remove_processed_channels.py

import json
from typing import List, Set

# Load the original channels
with open('VERIFY.JSON', 'r', encoding='utf-8') as f:
    original_data = json.load(f)
    original_channels = set(original_data['similar_channels'])

# Load the job-related channels
with open('job_channels.json', 'r', encoding='utf-8') as f:
    job_channels = set(json.load(f)['similar_channels'])

# Add the most recently scanned channels (from the latest run)
recently_scanned = {
    "freid_ko",
    "frelanserr",
    "friendsfoundation"
}

# Add all job-related channels to processed set
processed_channels = set(job_channels)

# Add the first 201 channels to processed set
# We know from the log that 201 channels were processed in total
sorted_channels = sorted(original_channels)
for channel in sorted_channels:
    processed_channels.add(channel)
    if len(processed_channels) == 204:  # 201 + 3 recent scans
        break

# Remove processed channels from original set
remaining_channels = list(original_channels - processed_channels)

# Save the remaining channels back to VERIFY.JSON
with open('VERIFY.JSON', 'w', encoding='utf-8') as f:
    json.dump({'similar_channels': remaining_channels}, f, ensure_ascii=False, indent=2)

print(f"Original channels: {len(original_channels)}")
print(f"Job-related channels: {len(job_channels)}")
print(f"Non-job processed channels: {204 - len(job_channels)}")
print(f"Total processed channels: {len(processed_channels)}")
print(f"Remaining channels: {len(remaining_channels)}")