`revision_pattern_from_build_bucket_path` can return a bad revision pattern if bucket name contains numerics

[jasocrow]

Consider:

RELEASE_BUILD_BUCKET_PATH = gs://builds.fuzz-nprd-53709/([0-9]+).*.zip

def revision_pattern_from_build_bucket_path(bucket_path):
  """Get the revision pattern from a build bucket path."""
  return '.*?' + os.path.basename(bucket_path)

revision_pattern_from_build_bucket_path will return .*?([0-9]+).*.zip

for a path like gs://builds.fuzz-nprd-53709/123_01bced44.zip, the revision will be 53709 instead of 123.

Some code that shows this:

import re
import os

RELEASE_BUILD_BUCKET_PATH = "gs://builds.fuzz-nprd-53709/([0-9]+).*.zip"

def revision_pattern_from_build_bucket_path(bucket_path):
  """Get the revision pattern from a build bucket path."""
  return '.*?' + os.path.basename(bucket_path)

build_url = 'gs://builds.fuzz-nprd-53709/123_01bced44.zip'

revision_pattern = revision_pattern_from_build_bucket_path(RELEASE_BUILD_BUCKET_PATH)
print(f"{revision_pattern}")

match = re.match(revision_pattern, build_url)
revision = match.group(1)
print(f"{revision}")

% python3 show-bad-revision-number.py
.*?([0-9]+).*.zip
53709