|
| 1 | +""" |
| 2 | +TFB Fail Detector Web Scraper |
| 3 | +Scrapes data from the last 3 complete runs and outputs the frameworks that have failed all of them with the same errors. |
| 4 | +
|
| 5 | +Developed using selenium==4.10.0 and the selenium/standalone-chrome:4.10.0 docker image. |
| 6 | +
|
| 7 | +USAGE: python tfb-fail-detector.py |
| 8 | +""" |
| 9 | + |
| 10 | +from collections import defaultdict |
| 11 | +import re |
| 12 | + |
| 13 | +from selenium import webdriver |
| 14 | +from selenium.webdriver.chrome.options import Options |
| 15 | +from selenium.webdriver.common.by import By |
| 16 | + |
| 17 | +NUM_RUNS = 3 |
| 18 | + |
| 19 | +def get_driver(): |
| 20 | + """Gets the selenium webdriver for interacting with the website""" |
| 21 | + chrome_options = Options() |
| 22 | + chrome_options.add_argument("--headless") |
| 23 | + chrome_options.add_argument("--no-sandbox") |
| 24 | + chrome_options.add_argument("--disable-dev-shm-usage") |
| 25 | + chrome_prefs = {} |
| 26 | + chrome_options.experimental_options["prefs"] = chrome_prefs |
| 27 | + chrome_prefs["profile.default_content_settings"] = {"images": 2} |
| 28 | + driver = webdriver.Chrome(options=chrome_options) |
| 29 | + return driver |
| 30 | + |
| 31 | +def get_last_n_complete_runs(driver, n=3): |
| 32 | + """Scrape the last n complete runs""" |
| 33 | + rows = driver.find_elements(By.CSS_SELECTOR, "table.resultsTable > tbody > tr") |
| 34 | + complete_runs = [] |
| 35 | + for row in rows: |
| 36 | + row_uuid = row.get_dom_attribute("data-uuid") |
| 37 | + run_stats = row.find_element(By.CSS_SELECTOR, "td:nth-of-type(2)") |
| 38 | + frameworks_tested_stats = re.search(r"([0-9]+)/([0-9]+) frameworks tested", run_stats.text) |
| 39 | + if not frameworks_tested_stats: |
| 40 | + # print("Unable to get info from run %s" % row_uuid) |
| 41 | + continue |
| 42 | + tested, total = frameworks_tested_stats.groups() |
| 43 | + if tested != total: |
| 44 | + # print("Found incomplete run %s. Tested: %s/%s" % (row_uuid, tested, total)) |
| 45 | + continue |
| 46 | + # print("Found complete run %s. Tested: %s/%s" % (row_uuid, tested, total)) |
| 47 | + complete_runs.append(row_uuid) |
| 48 | + if len(complete_runs) >= n: |
| 49 | + return complete_runs |
| 50 | + |
| 51 | +def find_failing_frameworks(driver, run_uuids): |
| 52 | + """Find frameworks that have failed all the given runs with the same error message""" |
| 53 | + failing_frameworks = defaultdict(lambda: 0) |
| 54 | + def process_failed_framework(framework_element): |
| 55 | + framework = re.search(r"\[\S+] [a-zA-Z][a-zA-Z:\_ ]*$", framework_element.text) |
| 56 | + framework_failure = framework.group(0) |
| 57 | + failing_frameworks[framework_failure] += 1 |
| 58 | + |
| 59 | + for run_uuid in run_uuids: |
| 60 | + driver.get("https://tfb-status.techempower.com/results/%s" % run_uuid) |
| 61 | + assert "TFB Status" in driver.title |
| 62 | + failure_list = driver.find_element(By.CLASS_NAME, "failures") |
| 63 | + failures = failure_list.find_elements(By.TAG_NAME, "li") |
| 64 | + for failure in failures: |
| 65 | + process_failed_framework(failure) |
| 66 | + return failing_frameworks |
| 67 | + |
| 68 | +if __name__ == '__main__': |
| 69 | + driver = get_driver() |
| 70 | + driver.get("https://tfb-status.techempower.com/") |
| 71 | + assert "TFB Status" in driver.title |
| 72 | + complete_runs = get_last_n_complete_runs(driver, NUM_RUNS) |
| 73 | + failed_frameworks = find_failing_frameworks(driver, complete_runs) |
| 74 | + |
| 75 | + for failure_info, fail_count in failed_frameworks.items(): |
| 76 | + if fail_count != NUM_RUNS: |
| 77 | + continue |
| 78 | + print(failure_info) |
0 commit comments