Fix published paths for transcoded media

repub/utils.py

@@ -2,7 +2,7 @@ import hashlib
 import mimetypes
 from enum import Enum
 from pathlib import Path
-from typing import Optional
+from typing import Any, Mapping, Optional, Sequence
 
 from scrapy.utils.python import to_bytes
 
@@ -42,6 +42,30 @@ def local_audio_path(s: str) -> str:
     return local_file_path(s)
 
 
+def variant_media_path(base_path: str, profile: Mapping[str, Any]) -> str:
+    return f"{base_path}-{profile['name']}.{profile['extension']}"
+
+
+def published_media_path(
+    file_type: FileType, source_url: str, profile: Mapping[str, Any]
+) -> str:
+    if file_type == FileType.AUDIO:
+        return variant_media_path(local_audio_path(source_url), profile)
+    if file_type == FileType.VIDEO:
+        return variant_media_path(local_video_path(source_url), profile)
+    raise ValueError(f"Unsupported file type for published media path: {file_type}")
+
+
+def canonical_published_media_path(
+    file_type: FileType, source_url: str, profiles: Sequence[Mapping[str, Any]]
+) -> str:
+    if not profiles:
+        raise ValueError(f"Missing transcode profiles for {file_type.value}")
+    # The first configured profile is the public URL contract. Reordering profiles
+    # changes published URLs for already-mirrored media.
+    return published_media_path(file_type, source_url, profiles[0])
+
+
 def determine_file_type(
     url: str, medium: Optional[str] = None, mimetype: Optional[str] = None
 ):