Take filenames with dots into account in get_sigmf_filenames() (#107)

* take filenames with dots into account in get_sigmf_filenames()

The function get_sigmf_filenames() does not work correctly if a
file contains dots in the filename. These are interpretted as suffixes
by pathlib, and a part of the filename can be removed when forming
the different SigMF filenames. This fixes the problem by only
considering as file extensions the canonical SigMF file extensions.
Any other suffixes are treated as part of the filename and not removed.

This fixes #106.

* fix SigMFCollection.get_SigMFFile()

The get_SigMFFile() method was broken because it used ".sigmf_meta"
instead of ".sigmf-meta". This rewrites the method to use
get_sigmf_filenames().

* test that SigMF files with . in the filename are loaded correctly

This adds a unit test that shows that SigMF files like
a.sigmf-meta, b.c.sigmf-meta, and d.e.f.sigmf-meta are all loaded
correctly.

* increment patch version

---------

Co-authored-by: Teque5 <[email protected]>

Author: daniestevez

sigmf/__init__.py

@@ -5,7 +5,7 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 
 # version of this python module
-__version__ = "1.2.8"
+__version__ = "1.2.9"
 # matching version of the SigMF specification
 __specification__ = "1.2.3"
 

sigmf/sigmffile.py

@@ -915,13 +915,13 @@ def get_SigMFFile(self, stream_name=None, stream_index=None):
         """
         Returns the SigMFFile instance of the specified stream if it exists
         """
-        metafile = None
-        if stream_name is not None:
-            if stream_name in self.get_stream_names():
-                metafile = stream_name + ".sigmf_meta"
+        if stream_name is not None and stream_name not in self.get_stream_names():
+            # invalid stream name
+            return
         if stream_index is not None and stream_index < len(self):
-            metafile = self.get_stream_names()[stream_index] + ".sigmf_meta"
-        if metafile is not None:
+            stream_name = self.get_stream_names()[stream_index]
+        if stream_name is not None:
+            metafile = get_sigmf_filenames(stream_name)["meta_fn"]
             metafile_path = self.base_path / metafile
             return fromfile(metafile_path, skip_checksum=self.skip_checksums)
 
@@ -1111,11 +1111,26 @@ def get_sigmf_filenames(filename):
     -------
     dict with 'data_fn', 'meta_fn', and 'archive_fn' as keys.
     """
-    stem_path = Path(filename).with_suffix("")
+    stem_path = Path(filename)
+    # If the path has a sigmf suffix, remove it. Otherwise do not remove the
+    # suffix, because the filename might contain '.' characters which are part
+    # of the filename rather than an extension.
+    sigmf_suffixes = [
+        SIGMF_DATASET_EXT, SIGMF_METADATA_EXT,
+        SIGMF_ARCHIVE_EXT, SIGMF_COLLECTION_EXT,
+    ]
+    if stem_path.suffix in sigmf_suffixes:
+        with_suffix_path = stem_path
+        stem_path = stem_path.with_suffix("")
+    else:
+        # Add a dummy suffix to prevent the .with_suffix() calls below from
+        # overriding part of the filename which is interpreted as a suffix
+        with_suffix_path = stem_path.with_name(f"{stem_path.name}{SIGMF_DATASET_EXT}")
+
     return {
         "base_fn": stem_path,
-        "data_fn": stem_path.with_suffix(SIGMF_DATASET_EXT),
-        "meta_fn": stem_path.with_suffix(SIGMF_METADATA_EXT),
-        "archive_fn": stem_path.with_suffix(SIGMF_ARCHIVE_EXT),
-        "collection_fn": stem_path.with_suffix(SIGMF_COLLECTION_EXT),
+        "data_fn": with_suffix_path.with_suffix(SIGMF_DATASET_EXT),
+        "meta_fn": with_suffix_path.with_suffix(SIGMF_METADATA_EXT),
+        "archive_fn": with_suffix_path.with_suffix(SIGMF_ARCHIVE_EXT),
+        "collection_fn": with_suffix_path.with_suffix(SIGMF_COLLECTION_EXT),
     }

tests/test_sigmffile.py

@@ -43,6 +43,20 @@ def test_pathlib_handle(self):
         obj_pth = sigmffile.fromfile(self.temp_path_data)
         obj_pth.validate()
 
+    def test_filenames_with_dots(self):
+        """test that filenames with non-extension . characters are handled correctly"""
+        filenames = ["a", "b.c", "d.e.f"]
+        for filename in filenames:
+            temp_path_data = self.temp_dir / f"{filename}.sigmf-data"
+            temp_path_meta = self.temp_dir / f"{filename}.sigmf-meta"
+            TEST_FLOAT32_DATA.tofile(temp_path_data)
+            self.sigmf_object = SigMFFile(TEST_METADATA, data_file=temp_path_data)
+            self.sigmf_object.tofile(temp_path_meta)
+            files = [str(temp_path_data), temp_path_data, str(temp_path_meta), temp_path_meta]
+            for filename in files:
+                obj = sigmffile.fromfile(filename)
+                obj.validate()
+
     def test_iterator_basic(self):
         """make sure default batch_size works"""
         count = 0