provisioning/pelican_scheduler.py

   1 #!/usr/bin/env python3
   2
   3 """A script to schedule Pelican posts in advance, appropriate for a Git
   4 post-receive hook. Requires the `at` job-scheduling utility."""
   5
   6 import datetime
   7 import os
   8 import re
   9 import subprocess
  10
  11 WORKING_REPO = "/home/mtsw/working"
  12 INPUT_DIR = os.path.join(WORKING_REPO, "content")
  13 OUTPUT_DIR = "/var/www/html"
  14 PUBLISH_CONF = os.path.join(WORKING_REPO, "publishconf.py")
  15 SITEGEN_COMMAND = "bash -c 'cd {} && source bin/activate && pelican {} -o {} -s {}'".format(
  16     WORKING_REPO, INPUT_DIR, OUTPUT_DIR, PUBLISH_CONF)
  17
  18 DATELINE_REGEX = re.compile(r"^Date: *(\d{4}-\d{2}-\d{2} \d{2}:\d{2}) *$",
  19                             re.MULTILINE)
  20 JOBLINE_REGEX = re.compile(r"\d+\s\w{3} (\w{3} +\d{1,2} \d{2}:\d{2}:\d{2} \d{4})")
  21
  22 def get_future_publication_times():
  23     now = datetime.datetime.now()
  24     times = set()
  25     for path, _dirnames, filenames in os.walk(INPUT_DIR):
  26         if path.endswith("drafts"):
  27             continue
  28         for filename in filenames:
  29             if not filename.endswith(".md"):
  30                 continue
  31             with open(os.path.join(path, filename)) as post_file:
  32                 match = DATELINE_REGEX.search(post_file.read())
  33                 if match:
  34                     time = datetime.datetime.strptime(match.group(1),
  35                                                       "%Y-%m-%d %H:%M")
  36                     if time > now:
  37                         times.add(time)
  38     return times
  39
  40
  41 def get_extant_at_job_times():
  42     times = set()
  43     result = subprocess.run(["atq"], stdout=subprocess.PIPE)
  44     job_lines = result.stdout.decode('utf8').split('\n')
  45     for job_line in job_lines:
  46         match = JOBLINE_REGEX.match(job_line)
  47         if match:
  48             times.add(datetime.datetime.strptime(match.group(1),
  49                                                  "%b %d %H:%M:%S %Y"))
  50     return times
  51
  52
  53 def schedule(command, when):
  54     timestamp = when.strftime("%H:%M %Y-%m-%d")
  55     at_command = ['at', timestamp]
  56     at = subprocess.Popen(
  57         at_command,
  58         stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE
  59     )
  60     at.communicate(command.encode())
  61
  62
  63 def main():
  64     # sync our "working" repo with the bare one
  65
  66     # but first, don't let the fact that this is running as a hook on the bare
  67     # repo confuse us
  68     our_env = os.environ.copy()
  69     del our_env['GIT_DIR']
  70
  71     for git_cmd in [["git", "fetch", "origin"],
  72                     ["git", "reset", "--hard", "origin/master"]]:
  73         subprocess.run(git_cmd, cwd=WORKING_REPO, env=our_env)
  74
  75     # sitegen now! (even if there are no posts to queue, we can at least update
  76     # /drafts/) and make any back-edits to published posts live
  77     subprocess.run(SITEGEN_COMMAND,
  78                    # XXX: `shell=True` is contrary to the moral law
  79                    shell=True)
  80
  81     # look for scheduled future posts
  82     future_publication_times = get_future_publication_times()
  83
  84     # look at atq
  85     extant_at_job_times = get_extant_at_job_times()
  86
  87     # if there are future posts that don't have an atq entry, schedule a
  88     # site-regen at that time
  89     to_schedule = future_publication_times - extant_at_job_times
  90     for time in to_schedule:
  91         schedule(SITEGEN_COMMAND, time)
  92
  93
  94 if __name__ == "__main__":
  95     main()