背景
Jenkins借助釘釘插件，實(shí)現(xiàn)當(dāng)構(gòu)建失敗時(shí)，自動(dòng)觸發(fā)釘釘預(yù)警。雖然插件允許自定義消息主體，支持使用 Jenkins環(huán)境變量，但是局限性依舊很大。當(dāng)接收到釘釘通知后，若想進(jìn)一步查看報(bào)錯(cuò)具體原因，仍完全依賴郵件通知，很影響效率。
解決思路
Jenkins + RobotFramework 項(xiàng)目
思路如下：
- 使用
Startup Trigger插件和Groovy plugin插件, 在服務(wù)重啟時(shí)自動(dòng)修改Jenkins CSP配置, 解決Robot Framework項(xiàng)目測(cè)試報(bào)告在Jenkins服務(wù)器上無(wú)法預(yù)覽的問(wèn)題, 如下:
System.setProperty("hudson.model.DirectoryBrowserSupport.CSP", "")
- Jenkins項(xiàng)目構(gòu)建完成后, 在構(gòu)建后操作中，利用
Post Build Task插件提供的命令行功能, 傳入相關(guān)參數(shù)執(zhí)行封裝的腳本, 如下：
if "%構(gòu)建失敗時(shí)觸發(fā)釘釘通知%"=="true" (
echo ************推送釘釘消息通知************
cd %WORKSPACE%/test
C:/python38/python.exe dingtalk.py --id=%BUILD_NUMBER% --name=%JOB_NAME% --server=xxxxxx
echo *******************************************
exit 0
)
說(shuō)明幾點(diǎn)：
- “構(gòu)建失敗時(shí)觸發(fā)釘釘通知” 是我自定義的布爾類型的項(xiàng)目構(gòu)建參數(shù)，用于更方便的控制是否觸發(fā)釘釘消息通知；
- dingtalk.py是封裝的腳本，其內(nèi)容在下面會(huì)詳細(xì)說(shuō)明；
-
Post Build Task插件的位置必須位于Robot Framework插 件之后, 否則會(huì)引發(fā)以下異常:
robot.errors.DataError: Reading XML source '<in-memory file>' failed: Incompatible XML element 'html'
- 編寫(xiě)第二步中調(diào)用的釘釘消息通知腳本，其核心思路可分為兩點(diǎn)：
1. 獲取并分析本次構(gòu)建結(jié)果文件(如: output.xml), 提取數(shù)據(jù), 如: 失敗占比、失敗用例報(bào)錯(cuò)詳情等;
2. 組裝 Markdown 格式的消息主體, 然后調(diào)用 DingTalk Webhook 接口, 推送消息通知;
核心代碼如下：
def fetch_robot_output_xml(xml_name=None, save=False, cache=True) -> str:
""" 從服務(wù)器上獲取 Robot Framework 項(xiàng)目構(gòu)建完成后生成的xml文件
@param xml_name: xml文件名稱, 部分項(xiàng)目會(huì)合并測(cè)試報(bào)告, 默認(rèn)`merge.xml`
@param save: 是否將文件保存到本地, 默認(rèn) False
@param cache: 默認(rèn)為T(mén)rue, 會(huì)首先遍歷本地已保存的構(gòu)建文件, 若文件名稱不存在才會(huì)向服務(wù)器發(fā)起請(qǐng)求;
@Returns : xml文件內(nèi)容
"""
xml_name = xml_name if xml_name else "merge.xml"
url = url_prefix + f"{name}/{id}" + f"/robot/report/{xml_name}"
output_xml = None
if cache:
for xml in robot_outputs.glob('*.xml'):
if name in xml.stem and id in xml.stem:
output_xml = xml.read_text(encoding='utf-8')
break
if not output_xml:
print(f"發(fā)起請(qǐng)求獲取測(cè)試報(bào)告, 目標(biāo)URL: {url}")
# curl -X GET {URL} --user {USER}:{TOKEN}
resp = requests.get(url, auth=HTTPBasicAuth(jenkins_user.encode('utf-8'), jenkins_token.encode('utf-8')))
resp.encoding = 'utf-8'
output_xml = resp.text
if save and not cache:
save_output_xml(output_xml, f"{server}_{name}_{id}_output.xml")
print("Done.")
return output_xml
def parse_robot_result(xml) -> dict:
"""解析用例運(yùn)行結(jié)果
pip install robotframework==3.2.1
獲取統(tǒng)計(jì)信息, 優(yōu)先使用此方法, 因?yàn)閤ml文件中可能會(huì)不包含有效的測(cè)試報(bào)告信息, 如請(qǐng)求獲取文件服務(wù)器返回404的情況
"""
print('解析用例運(yùn)行結(jié)果...')
try:
suite = ExecutionResult(xml).suite
except DataError as e:
raise e('解析xml文件失敗, 請(qǐng)檢查構(gòu)建文件內(nèi)容是否完整, 確認(rèn)服務(wù)器上的構(gòu)建文件是否已被刪除')
all_tests = suite.statistics.critical
total, passed, failed = all_tests.total, all_tests.passed, all_tests.failed
pass_rate = round((passed / total) * 100, 2)
is_success = True if failed == 0 else False
stat = {'server': server, 'job': name, 'build_id': id, 'total': total, 'passed': passed, 'failed': failed, 'pass_rate': str(pass_rate) + "%", 'is_success': is_success}
print('Done')
return stat
def pick_fail_cases(xml: str) -> list:
print('獲取所有失敗用例及其報(bào)錯(cuò)信息...')
root = ET.fromstring(xml)
fail_cases = []
for case in root.iter('test'):
status = case.find('status')
if status.get('status') == "FAIL":
fail_cases.append({"case": case.get('name'), "message": truncate_text(status.text)})
print('Done.')
return fail_cases
def package_dingtalk_text(stat: dict) -> dict:
print("組裝釘釘通知信息主體...")
# 項(xiàng)目頁(yè)面地址
job_url = url_prefix + stat['job'] + '/'
# 本次構(gòu)建ID以及詳情頁(yè)面地址
build_id = stat['build_id']
build_url = job_url + build_id + '/'
# 統(tǒng)計(jì)信息
summary_text = f"用例總數(shù)：{stat['total']}，失敗用例數(shù)：{stat['failed']}， 通過(guò)率： {stat['pass_rate']}\n"
# 用例報(bào)錯(cuò)詳情
case_error_details = ""
num = 1 # 序號(hào)
if len(stat['fail_cases']) > 0:
for case in stat['fail_cases']:
case_error_details = case_error_details + f"{num}. 用例名稱: {case['case']}, 失敗原因: {case['message']}\n"
num += 1
print("Done.")
return {
"server_show_name": server_show_name,
"job_show_name": job_show_name,
"job_url": job_url,
"build_id": build_id,
"build_url": build_url,
"summary_text": summary_text,
"case_error_details": case_error_details
}
def ding_talk(message=None):
"""
目前只支持markdown語(yǔ)法的子集, 參考官方文檔: https://developers.dingtalk.com/document/app/custom-robot-access
"""
print("推送釘釘消息通知...")
webhook = "https://oapi.dingtalk.com/robot/send?access_token=1048af6c9a25ea5e924ae328c6782b68c48a40bcea773df96eadce783f2941ca"
headers = {"Content-Type": "application/json", "Charset": "UTF-8"}
payload = {
"msgtype": "markdown",
"markdown": {
"title":
"心跳測(cè)試",
"text":
f"# [[{message['server_show_name']}] 心跳測(cè)試-{message['job_show_name']}]({message['job_url']})\n***\n>*失敗用例會(huì)自動(dòng)重復(fù)執(zhí)行一次, 若仍然失敗, 則標(biāo)記為失敗*\n***\n- 任務(wù)：#[{message['build_id']}]({message['build_url']})\n- 狀態(tài)：失敗\n- 統(tǒng)計(jì)信息：{message['summary_text']}\n - 失敗用例報(bào)錯(cuò)詳情： \n\n{message['case_error_details']}\n"
},
"at": {
"isAtAll": True
}
}
resp = requests.post(webhook, json=payload, headers=headers)
print(resp.text)
print("Done.")
return resp
Jenkins + Jmeter 項(xiàng)目
前兩步是共同的，也是通過(guò)解析result.jtl文件提取需要的數(shù)據(jù)，直接貼代碼：
def fetch(xml_name=None, save_file=False) -> str:
xml_name = xml_name if xml_name else "result.jtl"
url = url_prefix + f"{name}/{id}" + f"/artifact/{name}result/{xml_name}"
print(url)
resp = requests.get(url, auth=HTTPBasicAuth(jenkins_user.encode('utf-8'), jenkins_token.encode('utf-8')))
status_code = resp.status_code
if status_code == 200:
resp.encoding = 'utf-8'
xml = resp.text
if save_file:
save(xml, f"{server}_{name}_{id}_{xml_name}")
return xml
elif status_code == 404:
print(f"{status_code}, 指定的測(cè)試報(bào)告不存在, 可能已被刪除, 請(qǐng)去服務(wù)器上檢查")
raise Exception("{status_code}, 指定的測(cè)試報(bào)告不存在, 可能已被刪除, 請(qǐng)去服務(wù)器上檢查")
elif status_code == 500:
print(f"獲取測(cè)試報(bào)告失敗, 詳情: {status_code} >> {resp.text}")
raise Exception(f"獲取測(cè)試報(bào)告失敗, 詳情: {status_code} >> {resp.text}")
else:
return None
def parse(xml):
"""分析xml測(cè)試報(bào)告, 提取數(shù)據(jù)"""
result = {}
try:
root = ET.fromstring(xml)
except ParseError:
raise ParseError("解析XML文件失敗, 請(qǐng)檢查服務(wù)器上文件是否存在")
# 獲取統(tǒng)計(jì)信息
samples = root.findall('httpSample')
failures = [sample for sample in samples if sample.get('s') == "false"]
failures_cnt = len(failures)
success = [sample for sample in samples if sample.get('s') == "true"]
success_cnt = len(success)
is_success = True if failures_cnt == 0 else False
success_rate = "100%" if is_success else str(round((success_cnt / len(samples)) * 100, 2)) + "%"
result['summary'] = {"samples": len(samples), "is_success": is_success, "failures_cnt": failures_cnt, "success_rate": success_rate}
# 數(shù)據(jù)清洗, 獲取失敗用例的報(bào)錯(cuò)信息
fail_details = []
if failures_cnt > 0:
pat = re.compile(".*R-TraceId:(.*?)\n.*", re.DOTALL)
for sample in failures:
msg = []
for res in sample.findall('assertionResult'):
for child in res:
if child.tag in ("failure", "error") and child.text == "true":
try:
msg.append(res.find('failureMessage').text) # ?這個(gè)節(jié)點(diǎn)可能會(huì)不存在
except AttributeError:
msg.append("")
response_header = sample.find('responseHeader').text
try:
traceid = pat.match(response_header).group(1).strip()
except AttributeError:
traceid = None
response_data = sample.find('responseData').text.replace('\r\n', '')
fail_details.append({
'sample': sample.get('lb'),
"traceid": traceid,
'data': truncate_text(response_data),
'msg': truncate_text(",".join(msg))
})
result['fail_samples'] = fail_details
result.update({
"server": server,
'job': name,
'build_id': id,
})
return result
參考文檔:
- Jenkins DingTalk插件文檔: https://jenkinsci.github.io/dingtalk-plugin/guide/getting-started.html
- Jenkins Robot Framework插件文檔: https://plugins.jenkins.io/robot/
- 釘釘自定義機(jī)器人接入官方文檔: https://developers.dingtalk.com/document/app/custom-robot-access/title-nfv-794-g71
- RobotFramework測(cè)試用例的自行解析： http://www.itdecent.cn/p/3e1b7ba3a48c
