Quelle  upload.py   Sprache: Python

# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, # You can obtain one at http://mozilla.org/MPL/2.0/.

import io
import mimetypes
import os
import sys
from concurrent import futures
from pprint import pprint

import boto3
import botocore
import requests
from mozbuild.util import memoize


@memoize
def create_aws_session():
    """
    This function creates an aws session that is
    shared between upload and delete both.
    """
    region = "us-west-2"
    level = os.environ.get("MOZ_SCM_LEVEL", "1")
    bucket = {
        "1": "gecko-docs.mozilla.org-l1",
        "2": "gecko-docs.mozilla.org-l2",
        "3": "gecko-docs.mozilla.org",
    }[level]
    secrets_url = "http://taskcluster/secrets/v1/secret/"
    secrets_url += "project/releng/gecko/build/level-{}/gecko-docs-upload".format(level)

    # Get the credentials from the TC secrets service.  Note that these
    # differ per SCM level
    if "TASK_ID" in os.environ:
        print("Using AWS credentials from the secrets service")
        session = requests.Session()
        res = session.get(secrets_url)
        res.raise_for_status()
        secret = res.json()["secret"]
        session = boto3.session.Session(
            aws_access_key_id=secret["AWS_ACCESS_KEY_ID"],
            aws_secret_access_key=secret["AWS_SECRET_ACCESS_KEY"],
            region_name=region,
        )
    else:
        print("Trying to use your AWS credentials..")
        session = boto3.session.Session(region_name=region)

    s3 = session.client("s3", config=botocore.client.Config(max_pool_connections=20))

    return s3, bucket


@memoize
def get_s3_keys(s3, bucket):
    kwargs = {"Bucket": bucket}
    all_keys = []
    while True:
        response = s3.list_objects_v2(**kwargs)
        for obj in response["Contents"]:
            all_keys.append(obj["Key"])

        try:
            kwargs["ContinuationToken"] = response["NextContinuationToken"]
        except KeyError:
            break

    return all_keys


def s3_set_redirects(redirects):
    s3, bucket = create_aws_session()

    configuration = {"IndexDocument": {"Suffix": "index.html"}, "RoutingRules": []}

    for path, redirect in redirects.items():
        rule = {
            "Condition": {"KeyPrefixEquals": path},
            "Redirect": {"ReplaceKeyPrefixWith": redirect},
        }
        if os.environ.get("MOZ_SCM_LEVEL") == "3":
            rule["Redirect"]["HostName"] = "firefox-source-docs.mozilla.org"

        configuration["RoutingRules"].append(rule)

    s3.put_bucket_website(
        Bucket=bucket,
        WebsiteConfiguration=configuration,
    )


def s3_delete_missing(files, key_prefix=None):
    """Delete files in the S3 bucket.

    Delete files on the S3 bucket that doesn't match the files
    given as the param. If the key_prefix is not specified, missing
    files that has main/ as a prefix will be removed. Otherwise, it
    will remove files with the same prefix as key_prefix.
    """
    s3, bucket = create_aws_session()
    files_on_server = get_s3_keys(s3, bucket)
    if key_prefix:
        files_on_server = [
            path for path in files_on_server if path.startswith(key_prefix)
        ]
    else:
        files_on_server = [
            path for path in files_on_server if not path.startswith("main/")
        ]
    files = [key_prefix + "/" + path if key_prefix else path for path, f in files]
    files_to_delete = [path for path in files_on_server if path not in files]

    query_size = 1000
    while files_to_delete:
        keys_to_remove = [{"Key": key} for key in files_to_delete[:query_size]]
        response = s3.delete_objects(
            Bucket=bucket,
            Delete={
                "Objects": keys_to_remove,
            },  # NOQA
        )
        pprint(response, indent=2)
        files_to_delete = files_to_delete[query_size:]


def s3_upload(files, key_prefix=None):
    """Upload files to an S3 bucket.

    ``files`` is an iterable of ``(path, BaseFile)`` (typically from a
    mozpack Finder).

    Keys in the bucket correspond to source filenames. If ``key_prefix`` is
    defined, key names will be ``<key_prefix>/<path>``.
    """
    s3, bucket = create_aws_session()

    def upload(f, path, bucket, key, extra_args):
        # Need to flush to avoid buffering/interleaving from multiple threads.
        sys.stdout.write("uploading %s to %s\n" % (path, key))
        sys.stdout.flush()
        s3.upload_fileobj(f, bucket, key, ExtraArgs=extra_args)

    fs = []
    with futures.ThreadPoolExecutor(20) as e:
        for path, f in files:
            content_type, content_encoding = mimetypes.guess_type(path)
            extra_args = {}
            if content_type:
                if content_type.startswith("text/"):
                    content_type += '; charset="utf-8"'
                extra_args["ContentType"] = content_type
            if content_encoding:
                extra_args["ContentEncoding"] = content_encoding

            if key_prefix:
                key = "%s/%s" % (key_prefix, path)
            else:
                key = path

            # The file types returned by mozpack behave like file objects. But
            # they don't accept an argument to read(). So we wrap in a BytesIO.
            fs.append(
                e.submit(upload, io.BytesIO(f.read()), path, bucket, key, extra_args)
            )

    s3_delete_missing(files, key_prefix)
    # Need to do this to catch any exceptions.
    for f in fs:
        f.result()