def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
directory = base_path + "/" + login
files = os.listdir(directory)
for file in files:
file_path = directory + "/" + file
text = base.get_info_from_file(file_path)
if text is False:
logging.warn("file not existed: " + file_path)
else:
obj = json.loads(text)
logging.info("read file: " + file_path)
count = 1
for edge in obj["data"]["user"][
"sponsorshipsAsMaintainer"]["edges"]:
if edge["node"]["privacyLevel"] == "PRIVATE":
cur.execute(
"insert into github_sponsorships_as_maintainer "
"(login, flag, created_at) "
"values (%s, %s, %s)",
(obj["data"]["user"]["login"], base.flag2,
base.time_handler(
edge["node"]["createdAt"])))
else:
if "company" in edge["node"]["sponsorEntity"]:
flag = base.flag0
else:
flag = base.flag1
cur.execute(
"insert into github_sponsorships_as_maintainer "
"(login, sponsor_login, flag, created_at) "
"values (%s, %s, %s, %s)",
(obj["data"]["user"]["login"],
edge["node"]["sponsorEntity"]["login"],
flag,
base.time_handler(
edge["node"]["createdAt"])))
db.commit()
logging.info("the " + str(count) +
"th record in file: " + file_path)
count += 1
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
def run(self):
while not self.q.empty():
work = self.q.get()
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
file = base_path + "/" + login + ".json"
text = base.get_info_from_file(file)
if text is False:
logging.warn("file not existed: " + file)
else:
obj = json.loads(text)
logging.info("writing login data: " + login)
if obj["data"]["user"]["hasSponsorsListing"] is True:
cur.execute(
"insert into github_user "
"(database_id, login, name, email,spon_maintainer_count,"
" spon_sponsor_count, created_at, updated_at, has_sponsors_listing) "
"values (%s, %s, %s, %s, %s, %s, %s, %s, %s)",
(obj["data"]["user"]["databaseId"],
obj["data"]["user"]["login"],
obj["data"]["user"]["name"],
obj["data"]["user"]["email"], obj["data"]["user"]
["sponsorshipsAsMaintainer"]["totalCount"],
obj["data"]["user"]["sponsorshipsAsSponsor"]
["totalCount"],
base.time_handler(obj["data"]["user"]["createdAt"]),
base.time_handler(
obj["data"]["user"]["updatedAt"]), "1"))
else:
cur.execute(
"insert into github_user "
"(database_id, login, name, email,spon_maintainer_count,"
" spon_sponsor_count, created_at, updated_at, has_sponsors_listing) "
"values (%s, %s, %s, %s, %s, %s, %s, %s, %s)",
(obj["data"]["user"]["databaseId"],
obj["data"]["user"]["login"],
obj["data"]["user"]["name"],
obj["data"]["user"]["email"], obj["data"]["user"]
["sponsorshipsAsMaintainer"]["totalCount"],
obj["data"]["user"]["sponsorshipsAsSponsor"]
["totalCount"],
base.time_handler(obj["data"]["user"]["createdAt"]),
base.time_handler(
obj["data"]["user"]["updatedAt"]), "0"))
db.commit()
logging.info(login +
" ~~~~~~~~~ data commit into dababase success!!")
self.q.task_done()
cur.close()
db.close()
def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
file = base_path + "/" + login + ".json"
text = base.get_info_from_file(file)
if text is False:
logging.warn("file not existed: " + file)
else:
obj = json.loads(text)
if obj["data"]["user"]["sponsorsListing"] is not None:
logging.info(login + " ~~~~~~~~~ has " +
str(obj["data"]["user"]["sponsorsListing"]
["tiers"]["totalCount"]) + " tiers")
count = 1
for edge in obj["data"]["user"]["sponsorsListing"][
"tiers"]["edges"]:
cur.execute(
"insert into github_sponsor_listing_tiers "
"(login, slug, monthly_price_in_cents, monthly_price_in_dollars, name, created_at, updated_at, description) "
"values (%s, %s, %s, %s, %s, %s, %s, %s)",
(obj["data"]["user"]["login"], obj["data"]
["user"]["sponsorsListing"]["slug"],
edge["node"]["monthlyPriceInCents"],
edge["node"]["monthlyPriceInDollars"],
edge["node"]["name"],
base.time_handler(edge["node"]["createdAt"]),
base.time_handler(edge["node"]["updatedAt"]),
edge["node"]["description"]))
db.commit()
# logging.info("the " + str(count) + "th tier data commit into dababase success!!")
count += 1
else:
logging.warn("login: "******" don't have sponsor_listing")
logging.warn("sponsor_listing: " + str(obj))
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
logging.error(e)
def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
directory = base_path + "/" + login
files = base.read_all_filename_in_directory(directory)
for file in files:
text = base.get_info_from_file(file)
if text is False:
logging.warn("file not existed: " + file)
else:
obj = json.loads(text)
logging.info("read file: " + file)
count = 1
if "edges" not in obj["data"]["user"][
"commitComments"]:
continue
for node in obj["data"]["user"]["commitComments"][
"edges"]:
logging.info("the " + str(count) +
"th record in file: " + file)
if node["node"]["commit"] is not None:
oid = node["node"]["commit"]["oid"]
else:
oid = ""
cur.execute(
"insert into github_commit_comment "
"(comm_database_id, login, created_at, updated_at, body, commit_oid) "
"values (%s, %s, %s, %s, %s, %s)",
(node["node"]["databaseId"],
obj["data"]["user"]["login"],
base.time_handler(node["node"]["createdAt"]),
base.time_handler(node["node"]["updatedAt"]),
node["node"]["body"], oid))
db.commit()
count += 1
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
return
def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
file = base_path + "/" + login + ".json"
text = base.get_info_from_file(file)
if text is False:
logging.warn("file not existed: " + file)
else:
obj = json.loads(text)
if obj["data"]["user"]["sponsorsListing"] is None:
logging.info("user: "******" don't create sponsors")
else:
cur.execute(
"SELECT * FROM github_sponsor_listing WHERE login='******'")
items = cur.fetchall()
if len(items) == 1:
logging.info("user: "******" had been inserted into database!")
else:
cur.execute(
"insert into github_sponsor_listing "
"(login, slug, name, tiers_total_count, created_at, short_description) "
"values (%s, %s, %s, %s, %s, %s)",
(obj["data"]["user"]["login"], obj["data"]
["user"]["sponsorsListing"]["slug"],
obj["data"]["user"]["sponsorsListing"]
["name"], obj["data"]["user"]
["sponsorsListing"]["tiers"]["totalCount"],
base.time_handler(
obj["data"]["user"]["sponsorsListing"]
["createdAt"]), obj["data"]["user"]
["sponsorsListing"]["shortDescription"]))
db.commit()
logging.info(
login +
" ~~~~~~~~~ data commit into dababase success!!"
)
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
return
def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
directory = base_path + "/" + login
files = base.read_all_filename_in_directory(directory)
for file in files:
text = base.get_info_from_file(file)
if text is False:
logging.warn("file not existed: " + file)
else:
obj = json.loads(text)
logging.info("read file: " + file)
count = 1
for node in obj["data"]["user"][
"contributionsCollection"][
"pullRequestReviewContributions"]["edges"]:
try: # maybe happen duplicate key when insert data
cur.execute(
"insert ignore into github_user_pr_review "
"(pr_database_id, login, created_at, body) "
"values (%s, %s, %s, %s)",
(node["node"]["pullRequestReview"]
["databaseId"], node["node"]
["pullRequestReview"]["author"]["login"],
base.time_handler(
node["node"]["pullRequestReview"]
["createdAt"]), node["node"]
["pullRequestReview"]["body"]))
db.commit()
# logging.info("the " + str(count) + "th record in file: " + file)
except Exception as e:
logging.error(e)
count += 1
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
return
def run(self):
# get db connection
db = connectMysqlDB(config, autocommit=False)
cur = db.cursor()
while True:
try:
print ""
work = self.q.get(timeout=0)
print "the number of work in queue: " + str(self.q.qsize())
number = work["number"]
owner = work["owner"]
repo = work["repo"]
page = 1
sum = 0 # sum of inserted db
# every comments has several pages of reaction
while True:
# get a suitable token and combine header
github_token = get_token()
headers = {
'User-Agent':
'Mozilla/5.0',
'Authorization':
'token ' + github_token,
'Content-Type':
'application/json',
'method':
'GET',
'Accept':
'application/vnd.github.squirrel-girl-preview+json'
}
# print "headers is: " + str(headers)
# combine url
url = "https://api.github.com/repos/" + owner + "/" + repo + "/issues" + "/" + str(
number) + "/reactions"
url = url + "?page=" + str(page)
print "url is: " + url
try:
# request data and parse response
req = urllib2.Request(url=url, headers=headers)
response = urllib2.urlopen(req)
result = json.loads(response.read().decode("utf-8"))
# print result
length = len(result)
sum += length
if length == 0:
print "finish, comment " + str(
number) + " has reactions: " + str(sum)
self.q.task_done()
break
# write file
json_str = json.dumps(result)
# print "json format data: " + json_str
filename = base_path + "/" + owner + "&" + repo + "/" + str(
number) + "/" + str(page) + ".json"
flag = base.generate_file(filename, json_str)
if flag is True:
print "create file successfully: " + filename
elif flag is False:
print "file is already existed: " + filename
else:
print "create file failed: " + flag + " filename: " + filename
continue
page += 1 # page++
# handle response json data
num = 0
while num < length:
insert_dict = {}
if "id" not in result[num]:
insert_dict["id"] = None
else:
insert_dict["id"] = result[num]["id"]
if "login" not in result[num]["user"]:
insert_dict["user_login"] = None
else:
insert_dict["user_login"] = result[num][
"user"]["login"]
if "created_at" not in result[num]:
insert_dict["created_at"] = None
else:
insert_dict["created_at"] = result[num][
"created_at"]
if "content" not in result[num]:
insert_dict["content"] = None
else:
insert_dict["content"] = result[num]["content"]
print "insert info: " + str(insert_dict)
# insert data to database table
try:
if insert_dict is not None:
cur.execute(
"insert into github_reaction "
"(id, number, user_login, owner_login, repo, created_at, flag, content) "
"values (%s, %s, %s, %s, %s, %s, %s, %s)",
(insert_dict["id"], number,
insert_dict["user_login"], owner,
repo,
base.time_handler(
insert_dict["created_at"]), 1,
insert_dict["content"]))
db.commit()
except Exception as e:
print str(e)
num += 1
except Exception as e:
print str(e) + " error with this page: " + url
# if e.code == 403:
# break
# if e.code != 404:
# # mainly 403, sometimes 503
# # token rate limit
# self.q.put(work) # put into the queue again
# sleep_time_tokens[github_token] = time.time() # set sleep time for that token
# insert_dict = None
# else:
# insert_dict["body"] = "404 error"
# insert_dict["created_at"] = None
# insert_dict["updated_at"] = None
else:
pass # 403... error
except Queue.Empty:
cur.close()
db.close()
return
except Exception as e:
print str(
e
) + "qiubing" # unexpected error, don't interrupt the program
def run(self):
while not self.q.empty():
work = self.q.get(timeout=0)
logging.info("the number of work in queue: " + str(self.q.qsize()))
login = work["login"]
# get db connection
db = base.connectMysqlDB(config, autocommit=False)
cur = db.cursor()
# read data from file
try:
directory = base_path + "/" + login
files = os.listdir(directory)
for file in files:
file_path = directory + "/" + file
text = base.get_info_from_file(file_path)
if text is False:
logging.warn("file not existed: " + file_path)
continue
obj = json.loads(text)
print "read file: " + file_path
count = 1
# github user 接受了打赏，但是没有打赏过别人。
# 之所以将这部分数据写入 github_sponsorships_as_sponsor 表中，是为了做筛选
if len(obj["data"]["user"]["sponsorshipsAsSponsor"]
["edges"]) == 0:
logging.warn("the user " + login +
" doesn't sponsor others")
cur.execute(
"insert into github_sponsorships_as_sponsor "
"(login, sponsor_login, flag) "
"values (%s, %s, %s)",
(login, login, str(base.flag4)))
db.commit()
continue
for edge in obj["data"]["user"]["sponsorshipsAsSponsor"][
"edges"]:
if edge["node"]["privacyLevel"] == "PRIVATE":
logging.info("the " + str(count) +
"th record is private in file: " +
file_path)
count += 1
continue
else:
slug = edge["node"]["sponsorable"][
"sponsorsListing"]["slug"].split("-")[1]
cur.execute(
"insert into github_sponsorships_as_sponsor "
"(login, slug, sponsor_login, flag, created_at) "
"values (%s, %s, %s, %s, %s)",
(slug, edge["node"]["sponsorable"]
["sponsorsListing"]["slug"],
obj["data"]["user"]["login"], str(3),
base.time_handler(edge["node"]["createdAt"])))
db.commit()
logging.info("the " + str(count) +
"th record in file: " + file_path)
count += 1
self.q.task_done()
cur.close()
db.close()
except Exception as e:
logging.fatal(e)
def run(self):
work = self.q.get(timeout=0)
print "the number of work in queue: " + str(self.q.qsize())
id = work["repo_id"]
owner = work["owner"]
repo = work["repo"]
page = 1
sum = 0 # sum of inserted db
# get db connection
db = connectMysqlDB(config, autocommit=False)
cur = db.cursor()
while True:
print ""
try:
# get a suitable token and combine header
github_token = get_token()
headers = {
'User-Agent': 'Mozilla/5.0',
'Authorization': 'token ' + github_token,
'Content-Type': 'application/json',
'method': 'GET',
'Accept':
'application/vnd.github.squirrel-girl-preview+json'
}
# print "headers is: " + str(headers)
# combine url, notice: per page is 30
url = "https://api.github.com/repos/" + owner + "/" + repo + "/issues"
url = url + "?state=all" + "&page=" + str(
page) + "&per_page=30"
print "url is: " + url
insert_dict = {}
try:
# request data and parse response
req = urllib2.Request(url=url, headers=headers)
response = urllib2.urlopen(req)
result = json.loads(response.read().decode("utf-8"))
# print result
# judge response info empty
length = len(result)
sum += length
if length == 0:
# close the db connection
cur.close()
db.close()
print "finish & the sum of issue of pull request is: " + str(
sum)
self.q.task_done()
return
# write file
json_str = json.dumps(result)
# print "json format data: " + json_str
filename = base_path + "/" + owner + "&" + repo + "&" + str(
id) + "/" + str(page) + ".json"
flag = base.generate_file(filename, json_str)
if flag is True:
print "create file successfully: " + filename
elif flag is False:
print "file is already existed: " + filename
else:
print "create file failed: " + flag + " filename: " + filename
continue
page += 1 # page++
# handle response json data
num = 0
while num < length:
insert_dict = {}
if "id" not in result[num]:
insert_dict["id"] = None
else:
insert_dict["id"] = result[num]["id"]
if "number" not in result[num]:
insert_dict["number"] = None
else:
insert_dict["number"] = result[num]["number"]
if "comments" not in result[num]:
insert_dict["comments"] = None
else:
insert_dict["comments"] = result[num]["comments"]
if "created_at" not in result[num]:
insert_dict["created_at"] = None
else:
insert_dict["created_at"] = result[num][
"created_at"]
if "updated_at" not in result[num]:
insert_dict["updated_at"] = None
else:
insert_dict["updated_at"] = result[num][
"updated_at"]
if "login" not in result[num]["user"]:
insert_dict["user_login"] = None
else:
insert_dict["user_login"] = result[num]["user"][
"login"]
if "heart" not in result[num]["reactions"]:
insert_dict["heart"] = None
else:
insert_dict["heart"] = result[num]["reactions"][
"heart"]
if "eyes" not in result[num]["reactions"]:
insert_dict["eyes"] = None
else:
insert_dict["eyes"] = result[num]["reactions"][
"eyes"]
if "rocket" not in result[num]["reactions"]:
insert_dict["rocket"] = None
else:
insert_dict["rocket"] = result[num]["reactions"][
"rocket"]
if "total_count" not in result[num]["reactions"]:
insert_dict["total_count"] = None
else:
insert_dict["total_count"] = result[num][
"reactions"]["total_count"]
if "confused" not in result[num]["reactions"]:
insert_dict["confused"] = None
else:
insert_dict["confused"] = result[num]["reactions"][
"confused"]
if "hooray" not in result[num]["reactions"]:
insert_dict["hooray"] = None
else:
insert_dict["hooray"] = result[num]["reactions"][
"hooray"]
if "+1" not in result[num]["reactions"]:
insert_dict["up"] = None
else:
insert_dict["up"] = result[num]["reactions"]["+1"]
if "laugh" not in result[num]["reactions"]:
insert_dict["laugh"] = None
else:
insert_dict["laugh"] = result[num]["reactions"][
"laugh"]
if "-1" not in result[num]["reactions"]:
insert_dict["down"] = None
else:
insert_dict["down"] = result[num]["reactions"][
"-1"]
# print "insert num: " + str(insert_dict)
# 0 represent issue, 1 represent pull request
if "pull_request" not in result[num]:
flag = 0
else:
flag = 1
print "the issue type: " + str(flag)
# insert data to database table
try:
if insert_dict is not None:
cur.execute(
"insert into github_issue "
"(id, number, user_login, owner_login, repo, created_at, updated_at, flag, comments, total_count, up, down, laugh, confused, heart, hooray, rocket, eyes) "
"values (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)",
(insert_dict["id"], insert_dict["number"],
insert_dict["user_login"], owner, repo,
base.time_handler(
insert_dict["created_at"]),
base.time_handler(
insert_dict["updated_at"]), flag,
insert_dict["comments"],
insert_dict["total_count"],
insert_dict["up"], insert_dict["down"],
insert_dict["laugh"],
insert_dict["confused"],
insert_dict["heart"],
insert_dict["hooray"],
insert_dict["rocket"],
insert_dict["eyes"]))
db.commit()
except Exception as e:
print str(e)
num += 1
except urllib2.HTTPError as e:
print str(e.code) + " error with this page: " + url
if e.code != 404:
# mainly 403, sometimes 503
# token rate limit
self.q.put(work) # put into the queue again
sleep_time_tokens[github_token] = time.time(
) # set sleep time for that token
insert_dict = None
else:
insert_dict["body"] = "404 error"
insert_dict["created_at"] = None
insert_dict["updated_at"] = None
else:
pass # 403... error
except Exception as e:
print str(e) # unexpected error, don't interrupt the program