Handling of Windows (and legacy Mac) line endings (#47)

Co-authored-by: Peter Cock <p.j.a.cock@googlemail.com>

doc8/checks.py

@@ -60,12 +60,13 @@ def report_iter(self, line):
                 yield ("D003", "Tabulation used for indentation")
 
 
-class CheckCarriageReturn(LineCheck):
+class CheckCarriageReturn(ContentCheck):
     REPORTS = frozenset(["D004"])
 
-    def report_iter(self, line):
-        if "\r" in line:
-            yield ("D004", "Found literal carriage return")
+    def report_iter(self, parsed_file):
+        for i, line in enumerate(parsed_file.lines):
+            if b"\r" in line:
+                yield (i + 1, "D004", "Found literal carriage return")
 
 
 class CheckNewlineEndOfFile(ContentCheck):
@@ -75,7 +76,10 @@ def __init__(self, cfg):
         super(CheckNewlineEndOfFile, self).__init__(cfg)
 
     def report_iter(self, parsed_file):
-        if parsed_file.lines and not parsed_file.lines[-1].endswith(b"\n"):
+        if parsed_file.lines and not (
+            parsed_file.lines[-1].endswith(b"\n")
+            or parsed_file._lines[-1].endswith(b"\r")
+        ):
             yield (len(parsed_file.lines), "D005", "No newline at end of file")
 
 

doc8/parser.py

@@ -80,17 +80,20 @@ def _read(self):
         with self._read_lock:
             if not self._has_read:
                 with open(self.filename, "rb") as fh:
-                    self._lines = list(fh)
-                    fh.seek(0)
                     self._raw_content = fh.read()
+                    self._lines = self._raw_content.splitlines(True)
                 self._has_read = True
 
     def lines_iter(self, remove_trailing_newline=True):
         self._read()
         for line in self._lines:
             line = str(line, encoding=self.encoding)
-            if remove_trailing_newline and line.endswith("\n"):
-                line = line[0:-1]
+            if remove_trailing_newline:
+                # Cope with various OS new line conventions
+                if line.endswith("\n"):
+                    line = line[:-1]
+                if line.endswith("\r"):
+                    line = line[:-1]
             yield line
 
     @property

doc8/tests/test_checks.py

@@ -48,14 +48,18 @@ def test_tabs(self):
 
 class TestCarriageReturn(testtools.TestCase):
     def test_cr(self):
-        lines = ["\tabc", "efg", "\r\n"]
-        check = checks.CheckCarriageReturn({})
-        errors = []
-        for line in lines:
-            errors.extend(check.report_iter(line))
-        self.assertEqual(1, len(errors))
-        (code, msg) = errors[0]
-        self.assertIn(code, check.REPORTS)
+        content = b"Windows line ending\r\nLegacy Mac line ending\r"
+        content += (b"a" * 79) + b"\r\n" + b"\r"
+        conf = {"max_line_length": 79}
+        with tempfile.NamedTemporaryFile() as fh:
+            fh.write(content)
+            fh.flush()
+            parsed_file = parser.ParsedFile(fh.name)
+            check = checks.CheckCarriageReturn(conf)
+            errors = list(check.report_iter(parsed_file))
+            self.assertEqual(4, len(errors))
+            (line, code, msg) = errors[0]
+            self.assertIn(code, check.REPORTS)
 
 
 class TestLineLength(testtools.TestCase):
@@ -172,6 +176,8 @@ def test_newline(self):
             (1, b"testing\ntesting"),
             (0, b"testing\n"),
             (0, b"testing\ntesting\n"),
+            (0, b"testing\r\n"),
+            (0, b"testing\r"),
         ]
 
         for expected_errors, line in tests: