Messenger_1Million_Stats/scripts/find_holes.py

from datetime import datetime
import million.analyze.find_holes as fh
from million.analyze.word_finder import retain_counts
import million.parse.fb_exports as fb


DATA_PATH = "./data/"

export = fb.parse_dirfiles(DATA_PATH)

filtered = retain_counts(export.messages)

sequences = fh.compute_sequences(filtered)

actual_counted = sum([s.length() for s in sequences])

print(f"Actual counted: {actual_counted}")

merged = fh.merge_duplicates(sequences)
merged = [s for s in merged if s.length() > 1]
holes = fh.find_holes(filtered)

print(len(holes))

for hole in holes:
    print(f"{hole.start() + 1} -> {hole.end() - 1} ({hole.length() - 2})")


# lets export a csv file of the holes and the people responsible for them
with open("output/holes.csv", "w") as f:
    f.write("début,fin,taille,responsable1,responsable2,date1,date2\n")
    for hole in holes:
        date_start = hole.start_message.date_time.strftime("%Y-%m-%d %H:%M:%S")
        date_end = hole.end_message.date_time.strftime("%Y-%m-%d %H:%M:%S")
        f.write(
            f"{hole.start()},"
            f"{hole.end()},"
            f"{hole.length()},"
            f"{hole.start_message.sender_name},"
            f"{hole.end_message.sender_name},"
            f"{date_start},{date_end}\n"
        )