From 478c0202110da39a18fc0fc16b4f2b6cb96ef7d7 Mon Sep 17 00:00:00 2001
From: Stepan Oksanichenko <soksanichenko@almalinux.org>
Date: Tue, 11 Jun 2024 23:43:51 +0300
Subject: [PATCH] ALBS-1216: Implement support for new SHA512 external sources
 file format in almalinux-git-utils

- New key argument `--sha512` for the sources uploader if you need to upload a source in new format (using sha512 checksum)
- A separate recognizer of a checksum type in metadata file
- The sources downloader can automatically detect a type of checksum
---
 almalinux/gitutils/blob_upload.py | 32 +++++++++++++++++++++++++------
 almalinux/gitutils/common.py      | 12 +++++++++++-
 2 files changed, 37 insertions(+), 7 deletions(-)

diff --git a/almalinux/gitutils/blob_upload.py b/almalinux/gitutils/blob_upload.py
index f378bd6..d7b7f88 100644
--- a/almalinux/gitutils/blob_upload.py
+++ b/almalinux/gitutils/blob_upload.py
@@ -40,6 +40,8 @@ def init_arg_parser() -> argparse.ArgumentParser:
                             help='output metadata file path')
     arg_parser.add_argument('-a', '--append-metadata', action='store_true',
                             help='Append to an output metadata')
+    arg_parser.add_argument('--sha512', action='store_true',
+                            help='Use a new format of hashsum')
     arg_parser.add_argument('-p', '--private', action='store_true',
                             help='set uploaded file mode to private. All '
                                  'uploaded files are public by default')
@@ -51,17 +53,21 @@ def init_arg_parser() -> argparse.ArgumentParser:
     return arg_parser
 
 
-def iter_files(files: List[str]) -> Iterator[Tuple[str, str, str]]:
+def iter_files(
+    files: List[str],
+    sha512: bool
+) -> Iterator[Tuple[str, str, str]]:
     """
     Iterates over a list of files and calculates checksums for them.
 
     Args:
         files: List of files.
+        sha512: True if we use new format of hashsum.
 
     Returns:
         Iterator over files and their checksums.
     """
-    checksum_type = 'sha1'
+    checksum_type = 'sha512' if sha512 else 'sha1'
     for rel_path in files:
         file_path = normalize_path(rel_path)
         checksum = get_file_checksum(file_path, checksum_type)
@@ -104,7 +110,9 @@ def upload_file(s3_client, bucket_name: str, file_path: str,
 
 
 def get_file_iterator(
-        files: List[str], metadata_path: Optional[str]
+    files: List[str],
+    metadata_path: Optional[str],
+    sha512: bool,
 ) -> Iterator[Tuple[str, str, str]]:
     """
     Finds a suitable file iterator for given arguments.
@@ -112,12 +120,16 @@ def get_file_iterator(
     Args:
         files: List of files.
         metadata_path: Metadata file path.
+        sha512: True if we use new format of hashsum.
 
     Returns:
         File iterator.
     """
     if files:
-        iterator = iter_files(files)
+        iterator = iter_files(
+            files=files,
+            sha512=sha512,
+        )
     else:
         if not metadata_path:
             metadata_path = find_metadata_file(os.getcwd())
@@ -130,10 +142,18 @@ def main():
     args = arg_parser.parse_args(sys.argv[1:])
     configure_logger(args.verbose)
     s3_client = boto3.client('s3')
-    iterator = get_file_iterator(args.file, args.input_metadata)
+    iterator = get_file_iterator(
+        args.file,
+        args.input_metadata,
+        args.sha512,
+    )
     out_fd = None
+    if args.append_metadata:
+        file_mdoe = 'a'
+    else:
+        file_mdoe = 'w'
     if args.output_metadata:
-        out_fd = open(args.output_metadata, 'a' if args.append_metadata else 'w')
+        out_fd = open(args.output_metadata, file_mdoe)
     try:
         for rel_path, checksum, checksum_type in iterator:
             file_path = normalize_path(rel_path)
diff --git a/almalinux/gitutils/common.py b/almalinux/gitutils/common.py
index b35085f..34e2fa9 100644
--- a/almalinux/gitutils/common.py
+++ b/almalinux/gitutils/common.py
@@ -90,6 +90,16 @@ def get_file_checksum(file_path: str, checksum_type: str = 'sha1',
     return hasher.hexdigest()
 
 
+def extract_checksum_and_file_path(line: str) -> Tuple[str, str]:
+    if 'SHA512' in line.upper():
+        file_path, checksum = line.split('=')
+        file_path = file_path.strip().split(' ')[1].strip('(').strip(')')
+        checksum = checksum.strip()
+    else:
+        checksum, file_path = line.split()
+    return checksum, file_path
+
+
 def iter_metadata(metadata_path: str) -> Iterator[Tuple[str, str, str]]:
     """
     Iterates over records in a CentOS git repository-compatible metadata file.
@@ -102,7 +112,7 @@ def iter_metadata(metadata_path: str) -> Iterator[Tuple[str, str, str]]:
     """
     with open(metadata_path, 'r') as fd:
         for line in fd:
-            checksum, file_path = line.split()
+            checksum, file_path = extract_checksum_and_file_path(line)
             checksum_type = detect_checksum_type(checksum)
             yield file_path, checksum, checksum_type