Fail gracefully on undecodable install output.

pre_commit/error_handler.py

@@ -18,8 +18,19 @@ class PreCommitSystemExit(SystemExit):
     pass
 
 
+def _to_bytes(exc):
+    try:
+        return bytes(exc)
+    except Exception:
+        return five.text(exc).encode('UTF-8')
+
+
 def _log_and_exit(msg, exc, formatted, write_fn=sys_stdout_write_wrapper):
-    error_msg = '{0}: {1}: {2}\n'.format(msg, type(exc).__name__, exc)
+    error_msg = b''.join((
+        five.to_bytes(msg), b': ',
+        five.to_bytes(type(exc).__name__), b': ',
+        _to_bytes(exc), b'\n',
+    ))
     write_fn(error_msg)
     write_fn('Check the log at ~/.pre-commit/pre-commit.log\n')
     store = Store()

pre_commit/languages/node.py

@@ -45,13 +45,14 @@ def install_environment(
         repo_cmd_runner.run(cmd)
 
         with in_env(repo_cmd_runner, version) as node_env:
-            node_env.run("cd '{prefix}' && npm install -g")
+            node_env.run("cd '{prefix}' && npm install -g", encoding=None)
             if additional_dependencies:
                 node_env.run(
                     "cd '{prefix}' && npm install -g " +
                     ' '.join(
                         shell_escape(dep) for dep in additional_dependencies
-                    )
+                    ),
+                    encoding=None,
                 )
 
 

pre_commit/languages/python.py

@@ -63,13 +63,14 @@ def install_environment(
             venv_cmd.extend(['-p', norm_version(version)])
         repo_cmd_runner.run(venv_cmd)
         with in_env(repo_cmd_runner, version) as env:
-            env.run("cd '{prefix}' && pip install .")
+            env.run("cd '{prefix}' && pip install .", encoding=None)
             if additional_dependencies:
                 env.run(
                     "cd '{prefix}' && pip install " +
                     ' '.join(
                         shell_escape(dep) for dep in additional_dependencies
-                    )
+                    ),
+                    encoding=None,
                 )
 
 

pre_commit/languages/ruby.py

@@ -95,13 +95,15 @@ def install_environment(
             ruby_env.run(
                 'cd {prefix} && gem build *.gemspec && '
                 'gem install --no-ri --no-rdoc *.gem',
+                encoding=None,
             )
             if additional_dependencies:
                 ruby_env.run(
                     'cd {prefix} && gem install --no-ri --no-rdoc ' +
                     ' '.join(
                         shell_escape(dep) for dep in additional_dependencies
-                    )
+                    ),
+                    encoding=None,
                 )
 
 

pre_commit/util.py

@@ -124,26 +124,38 @@ class CalledProcessError(RuntimeError):
         self.expected_returncode = expected_returncode
         self.output = output
 
-    def __str__(self):
+    def to_bytes(self):
         output = []
-        for text in self.output:
-            if text:
-                output.append('\n    ' + text.replace('\n', '\n    '))
+        for maybe_text in self.output:
+            if maybe_text:
+                output.append(
+                    b'\n    ' +
+                    five.to_bytes(maybe_text).replace(b'\n', b'\n    ')
+                )
             else:
-                output.append('(none)')
+                output.append(b'(none)')
 
-        return (
-            'Command: {0!r}\n'
-            'Return code: {1}\n'
-            'Expected return code: {2}\n'
-            'Output: {3}\n'
-            'Errors: {4}\n'.format(
-                self.cmd,
-                self.returncode,
-                self.expected_returncode,
-                *output
-            )
-        )
+        return b''.join((
+            five.to_bytes(
+                'Command: {0!r}\n'
+                'Return code: {1}\n'
+                'Expected return code: {2}\n'.format(
+                    self.cmd, self.returncode, self.expected_returncode
+                )
+            ),
+            b'Output: ', output[0], b'\n',
+            b'Errors: ', output[1], b'\n',
+        ))
+
+    def to_text(self):
+        return self.to_bytes().decode('UTF-8')
+
+    if five.PY3:  # pragma: no cover
+        __bytes__ = to_bytes
+        __str__ = to_text
+    else:
+        __str__ = to_bytes
+        __unicode__ = to_text
 
 
 def cmd_output(*cmd, **kwargs):