def sourcename(url, cat1=None, cat2=None, cat3=None, d=True):
html = getPage(url, "page1.html")
os.remove('pages/page1.html')
title = html.select('h1')[0].text
first_para = html.select('.story-content > p:nth-of-type(1)')[0].text
second_para = html.select('.story-content > p:nth-of-type(2)')[0].text
try:
image = html.select('.image > img')[0].get('src')
except Exception:
image = None
wp = Client('http://www.domain.com/xml-rpc.php', 'username', 'password')
if image:
filename = 'http://www.livemint.com' + image
path = os.getcwd() + "\\00000001.jpg"
f = open(path, 'wb')
f.write(urllib.urlopen(filename).read())
f.close()
# prepare metadata
data = {
'name': 'picture.jpeg',
'type': 'image/jpeg', # mimetype
}
with open(path, 'rb') as img:
data['bits'] = xmlrpc_client.Binary(img.read())
response = wp.call(media.UploadFile(data))
post = WordPressPost()
post.title = title
post.user = 14
post.post_type = "post"
post.content = first_para + '\n' + '\n' + second_para
if d:
post.post_status = "draft"
else:
post.post_status = "publish"
if image:
attachment_id = response['id']
post.thumbnail = attachment_id
post.custom_fields = []
post.custom_fields.append({
'key': 'custom_source_url',
'value': url
})
if cat1:
cat1 = wp.call(taxonomies.GetTerm('category', cat1))
post.terms.append(cat1)
if cat2:
cat2 = wp.call(taxonomies.GetTerm('category', cat2))
post.terms.append(cat2)
if cat3:
cat3 = wp.call(taxonomies.GetTerm('category', cat3))
post.terms.append(cat3)
addpost = wp.call(posts.NewPost(post))
def process_item(self, item, spider):
wp = Client('https://www.along.party/xmlrpc.php', '', '')
post = WordPressPost()
post.title = item['title']
post.user = item['author']
post.link = item['url']
# post.date = item['publish_time']
post.content = item['body']
post.content = u"%s \n 本文转载自 <a href='%s'> %s</a> " % (
item['body'], item['url'], item['title'])
post.post_status = 'publish'
post.terms_names = {'post_tag': 'Python', 'category': 'Python'}
wp.call(NewPost(post))
def create_wordpress_draft(publish_target, title, html, tags):
post = WordPressPost()
today = datetime.date.today()
post.title = title
post.content = html
client = Client( publish_target["url"] + "/xmlrpc.php", publish_target["username"], publish_target["password"])
category = client.call(taxonomies.GetTerm('category', publish_target["default_category_id"]))
post.terms.append(category)
post.user = publish_target["default_user_id"]
post.terms_names = {'post_tag': tags}
post.comment_status = 'open'
post.id = client.call(posts.NewPost(post))
return post
def sourcename(url, cat1=None, cat2=None, cat3=None, d=True):
html = getPage(url, "page1.html")
os.remove('pages/page1.html')
title = html.select('h1')[0].text
first_para = html.select('.story-content > p:nth-of-type(1)')[0].text
second_para = html.select('.story-content > p:nth-of-type(2)')[0].text
try:
image = html.select('.image > img')[0].get('src')
except Exception:
image = None
wp = Client('http://www.domain.com/xml-rpc.php', 'username', 'password')
if image:
filename = 'http://www.livemint.com' + image
path = os.getcwd() + "\\00000001.jpg"
f = open(path, 'wb')
f.write(urllib.urlopen(filename).read())
f.close()
# prepare metadata
data = {
'name': 'picture.jpeg',
'type': 'image/jpeg', # mimetype
}
with open(path, 'rb') as img:
data['bits'] = xmlrpc_client.Binary(img.read())
response = wp.call(media.UploadFile(data))
post = WordPressPost()
post.title = title
post.user = 14
post.post_type = "post"
post.content = first_para + '\n' + '\n' + second_para
if d:
post.post_status = "draft"
else:
post.post_status = "publish"
if image:
attachment_id = response['id']
post.thumbnail = attachment_id
post.custom_fields = []
post.custom_fields.append({'key': 'custom_source_url', 'value': url})
if cat1:
cat1 = wp.call(taxonomies.GetTerm('category', cat1))
post.terms.append(cat1)
if cat2:
cat2 = wp.call(taxonomies.GetTerm('category', cat2))
post.terms.append(cat2)
if cat3:
cat3 = wp.call(taxonomies.GetTerm('category', cat3))
post.terms.append(cat3)
addpost = wp.call(posts.NewPost(post))
def create_wordpress_draft(publish_target, title, html, tags):
post = WordPressPost()
today = datetime.date.today()
post.title = title
post.content = html
client = Client(publish_target["url"] + "/xmlrpc.php",
publish_target["username"], publish_target["password"])
category = client.call(
taxonomies.GetTerm('category', publish_target["default_category_id"]))
post.terms.append(category)
post.user = publish_target["default_user_id"]
post.terms_names = {'post_tag': tags}
post.comment_status = 'open'
post.id = client.call(posts.NewPost(post))
return post
def publish(pic, title, content):
'''
publish a post and set to open to comment and ping(trackback)
'''
attachment = upload(pic)
wp_client = Client(rpc_service_url, user, password)
post = WordPressPost()
post.title = title
post.content = "%s\n\n<a href='%s'><img class='alignnone size-full wp-image-%s' alt='%s' src='%s' /></a>" % (content, attachment['url'], attachment['id'], attachment['file'], attachment['url'])
#post.tags='test, test2'
#post.categories=['pet','picture']
post.thumbnail = attachment['id']
#change status to publish
post.id = wp_client.call(posts.NewPost(post))
post.post_status = 'publish'
post.comment_status = 'open'
post.ping_status = 'open'
post.user = account_ids[r.randint(0, len(account_ids)-1)]
wp_client.call(posts.EditPost(post.id, post))
return post.id
def _add_blogpost(post_data):
'''
Adds a blog post parsed from blog.blogpost
Model published status codes are as follows: PUBLISHED = 1, DRAFT = 2, DELETED = 3
:param post_data: Python dict of post structure. See modelsamples/blogpost_sample.txt for structure.
:return:
'''
post = WordPressPost()
post.post_type = 'news'
post.title = post_data['fields']['title']
post.content = post_data['fields']['content']
post.date = _return_datetime(post_data['fields']['publish_date'])
post.date_modified = _return_datetime(post_data['fields']['updated'])
post.slug = post_data['fields']['slug']
if post_data['fields']['status'] == 2:
post.post_status = 'publish'
# Assign Author
if post_data['fields']['user']:
wp_userid = _get_wordpress_user_id_by_email(
_get_django_user_email_by_id(post_data['fields']['user']))
if wp_userid:
post.user = wp_userid
# TODO set catagories and tags to proper taxonomy
# post.terms_names = {
# 'category': ['Blogpost']
# }
# if post_data['fields']['categories']:
# categories = []
# for category in dp.get_content(DJANGO_DATA, 'blog.blogcategory'):
# if category['pk'] in post_data['fields']['categories']:
# categories.append(category['fields']['title'])
# post.terms_names['post_tag'] = categories
try:
if post_data['fields']['status'] != 3:
post.id = API.call(NewPost(post))
print("created post", post.id, post.title)
except Fault as err:
pprint(post)
print(err.faultCode, err.faultString)
def entry_to_wppost(entry, client):
post = WordPressPost()
# Convert entry values to Post value
post.user = get_author_by_display_name(entry.author, client)
post.date = entry.published_parsed
post.post_status = "draft"
post.title = entry.title
post.content = entry.content[0].value
post.excerpt = entry.summary
post.link = entry.link
# There is some given tags
if len(entry.tags):
entry.tags = [t.term for t in entry.tags]
# Add category (with the first tag)
post.terms_names = {
'category': entry.tags[0:1],
'post_tag': [],
}
# Add tags
if len(entry.tags) > 1: post.terms_names['post_tag'] = entry.tags[1:]
return post
def _add_application(app_data):
'''
Adds an application parse from apps.application
:param app_data: Python dict of application structure. See modelsamples/application_sample.txt
:return:
'''
post = WordPressPost()
post.post_type = 'application'
post.title = app_data['fields']['name']
post.slug = app_data['fields']['slug']
post.content = _format_app_content(app_data)
post.date = _return_datetime(app_data['fields']['created'])
post.date_modified = _return_datetime(app_data['fields']['updated'])
# TODO assign to proper taxonomies once those are in.
# Assign Author
if app_data['fields']['owner']:
wp_userid = _get_wordpress_user_id_by_email(
_get_django_user_email_by_id(app_data['fields']['owner']))
if wp_userid:
post.user = wp_userid
# Assign Categories and Tags
post.terms_names = _parse_taxonomies(app_data)
# Put the previous page url in a custom field.
legacy_url = "https://www.us-ignite.org/apps/%s/" % (
app_data['fields']['slug'])
post.custom_fields = [{'key': 'legacy_url', 'value': legacy_url}]
# Set publish status and push to site
if app_data['fields']['status'] == 1:
post.post_status = 'publish'
try:
if app_data['fields']['status'] != 3:
post.id = API.call(NewPost(post))
except Fault as err:
pprint(err.faultString)
def create_blog_post(message):
pprint(message['username'] + str(message['message_id']))
username = message['username']
if username in username_converts: username = username_converts[username]
url = "http://" + username + ".plataforma.cc/xmlrpc.php"
client = Client(url, admin_user, admin_pass)
post = WordPressPost()
if not username in username_ids:
blogusers = client.call(users.GetUsers())
username_ids[username] = next(u for u in blogusers
if u.username == username).id
post.user = username_ids[username]
post.title = ''
post.content = message['text']
post.post_status = 'publish'
if len(message['photo']):
if post.content.find('[plataformacc_gallery]') == -1:
post.content = post.content + '[plataformacc_gallery]'
if message['caption']:
post.content = post.content + message[
'caption'] + '[/plataformacc_gallery]'
elif not len(message['text']):
data = dict(id=str(message['message_id'][-1]),
channel=message['username'])
post.content = Template(
'[telegram_embed id="$id" channel="$channel"]').substitute(data)
post_id = client.call(GetPostByTelegramID(message['message_id']))
if (post_id == -1):
print('multiple, wont update')
return
if (post_id == -2):
print('has images, wont update')
return
if post_id and not message['edit']:
print('already exists')
return
if not post_id:
print('adding new')
post_id = client.call(posts.NewPost(post))
post.custom_fields = []
for telegram_id in message['message_id']:
post.custom_fields.append({
'key': 'telegram_id',
'value': telegram_id
})
print('custom field', {'key': 'telegram_id', 'value': telegram_id})
if len(message['photo']):
for photo in message['photo']:
mimetype = mime.from_file(photo)
ext = mimetype.split('/')[-1]
data = {
'name': photo + '.' + ext,
'type': mimetype, # mimetype
'post_id': post_id,
}
with open(photo, 'rb') as img:
data['bits'] = xmlrpc_client.Binary(img.read())
response = client.call(media.UploadFile(data))
post.wp_post_thumbnail = response['id']
os.remove(photo)
print('calling edit for photos', post_id)
print(post)
print('Updating', post_id, post)
client.call(posts.EditPost(post_id, post))
continue
if not tag: continue
if getImageSrc(tag) != '':
img_id = imageUpload(getImageSrc(tag), alias, i=i)
else:
continue
img_tag = '<!-- wp:image {"id":%s,"sizeSlug":"large"} --><figure class="wp-block-image size-large"><img src="%s" alt="" class="wp-image-%s"/></figure><!-- /wp:image -->' % (
img_id['id'], img_id['link'], img_id['id'])
content += img_tag
i += 1
continue
try:
content += tag.prettify()
except Exception as e:
print('Exception: %s' % e)
img_id = imageUpload(image, alias)
post = WordPressPost()
post.title = soup.find(class_='article__title').text
post.content = content
post.mime_type = "text/html"
post.user = 2 # WP Author ID
post.link = alias
post.post_type = "post"
post.thumbnail = img_id['id']
post.post_status = "publish" # "draft"
post_id = client.call(posts.NewPost(post))
print("Post: %s (%s) Done!" % (post_id, alias))
