Unify and clarify reading Python source. Probably broke .pyw files

7 files changed, 34 insertions, 34 deletions

diff --git a/coverage/codeunit.py b/coverage/codeunit.py
index 09b86fe..1182398 100644
--- a/coverage/codeunit.py
+++ b/coverage/codeunit.py
@@ -169,6 +169,8 @@ class PythonCodeUnit(CodeUnit):
         return self._source
 
     def get_parser(self, exclude=None):
+        return PythonParser(filename=self.filename, exclude=exclude)
+        # TODO: REMOVE THIS USELESS CODE!
         actual_filename, source = self._find_source(self.filename)
         return PythonParser(
             text=source, filename=actual_filename, exclude=exclude,
diff --git a/coverage/execfile.py b/coverage/execfile.py
index 246d79a..ecb0337 100644
--- a/coverage/execfile.py
+++ b/coverage/execfile.py
@@ -8,7 +8,7 @@ import types
 from coverage.backward import BUILTINS
 from coverage.backward import PYC_MAGIC_NUMBER, imp, importlib_util_find_spec
 from coverage.misc import ExceptionDuringRun, NoCode, NoSource
-from coverage.phystokens import open_python_source
+from coverage.phystokens import read_python_source
 
 
 class DummyLoader(object):
@@ -178,13 +178,10 @@ def make_code_from_py(filename):
     """Get source from `filename` and make a code object of it."""
     # Open the source file.
     try:
-        source_file = open_python_source(filename)
+        source = read_python_source(filename)
     except IOError:
         raise NoSource("No file to run: %r" % filename)
 
-    with source_file:
-        source = source_file.read()
-
     # We have the source.  `compile` still needs the last line to be clean,
     # so make sure it is, then compile a code object from it.
     if not source or source[-1] != '\n':
diff --git a/coverage/files.py b/coverage/files.py
index 6ba7983..2945054 100644
--- a/coverage/files.py
+++ b/coverage/files.py
@@ -8,8 +8,8 @@ import sys
 import ntpath
 import posixpath
 
-from coverage.misc import CoverageException, join_regex
-from coverage.phystokens import open_python_source
+from coverage.misc import CoverageException, NoSource, join_regex
+from coverage.phystokens import read_python_source, source_encoding
 
 
 class FileLocator(object):
@@ -56,19 +56,20 @@ class FileLocator(object):
 
 
 def get_python_source(filename):
-    """Return the source code, as a string."""
+    """Return the source code, as a str."""
     if os.path.exists(filename):
         # A regular text file: open it.
-        with open_python_source(filename) as f:
-            return f.read()
+        return read_python_source(filename)
 
     # Maybe it's in a zip file?
     source = get_zip_bytes(filename)
     if source is not None:
+        if sys.version_info >= (3, 0):
+            source = source.decode(source_encoding(source))
         return source
 
     # Couldn't find source.
-    raise CoverageException(
+    raise NoSource(
         "No source for code: '%s'." % filename
         )
 
diff --git a/coverage/parser.py b/coverage/parser.py
index 317f7ec..b50bc57 100644
--- a/coverage/parser.py
+++ b/coverage/parser.py
@@ -54,6 +54,7 @@ class PythonParser(CodeParser):
                     )
 
         if self.text:
+            assert isinstance(self.text, str)
             # Scrap the BOM if it exists.
             if ord(self.text[0]) == 0xfeff:
                 self.text = self.text[1:]
@@ -90,8 +91,7 @@ class PythonParser(CodeParser):
     def byte_parser(self):
         """Create a ByteParser on demand."""
         if not self._byte_parser:
-            self._byte_parser = \
-                            ByteParser(text=self.text, filename=self.filename)
+            self._byte_parser = ByteParser(self.text, filename=self.filename)
         return self._byte_parser
 
     def lines_matching(self, *regexes):
@@ -343,16 +343,11 @@ OP_RETURN_VALUE = _opcode('RETURN_VALUE')
 class ByteParser(object):
     """Parse byte codes to understand the structure of code."""
 
-    def __init__(self, code=None, text=None, filename=None):
+    def __init__(self, text, code=None, filename=None):
+        self.text = text
         if code:
             self.code = code
-            self.text = text
         else:
-            if not text:
-                assert filename, "If no code or text, need a filename"
-                text = get_python_source(filename)
-            self.text = text
-
             try:
                 self.code = compile(text, filename, "exec")
             except SyntaxError as synerr:
@@ -378,7 +373,7 @@ class ByteParser(object):
 
         """
         children = CodeObjects(self.code)
-        return (ByteParser(code=c, text=self.text) for c in children)
+        return (ByteParser(self.text, code=c) for c in children)
 
     def _bytes_lines(self):
         """Map byte offsets to line numbers in `code`.
diff --git a/coverage/phystokens.py b/coverage/phystokens.py
index 4faa3c3..70deb80 100644
--- a/coverage/phystokens.py
+++ b/coverage/phystokens.py
@@ -153,17 +153,17 @@ def source_encoding(source):
     Returns a string, the name of the encoding.
 
     """
-    # Note: this function should never be called on Python 3, since py3 has
-    # built-in tools to do this.
-    assert sys.version_info < (3, 0)
+    if sys.version_info >= (3, 0):
+        readline = iter(source.splitlines(True)).__next__
+        return tokenize.detect_encoding(readline)[0]
+
+    # Do this so the detect_encode code we copied will work.
+    readline = iter(source.splitlines(True)).next
 
     # This is mostly code adapted from Py3.2's tokenize module.
 
     cookie_re = re.compile(r"^\s*#.*coding[:=]\s*([-\w.]+)")
 
-    # Do this so the detect_encode code we copied will work.
-    readline = iter(source.splitlines(True)).next
-
     def _get_normal_name(orig_enc):
         """Imitates get_normal_name in tokenizer.c."""
         # Only care about the first 12 characters.
@@ -246,9 +246,12 @@ def source_encoding(source):
 # Reading Python source and interpreting the coding comment is a big deal.
 if sys.version_info >= (3, 0):
     # Python 3.2 provides `tokenize.open`, the best way to open source files.
-    import tokenize
-    open_python_source = tokenize.open
+    def read_python_source(filename):
+        # Returns unicode on Py3, bytes on Py2
+        with tokenize.open(filename) as f:
+            return f.read()
 else:
-    def open_python_source(fname):
-        """Open a source file the best way."""
-        return open(fname, "rU")
+    def read_python_source(filename):
+        # Returns unicode on Py3, bytes on Py2
+        with open(filename, "rU") as f:
+            return f.read()
diff --git a/coverage/summary.py b/coverage/summary.py
index 10ac7e2..cfcdfcd 100644
--- a/coverage/summary.py
+++ b/coverage/summary.py
@@ -4,7 +4,7 @@ import sys
 
 from coverage.report import Reporter
 from coverage.results import Numbers
-from coverage.misc import NotPython
+from coverage.misc import CoverageException, NotPython
 
 
 class SummaryReporter(Reporter):
diff --git a/lab/parser.py b/lab/parser.py
index 932480d..1783468 100644
--- a/lab/parser.py
+++ b/lab/parser.py
@@ -9,6 +9,7 @@ from optparse import OptionParser
 import disgen
 
 from coverage.misc import CoverageException
+from coverage.files import get_python_source
 from coverage.parser import ByteParser, PythonParser
 
 opcode_counts = collections.Counter()
@@ -70,7 +71,8 @@ class ParserMain(object):
         """Process just one file."""
 
         try:
-            bp = ByteParser(filename=filename)
+            text = get_python_source(filename)
+            bp = ByteParser(text, filename=filename)
         except Exception as err:
             print("%s" % (err,))
             return