#650: make cmdline() handle unicode on python 2

4 files changed, 38 insertions, 42 deletions

diff --git a/docs/index.rst b/docs/index.rst
index 69edcead..f7dad748 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -697,6 +697,10 @@ Process class
 
      The command line this process has been called with.
 
+     *Changed in 3.2.0:* (Windows, Python 2) in case one or more parts of the
+     cmdline contains non ASCII characters the returned type is a list of
+     unicode strings.
+
   .. method:: create_time()
 
      The process creation time as a floating point number expressed in seconds
diff --git a/psutil/_pswindows.py b/psutil/_pswindows.py
index 479cf770..3c86d994 100644
--- a/psutil/_pswindows.py
+++ b/psutil/_pswindows.py
@@ -315,7 +315,20 @@ class Process(object):
 
     @wrap_exceptions
     def cmdline(self):
-        return cext.proc_cmdline(self.pid)
+        ret = cext.proc_cmdline(self.pid)
+        if PY3:
+            return ret
+        else:
+            # On Python 2, if one or more bits of the cmdline is unicode
+            # we return a list of unicode strings.
+            new = []
+            for x in ret:
+                x = py2_stringify(x)
+                if isinstance(x, unicode):
+                    return ret
+                else:
+                    new.append(x)
+            return new
 
     def ppid(self):
         try:
diff --git a/psutil/arch/windows/process_info.c b/psutil/arch/windows/process_info.c
index b925b05f..84860413 100644
--- a/psutil/arch/windows/process_info.c
+++ b/psutil/arch/windows/process_info.c
@@ -217,9 +217,8 @@ psutil_get_arg_list(long pid) {
     PVOID rtlUserProcParamsAddress;
     UNICODE_STRING commandLine;
     WCHAR *commandLineContents = NULL;
-    PyObject *arg = NULL;
-    PyObject *arg_from_wchar = NULL;
-    PyObject *argList = NULL;
+    PyObject *py_arg = NULL;
+    PyObject *py_retlist = NULL;
 
     hProcess = psutil_handle_from_pid(pid);
     if (hProcess == NULL)
@@ -274,48 +273,27 @@ psutil_get_arg_list(long pid) {
     // commandLine.Length is in bytes.
     commandLineContents[(commandLine.Length / sizeof(WCHAR))] = '\0';
 
-    // attempt tp parse the command line using Win32 API, fall back
+    // attempt to parse the command line using Win32 API, fall back
     // on string cmdline version otherwise
     szArglist = CommandLineToArgvW(commandLineContents, &nArgs);
-    if (NULL == szArglist) {
-        // failed to parse arglist
-        // encode as a UTF8 Python string object from WCHAR string
-        arg_from_wchar = PyUnicode_FromWideChar(commandLineContents,
-                                                commandLine.Length / 2);
-        if (arg_from_wchar == NULL)
-            goto error;
-#if PY_MAJOR_VERSION >= 3
-        argList = Py_BuildValue("N", PyUnicode_AsUTF8String(arg_from_wchar));
-#else
-        argList = Py_BuildValue("N", PyUnicode_FromObject(arg_from_wchar));
-#endif
-        if (!argList)
-            goto error;
+    if (szArglist == NULL) {
+        PyErr_SetFromWindowsErr(0);
+        goto error;
     }
     else {
         // arglist parsed as array of UNICODE_STRING, so convert each to
         // Python string object and add to arg list
-        argList = Py_BuildValue("[]");
-        if (argList == NULL)
+        py_retlist = Py_BuildValue("[]");
+        if (py_retlist == NULL)
             goto error;
         for (i = 0; i < nArgs; i++) {
-            arg_from_wchar = NULL;
-            arg = NULL;
-            arg_from_wchar = PyUnicode_FromWideChar(szArglist[i],
-                                                    wcslen(szArglist[i]));
-            if (arg_from_wchar == NULL)
-                goto error;
-#if PY_MAJOR_VERSION >= 3
-            arg = PyUnicode_FromObject(arg_from_wchar);
-#else
-            arg = PyUnicode_AsUTF8String(arg_from_wchar);
-#endif
-            if (arg == NULL)
+            py_arg = PyUnicode_FromWideChar(
+                szArglist[i], wcslen(szArglist[i]));
+            if (py_arg == NULL)
                 goto error;
-            Py_XDECREF(arg_from_wchar);
-            if (PyList_Append(argList, arg))
+            if (PyList_Append(py_retlist, py_arg))
                 goto error;
-            Py_XDECREF(arg);
+            Py_XDECREF(py_arg);
         }
     }
 
@@ -323,12 +301,11 @@ psutil_get_arg_list(long pid) {
         LocalFree(szArglist);
     free(commandLineContents);
     CloseHandle(hProcess);
-    return argList;
+    return py_retlist;
 
 error:
-    Py_XDECREF(arg);
-    Py_XDECREF(arg_from_wchar);
-    Py_XDECREF(argList);
+    Py_XDECREF(py_arg);
+    Py_XDECREF(py_retlist);
     if (hProcess != NULL)
         CloseHandle(hProcess);
     if (commandLineContents != NULL)
diff --git a/test/_windows.py b/test/_windows.py
index 8f2c7a52..70101726 100644
--- a/test/_windows.py
+++ b/test/_windows.py
@@ -491,8 +491,10 @@ class TestUnicode(unittest.TestCase):
         shutil.copyfile(sys.executable, self.uexe)
         subp = get_test_subprocess(cmd=[self.uexe])
         p = psutil.Process(subp.pid)
-        self.assertIsInstance("".join(p.cmdline()), unicode)
-        self.assertEqual(p.cmdline(), [self.uexe])
+        self.assertIsInstance(u("").join(p.cmdline()), unicode)
+        uexe = self.uexe if PY3 else \
+            unicode(self.uexe, sys.getfilesystemencoding())
+        self.assertEqual(p.cmdline(), [uexe])
 
     def test_proc_cwd(self):
         tdir = tempfile.mkdtemp(prefix=u("psutil-è-"))