# find number of top 1000 python packages that have one or more CVEs

analyze_pyup_io_safety_db.py

# find number of top 1000 python packages that have one or more CVEs

import json

# find all packages with a CVE

# Open the JSON file
with open("pyup_io_safety_db_python_cve.json") as file:
    # Load the JSON data
    data = json.load(file)

pkgs_with_cves = []
for record in data:
    try:
        # only include records indicating CVEs
        for advisory in data[record]:
            if "CVE" in advisory["cve"]:
                pkgs_with_cves.append(record)
                break
    except:
        continue

pkgs_with_cves_set = set(pkgs_with_cves)

# import list of top 1000 packages

with open("top_1000_pkgs.txt") as file:
    # Load the JSON data
    top_repos = file.readlines()

# remove new lines to enable precise matching in boolean algebra below
top_repos = [repo.rstrip("\n") for repo in top_repos]

top_repos_set = set(top_repos)
print(top_repos_set)

# find intersection of packages with a CVE and top repos
top_pkgs_with_a_cve = pkgs_with_cves_set & top_repos_set

print(len(top_pkgs_with_a_cve))