Further consolidation of code reading Python source.

4 files changed, 41 insertions, 38 deletions

diff --git a/coverage/execfile.py b/coverage/execfile.py
index ecb03372..beaf1913 100644
--- a/coverage/execfile.py
+++ b/coverage/execfile.py
@@ -7,8 +7,8 @@ import types
 
 from coverage.backward import BUILTINS
 from coverage.backward import PYC_MAGIC_NUMBER, imp, importlib_util_find_spec
+from coverage.files import get_python_source
 from coverage.misc import ExceptionDuringRun, NoCode, NoSource
-from coverage.phystokens import read_python_source
 
 
 class DummyLoader(object):
@@ -178,16 +178,11 @@ def make_code_from_py(filename):
     """Get source from `filename` and make a code object of it."""
     # Open the source file.
     try:
-        source = read_python_source(filename)
-    except IOError:
+        source = get_python_source(filename)
+    except (IOError, NoSource):
         raise NoSource("No file to run: %r" % filename)
 
-    # 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':
-        source += '\n'
     code = compile(source, filename, "exec")
-
     return code
 
 
diff --git a/coverage/files.py b/coverage/files.py
index 7f2431d3..8fa8067f 100644
--- a/coverage/files.py
+++ b/coverage/files.py
@@ -1,15 +1,16 @@
 """File wrangling."""
 
 import fnmatch
+import ntpath
 import os
 import os.path
+import posixpath
 import re
 import sys
-import ntpath
-import posixpath
+import tokenize
 
 from coverage.misc import CoverageException, NoSource, join_regex
-from coverage.phystokens import read_python_source, source_encoding
+from coverage.phystokens import source_encoding
 
 
 class FileLocator(object):
@@ -55,6 +56,22 @@ class FileLocator(object):
         return self.canonical_filename_cache[filename]
 
 
+def read_python_source(filename):
+    """Read the Python source text from `filename`.
+
+    Returns a str: unicode on Python 3, bytes on Python 2.
+
+    """
+    # Python 3.2 provides `tokenize.open`, the best way to open source files.
+    if sys.version_info >= (3, 2):
+        f = tokenize.open(filename)
+    else:
+        f = open(filename, "rU")
+
+    with f:
+        return f.read()
+
+
 def get_python_source(filename):
     """Return the source code, as a str."""
     base, ext = os.path.splitext(filename)
@@ -67,17 +84,24 @@ def get_python_source(filename):
         try_filename = base + ext
         if os.path.exists(try_filename):
             # A regular text file: open it.
-            return read_python_source(try_filename)
+            source = read_python_source(try_filename)
+            break
 
         # Maybe it's in a zip file?
         source = get_zip_bytes(try_filename)
         if source is not None:
             if sys.version_info >= (3, 0):
                 source = source.decode(source_encoding(source))
-            return source
+            break
+    else:
+        # Couldn't find source.
+        raise NoSource("No source for code: %r." % filename)
+
+    # Python code should always end with a line with a newline.
+    if source and source[-1] != '\n':
+        source += '\n'
 
-    # Couldn't find source.
-    raise NoSource("No source for code: %r." % filename)
+    return source
 
 
 def get_zip_bytes(filename):
diff --git a/coverage/parser.py b/coverage/parser.py
index b50bc578..05c0d0c2 100644
--- a/coverage/parser.py
+++ b/coverage/parser.py
@@ -56,11 +56,10 @@ 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:]
-            # Python source should always have a final newline.
-            if self.text[-1] != "\n":
-                self.text += "\n"
+            # (Used to do this, but no longer.  Not sure what bad will happen
+            # if we don't do it.)
+            #   if ord(self.text[0]) == 0xfeff:
+            #       self.text = self.text[1:]
 
         self.exclude = exclude
 
@@ -352,9 +351,10 @@ class ByteParser(object):
                 self.code = compile(text, filename, "exec")
             except SyntaxError as synerr:
                 raise NotPython(
-                    "Couldn't parse '%s' as Python source: '%s' at line %d" %
-                        (filename, synerr.msg, synerr.lineno)
+                    "Couldn't parse %r as Python source: '%s' at line %d" % (
+                        filename, synerr.msg, synerr.lineno
                     )
+                )
 
         # Alternative Python implementations don't always provide all the
         # attributes on code objects that we need to do the analysis.
diff --git a/coverage/phystokens.py b/coverage/phystokens.py
index f3f633dd..c52e28ae 100644
--- a/coverage/phystokens.py
+++ b/coverage/phystokens.py
@@ -262,19 +262,3 @@ if sys.version_info >= (3, 0):
     source_encoding = _source_encoding_py3
 else:
     source_encoding = _source_encoding_py2
-
-
-def read_python_source(filename):
-    """Read the Python source text from `filename`.
-
-    Returns unicode on Python 3, bytes on Python 2.
-
-    """
-    # Python 3.2 provides `tokenize.open`, the best way to open source files.
-    if sys.version_info >= (3, 2):
-        f = tokenize.open(filename)
-    else:
-        f = open(filename, "rU")
-
-    with f:
-        return f.read()