compare_versions_to_manual.py

import boto3
import sys
from pathlib import Path
import json

BUCKET_NAME = "docs-mongodb-org-dotcomprd"
versions = sys.argv[1:]

def remove_prefix(s: str, version: str) -> str:
    return s.split(f"docs/{version}/", 1)[-1]

s3 = boto3.resource("s3")
my_bucket = s3.Bucket(BUCKET_NAME)
version_db = {}
for version in versions + ["manual"]:
    output_pathname = Path(version + ".txt")
    if output_pathname.exists():
        version_db[version] = {remove_prefix(k, version): v for k, v in json.loads(output_pathname.read_text())}
        continue

    print("Downloading " + version)
    files = []
    for obj in my_bucket.objects.filter(Prefix=f"docs/{version}/"):
        if not obj.key.endswith("index.html"):
            continue
        redirect_target = None
        if obj.size == 0:
            real_obj = s3.Object(BUCKET_NAME, obj.key)
            redirect_target = real_obj.website_redirect_location
        files.append([obj.key, redirect_target])
    output_pathname.write_text(json.dumps(files, indent=2))
    version_db[version] = {remove_prefix(k, version): v for k,v in files}

for version in versions:
    for key, _ in version_db[version].items():
        fully_qualified_key = f"docs/{version}/{key}"
        if key in version_db["manual"]:
            redirect_target = version_db["manual"][key]
            if redirect_target:
                print(f"{fully_qualified_key}\t{redirect_target}")
            else:
                key_without_indexhtml = key.removesuffix("/index.html") + "/"
                print(f"{fully_qualified_key}\thttps://www.mongodb.com/docs/manual/{key_without_indexhtml}")
        else:
            print(f"{key}\thttps://www.mongodb.com/docs/manual/")