Add a script to compare reviews CI artifacts

The "compare-reviews.py" is a manual execution script that needs a left
hand and right hand review numbers and optionally you can override the
list of files that it's going to compare.

By default the list of files that compare is:
- playbook_executions.log
- reproducer-quickstart.sh
- collect_logs.sh

An example of execution would be "./compare-reviews.py 584508  589068"
It will compare the CI artifacts stated before between the two reviews.

The output is a diff with the differences, for each file at each job.

Change-Id: Ib3c05f327e3d6773990e4e39b1892a61e41cb044

requirements.txt

@@ -2,3 +2,4 @@ pbr
 # required to run scripts/tripleo-jobs.py
 sqlalchemy
 jenkinsapi
+colorama

scripts/compare-reviews.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python
+from __future__ import print_function
+
+import argparse
+import difflib
+import json
+import requests
+import os
+
+from colorama import init
+from colorama import Fore
+
+GERRIT_DETAIL_API = "https://review.openstack.org/changes/{}/detail"
+GERRIT_USER_NAME = "zuul"
+ZUUL_PIPELINE = "check"
+
+
+def parse_ci_message(message):
+    """Convert zuul's gerrit message into a dict with job name as key and
+    job url as value
+    """
+
+    jobs = {}
+    for line in message.split("\n"):
+        if line[:1] == '-':
+            splitted_line = line.split()
+            jobs[splitted_line[1]] = splitted_line[2]
+    return jobs
+
+
+def get_file(logs_url, file):
+    """Download a file from logs server for this job
+    """
+
+    response = requests.get(logs_url + '/logs/' + file)
+    if response.ok:
+        return response.content
+    return None
+
+
+def get_last_jobs(change):
+    """Get the last CI jobs execution at check pipeline for this review
+    """
+
+    last_jobs = {}
+    detail_url = GERRIT_DETAIL_API.format(change)
+    response = requests.get(detail_url)
+    if response.ok:
+        sanitized_content = "\n".join(response.content.split("\n")[1:])
+        detail = json.loads(sanitized_content)
+        zuul_messages = [
+            message for message in detail['messages']
+            if message['author']['username'] == GERRIT_USER_NAME
+            and "({} pipeline)".format(ZUUL_PIPELINE) in message['message']
+        ]
+        last_message = zuul_messages[-1]
+        last_jobs = parse_ci_message(last_message['message'])
+        date = last_message['date']
+    else:
+        raise RuntimeError(response.content)
+    return last_jobs, date
+
+
+def download(jobs, file_path):
+    """Download a file from all the specified jobs and return them as a
+    dictionary with job name as key and file content as value
+    """
+    downloaded_files = {}
+    for job, logs in jobs.iteritems():
+        downloaded_file = get_file(logs, file_path)
+        if downloaded_file:
+            downloaded_files[job] = downloaded_file
+        else:
+            print("WARNING: {} not found at {}".format(file_path, job))
+    return downloaded_files
+
+
+def is_equal(lho_jobs, rho_jobs, file_path):
+    """Check the differences of file_path between the lho and rho job sets and
+    print out them
+    """
+
+    lho_files = download(lho_jobs, file_path)
+    rho_files = download(rho_jobs, file_path)
+    print(">>>>>>> Comparing {}".format(file_path))
+    if lho_files != rho_files:
+        diffkeys = [
+            k for k in lho_files
+            if lho_files[k] != rho_files.get(k, None)
+        ]
+        print("{} are different at the following jobs:".format(file_path))
+        for key in diffkeys:
+            print(Fore.BLUE + key)
+            print(Fore.BLUE + lho + ": " + lho_jobs[key])
+            print(Fore.BLUE + rho + ": " + rho_jobs[key])
+            for line in difflib.unified_diff(
+                    lho_files[key].splitlines(),
+                    rho_files.get(key, '').splitlines()):
+                print(colors.get(line[0], Fore.BLACK) + line)
+        return False
+    print("{} files are the same".format(file_path))
+    return True
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(
+        description='Compares files at logs.o.o from two reviews')
+
+    parser.add_argument(
+        'reviews',
+        metavar='review',
+        nargs=2,
+        help='left-side and right-side review to compare')
+
+    parser.add_argument(
+        '--files',
+        type=str,
+        default='playbook_executions.log,reproducer-quickstart.sh,'
+                'collect_logs.sh',
+        help='Comma separated list of files to compare at logs.o.o '
+             '(default: %(default)s)')
+
+    args = parser.parse_args()
+
+    colors = {'-': Fore.RED, '+': Fore.GREEN, '@': Fore.YELLOW}
+
+    # When piping colors are disabled unless you define PY_COLORS variable.
+    strip = os.environ.get('PY_COLORS', '0') != '1'
+    init(autoreset=True, strip=strip)
+    lho = args.reviews[0]
+    rho = args.reviews[1]
+    lho_jobs, lho_date = get_last_jobs(lho)
+    rho_jobs, rho_date = get_last_jobs(rho)
+
+    # Compare only the job at both reviews
+    jobs_intersection = set(lho_jobs.keys()) & set(rho_jobs.keys())
+    lho_jobs = {job: lho_jobs[job] for job in jobs_intersection}
+    rho_jobs = {job: rho_jobs[job] for job in jobs_intersection}
+
+    print("- {}, {}".format(lho, lho_date))
+    print("+ {}, {}".format(rho, rho_date))
+
+    for file_to_compare in args.files.split(','):
+        is_equal(lho_jobs, rho_jobs, file_to_compare)