Support binary contents in FileContains matcher

Fixes #538

Author: jelmer

NEWS

@@ -11,6 +11,12 @@ Changes
 
 Fully migrate away from Launchpad to GitHub.
 
+Improvements
+------------
+
+* Support binary contents in ``FileContains`` matcher.
+  (Jelmer Vernooij, #538)
+
 2.8.1
 ~~~~~
 

tests/matchers/test_filesystem.py

@@ -175,6 +175,41 @@ def test_does_not_contain(self):
             Equals(mismatch.describe()),
         )
 
+    def test_binary_content(self):
+        tempdir = self.mkdtemp()
+        filename = os.path.join(tempdir, "binary_file")
+        binary_data = b"\x00\x01\x02\x03\xff\xfe"
+        with open(filename, "wb") as f:
+            f.write(binary_data)
+        self.assertThat(filename, FileContains(binary_data))
+
+    def test_binary_content_mismatch(self):
+        tempdir = self.mkdtemp()
+        filename = os.path.join(tempdir, "binary_file")
+        with open(filename, "wb") as f:
+            f.write(b"\x00\x01\x02")
+        mismatch = FileContains(b"\xff\xfe\xfd").match(filename)
+        self.assertThat(
+            Equals(b"\xff\xfe\xfd").match(b"\x00\x01\x02").describe(),
+            Equals(mismatch.describe()),
+        )
+
+    def test_text_with_encoding(self):
+        tempdir = self.mkdtemp()
+        filename = os.path.join(tempdir, "utf8_file")
+        text_data = "Hello 世界!"
+        with open(filename, "w", encoding="utf-8") as f:
+            f.write(text_data)
+        self.assertThat(filename, FileContains(text_data, encoding="utf-8"))
+
+    def test_text_default_encoding(self):
+        tempdir = self.mkdtemp()
+        filename = os.path.join(tempdir, "text_file")
+        text_data = "Hello World!"
+        with open(filename, "w") as f:
+            f.write(text_data)
+        self.assertThat(filename, FileContains(text_data))
+
 
 class TestTarballContains(TestCase, PathHelpers):
     def test_match(self):

testtools/matchers/_filesystem.py

@@ -94,7 +94,7 @@ def match(self, path):
 class FileContains(Matcher):
     """Matches if the given file has the specified contents."""
 
-    def __init__(self, contents=None, matcher=None):
+    def __init__(self, contents=None, matcher=None, encoding=None):
         """Construct a ``FileContains`` matcher.
 
         Can be used in a basic mode where the file contents are compared for
@@ -103,9 +103,14 @@ def __init__(self, contents=None, matcher=None):
         matched against an arbitrary matcher (by passing ``matcher`` instead).
 
         :param contents: If specified, match the contents of the file with
-            these contents.
+            these contents. If bytes, the file will be opened in binary mode.
+            If str, the file will be opened in text mode using the specified
+            encoding (or the default encoding if not specified).
         :param matcher: If specified, match the contents of the file against
             this matcher.
+        :param encoding: Optional text encoding to use when opening the file
+            in text mode. Only used when contents is a str (or when using a
+            matcher for text content). Defaults to the system default encoding.
         """
         if contents == matcher is None:
             raise AssertionError("Must provide one of `contents` or `matcher`.")
@@ -115,19 +120,23 @@ def __init__(self, contents=None, matcher=None):
             )
         if matcher is None:
             self.matcher = Equals(contents)
+            self._binary_mode = isinstance(contents, bytes)
         else:
             self.matcher = matcher
+            self._binary_mode = False
+        self.encoding = encoding
 
     def match(self, path):
         mismatch = PathExists().match(path)
         if mismatch is not None:
             return mismatch
-        f = open(path)
-        try:
-            actual_contents = f.read()
-            return self.matcher.match(actual_contents)
-        finally:
-            f.close()
+        if self._binary_mode:
+            with open(path, "rb") as f:
+                actual_contents: bytes | str = f.read()
+        else:
+            with open(path, encoding=self.encoding) as f:
+                actual_contents = f.read()
+        return self.matcher.match(actual_contents)
 
     def __str__(self):
         return f"File at path exists and contains {self.matcher}"