Merge pull request #153 from kmvanbrunt/text_file

Verifying a file to be loaded as a text script is either ASCII or UTF-8

Author: tleonhardt

cmd2.py

@@ -26,6 +26,7 @@
 Git repository on GitHub at https://github.com/python-cmd2/cmd2
 """
 import cmd
+import codecs
 import collections
 import datetime
 import glob
@@ -1643,31 +1644,31 @@ def do_edit(self, arg):
         filename = None
         if arg:
             try:
-                buffer = self._last_matching(int(arg))
+                history_item = self._last_matching(int(arg))
             except ValueError:
                 filename = arg
-                buffer = ''
+                history_item = ''
         else:
             try:
-                buffer = self.history[-1]
+                history_item = self.history[-1]
             except IndexError:
                 self.perror('edit must be called with argument if history is empty', traceback_war=False)
                 return
 
         delete_tempfile = False
-        if buffer:
+        if history_item:
             if filename is None:
                 fd, filename = tempfile.mkstemp(suffix='.txt', text=True)
                 os.close(fd)
                 delete_tempfile = True
 
             f = open(os.path.expanduser(filename), 'w')
-            f.write(buffer or '')
+            f.write(history_item or '')
             f.close()
 
         os.system('%s %s' % (self.editor, filename))
 
-        if self.autorun_on_edit or buffer:
+        if self.autorun_on_edit or history_item:
             self.do_load(filename)
 
         if delete_tempfile:
@@ -1758,6 +1759,22 @@ def do_load(self, file_path):
             return
 
         expanded_path = os.path.abspath(os.path.expanduser(file_path.strip()))
+
+        # Make sure expanded_path points to a file
+        if not os.path.isfile(expanded_path):
+            self.perror('{} does not exist or is not a file\n'.format(expanded_path), traceback_war=False)
+            return
+
+        # Make sure the file is not empty
+        if os.path.getsize(expanded_path) == 0:
+            self.perror('{} is empty\n'.format(expanded_path), traceback_war=False)
+            return
+
+        # Make sure the file is ASCII or UTF-8 encoded text
+        if not self.is_text_file(expanded_path):
+            self.perror('{} is not an ASCII or UTF-8 encoded text file\n'.format(expanded_path), traceback_war=False)
+            return
+
         try:
             target = open(expanded_path)
         except IOError as e:
@@ -1788,6 +1805,40 @@ def do_run(self, arg):
         if runme:
             return self.onecmd_plus_hooks(runme)
 
+    @staticmethod
+    def is_text_file(file_path):
+        """
+        Returns if a file contains only ASCII or UTF-8 encoded text
+        :param file_path: path to the file being checked
+        """
+        expanded_path = os.path.abspath(os.path.expanduser(file_path.strip()))
+        valid_text_file = False
+
+        # Check if the file is ASCII
+        try:
+            with codecs.open(expanded_path, encoding='ascii', errors='strict') as f:
+                # Make sure the file has at least one line of text
+                # noinspection PyUnusedLocal
+                if sum(1 for line in f) > 0:
+                    valid_text_file = True
+        except IOError:
+            pass
+        except UnicodeDecodeError:
+            # The file is not ASCII. Check if it is UTF-8.
+            try:
+                with codecs.open(expanded_path, encoding='utf-8', errors='strict') as f:
+                    # Make sure the file has at least one line of text
+                    # noinspection PyUnusedLocal
+                    if sum(1 for line in f) > 0:
+                        valid_text_file = True
+            except IOError:
+                pass
+            except UnicodeDecodeError:
+                # Not UTF-8
+                pass
+
+        return valid_text_file
+
     def run_transcript_tests(self, callargs):
         """Runs transcript tests for provided file(s).
 
@@ -2357,5 +2408,4 @@ def __bool__(self):
 if __name__ == '__main__':
     # If run as the main application, simply start a bare-bones cmd2 application with only built-in functionality.
     app = Cmd()
-    app.debug = True
     app.cmdloop()

tests/scripts/binary.bin

@@ -0,0 +1 @@
+!echo γνωρίζω

tests/scripts/empty.txt

@@ -293,6 +293,54 @@ def test_base_load_with_empty_args(base_app, capsys):
     assert normalize(str(err)) == expected
 
 
+def test_base_load_with_nonexistent_file(base_app, capsys):
+    # The way the load command works, we can't directly capture its stdout or stderr
+    run_cmd(base_app, 'load does_not_exist.txt')
+    out, err = capsys.readouterr()
+
+    # The load command requires a path to an existing file
+    assert str(err).startswith("ERROR")
+    assert "does not exist or is not a file" in str(err)
+
+
+def test_base_load_with_empty_file(base_app, capsys, request):
+    test_dir = os.path.dirname(request.module.__file__)
+    filename = os.path.join(test_dir, 'scripts', 'empty.txt')
+
+    # The way the load command works, we can't directly capture its stdout or stderr
+    run_cmd(base_app, 'load {}'.format(filename))
+    out, err = capsys.readouterr()
+
+    # The load command requires non-empty scripts files
+    assert str(err).startswith("ERROR")
+    assert "is empty" in str(err)
+
+
+def test_base_load_with_binary_file(base_app, capsys, request):
+    test_dir = os.path.dirname(request.module.__file__)
+    filename = os.path.join(test_dir, 'scripts', 'binary.bin')
+
+    # The way the load command works, we can't directly capture its stdout or stderr
+    run_cmd(base_app, 'load {}'.format(filename))
+    out, err = capsys.readouterr()
+
+    # The load command requires non-empty scripts files
+    assert str(err).startswith("ERROR")
+    assert "is not an ASCII or UTF-8 encoded text file" in str(err)
+
+
+def test_base_load_with_utf8_file(base_app, capsys, request):
+    test_dir = os.path.dirname(request.module.__file__)
+    filename = os.path.join(test_dir, 'scripts', 'utf8.txt')
+
+    # The way the load command works, we can't directly capture its stdout or stderr
+    run_cmd(base_app, 'load {}'.format(filename))
+    out, err = capsys.readouterr()
+
+    # TODO Make this test better once shell command is fixed to used cmd2's stdout
+    assert str(err) == ''
+
+
 def test_base_relative_load(base_app, request):
     test_dir = os.path.dirname(request.module.__file__)
     filename = os.path.join(test_dir, 'script.txt')
@@ -458,10 +506,10 @@ def test_base_timing(base_app, capsys):
 
 
 def test_base_debug(base_app, capsys):
-    # Try to load a non-existent file with debug set to False by default
-    run_cmd(base_app, 'load does_not_exist.txt')
+    # Try to set a non-existent parameter with debug set to False by default
+    run_cmd(base_app, 'set does_not_exist 5')
     out, err = capsys.readouterr()
-    assert err.startswith('ERROR')
+    assert err.startswith('EXCEPTION')
 
     # Set debug true
     out = run_cmd(base_app, 'set debug True')
@@ -472,7 +520,7 @@ def test_base_debug(base_app, capsys):
     assert out == expected
 
     # Verify that we now see the exception traceback
-    run_cmd(base_app, 'load does_not_exist.txt')
+    run_cmd(base_app, 'set does_not_exist 5')
     out, err = capsys.readouterr()
     assert str(err).startswith('Traceback (most recent call last):')