Develop

Resolves #18 (closed) #19 (closed) #20 (closed)

lpmn_client/file_handler.py

@@ -60,15 +60,18 @@ def upload_file(file_path: str) -> FileID:
 
 def download_file(
     file_id: FileID,
-    output_path: tp.Optional[str] = None
+    output_path: tp.Optional[str] = None,
+    filename: tp.Optional[str] = None
 ) -> str:
     """Downloads file by its FileID.
 
     Args:
         file_id (FileID): File id of a file to download
-        path (Optional[str]): Path for downloaded file.
+        output_path (Optional[str]): Directory to save downloaded file in.
             If not given the file will be saved in default location
             (config.ini).
+        filename (Optional[str]): Output filename without .zip format.
+            If not specified, filename is random.
     Returns:
         str. A path to downloaded file.
 
@@ -76,8 +79,8 @@ def download_file(
     response = requests.get(url=f"{Config.get('base_url')}/download{file_id}")
     if response.status_code == 404 or response.status_code == 500:
         raise FileNotFoundError("File not found")
-
-    filename = file_id.split("/")[-1] + ".zip"
+    if not filename:
+        filename = f"{file_id.split('/')[-1]}.zip"
     if not output_path:
         output_path = Config.get("output_path")
     os.makedirs(output_path, exist_ok=True)
@@ -86,3 +89,42 @@ def download_file(
     with open(output_path, "wb") as f:
         f.write(response.content)
     return output_path
+
+
+def download_file_as_dict(
+    file_id: FileID,
+) -> tp.Dict[str, str]:
+    """Downloads file by its FileID.
+
+    Args:
+        file_id (FileID): File id of a file to download
+
+    Returns:
+        dict. Dictionary mapping filenames in output zipfile to file contents.
+
+    """
+    def try_decode(bytes_data):
+        try:
+            bytes_data.decode("utf-8")
+            return True
+        except UnicodeDecodeError:
+            return False
+
+    response = requests.get(url=f"{Config.get('base_url')}/download{file_id}")
+    if response.status_code == 404 or response.status_code == 500:
+        raise FileNotFoundError("File not found")
+
+    zip_buffer = io.BytesIO(initial_bytes=response.content)
+    with zipfile.ZipFile(
+        zip_buffer,
+        mode="r",
+        compression=zipfile.ZIP_DEFLATED,
+    ) as zip_file:
+        return {
+            i.filename: zip_file.open(i).read().decode("utf-8")
+            for i in filter(
+                lambda x:
+                    try_decode(zip_file.read(x)) and not x.is_dir(),
+                zip_file.infolist()
+            )
+        }

[Mateusz Gniewkowski]

File is not longer a keyword in python3.*, but it is still not a great idea to call a variable this way. Many interpreters (gitlab included) use different color schema for those and therefore code is harder to read.

[Mateusz Gniewkowski]

Indentation.

Also, it looks cool, but I'm not entirely certain if not closing a file here is ok (I think it should be closed when you leave with statement, but it is rather python implementation dependent). Maybe consider using regular loop and closing the file?

[Szymon Ciombor]

Dict comprehension with filename keys and file content values. If file can be decoded to UTF-8 -> append to dictionary, else ignore