python/tskit/util.py: Add a file name to FileFormatError

.gitignore

@@ -7,3 +7,6 @@ python/benchmark/*.html
 .env
 .vscode
 env
+# pixi environments
+.pixi/*
+!.pixi/config.toml

python/.gitignore

@@ -5,3 +5,6 @@ build
 .*.swp
 .*.swo
 */.ipynb_checkpoints
+# pixi environments
+.pixi/*
+!.pixi/config.toml

python/tests/test_file_format.py

@@ -261,12 +261,12 @@ def test_format_too_old_raised_for_hdf5(self):
 
             with pytest.raises(
                 exceptions.FileFormatError,
-                match="appears to be in HDF5 format",
+                match=f"{filename}.*appears to be in HDF5 format",
             ):
                 tskit.load(path)
             with pytest.raises(
                 exceptions.FileFormatError,
-                match="appears to be in HDF5 format",
+                match=f"{filename}.*appears to be in HDF5 format",
             ):
                 tskit.TableCollection.load(path)
 
@@ -284,9 +284,15 @@ class TestErrors(TestFileFormat):
     def test_tszip_file(self):
         ts = msprime.simulate(5)
         tszip.compress(ts, self.temp_file)
-        with pytest.raises(tskit.FileFormatError, match="appears to be in zip format"):
+        with pytest.raises(
+            tskit.FileFormatError,
+            match=f"{self.temp_file}.*appears to be in zip format",
+        ):
             tskit.load(self.temp_file)
-        with pytest.raises(tskit.FileFormatError, match="appears to be in zip format"):
+        with pytest.raises(
+            tskit.FileFormatError,
+            match=f"{self.temp_file}.*appears to be in zip format",
+        ):
             tskit.TableCollection.load(self.temp_file)
 
 
@@ -897,7 +903,10 @@ def test_format_name_error(self):
                 data = dict(store)
             data["format/name"] = np.array(bytearray(bad_name.encode()), dtype=np.int8)
             kastore.dump(data, self.temp_file)
-            with pytest.raises(exceptions.FileFormatError):
+            with pytest.raises(
+                exceptions.FileFormatError,
+                match=f"While trying to load {self.temp_file}",
+            ):
                 tskit.load(self.temp_file)
 
     def test_load_bad_formats(self):
@@ -908,12 +917,16 @@ def test_load_bad_formats(self):
         # Now some ascii text
         with open(self.temp_file, "wb") as f:
             f.write(b"Some ASCII text")
-        with pytest.raises(exceptions.FileFormatError):
+        with pytest.raises(
+            exceptions.FileFormatError, match=f"While trying to load {self.temp_file}"
+        ):
             tskit.load(self.temp_file)
         # Now write 8k of random bytes
         with open(self.temp_file, "wb") as f:
             f.write(os.urandom(8192))
-        with pytest.raises(exceptions.FileFormatError):
+        with pytest.raises(
+            exceptions.FileFormatError, match=f"While trying to load {self.temp_file}"
+        ):
             tskit.load(self.temp_file)
 
     def test_load_bad_formats_fileobj(self):
@@ -925,7 +938,9 @@ def load():
             load()
         with open(self.temp_file, "wb") as f:
             f.write(b"Some ASCII text")
-        with pytest.raises(exceptions.FileFormatError):
+        with pytest.raises(
+            exceptions.FileFormatError, match=f"While trying to load {self.temp_file}"
+        ):
             load()
 
 

python/tskit/util.py

@@ -878,6 +878,7 @@ def raise_known_file_format_errors(open_file, existing_exception):
     Sniffs the file for pk-zip or hdf header bytes, then raises an exception
     if these are detected, if not raises the existing exception.
     """
+    existing_exception.add_note(f"While trying to load {open_file.name}")
     # Check for HDF5 header bytes
     try:
         open_file.seek(0)
@@ -887,16 +888,16 @@ def raise_known_file_format_errors(open_file, existing_exception):
         raise existing_exception
     if header == b"\x89HDF":
         raise tskit.FileFormatError(
-            "The specified file appears to be in HDF5 format. This file "
+            f"The file {open_file.name} appears to be in HDF5 format. This file "
             "may have been generated by msprime < 0.6.0 (June 2018) which "
             "can no longer be read directly. Please convert to the new "
             "kastore format using the ``tskit upgrade`` command from tskit version<0.6.2"
         ) from existing_exception
     if header[:2] == b"\x50\x4b":
         raise tskit.FileFormatError(
-            "The specified file appears to be in zip format, so may be a compressed "
-            "tree sequence. Try using the tszip module to decompress this file before "
-            "loading. `pip install tszip; tsunzip <filename>` or use "
+            f"The file {open_file.name} appears to be in zip format, so may be a "
+            "compressed tree sequence. Try using the tszip module to decompress this "
+            "file before loading. `pip install tszip; tsunzip <filename>` or use "
             "`tszip.decompress` in Python code."
         ) from existing_exception
     raise existing_exception