fix: removed download_file (#121)

## changes

- removed `download_file`
- refactored rest of code to use `download_files`

src/data_hub/api.py

@@ -239,48 +239,6 @@ def list_rows(
     )
 
 
-@beartype
-@ensure_client
-def download_file(
-    file_id: str,
-    *,
-    destination: str | Path = os.getcwd(),
-    client=None,
-    _stash: bool = False,
-) -> None:
-    """Download a file to a destination folder (workspace).
-
-    Download a file synchronously from Deep Origin
-    to folder on the local file system.
-
-    Args:
-        file_id: ID of the file on Deep Origin
-        destination: Path to the destination folder
-
-    """
-
-    if not os.path.isdir(destination):
-        raise DeepOriginException(
-            message=f"Destination `{destination}` should be a path for a folder."
-        )
-
-    file_name = _api.describe_file(
-        file_id=file_id,
-        client=client,
-        _stash=_stash,
-    ).name
-
-    url = _api.create_file_download_url(
-        file_id=file_id,
-        client=client,
-        _stash=_stash,
-    ).downloadUrl
-
-    save_path = os.path.join(destination, file_name)
-
-    download_sync(url, save_path)
-
-
 @beartype
 @ensure_client
 def upload_file(
@@ -929,9 +887,9 @@ def download(
     if PREFIXES.FILE in source:
         # this is a file
 
-        download_file(
-            file_id=source,
-            destination=destination,
+        download_files(
+            file_ids=[source],
+            save_to_dir=destination,
             client=client,
             _stash=_stash,
         )
@@ -1002,10 +960,11 @@ def download_database(
 
     # now download all files in the database
     if include_files:
-        file_ids = df.attrs["file_ids"]
-
-        for file_id in file_ids:
-            download_file(file_id, destination, client=client)
+        download_files(
+            file_ids=df.attrs["file_ids"],
+            save_to_dir=destination,
+            client=client,
+        )
 
     df.to_csv(os.path.join(destination, database_hid + ".csv"))
 
@@ -1084,6 +1043,7 @@ def get_dataframe(
             df = _make_deeporigin_dataframe(
                 data=data,
                 reference_ids=None,
+                file_ids=None,
                 db_row=db_row,
                 rows=None,
                 columns=None,
@@ -1169,6 +1129,7 @@ def get_dataframe(
         df = _make_deeporigin_dataframe(
             data=data,
             reference_ids=reference_ids,
+            file_ids=file_ids,
             db_row=db_row,
             rows=rows,
             columns=columns,
@@ -1192,6 +1153,7 @@ def _make_deeporigin_dataframe(
     *,
     data: dict,
     reference_ids: Optional[list],
+    file_ids: Optional[list],
     db_row: dict,
     columns: Optional[list],
     rows: Optional[list],
@@ -1203,6 +1165,12 @@ def _make_deeporigin_dataframe(
     df = DataFrame(data)
     if reference_ids is not None:
         df.attrs["reference_ids"] = list(set(reference_ids))
+        df.attrs["reference_ids"].sort()
+
+    if file_ids is not None:
+        df.attrs["file_ids"] = list(set(file_ids))
+        df.attrs["file_ids"].sort()
+
     df.attrs["id"] = db_row.id
     df.attrs["metadata"] = dict(db_row)
 
@@ -1223,8 +1191,9 @@ def _make_deeporigin_dataframe(
 @beartype
 @ensure_client
 def download_files(
-    files: Optional[list | dict] = None,
     *,
+    files: Optional[list[dict]] = None,
+    file_ids: Optional[list[str]] = None,
     save_to_dir: Path | str = Path("."),
     use_file_names: bool = True,
     client=None,
@@ -1233,21 +1202,37 @@ def download_files(
     """download multiple files in parallel to local disk
 
     Args:
-        files: list of files to download. These can be of type `types.list_files_response.Data` (as returned by api.list_files) or can be a list of strings of file IDs.
+        files: list of files to download. These can be a list of file_ids or a list of files as returned by api.list_files
         save_to_dir: directory to save files to on local computer
+        use_file_names: If `True`, refer to files by name rather than ID.
     """
 
-    if files is None:
-        files = list_files(client=client)
+    if not os.path.isdir(save_to_dir):
+        raise DeepOriginException(
+            message=f"Destination `{save_to_dir}` should be a path for a folder."
+        )
+
+    if files is None and file_ids is None:
+        # nothing provided, download everything
+        files = list_files(client=client, _stash=_stash)
+    elif files is not None and file_ids is None:
+        # list of files provided
+        pass
+    elif files is None and file_ids is not None:
+        # list of file IDs provided
+        files = list_files(
+            file_ids=file_ids,
+            client=client,
+            _stash=_stash,
+        )
 
-    if isinstance(files, dict):
-        files = [files]
+    else:
+        raise DeepOriginException("Only one of `files` or `file_ids` can be provided")
 
     if isinstance(save_to_dir, str):
         save_to_dir = Path(save_to_dir)
 
-    if isinstance(files[0], dict):
-        file_ids = [item.file.id for item in files]
+    file_ids = [item.file.id for item in files]
 
     if use_file_names:
         save_paths = [save_to_dir / item.file.name for item in files]

tests/fixtures/responses/convert_id_format.json

@@ -4,16 +4,30 @@
             "id": "_row:cbOepriiytSaPfpqetT4C",
             "hid": "dfsd-306"
         },
+        {
+            "id": "_row:rakyaiwKrKsS4JT9IJsXn",
+            "hid": "exp-1"
+        },
         {
             "id": "_row:ku2ZdpZ0VkYgOj1Y37rSi",
             "hid": "dna-8"
+        }
+    ],
+    "{\"conversions\": [{\"id\": \"_row:cbOepriiytSaPfpqetT4C\"}, {\"id\": \"_row:ku2ZdpZ0VkYgOj1Y37rSi\"}, {\"id\": \"_row:rakyaiwKrKsS4JT9IJsXn\"}]}": [
+        {
+            "id": "_row:cbOepriiytSaPfpqetT4C",
+            "hid": "dfsd-306"
         },
         {
             "id": "_row:rakyaiwKrKsS4JT9IJsXn",
             "hid": "exp-1"
+        },
+        {
+            "id": "_row:ku2ZdpZ0VkYgOj1Y37rSi",
+            "hid": "dna-8"
         }
     ],
-    "{\"conversions\": [{\"id\": \"_row:cbOepriiytSaPfpqetT4C\"}, {\"id\": \"_row:ku2ZdpZ0VkYgOj1Y37rSi\"}, {\"id\": \"_row:rakyaiwKrKsS4JT9IJsXn\"}]}": [
+    "{\"conversions\": [{\"id\": \"_row:cbOepriiytSaPfpqetT4C\"}, {\"id\": \"_row:rakyaiwKrKsS4JT9IJsXn\"}, {\"id\": \"_row:ku2ZdpZ0VkYgOj1Y37rSi\"}]}": [
         {
             "id": "_row:cbOepriiytSaPfpqetT4C",
             "hid": "dfsd-306"

tests/fixtures/responses/create_database.json

@@ -1,11 +1,11 @@
 {
     "{\"database\": {\"hid\": \"tc-4Qzkrn57rM-db\", \"hidPrefix\": \"tc-4Qzkrn57rM-db\", \"name\": \"tc-4Qzkrn57rM-db\", \"parentId\": \"registry\"}}": {
-        "id": "_database:zN9taqV6WxLzjWxKtxwce",
+        "id": "_database:LanvJgEqhfNVHtgeZgsnT",
         "type": "database",
         "hid": "tc-4Qzkrn57rM-db",
         "name": "tc-4Qzkrn57rM-db",
-        "dateCreated": "2024-11-19 03:19:28.539577",
-        "dateUpdated": "2024-11-19 03:19:28.539577",
+        "dateCreated": "2024-11-25 19:08:42.989402",
+        "dateUpdated": "2024-11-25 19:08:42.989402",
         "createdByUserDrn": "<redacted>",
         "parentId": "_workspace:IYTHdIjtuSvnOekwQjUlN",
         "hidPrefix": "tc-4Qzkrn57rM-db",

tests/fixtures/responses/create_file_upload.json

@@ -43,5 +43,20 @@
             "dateUpdated": "2024-11-19T03:19:28.397Z",
             "createdByUserDrn": "<redacted>"
         }
+    },
+    "{\"checksum_sha256\": \"rRS/5tR0d/MjICI1AdOr7CAyl1JIvgc79n3isXNAf9g=\", \"content_length\": \"10682\", \"content_type\": \"text/x-python\", \"name\": \"test_data_hub.py\"}": {
+        "uploadUrl": "<redacted>",
+        "file": {
+            "name": "test_data_hub.py",
+            "contentType": "text/x-python",
+            "contentLength": 10682,
+            "checksumSha256": "rRS/5tR0d/MjICI1AdOr7CAyl1JIvgc79n3isXNAf9g=",
+            "id": "_file:ZKPAZZ8H8eKtrXRwZ9CON",
+            "status": "ready",
+            "uri": "<redacted>",
+            "dateCreated": "2024-11-25T19:08:43.221Z",
+            "dateUpdated": "2024-11-25T19:08:43.221Z",
+            "createdByUserDrn": "<redacted>"
+        }
     }
 }

tests/fixtures/responses/create_workspace.json

@@ -1,11 +1,11 @@
 {
     "{\"workspace\": {\"hid\": \"tc-4Qzkrn57rM-ws\", \"name\": \"tc-4Qzkrn57rM-ws\", \"parentId\": null}}": {
-        "id": "_workspace:WGKwIfOV6ySsJMkbtvRvC",
+        "id": "_workspace:MptBTTL9qUbbSbUiAydsN",
         "type": "workspace",
         "hid": "tc-4Qzkrn57rM-ws",
         "name": "tc-4Qzkrn57rM-ws",
-        "dateCreated": "2024-11-19 03:19:28.75955",
-        "dateUpdated": "2024-11-19 03:19:28.75955",
+        "dateCreated": "2024-11-25 19:08:43.22463",
+        "dateUpdated": "2024-11-25 19:08:43.22463",
         "createdByUserDrn": "<redacted>"
     }
 }

tests/fixtures/responses/describe_database_stats.json

@@ -1,5 +1,5 @@
 {
     "{\"database_id\": \"kitchen-sink\"}": {
-        "rowCount": 127
+        "rowCount": 0
     }
 }

tests/fixtures/responses/describe_row.json

@@ -329,6 +329,7 @@
         ]
     },
     "{\"fields\": true, \"row_id\": \"dfsd-306\"}": {
+        "parentHid": "ligands",
         "type": "row",
         "id": "_row:cbOepriiytSaPfpqetT4C",
         "hid": "dfsd-306",
@@ -729,6 +730,7 @@
         ]
     },
     "{\"fields\": false, \"row_id\": \"dfsd-306\"}": {
+        "parentHid": "ligands",
         "type": "row",
         "id": "_row:cbOepriiytSaPfpqetT4C",
         "hid": "dfsd-306",

tests/fixtures/responses/ensure_rows.json

@@ -2,12 +2,13 @@
     "{\"database_id\": \"kitchen-sink\", \"rows\": [{\"row\": {}}]}": {
         "rows": [
             {
+                "parentHid": "kitchen-sink",
                 "type": "row",
-                "id": "_row:OIy39TGcV7jqLbCHsdhTO",
-                "hid": "ks-136",
+                "id": "_row:_database:swAKhqQzxqTB7cm2FeVcW:iqZurZ4W4UzDwpRgo5Vuj",
+                "hid": "ks-155",
                 "parentId": "_database:swAKhqQzxqTB7cm2FeVcW",
-                "dateCreated": "2024-11-19 03:19:29.007535",
-                "dateUpdated": "2024-11-19 03:19:29.007535",
+                "dateCreated": "2024-11-25 19:08:43.711226",
+                "dateUpdated": "2024-11-25 19:08:43.711226",
                 "createdByUserDrn": "<redacted>",
                 "validationStatus": "valid"
             }