def computeQualitySub():
# 缺陷修复比例，平均修复时间
repair_ratio, repair_time = [], []
metrics = [repair_ratio, repair_time]
for repo in REPOS:
if repo in NONE_GH:
repair_ratio.append(None)
repair_time.append(None)
else:
# issue_total,done
result = dbop.select_all(
"select closed_at,created_at from issues_info where repo_id=%s and is_pr=0",
(repo, ))
total_num = len(result)
if total_num == 0:
tmp_repair_ratio = 0
tmp_repair_time = 0
else:
issue_done = [item for item in result if item[0] is not None]
tmp_repair_ratio = len(issue_done) * 1.0 / total_num
tmp_repair_time = sum([
_datetime2int(item[0]) - _datetime2int(item[1])
for item in issue_done
]) * 1.0 / len(issue_done)
repair_ratio.append(tmp_repair_ratio)
repair_time.append(1.0 / (tmp_repair_time + 1))
repair_time = _nor_data(repair_time)
for i in range(0, len(REPOS)):
dbop.execute(
"insert into quality_sub(repo_id,repair_ratio,repair_time) values(%s,%s,%s)",
(REPOS[i], repair_ratio[i], repair_time[i]))
def readPrjLists():
result = dbop.select_all("select prj_id,github_url from prj_list")
for prj in result:
if prj[1] is not None:
repo_name = prj[1][19:]
PRJS.put(repo_name)
REPO_ID[repo_name] = int(prj[0])
def _get_last_commit_fetch(prj):
last_page = dbop.select_one("select page from commits_json_raw where repo_id=%s order by id desc limit 1",
(REPO_ID[prj],), (1,))[0]
last_data_set = set([ item[0] for item in
dbop.select_all("select sha from commits_info where repo_id=%s and page =%s", (
REPO_ID[prj],last_page))])
return last_page, last_data_set
def readPrjLists():
prjs = []
result = dbop.select_all("select prj_id,github_url from prj_list")
for prj in result:
if prj[1] is not None:
repo_name = prj[1][19:]
prjs.append(repo_name)
REPO_ID[repo_name] = int(prj[0])
return prjs
def _get_last_issue_fetch(prj):
# 获取上次记录，以及上次获得数据集合
last_page = dbop.select_one("select page from issues_json_raw where repo_id=%s order by id desc limit 1",
(REPO_ID[prj],), (1,))[0]
last_data_set = set([ item[0] for item in
dbop.select_all("select number from issues_info where repo_id=%s and page =%s", (
REPO_ID[prj],last_page))
])
return last_page, last_data_set
def readPrjLists():
prjs = dbop.select_all(
"select prj_id,github_url,facebook_url,twitter_url from prj_list")
for prj in prjs:
REPOS.append(prj[0])
if prj[1] is None:
NONE_GH.add(prj[0])
if prj[2] is None or len(prj[2].strip()) == 0:
NONE_FB.add(prj[0])
if prj[3] is None or len(prj[3].strip()) == 0:
NONE_TW.add(prj[0])
def computeTeamHealth():
# 几个重要时间点
time_now = time.time()
time_now_str = _strtime_before_days(time_now, 0)
time_before_1_window = _strtime_before_days(time_now, EXAMINE_WINDOW)
time_before_2_window = _strtime_before_days(time_now, 2 * EXAMINE_WINDOW)
time_before_3_window = _strtime_before_days(time_now, 3 * EXAMINE_WINDOW)
ccrs, ngrs, tbrs = [], [], []
for repo in REPOS:
if repo in NONE_GH:
ccrs.append(None)
ngrs.append(None)
tbrs.append(None)
else:
# 几个重要集合
data_before_1_window = set([
item[0] for item in dbop.select_all(
"select author_id from commits_info where repo_id=%s and author_id is not null and (author_date>%s and author_date<%s)",
(repo, time_before_1_window, time_now_str))
])
data_before_2_window = set([
item[0] for item in dbop.select_all(
"select author_id from commits_info where repo_id=%s and author_id is not null and (author_date>%s and author_date<%s)",
(repo, time_before_2_window, time_before_1_window))
])
data_before_3_window = set([
item[0] for item in dbop.select_all(
"select author_id from commits_info where repo_id=%s and author_id is not null and (author_date>%s and author_date<%s)",
(repo, time_before_3_window, time_before_2_window))
])
# ccr
data_common = _common_num(data_before_1_window,
data_before_2_window)
ccrs.append(data_common * 1.0 /
(len(data_before_2_window) + 1)) #避免分母为0
# ngr
new_users_1 = len(data_before_1_window) - data_common + 1 #避免分母为0
data_common_2 = _common_num(data_before_3_window,
data_before_2_window)
new_users_2 = len(
data_before_2_window) - data_common_2 + 1 #避免分母为0
ngrs.append((new_users_1 - new_users_2) * 1.0 / new_users_2)
# tbr 上一个窗口期的
commits_dis = dbop.select_all(
"select count(*) from commits_info where repo_id=%s and author_id is not null group by author_id",
(repo, ))
issues_dis = dbop.select_all(
"select count(*) from issues_info where repo_id=%s and user_id is not null group by user_id",
(repo, ))
tbrs.append(1.0 / (_gini([item[0] for item in commits_dis]) +
_gini([item[0] for item in issues_dis]) + 1))
metrics = []
metrics.append(_nor_data(ccrs))
metrics.append(_nor_data(ngrs))
metrics.append(_nor_data(tbrs))
for i in range(0, len(REPOS)):
tmp_row = [REPOS[i]]
for j in range(0, len(metrics)):
tmp_row.append(metrics[j][i])
dbop.execute(
"insert into team_health(repo_id, ccr,ngr,tbr) values(%s,%s,%s,%s)",
tmp_row)
def computeMaturity():
# maturity: repo_id, issue_done, commit_total, age_dev, fans_dev
issue_done, commit_total, age_dev = [], [], []
stars, watchs, forks = [], [], []
fans_fb, fans_tw = [], []
metrics = [
issue_done, commit_total, age_dev, stars, watchs, forks, fans_fb,
fans_tw
]
# 获取每个指标
for repo_id in REPOS:
if repo_id in NONE_GH:
issue_done.append(None)
commit_total.append(None)
age_dev.append(None)
stars.append(None)
watchs.append(None)
forks.append(None)
else:
# issue_done
result = dbop.select_one(
"select count(*) from issues_info where repo_id=%s and is_pr=0 and closed_at is not NULL",
(repo_id, ))
issue_done.append(result[0])
# commit_total
result = dbop.select_one(
"select count(*) from commits_info where repo_id=%s",
(repo_id, ))
commit_total.append(result[0])
# age_dev
result = dbop.select_all(
"select author_date from commits_info where repo_id =%s",
(repo_id, ))
age_dev.append(_continuous_dev_month(result))
# fans_dev
result = dbop.select_one(
"select watch,star,fork from html_info where repo_id=%s order by id desc limit 1",
(repo_id, ), (0, 0, 0))
stars.append(result[0])
watchs.append(result[1])
forks.append(result[2])
if repo_id in NONE_FB:
fans_fb.append(None)
else:
# fans_social
result = dbop.select_one(
"select watches_num from facebook_data where coin_id=%s order by id desc limit 1",
(repo_id, ), (0, ))
fans_fb.append(result[0])
if repo_id in NONE_TW:
fans_tw.append(None)
else:
result = dbop.select_one(
"select followers_num from twitters_data where coin_id=%s order by id desc limit 1",
(repo_id, ), (0, ))
fans_tw.append(result[0])
# 归一化
nor_data = []
for metric in metrics:
nor_data.append(_nor_data(metric))
for i in range(0, len(REPOS)):
tmp_row = [nor_metric[i] for nor_metric in nor_data]
dbop.execute(
"insert into maturity(repo_id, issue_done, commit_total, age_dev, fans_dev, fans_social) values(%s,%s,%s,%s,%s,%s)",
(REPOS[i], tmp_row[0], tmp_row[1], tmp_row[2],
_my_avg(tmp_row[3:-2]), _my_avg(tmp_row[-2:])))