bpo-36443: Disable C locale coercion and UTF-8 Mode by default (GH-12589)

bpo-36443, bpo-36202: Since Python 3.7.0, calling Py_DecodeLocale()
before Py_Initialize() produces mojibake if the LC_CTYPE locale is
coerced and/or if the UTF-8 Mode is enabled by the user
configuration. This change fix the issue by disabling LC_CTYPE
coercion and UTF-8 Mode by default. They must now be enabled
explicitly (opt-in) using the new _Py_PreInitialize() API with
_PyPreConfig.

When embedding Python, set coerce_c_locale and utf8_mode attributes
of _PyPreConfig to -1 to enable automatically these parameters
depending on the LC_CTYPE locale, environment variables and command
line arguments

Alternative: Setting Py_UTF8Mode to 1 always explicitly enables the
UTF-8 Mode.

Changes:

* _PyPreConfig_INIT now sets coerce_c_locale and utf8_mode to 0 by
  default.
* _Py_InitializeFromArgs() and _Py_InitializeFromWideArgs() can now
  be called with config=NULL.

7 files changed, 58 insertions, 46 deletions

diff --git a/Include/cpython/coreconfig.h b/Include/cpython/coreconfig.h
index 27ee1f4..7ce1a02 100644
--- a/Include/cpython/coreconfig.h
+++ b/Include/cpython/coreconfig.h
@@ -63,13 +63,20 @@ typedef struct {
        set to !Py_IgnoreEnvironmentFlag. */
     int use_environment;
 
-    /* PYTHONCOERCECLOCALE, -1 means unknown.
+    /* Coerce the LC_CTYPE locale if it's equal to "C"? (PEP 538)
+
+       Set to 0 by PYTHONCOERCECLOCALE=0. Set to 1 by PYTHONCOERCECLOCALE=1.
+       Set to 2 if the user preferred LC_CTYPE locale is "C".
 
        If it is equal to 1, LC_CTYPE locale is read to decide it it should be
        coerced or not (ex: PYTHONCOERCECLOCALE=1). Internally, it is set to 2
        if the LC_CTYPE locale must be coerced. */
     int coerce_c_locale;
-    int coerce_c_locale_warn; /* PYTHONCOERCECLOCALE=warn */
+
+    /* Emit a warning if the LC_CTYPE locale is coerced?
+
+       Disabled by default. Set to 1 by PYTHONCOERCECLOCALE=warn. */
+    int coerce_c_locale_warn;
 
 #ifdef MS_WINDOWS
     /* If greater than 1, use the "mbcs" encoding instead of the UTF-8
@@ -83,9 +90,17 @@ typedef struct {
     int legacy_windows_fs_encoding;
 #endif
 
-    /* Enable UTF-8 mode?
-       Set by -X utf8 command line option and PYTHONUTF8 environment variable.
-       If set to -1 (default), inherit Py_UTF8Mode value. */
+    /* Enable UTF-8 mode? (PEP 540)
+
+       Disabled by default (equals to 0).
+
+       Set to 1 by "-X utf8" and "-X utf8=1" command line options.
+       Set to 1 by PYTHONUTF8=1 environment variable.
+
+       Set to 0 by "-X utf8=0" and PYTHONUTF8=0.
+
+       If equals to -1, it is set to 1 if the LC_CTYPE locale is "C" or
+       "POSIX", otherwise inherit Py_UTF8Mode value. */
     int utf8_mode;
 
     int dev_mode;           /* Development mode. PYTHONDEVMODE, -X dev */
@@ -104,8 +119,6 @@ typedef struct {
         _PyPreConfig_WINDOWS_INIT \
         .isolated = -1, \
         .use_environment = -1, \
-        .coerce_c_locale = -1, \
-        .utf8_mode = -1, \
         .dev_mode = -1, \
         .allocator = NULL}
 
diff --git a/Lib/test/test_embed.py b/Lib/test/test_embed.py
index c63ea5a..164527a 100644
--- a/Lib/test/test_embed.py
+++ b/Lib/test/test_embed.py
@@ -494,8 +494,8 @@ class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
             if key not in expected_preconfig:
                 expected_preconfig[key] = expected_config[key]
 
-        self.check_core_config(config, expected_config)
         self.check_pre_config(config, expected_preconfig)
+        self.check_core_config(config, expected_config)
         self.check_global_config(config)
 
     def test_init_default_config(self):
@@ -573,7 +573,6 @@ class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
 
     INIT_ENV_PRECONFIG = {
         'allocator': 'malloc',
-        'utf8_mode': 1,
     }
     INIT_ENV_CONFIG = {
         'use_hash_seed': 1,
@@ -581,8 +580,6 @@ class InitConfigTests(EmbeddingTestsMixin, unittest.TestCase):
         'tracemalloc': 2,
         'import_time': 1,
         'malloc_stats': 1,
-        'filesystem_encoding': 'utf-8',
-        'filesystem_errors': UTF8_MODE_ERRORS,
         'inspect': 1,
         'optimization_level': 2,
         'pycache_prefix': 'env_pycache_prefix',
diff --git a/Misc/NEWS.d/next/C API/2019-03-27-15-58-23.bpo-36443.tAfZR9.rst b/Misc/NEWS.d/next/C API/2019-03-27-15-58-23.bpo-36443.tAfZR9.rst
new file mode 100644
index 0000000..3d98c31
--- /dev/null
+++ b/Misc/NEWS.d/next/C API/2019-03-27-15-58-23.bpo-36443.tAfZR9.rst
@@ -0,0 +1,6 @@
+Since Python 3.7.0, calling :c:func:`Py_DecodeLocale` before
+:c:func:`Py_Initialize` produces mojibake if the ``LC_CTYPE`` locale is coerced
+and/or if the UTF-8 Mode is enabled by the user configuration. The LC_CTYPE
+coercion and UTF-8 Mode are now disabled by default to fix the mojibake issue.
+They must now be enabled explicitly (opt-in) using the new
+:c:func:`_Py_PreInitialize` API with ``_PyPreConfig``.
diff --git a/Modules/main.c b/Modules/main.c
index ff79edb..7665769 100644
--- a/Modules/main.c
+++ b/Modules/main.c
@@ -52,23 +52,30 @@ pymain_init(const _PyArgv *args)
     fedisableexcept(FE_OVERFLOW);
 #endif
 
-    _PyCoreConfig config = _PyCoreConfig_INIT;
-
+    _PyPreConfig preconfig = _PyPreConfig_INIT;
+    /* Set to -1 to enable them depending on the LC_CTYPE locale and the
+       environment variables (PYTHONUTF8 and PYTHONCOERCECLOCALE)  */
+    preconfig.coerce_c_locale = -1;
+    preconfig.utf8_mode = -1;
     if (args->use_bytes_argv) {
-        err = _Py_PreInitializeFromArgs(NULL, args->argc, args->bytes_argv);
+        err = _Py_PreInitializeFromArgs(&preconfig,
+                                        args->argc, args->bytes_argv);
     }
     else {
-        err = _Py_PreInitializeFromWideArgs(NULL, args->argc, args->wchar_argv);
+        err = _Py_PreInitializeFromWideArgs(&preconfig,
+                                            args->argc, args->wchar_argv);
     }
     if (_Py_INIT_FAILED(err)) {
         return err;
     }
 
+    /* pass NULL as the config: config is read from command line arguments,
+       environment variables, configuration files */
     if (args->use_bytes_argv) {
-        return _Py_InitializeFromArgs(&config, args->argc, args->bytes_argv);
+        return _Py_InitializeFromArgs(NULL, args->argc, args->bytes_argv);
     }
     else {
-        return _Py_InitializeFromWideArgs(&config, args->argc, args->wchar_argv);
+        return _Py_InitializeFromWideArgs(NULL, args->argc, args->wchar_argv);
     }
 }
 
diff --git a/Programs/_testembed.c b/Programs/_testembed.c
index 425954c..d8e12cf 100644
--- a/Programs/_testembed.c
+++ b/Programs/_testembed.c
@@ -441,8 +441,6 @@ static int test_init_from_config(void)
     putenv("PYTHONMALLOCSTATS=0");
     config.malloc_stats = 1;
 
-    /* FIXME: test coerce_c_locale and coerce_c_locale_warn */
-
     putenv("PYTHONPYCACHEPREFIX=env_pycache_prefix");
     config.pycache_prefix = L"conf_pycache_prefix";
 
@@ -617,17 +615,6 @@ static int test_init_isolated(void)
 {
     _PyInitError err;
 
-    _PyPreConfig preconfig = _PyPreConfig_INIT;
-
-    /* Set coerce_c_locale and utf8_mode to not depend on the locale */
-    preconfig.coerce_c_locale = 0;
-    preconfig.utf8_mode = 0;
-
-    err = _Py_PreInitialize(&preconfig);
-    if (_Py_INIT_FAILED(err)) {
-        _Py_ExitInitError(err);
-    }
-
     /* Test _PyCoreConfig.isolated=1 */
     _PyCoreConfig config = _PyCoreConfig_INIT;
 
@@ -654,10 +641,6 @@ static int test_preinit_isolated1(void)
     _PyInitError err;
 
     _PyPreConfig preconfig = _PyPreConfig_INIT;
-
-    /* Set coerce_c_locale and utf8_mode to not depend on the locale */
-    preconfig.coerce_c_locale = 0;
-    preconfig.utf8_mode = 0;
     preconfig.isolated = 1;
 
     err = _Py_PreInitialize(&preconfig);
@@ -685,10 +668,6 @@ static int test_preinit_isolated2(void)
     _PyInitError err;
 
     _PyPreConfig preconfig = _PyPreConfig_INIT;
-
-    /* Set coerce_c_locale and utf8_mode to not depend on the locale */
-    preconfig.coerce_c_locale = 0;
-    preconfig.utf8_mode = 0;
     preconfig.isolated = 0;
 
     err = _Py_PreInitialize(&preconfig);
diff --git a/Python/preconfig.c b/Python/preconfig.c
index 011ed53..7ac645d 100644
--- a/Python/preconfig.c
+++ b/Python/preconfig.c
@@ -386,7 +386,9 @@ _PyPreConfig_GetGlobalConfig(_PyPreConfig *config)
 #ifdef MS_WINDOWS
     COPY_FLAG(legacy_windows_fs_encoding, Py_LegacyWindowsFSEncodingFlag);
 #endif
-    COPY_FLAG(utf8_mode, Py_UTF8Mode);
+    if (Py_UTF8Mode > 0) {
+        config->utf8_mode = 1;
+    }
 
 #undef COPY_FLAG
 #undef COPY_NOT_FLAG
diff --git a/Python/pylifecycle.c b/Python/pylifecycle.c
index 7c6948e..ad14472 100644
--- a/Python/pylifecycle.c
+++ b/Python/pylifecycle.c
@@ -485,7 +485,7 @@ _Py_Initialize_ReconfigureCore(PyInterpreterState **interp_p,
     _PyCoreConfig_Write(core_config);
 
     if (_PyCoreConfig_Copy(&interp->core_config, core_config) < 0) {
-        return _Py_INIT_ERR("failed to copy core config");
+        return _Py_INIT_NO_MEMORY();
     }
     core_config = &interp->core_config;
 
@@ -548,7 +548,7 @@ pycore_create_interpreter(const _PyCoreConfig *core_config,
     *interp_p = interp;
 
     if (_PyCoreConfig_Copy(&interp->core_config, core_config) < 0) {
-        return _Py_INIT_ERR("failed to copy core config");
+        return _Py_INIT_NO_MEMORY();
     }
     core_config = &interp->core_config;
 
@@ -785,6 +785,7 @@ _Py_PreInitialize(const _PyPreConfig *src_config)
 _PyInitError
 _Py_PreInitializeFromCoreConfig(const _PyCoreConfig *coreconfig)
 {
+    assert(coreconfig != NULL);
     _PyPreConfig config = _PyPreConfig_INIT;
     _PyCoreConfig_GetCoreConfig(&config, coreconfig);
     return _Py_PreInitialize(&config);
@@ -799,8 +800,10 @@ pyinit_coreconfig(_PyCoreConfig *config,
                   const _PyArgv *args,
                   PyInterpreterState **interp_p)
 {
-    if (_PyCoreConfig_Copy(config, src_config) < 0) {
-        return _Py_INIT_ERR("failed to copy core config");
+    if (src_config) {
+        if (_PyCoreConfig_Copy(config, src_config) < 0) {
+            return _Py_INIT_NO_MEMORY();
+        }
     }
 
     _PyInitError err = _PyCoreConfig_Read(config, args);
@@ -839,9 +842,14 @@ _Py_InitializeCore(const _PyCoreConfig *src_config,
                    const _PyArgv *args,
                    PyInterpreterState **interp_p)
 {
-    assert(src_config != NULL);
+    _PyInitError err;
 
-    _PyInitError err = _Py_PreInitializeFromCoreConfig(src_config);
+    if (src_config) {
+        err = _Py_PreInitializeFromCoreConfig(src_config);
+    }
+    else {
+        err = _Py_PreInitialize(NULL);
+    }
     if (_Py_INIT_FAILED(err)) {
         return err;
     }
@@ -1395,7 +1403,7 @@ new_interpreter(PyThreadState **tstate_p)
     }
 
     if (_PyCoreConfig_Copy(&interp->core_config, core_config) < 0) {
-        return _Py_INIT_ERR("failed to copy core config");
+        return _Py_INIT_NO_MEMORY();
     }
     core_config = &interp->core_config;