def check_completion(file_list):
#loop all bookmarks
for bookmark in file_list:
finished_posts, complete = load_bookmark(bookmark)
#if this run didn't finish, return false
if complete == False:
return False
return True
stdout=f,
stderr=f)
#run corresponding baseline models in background - if don't have results already
if run_baseline:
for mode in baseline_modes:
#define output filename for baseline model
baseline_outfile = "sim_results/%s/run_results/%s_baseline_%s_%dtrain_%dtest_%d-%d%s%s" % (
subreddit, subreddit, mode[1:], arguments['-n_train'],
arguments['-n'], arguments['-y'], arguments['-m'],
size_class, "_run%d" % run if repeat_runs > 1 else "")
outfile_lists[mode].append(baseline_outfile)
#no data for this baseline configuration, run the test
#check the bookmark saved by the model to know if finished or not
finished_posts, complete = load_bookmark(baseline_outfile)
if complete:
print("skipping", baseline_outfile)
else:
#build command arguments list
#base first
baseline_command = [
'time', 'python3', 'baseline_model.py', '-s',
subreddit, '-o', baseline_outfile, mode, '-v'
]
#add the dict args - but only the ones that make sense for the baseline model
for arg in [
'-n', '-n_train', '-m', '-y', '-min', '-max',
'-timestamps'
]:
if len(test_posts) != len(test_cascades):
test_cascades = functions_gen_cascade_model.filter_dict_by_list(test_cascades, list(test_posts.keys()))
all_metrics = [] #keep all metrics, separate for each post/observed time run, dump them all at the end
#for outputting comment timestamps
timestamps = {} #post_id -> time (or true) -> list of timestamps
#how often do we want to dump? every 20 tests or so
#20 / number of observation settings = number of posts to finish before dumping
dump_count = 20 // len(observed_list) + (20 % len(observed_list) > 0)
if dump_count == 0: dump_count = 20 #make sure not modding by 0 if small run
#load list of finished posts for this run, so we can skip ones that are already done
#(if no bookmark, will get back empty set and False flag)
finished_posts, complete = functions_gen_cascade_model.load_bookmark(outfile)
#if finished all posts already, exit
if complete:
vprint("Entire post set already simulated, exiting")
exit(0)
else: vprint("Skipping %d already simulated posts" % len(finished_posts))
#process all posts (or just one, if doing that)
post_count = 0
vprint("Processing %d post" % len(test_posts) + ("s" if len(test_posts) > 1 else ""))
for sim_post_id, sim_post in test_posts.items():
#skip this post if we've already done it
if sim_post_id in finished_posts:
continue