def add_model_specific_args(arg_parser, root_dir):
BaseTransformer.add_model_specific_args(arg_parser, root_dir)
arg_parser.add_argument(
"--data_dir",
default=None,
type=str,
required=True,
help=
"The input data dir. Should contain the dataset files for the task.",
)
arg_parser.add_argument("--dataset", required=True, type=str)
return arg_parser
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
num_comments = obj.get('fields', {}).get('comment', {}).get('total', 0),
com_list = obj.get('fields', {}).get('comment', {}).get('comments', [])
# TODO what is with that [0]??
if num_comments[0] > 0 and len(com_list) == num_comments[0]:
author = com_list[num_comments[0]-1]['author']
last_com_name = author['displayName']
last_com_avatar = author['avatarUrls']['32x32']
last_com_un = author['name']
last_com_body = com_list[num_comments[0]-1]['body']
last_comment = {
'name': last_com_name,
'avatar': last_com_avatar,
'user': last_com_un,
'body': last_com_body
}
else:
last_comment = None
base.update({
'id': obj['_id'],
'status': obj.get('fields', {}).get('status', None),
'comments': num_comments[0],
'url': self.baseurl + obj['_id'],
'title': obj['fields'].get('summary', ''),
'snippet': obj['fields'].get('description', '')
})
if last_comment:
base.update({'last_comment': last_comment})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
num_comments = obj.get('fields', {}).get('comment',
{}).get('total', 0),
com_list = obj.get('fields', {}).get('comment', {}).get('comments', [])
# TODO what is with that [0]??
if num_comments[0] > 0 and len(com_list) == num_comments[0]:
author = com_list[num_comments[0] - 1]['author']
last_com_name = author['displayName']
last_com_avatar = author['avatarUrls']['32x32']
last_com_un = author['name']
last_com_body = com_list[num_comments[0] - 1]['body']
last_comment = {
'name': last_com_name,
'avatar': last_com_avatar,
'user': last_com_un,
'body': last_com_body
}
else:
last_comment = None
base.update({
'id': obj['_id'],
'status': obj.get('fields', {}).get('status', None),
'comments': num_comments[0],
'url': self.baseurl + obj['_id'],
'title': obj['fields'].get('summary', ''),
'snippet': obj['fields'].get('description', '')
})
if last_comment:
base.update({'last_comment': last_comment})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'url': obj['url'],
'title': obj['title'],
'snippet': obj['text']
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'url': obj['url'],
'title': obj['title'],
'snippet': obj['text']
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'url': self.baseurl + obj['subject'] + '&btnI',
'title': obj['subject'],
'snippet': obj['body'],
'group': obj['group'],
'sender': obj['sender'],
'from': obj['from']
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'title': obj['title'],
'snippet': obj['body'],
'url': obj['url'],
'space': obj['subsource'],
'source': 'confluence'
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'title': obj['title'],
'snippet': obj['body'],
'url': obj['url'],
'space': obj['subsource'],
'source': 'confluence'
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'score': obj['score'],
'answered': obj['is_answered'],
'answers': obj['answer_count'],
'url': obj['link'],
'title': obj['title'],
'snippet': re.sub('<[^<]+?>', '', obj['body'])
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['_id'],
'score': obj['score'],
'answered': obj['is_answered'],
'answers': obj['answer_count'],
'url': obj['link'],
'title': obj['title'],
'snippet': re.sub('<[^<]+?>', '', obj['body'])
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id': obj['crowd_id'],
'full_name': obj['full_name'],
'url': corp_url() + 'employees/' + obj['crowd_id'],
'team': obj['team']['name'],
'office': obj['office'],
'email': obj['primary_email'],
'github': obj['github'],
'title': obj['title'],
'md5': md5.new(obj['primary_email'].strip().lower()).hexdigest()
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
committer = obj['commit']['committer']
commit_msg_header = obj['commit']['message']
commit_msg_body = None
newline_pos = commit_msg_header.find('\n')
if newline_pos >= 0:
commit_msg_body = commit_msg_header[newline_pos:].strip()
commit_msg_header = commit_msg_header[0:newline_pos].strip()
base.update({
'url': obj['html_url'],
'committer': obj['commit']['committer']['name'],
'avatar': committer.get('avatar_url', ''),
'date': obj['commit']['committer']['date'],
'commit_msg': commit_msg_header,
'commit_body': commit_msg_body,
'repo_name': obj['repo']['full_name']
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
committer = obj['commit']['committer']
commit_msg_header = obj['commit']['message']
commit_msg_body = None
newline_pos = commit_msg_header.find('\n')
if newline_pos >= 0:
commit_msg_body = commit_msg_header[newline_pos:].strip()
commit_msg_header = commit_msg_header[0:newline_pos].strip()
base.update({
'url': obj['html_url'],
'committer': obj['commit']['committer']['name'],
'avatar': committer.get('avatar_url', ''),
'date': obj['commit']['committer']['date'],
'commit_msg': commit_msg_header,
'commit_body': commit_msg_body,
'repo_name': obj['repo']['full_name']
})
return base
def transform(self, obj):
base = BaseTransformer.transform(self, obj)
base.update({
'id':
obj['crowd_id'],
'full_name':
obj['full_name'],
'url':
corp_url() + 'employees/' + obj['crowd_id'],
'team':
obj['team']['name'],
'office':
obj['office'],
'email':
obj['primary_email'],
'github':
obj['github'],
'title':
obj['title'],
'md5':
md5.new(obj['primary_email'].strip().lower()).hexdigest()
})
return base
def __init__(self):
BaseTransformer.__init__(self)
self.baseurl = 'http://jira.mongodb.org/browse/'
def __init__(self):
BaseTransformer.__init__(self)
def __init__(self):
BaseTransformer.__init__(self)
def __init__(self):
BaseTransformer.__init__(self)
self.baseurl = 'http://jira.mongodb.org/browse/'
def __init__(self):
BaseTransformer.__init__(self)
self.baseurl = "http://www.google.com/search?q=site:groups.google.com "
