Adding support for gzipped ASC logging file (.asc.gz) (#1138)

can/__init__.py

@@ -25,7 +25,7 @@
 )
 
 from .io import Logger, SizedRotatingLogger, Printer, LogReader, MessageSync
-from .io import ASCWriter, ASCReader
+from .io import ASCWriter, ASCReader, GzipASCWriter, GzipASCReader
 from .io import BLFReader, BLFWriter
 from .io import CanutilsLogReader, CanutilsLogWriter
 from .io import CSVWriter, CSVReader

can/io/__init__.py

@@ -8,7 +8,7 @@
 from .player import LogReader, MessageSync
 
 # Format specific
-from .asc import ASCWriter, ASCReader
+from .asc import ASCWriter, ASCReader, GzipASCWriter, GzipASCReader
 from .blf import BLFReader, BLFWriter
 from .canutils import CanutilsLogReader, CanutilsLogWriter
 from .csv import CSVWriter, CSVReader

can/io/asc.py

@@ -5,13 +5,14 @@
     - https://bitbucket.org/tobylorenz/vector_asc/src/47556e1a6d32c859224ca62d075e1efcc67fa690/src/Vector/ASC/tests/unittests/data/CAN_Log_Trigger_3_2.asc?at=master&fileviewer=file-view-default
     - under `test/data/logfile.asc`
 """
-
-from typing import cast, Any, Generator, IO, List, Optional, Dict
+import gzip
+from typing import cast, Any, Generator, IO, List, Optional, Dict, Union
 
 from datetime import datetime
 import time
 import logging
 
+from .. import typechecking
 from ..message import Message
 from ..listener import Listener
 from ..util import channel2int
@@ -396,3 +397,69 @@ def on_message_received(self, msg: Message) -> None:
                 data=" ".join(data),
             )
         self.log_event(serialized, msg.timestamp)
+
+
+class GzipASCReader(ASCReader):
+    """Gzipped version of :class:`~can.ASCReader`"""
+
+    def __init__(
+        self,
+        file: Union[typechecking.FileLike, typechecking.StringPathLike],
+        base: str = "hex",
+        relative_timestamp: bool = True,
+    ):
+        """
+        :param file: a path-like object or as file-like object to read from
+                     If this is a file-like object, is has to opened in text
+                     read mode, not binary read mode.
+        :param base: Select the base(hex or dec) of id and data.
+                     If the header of the asc file contains base information,
+                     this value will be overwritten. Default "hex".
+        :param relative_timestamp: Select whether the timestamps are
+                     `relative` (starting at 0.0) or `absolute` (starting at
+                     the system time). Default `True = relative`.
+        """
+        self._fileobj = None
+        if file is not None and (hasattr(file, "read") and hasattr(file, "write")):
+            # file is None or some file-like object
+            self._fileobj = file
+        super(GzipASCReader, self).__init__(
+            gzip.open(file, mode="rt"), base, relative_timestamp
+        )
+
+    def stop(self) -> None:
+        super(GzipASCReader, self).stop()
+        if self._fileobj is not None:
+            self._fileobj.close()
+
+
+class GzipASCWriter(ASCWriter):
+    """Gzipped version of :class:`~can.ASCWriter`"""
+
+    def __init__(
+        self,
+        file: Union[typechecking.FileLike, typechecking.StringPathLike],
+        channel: int = 1,
+        compresslevel: int = 6,
+    ):
+        """
+        :param file: a path-like object or as file-like object to write to
+                     If this is a file-like object, is has to opened in text
+                     write mode, not binary write mode.
+        :param channel: a default channel to use when the message does not
+                        have a channel set
+        :param compresslevel: Gzip compresslevel, see
+                              :class:`~gzip.GzipFile` for details. The default is 6.
+        """
+        self._fileobj = None
+        if file is not None and (hasattr(file, "read") and hasattr(file, "write")):
+            # file is None or some file-like object
+            self._fileobj = file
+        super(GzipASCWriter, self).__init__(
+            gzip.open(file, mode="wt", compresslevel=compresslevel), channel
+        )
+
+    def stop(self) -> None:
+        super(GzipASCWriter, self).stop()
+        if self._fileobj is not None:
+            self._fileobj.close()

can/io/logger.py

@@ -21,7 +21,7 @@
 from ..message import Message
 from ..listener import Listener
 from .generic import BaseIOHandler, FileIOMessageWriter
-from .asc import ASCWriter
+from .asc import ASCWriter, GzipASCWriter
 from .blf import BLFWriter
 from .canutils import CanutilsLogWriter
 from .csv import CSVWriter
@@ -36,6 +36,7 @@ class Logger(BaseIOHandler, Listener):  # pylint: disable=abstract-method
 
     The format is determined from the file format which can be one of:
       * .asc: :class:`can.ASCWriter`
+      * .asc.gz: :class:`can.CompressedASCWriter`
       * .blf :class:`can.BLFWriter`
       * .csv: :class:`can.CSVWriter`
       * .db: :class:`can.SqliteWriter`
@@ -54,6 +55,7 @@ class Logger(BaseIOHandler, Listener):  # pylint: disable=abstract-method
     fetched_plugins = False
     message_writers = {
         ".asc": ASCWriter,
+        ".asc.gz": GzipASCWriter,
         ".blf": BLFWriter,
         ".csv": CSVWriter,
         ".db": SqliteWriter,
@@ -83,7 +85,7 @@ def __new__(  # type: ignore
             )
             Logger.fetched_plugins = True
 
-        suffix = pathlib.PurePath(filename).suffix.lower()
+        suffix = "".join(s.lower() for s in pathlib.PurePath(filename).suffixes)
         try:
             return cast(
                 Listener, Logger.message_writers[suffix](filename, *args, **kwargs)

can/io/player.py

@@ -14,7 +14,7 @@
     import can
 
 from .generic import BaseIOHandler, MessageReader
-from .asc import ASCReader
+from .asc import ASCReader, GzipASCReader
 from .blf import BLFReader
 from .canutils import CanutilsLogReader
 from .csv import CSVReader
@@ -27,6 +27,7 @@ class LogReader(BaseIOHandler):
 
     The format is determined from the file format which can be one of:
       * .asc
+      * .asc.gz
       * .blf
       * .csv
       * .db
@@ -49,6 +50,7 @@ class LogReader(BaseIOHandler):
     fetched_plugins = False
     message_readers = {
         ".asc": ASCReader,
+        ".asc.gz": GzipASCReader,
         ".blf": BLFReader,
         ".csv": CSVReader,
         ".db": SqliteReader,
@@ -75,7 +77,7 @@ def __new__(  # type: ignore
             )
             LogReader.fetched_plugins = True
 
-        suffix = pathlib.PurePath(filename).suffix.lower()
+        suffix = "".join(s.lower() for s in pathlib.PurePath(filename).suffixes)
         try:
             return typing.cast(
                 MessageReader,

test/logformats_test.py

@@ -11,7 +11,7 @@
 
 TODO: correctly set preserves_channel and adds_default_channel
 """
-
+import gzip
 import logging
 import unittest
 import tempfile
@@ -555,6 +555,34 @@ def test_can_and_canfd_error_frames(self):
         self.assertMessagesEqual(actual, expected_messages)
 
 
+class TestGzipASCFileFormat(ReaderWriterTest):
+    """Tests can.GzipASCWriter and can.GzipASCReader"""
+
+    def _setup_instance(self):
+        super()._setup_instance_helper(
+            can.GzipASCWriter,
+            can.GzipASCReader,
+            binary_file=True,
+            check_comments=True,
+            preserves_channel=False,
+            adds_default_channel=0,
+        )
+
+    def assertIncludesComments(self, filename):
+        """
+        Ensures that all comments are literally contained in the given file.
+
+        :param filename: the path-like object to use
+        """
+        if self.original_comments:
+            # read the entire outout file
+            with gzip.open(filename, "rt" if self.binary_file else "r") as file:
+                output_contents = file.read()
+            # check each, if they can be found in there literally
+            for comment in self.original_comments:
+                self.assertIn(comment, output_contents)
+
+
 class TestBlfFileFormat(ReaderWriterTest):
     """Tests can.BLFWriter and can.BLFReader.