standardize : allow users to specify output encoding

clevercsv/console/commands/standardize.py

@@ -69,6 +69,16 @@ def register(self) -> None:
             ),
             default=[],
         )
+        self.add_argument(
+            "-E",
+            "--target_encoding",
+            help="Set the encoding of the output file(s)",
+            description=(
+                "If ommited, the output file encoding while be the same "
+                "as that of the original file."            
+            ),
+            type=str
+        )
         self.add_argument(
             "-i",
             "--in-place",
@@ -115,6 +125,7 @@ def handle(self) -> int:
         encodings = self.args.encoding
         num_chars = parse_int(self.args.num_chars, "num-chars")
         in_place = self.args.in_place
+        target_encoding = self.args.target_encoding
 
         if in_place and outputs:
             print(
@@ -154,6 +165,7 @@ def handle(self) -> int:
                 encoding=encoding,
                 verbose=verbose,
                 num_chars=num_chars,
+                target_encoding=target_encoding
             )
             if retval > 0 and global_retval == 0:
                 global_retval = retval
@@ -168,6 +180,7 @@ def handle_path(
         encoding: Optional[str] = None,
         num_chars: Optional[int] = None,
         verbose: bool = False,
+        target_encoding: Optional[str] = None
     ) -> int:
         encoding = encoding or get_encoding(path)
         dialect = detect_dialect(
@@ -178,23 +191,24 @@ def handle_path(
             return 1
 
         if self.args.in_place:
-            return self._in_place(path, dialect, encoding)
+            return self._in_place(path, dialect, encoding, target_encoding)
         elif output is None:
-            return self._to_stdout(path, dialect, encoding)
-        return self._to_file(path, output, dialect, encoding)
+            return self._to_stdout(path, dialect, encoding, target_encoding)
+        return self._to_file(path, output, dialect, encoding, target_encoding)
 
     def _write_transposed(
         self,
         path: StrPath,
         stream: SupportsWrite[str],
         dialect: SimpleDialect,
         encoding: Optional[str],
+        target_encoding: Optional[str]
     ) -> None:
         with open(path, "r", newline="", encoding=encoding) as fp:
             read = reader(fp, dialect=dialect)
             rows = list(read)
         rows = list(map(list, zip(*rows)))
-        write = writer(stream, dialect="excel")
+        write = writer(stream, dialect="excel", encoding=target_encoding or encoding)
         for row in rows:
             write.writerow(row)
 
@@ -204,10 +218,11 @@ def _write_direct(
         stream: SupportsWrite[str],
         dialect: SimpleDialect,
         encoding: Optional[str],
+        target_encoding: Optional[str]
     ) -> None:
         with open(path, "r", newline="", encoding=encoding) as fp:
             read = reader(fp, dialect=dialect)
-            write = writer(stream, dialect="excel")
+            write = writer(stream, dialect="excel", encoding=target_encoding or encoding)
             for row in read:
                 write.writerow(row)
 
@@ -217,14 +232,15 @@ def _write_to_stream(
         stream: SupportsWrite[str],
         dialect: SimpleDialect,
         encoding: Optional[str],
+        target_encoding: Optional[str]
     ) -> None:
         if self.args.transpose:
-            self._write_transposed(path, stream, dialect, encoding)
+            self._write_transposed(path, stream, dialect, encoding, target_encoding)
         else:
-            self._write_direct(path, stream, dialect, encoding)
+            self._write_direct(path, stream, dialect, encoding, target_encoding)
 
     def _in_place(
-        self, path: StrPath, dialect: SimpleDialect, encoding: Optional[str]
+        self, path: StrPath, dialect: SimpleDialect, encoding: Optional[str], target_encoding: Optional[str]
     ) -> int:
         """In-place mode overwrites the input file, if necessary
 
@@ -235,8 +251,8 @@ def _in_place(
 
         """
         tmpfd, tmpfname = tempfile.mkstemp(prefix="clevercsv_", suffix=".csv")
-        tmpid = os.fdopen(tmpfd, "w", newline="", encoding=encoding)
-        self._write_to_stream(path, tmpid, dialect, encoding)
+        tmpid = os.fdopen(tmpfd, "w", newline="", encoding=target_encoding or encoding)
+        self._write_to_stream(path, tmpid, dialect, encoding, target_encoding)
         tmpid.close()
 
         previous_sha1 = sha1sum(path)
@@ -249,10 +265,10 @@ def _in_place(
         return 2
 
     def _to_stdout(
-        self, path: StrPath, dialect: SimpleDialect, encoding: Optional[str]
+        self, path: StrPath, dialect: SimpleDialect, encoding: Optional[str], target_encoding: Optional[str]
     ) -> int:
         stream = io.StringIO(newline="")
-        self._write_to_stream(path, stream, dialect, encoding)
+        self._write_to_stream(path, stream, dialect, encoding, target_encoding)
         print(stream.getvalue(), end="")
         stream.close()
         return 0
@@ -263,7 +279,8 @@ def _to_file(
         output: StrPath,
         dialect: SimpleDialect,
         encoding: Optional[str],
+        target_encoding: Optional[str]
     ) -> int:
-        with open(output, "w", newline="", encoding=encoding) as fp:
-            self._write_to_stream(path, fp, dialect, encoding)
+        with open(output, "w", newline="", encoding=target_encoding or encoding) as fp:
+            self._write_to_stream(path, fp, dialect, encoding, target_encoding)
         return 0