fix: adjust file ordering based on previewer ability

also some comments, note that the actual files.order array seems to be dropped by the
Invenio API, slight clean up in a few places

migrate/import.py

@@ -56,12 +56,13 @@ def create_draft(record: dict) -> dict:
     return draft_record
 
 
-def add_files(dir: Path, attachments: list[dict], draft: dict):
+def add_files(dir: Path, record: Record, draft: dict):
     # add files to draft record
     # three steps: initiate, upload, and commit
+    # ! Unable to set files order or default_preview like API docs suggest
 
     # initiate all at once
-    keys = [{"key": attachment["filename"]} for attachment in attachments]
+    keys = [{"key": att["filename"]} for att in record.files]
     init_response: requests.Response = requests.post(
         draft["links"]["files"],
         data=json.dumps(keys),
@@ -75,8 +76,8 @@ def add_files(dir: Path, attachments: list[dict], draft: dict):
 
     # upload one by one
     # TODO use httpx to do in parallel?
-    for attachment in attachments:
-        binary_headers: dict[str, str] = headers
+    for attachment in record.files:
+        binary_headers: dict[str, str] = headers.copy()
         binary_headers["Content-Type"] = "application/octet-stream"
         with open(dir / attachment["filename"], "rb") as f:
             upload_response: requests.Response = requests.put(
@@ -126,17 +127,24 @@ def publish(draft: dict) -> dict:
 )
 @click.help_option("-h", "--help")
 @click.argument("dir", type=click.Path(exists=True), required=True)
+# TODO option to ignore errors which skips assert statements & response.raise_for_status()
 # @click.option("--ignore-errors", "-i", help="Ignore errors and continue", is_flag=True)
 @click.option("--verbose", "-v", "is_verbose", help="Print more output", is_flag=True)
 def main(dir: str, is_verbose: bool):
     global verbose
     verbose = is_verbose
+
     item = get_item(Path(dir) / "metadata" / "item.json")
     record = Record(item)
+
     click.echo(f"Importing {record.title} from {dir}...")
     draft = create_draft(record.get())
-    add_files(Path(dir), record.files, draft)
+
+    if len(record.files):
+        add_files(Path(dir), record, draft)
+
     published_record = publish(draft)
+
     # TODO add to community
     click.echo(f"Published: {published_record['links']['self_html']}")
 

migrate/record.py

@@ -464,9 +464,11 @@ def get(self) -> dict[str, Any]:
             },
             # ! blocked until we know what custom fields we'll have
             "custom_fields": {},
-            # TODO add files, figure out best one to show first (prefer image formats?)
             "files": {
                 "enabled": bool(len(self.files)),
+                # ! API drops these, whether we define before adding files or after
+                "order": [att["filename"] for att in self.files],
+                "default_preview": self.files[0]["filename"] if len(self.files) else "",
             },
             # "files": {
             #     "enabled": bool(len(self.files)),

migrate/tests.py

@@ -27,8 +27,8 @@ def test_mklist(input, expect):
         ([], []),
         ([{"filename": "file.py"}], [{"filename": "file.py"}]),
         (
-            [{"filename": "binary.gzip"}, {"filename": "film.mov"}],
-            [{"filename": "film.mov"}, {"filename": "binary.gzip"}],
+            [{"filename": "binary.gzip"}, {"filename": "img.gif"}],
+            [{"filename": "img.gif"}, {"filename": "binary.gzip"}],
         ),
         (
             [{"filename": "print.pdf"}, {"filename": "word.docx"}],
@@ -60,14 +60,14 @@ def test_mklist(input, expect):
             ],
             [
                 {"filename": "img.tiff"},
-                {"filename": "img.webp"},
-                {"filename": "movie.mp4"},
                 {"filename": "doc.pdf"},
+                {"filename": "zip.zip"},
+                {"filename": "unknown"},
                 {"filename": "plain.txt"},
+                {"filename": "img.webp"},
+                {"filename": "movie.mp4"},
                 {"filename": "song.mp3"},
                 {"filename": "app.exe"},
-                {"filename": "zip.zip"},
-                {"filename": "unknown"},
             ],
         ),
     ],

migrate/utils.py

@@ -59,25 +59,25 @@ def to_edtf(s) -> str | None:
 
 
 def visual_mime_type_sort(attachment) -> int:
-    # Sort EQUELLA attachment dicts by MIME type, prefer visual types
-    # Order: TIFF > Other images > Video > PDF > Other Text > Binary ("application") > Unknown
+    # Sort EQUELLA attachment dicts by MIME type, types previewable in Invenio
+    # which is (according to readme): PDF, ZIP, CSV, MARKDOWN, XML, JSON, PNG, JPG, GIF
+    # https://github.com/inveniosoftware/invenio-previewer
+    # Order: TIFF > Non-HEIC/WBEP Images > PDF > Markdown, CSV, XML > JSON > ZIP > Everything else (not previewable)
     # https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/MIME_types#types
     mt: str | None = mimetypes.guess_type(attachment["filename"])[0]
     type, subtype = mt.split("/") if mt else ("unknown", "unknown")
     match type, subtype:
         case "image", "tiff":
             return 0
-        case "image", _:
+        case "image", _ if subtype not in ["heic", "webp"]:
             return 10
-        case "video", _:
-            return 20
         case "application", "pdf":
+            return 20
+        case "text", _ if subtype in ["csv", "markdown", "xml"]:
             return 30
-        case "text", _:
+        case "application", "json":
             return 40
-        case "audio", _:
+        case "application", _ if subtype in ["zip", "x-zip-compressed"]:
             return 50
-        case "application", _:
+        case _, _:  # model, font types, subtypes not covered above
             return 60
-        case _, _:  # model, font
-            return 70