | re.DOTALL)
if not m:
return
old_body = m.group(2)
p = re.compile(
u"""(\{\{
(?P<title>морфо\s*) # заголовок
\|(?P<params>[^{}]+?(\{\{[^{}]+?\}\})?[^{}]+?) # параметры
\}\})""",
flags=re.UNICODE + re.DOTALL + re.VERBOSE)
parts = p.findall(old_body)
new_body = old_body
# print title.encode('utf-8')
for part in parts:
old_params = part[2]
is_title = old_params
is_title = re.sub(u'(префд|прист|суфф-?|корень|инф|соед|оконч-?|интер|суффд|частица|постфикс|источник)\d*=', '', is_title)
is_title = re.sub(u'\|', '', is_title).strip()
if is_title == title or is_title == title + u'т' or is_title == title + u'к':
# print
# print title
# print part[0]
# {{морфо||адрес|ова|суфф2=нн|ый}}
try:
sres = get_sorted_list(get_old_list(old_params))
except ValueError as e:
print title.encode('utf-8')
# sent = False
# try:
# send_mail(subject="morpho exception in: " + title,
# message=title + "\n\n" + repr(e),
# from_email=settings.DEFAULT_FROM_EMAIL,
# recipient_list=['*****@*****.**'])
# sent = True
# except:
# pass
# if sent:
# return
# else:
# raise
raise
# for key, value in sres:
# print "- %s = '%s'" % (key, value)
res = get_new_list(sres)
ress = '|'.join(res)
# print ress
new_body = new_body.replace(part[0], u'{{морфо-ru|%s}}' % ress)
# params = part[2].strip().split('|')
# for value in params:
# value = value.strip()
# if '=' in value:
# name, value = value.split('=', 1)
# if '-' in value:
# append_dict_list(self.morpho_wrong_hyphens, title, part[0])
# break
elif not is_title:
new_body = new_body.replace(part[0], u'{{морфо-ru|}}')
if old_body != new_body:
# print '=' * 100
# print old_body
# print '-' * 100
# print new_body
# print '=' * 100
old_section_content = section_content
new_section_content = section_content.replace(old_body, new_body)
try:
old_content = get_wiki_page_content(title)
except NoPage:
send_wiki_mail(subject="pywikibot.exceptions.NoPage: " + title,
message=title)
return
except IsRedirectPage:
send_wiki_mail(
subject="pywikibot.exceptions.IsRedirectPage: " + title,
message=title
)
return
new_content = old_content.replace(old_section_content, new_section_content)
if new_content != old_content:
desc = u"Замена {{морфо}} на {{морфо-ru}}"
# self.n_step += 1
# if self.n_step >= 100:
# self.stop()
new_content = default_wikifier(new_content)
save_wiki_page(title, new_content, desc)
# todo: create special function for this:
d = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
append_file(os.path.join(settings.FILES_PATH, 'reports',
self.report_name),
"%s: %s" % (d, title),
encode='utf-8')
