Merge pull request #5171 from tannewt/unicode_filenames

Turn on unicode for FATFS

lib/oofatfs/ff.c

@@ -1175,6 +1175,7 @@ static DWORD get_fat (      /* 0xFFFFFFFF:Disk error, 1:Internal error, 2..0x7FF
                     break;
                 }
             }
+            MP_FALLTHROUGH
             /* go to default */
 #endif
         default:
@@ -5518,6 +5519,7 @@ FRESULT f_mkfs (
                 }
                 st = 1;         /* Do not compress short run */
                 /* go to next case */
+                MP_FALLTHROUGH
             case 1:
                 ch = si++;      /* Fill the short run */
                 if (--j == 0) st = 0;

lib/oofatfs/ffconf.h

@@ -163,8 +163,11 @@
 /  memory for the working buffer, memory management functions, ff_memalloc() and
 /  ff_memfree() in ffsystem.c, need to be added to the project. */
 
-
+#ifdef MICROPY_FATFS_LFN_UNICODE
+#define FF_LFN_UNICODE  (MICROPY_FATFS_LFN_UNICODE)
+#else
 #define FF_LFN_UNICODE  0
+#endif
 /* This option switches the character encoding on the API when LFN is enabled.
 /
 /   0: ANSI/OEM in current CP (TCHAR = char)

locale/fr.po

@@ -266,7 +266,7 @@ msgstr "l'objet %s ne supporte pas l'assignation d'éléments"
 #: py/obj.c
 #, c-format
 msgid "'%s' object doesn't support item deletion"
-msgstr "L’objet '%s' ne prend pas en charge la suppression d’éléments"
+msgstr "L'objet '%s' ne prend pas en charge la suppression d'éléments"
 
 #: py/runtime.c
 msgid "'%s' object has no attribute '%q'"

ports/atmel-samd/boards/arduino_zero/mpconfigboard.h

@@ -19,3 +19,11 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// SWD-only
+#define IGNORE_PIN_PA30     1
+#define IGNORE_PIN_PA31     1

ports/atmel-samd/boards/datum_distance/mpconfigboard.h

@@ -19,3 +19,24 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+// Schematic at: https://jandjstudios.io/assets/pdfs/datum-Distance.pdf
+#define IGNORE_PIN_PA03     1
+#define IGNORE_PIN_PA02     1
+#define IGNORE_PIN_PB08     1
+#define IGNORE_PIN_PB09     1
+#define IGNORE_PIN_PA04     1
+#define IGNORE_PIN_PA05     1
+#define IGNORE_PIN_PB02     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA15     1
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PA28     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA13     1

ports/atmel-samd/boards/datum_imu/mpconfigboard.h

@@ -19,3 +19,22 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+// Schematic at: https://jandjstudios.io/assets/pdfs/datum-IMU.pdf
+#define IGNORE_PIN_PB08     1
+#define IGNORE_PIN_PB09     1
+#define IGNORE_PIN_PA04     1
+#define IGNORE_PIN_PA05     1
+#define IGNORE_PIN_PB02     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA15     1
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PA28     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA13     1

ports/atmel-samd/boards/datum_light/mpconfigboard.h

@@ -19,3 +19,24 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+// Schematic at: https://jandjstudios.io/assets/pdfs/datum-Light.pdf
+#define IGNORE_PIN_PA03     1
+#define IGNORE_PIN_PA02     1
+#define IGNORE_PIN_PB08     1
+#define IGNORE_PIN_PB09     1
+#define IGNORE_PIN_PA04     1
+#define IGNORE_PIN_PA05     1
+#define IGNORE_PIN_PB02     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA15     1
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PA28     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA13     1

ports/atmel-samd/boards/datum_weather/mpconfigboard.h

@@ -19,3 +19,24 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+// Schematic at: https://jandjstudios.io/assets/pdfs/datum-Weather.pdf
+#define IGNORE_PIN_PA03     1
+#define IGNORE_PIN_PA02     1
+#define IGNORE_PIN_PB08     1
+#define IGNORE_PIN_PB09     1
+#define IGNORE_PIN_PA04     1
+#define IGNORE_PIN_PA05     1
+#define IGNORE_PIN_PB02     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA15     1
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PA28     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA13     1

ports/atmel-samd/boards/feather_m0_adalogger/mpconfigboard.h

@@ -17,3 +17,21 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA09     1
+#define IGNORE_PIN_PA13     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA27     1
+#define IGNORE_PIN_PA28     1
+
+// SWD-only
+#define IGNORE_PIN_PA30     1
+#define IGNORE_PIN_PA31     1

ports/atmel-samd/boards/feather_m0_basic/mpconfigboard.h

@@ -17,3 +17,24 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1
+
+// Unconnected
+#define IGNORE_PIN_PB03     1
+#define IGNORE_PIN_PB22     1
+#define IGNORE_PIN_PB23     1
+#define IGNORE_PIN_PA06     1
+#define IGNORE_PIN_PA08     1
+#define IGNORE_PIN_PA09     1
+#define IGNORE_PIN_PA13     1
+#define IGNORE_PIN_PA14     1
+#define IGNORE_PIN_PA21     1
+#define IGNORE_PIN_PA27     1
+#define IGNORE_PIN_PA28     1
+
+// SWD-only
+#define IGNORE_PIN_PA30     1
+#define IGNORE_PIN_PA31     1

ports/atmel-samd/boards/kicksat-sprite/mpconfigboard.h

@@ -19,3 +19,5 @@
 
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+#define MICROPY_FATFS_EXFAT 0

ports/atmel-samd/boards/kicksat-sprite/mpconfigboard.mk

@@ -15,6 +15,7 @@ CIRCUITPY_AUDIOMP3 = 0
 CIRCUITPY_BLEIO_HCI = 0
 CIRCUITPY_DISPLAYIO = 0
 CIRCUITPY_FRAMEBUFFERIO = 0
+CIRCUITPY_GETPASS = 0
 CIRCUITPY_KEYPAD = 0
 CIRCUITPY_MSGPACK = 0
 CIRCUITPY_PS2IO = 0

ports/atmel-samd/boards/sensebox_mcu/mpconfigboard.h

@@ -18,3 +18,7 @@
 // USB is always used internally so skip the pin objects for it.
 #define IGNORE_PIN_PA24     1
 #define IGNORE_PIN_PA25     1
+
+// Connected to a crystal
+#define IGNORE_PIN_PA00     1
+#define IGNORE_PIN_PA01     1

ports/atmel-samd/mpconfigport.h

@@ -59,6 +59,8 @@
     X(EISDIR) \
     X(EINVAL) \
 
+#define MICROPY_FATFS_EXFAT 0
+
 #endif // SAMD21
 
 ////////////////////////////////////////////////////////////////////////////////////////////////////
@@ -238,8 +240,13 @@
 
 #include "peripherals/samd/dma.h"
 
+#if CIRCUITPY_AUDIOCORE
 #define MICROPY_PORT_ROOT_POINTERS \
     CIRCUITPY_COMMON_ROOT_POINTERS \
     mp_obj_t playing_audio[AUDIO_DMA_CHANNEL_COUNT];
+#else
+#define MICROPY_PORT_ROOT_POINTERS \
+    CIRCUITPY_COMMON_ROOT_POINTERS
+#endif
 
 #endif  // __INCLUDED_MPCONFIGPORT_H

ports/nrf/boards/pca10100/mpconfigboard.h

@@ -48,3 +48,5 @@
 #define BLEIO_ATTR_TAB_SIZE (BLE_GATTS_ATTR_TAB_SIZE_DEFAULT * 2)
 
 #define SOFTDEVICE_RAM_SIZE (32 * 1024)
+
+#define MICROPY_FATFS_EXFAT 0

ports/nrf/boards/simmel/mpconfigboard.h

@@ -57,3 +57,5 @@
 #define BLEIO_ATTR_TAB_SIZE (BLE_GATTS_ATTR_TAB_SIZE_DEFAULT * 2)
 
 #define SOFTDEVICE_RAM_SIZE (32 * 1024)
+
+#define MICROPY_FATFS_EXFAT 0

ports/stm/boards/espruino_pico/mpconfigboard.h

@@ -41,3 +41,5 @@
 #define BOARD_HAS_LOW_SPEED_CRYSTAL (0)
 // #define BOARD_HAS_LOW_SPEED_CRYSTAL (1)
 // #define LSE_VALUE ((uint32_t)32768)
+
+#define MICROPY_FATFS_EXFAT 0

ports/stm/boards/pyb_nano_v2/mpconfigboard.h

@@ -47,3 +47,5 @@
 #define BOARD_FLASH_SIZE (FLASH_SIZE - 0x2000 - 0xC000)
 
 #define AUTORESET_DELAY_MS (500)
+
+#define MICROPY_FATFS_EXFAT 0

ports/stm/boards/stm32f411ce_blackpill_with_flash/mpconfigboard.h

@@ -50,3 +50,5 @@
 #define BOARD_FLASH_SIZE (FLASH_SIZE - 0x2000 - 0xC000)
 
 #define AUTORESET_DELAY_MS (500)
+
+#define MICROPY_FATFS_EXFAT 0

ports/stm/boards/stm32f411ve_discovery/mpconfigboard.h

@@ -39,3 +39,5 @@
 // The schematic has a 32k crystal that isn't fitted. Uncommented the line below if you add it.
 // #define BOARD_HAS_LOW_SPEED_CRYSTAL (1)
 // #define LSE_VALUE  ((uint32_t)32000U)
+
+#define MICROPY_FATFS_EXFAT 0

ports/stm/boards/thunderpack_v11/mpconfigboard.h

@@ -51,3 +51,5 @@
 
 #define DEFAULT_I2C_BUS_SCL         (&pin_PB06)
 #define DEFAULT_I2C_BUS_SDA         (&pin_PB07)
+
+#define MICROPY_FATFS_EXFAT 0

ports/unix/mpconfigport.h

@@ -60,6 +60,7 @@
 #define MICROPY_MALLOC_USES_ALLOCATED_SIZE (1)
 #define MICROPY_MEM_STATS           (1)
 #define MICROPY_DEBUG_PRINTERS      (1)
+#define CIRCUITPY_MICROPYTHON_ADVANCED (1)
 // Printing debug to stderr may give tests which
 // check stdout a chance to pass, etc.
 #define MICROPY_DEBUG_PRINTER       (&mp_stderr_print)

py/circuitpy_mpconfig.h

@@ -132,12 +132,14 @@
 //
 // 1 = SFN/ANSI 437=LFN/U.S.(OEM)
 #define MICROPY_FATFS_ENABLE_LFN      (1)
+// Code page is ignored because unicode is enabled.
 // Don't use parens on the value below because it gets combined with a prefix in
 // the preprocessor.
 #define MICROPY_FATFS_LFN_CODE_PAGE   437
 #define MICROPY_FATFS_USE_LABEL       (1)
 #define MICROPY_FATFS_RPATH           (2)
 #define MICROPY_FATFS_MULTI_PARTITION (1)
+#define MICROPY_FATFS_LFN_UNICODE      2  // UTF-8
 
 // Only enable this if you really need it. It allocates a byte cache of this size.
 // #define MICROPY_FATFS_MAX_SS           (4096)
@@ -209,6 +211,12 @@ typedef long mp_off_t;
 #define MICROPY_PY_URE_MATCH_SPAN_START_END   (CIRCUITPY_RE)
 #define MICROPY_PY_URE_SUB                    (CIRCUITPY_RE)
 
+#define CIRCUITPY_MICROPYTHON_ADVANCED        (CIRCUITPY_FULL_BUILD)
+
+#ifndef MICROPY_FATFS_EXFAT
+#define MICROPY_FATFS_EXFAT           (CIRCUITPY_FULL_BUILD)
+#endif
+
 // LONGINT_IMPL_xxx are defined in the Makefile.
 //
 #ifdef LONGINT_IMPL_NONE

py/modmicropython.c

@@ -37,7 +37,7 @@
 // Various builtins specific to MicroPython runtime,
 // living in micropython module
 
-#if MICROPY_ENABLE_COMPILER
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_COMPILER
 STATIC mp_obj_t mp_micropython_opt_level(size_t n_args, const mp_obj_t *args) {
     if (n_args == 0) {
         return MP_OBJ_NEW_SMALL_INT(MP_STATE_VM(mp_optimise_value));
@@ -49,7 +49,7 @@ STATIC mp_obj_t mp_micropython_opt_level(size_t n_args, const mp_obj_t *args) {
 STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(mp_micropython_opt_level_obj, 0, 1, mp_micropython_opt_level);
 #endif
 
-#if MICROPY_PY_MICROPYTHON_MEM_INFO
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_PY_MICROPYTHON_MEM_INFO
 
 #if MICROPY_MEM_STATS
 STATIC mp_obj_t mp_micropython_mem_total(void) {
@@ -109,21 +109,21 @@ STATIC MP_DEFINE_CONST_FUN_OBJ_VAR_BETWEEN(mp_micropython_qstr_info_obj, 0, 1, m
 
 #endif // MICROPY_PY_MICROPYTHON_MEM_INFO
 
-#if MICROPY_PY_MICROPYTHON_STACK_USE
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_PY_MICROPYTHON_STACK_USE
 STATIC mp_obj_t mp_micropython_stack_use(void) {
     return MP_OBJ_NEW_SMALL_INT(mp_stack_usage());
 }
 STATIC MP_DEFINE_CONST_FUN_OBJ_0(mp_micropython_stack_use_obj, mp_micropython_stack_use);
 #endif
 
-#if MICROPY_ENABLE_PYSTACK
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_PYSTACK
 STATIC mp_obj_t mp_micropython_pystack_use(void) {
     return MP_OBJ_NEW_SMALL_INT(mp_pystack_usage());
 }
 STATIC MP_DEFINE_CONST_FUN_OBJ_0(mp_micropython_pystack_use_obj, mp_micropython_pystack_use);
 #endif
 
-#if MICROPY_ENABLE_GC
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_GC
 STATIC mp_obj_t mp_micropython_heap_lock(void) {
     gc_lock();
     return mp_const_none;
@@ -144,11 +144,11 @@ STATIC MP_DEFINE_CONST_FUN_OBJ_0(mp_micropython_heap_locked_obj, mp_micropython_
 #endif
 #endif
 
-#if MICROPY_ENABLE_EMERGENCY_EXCEPTION_BUF && (MICROPY_EMERGENCY_EXCEPTION_BUF_SIZE == 0)
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_EMERGENCY_EXCEPTION_BUF && (MICROPY_EMERGENCY_EXCEPTION_BUF_SIZE == 0)
 STATIC MP_DEFINE_CONST_FUN_OBJ_1(mp_alloc_emergency_exception_buf_obj, mp_alloc_emergency_exception_buf);
 #endif
 
-#if MICROPY_KBD_EXCEPTION
+#if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_KBD_EXCEPTION
 STATIC mp_obj_t mp_micropython_kbd_intr(mp_obj_t int_chr_in) {
     mp_hal_set_interrupt_char(mp_obj_get_int(int_chr_in));
     return mp_const_none;
@@ -169,10 +169,10 @@ STATIC MP_DEFINE_CONST_FUN_OBJ_2(mp_micropython_schedule_obj, mp_micropython_sch
 STATIC const mp_rom_map_elem_t mp_module_micropython_globals_table[] = {
     { MP_ROM_QSTR(MP_QSTR___name__), MP_ROM_QSTR(MP_QSTR_micropython) },
     { MP_ROM_QSTR(MP_QSTR_const), MP_ROM_PTR(&mp_identity_obj) },
-    #if MICROPY_ENABLE_COMPILER
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_COMPILER
     { MP_ROM_QSTR(MP_QSTR_opt_level), MP_ROM_PTR(&mp_micropython_opt_level_obj) },
     #endif
-    #if MICROPY_PY_MICROPYTHON_MEM_INFO
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_PY_MICROPYTHON_MEM_INFO
     #if MICROPY_MEM_STATS
     { MP_ROM_QSTR(MP_QSTR_mem_total), MP_ROM_PTR(&mp_micropython_mem_total_obj) },
     { MP_ROM_QSTR(MP_QSTR_mem_current), MP_ROM_PTR(&mp_micropython_mem_current_obj) },
@@ -181,23 +181,23 @@ STATIC const mp_rom_map_elem_t mp_module_micropython_globals_table[] = {
     { MP_ROM_QSTR(MP_QSTR_mem_info), MP_ROM_PTR(&mp_micropython_mem_info_obj) },
     { MP_ROM_QSTR(MP_QSTR_qstr_info), MP_ROM_PTR(&mp_micropython_qstr_info_obj) },
     #endif
-    #if MICROPY_PY_MICROPYTHON_STACK_USE
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_PY_MICROPYTHON_STACK_USE
     { MP_ROM_QSTR(MP_QSTR_stack_use), MP_ROM_PTR(&mp_micropython_stack_use_obj) },
     #endif
-    #if MICROPY_ENABLE_EMERGENCY_EXCEPTION_BUF && (MICROPY_EMERGENCY_EXCEPTION_BUF_SIZE == 0)
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_EMERGENCY_EXCEPTION_BUF && (MICROPY_EMERGENCY_EXCEPTION_BUF_SIZE == 0)
     { MP_ROM_QSTR(MP_QSTR_alloc_emergency_exception_buf), MP_ROM_PTR(&mp_alloc_emergency_exception_buf_obj) },
     #endif
-    #if MICROPY_ENABLE_PYSTACK
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_PYSTACK
     { MP_ROM_QSTR(MP_QSTR_pystack_use), MP_ROM_PTR(&mp_micropython_pystack_use_obj) },
     #endif
-    #if MICROPY_ENABLE_GC
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_ENABLE_GC
     { MP_ROM_QSTR(MP_QSTR_heap_lock), MP_ROM_PTR(&mp_micropython_heap_lock_obj) },
     { MP_ROM_QSTR(MP_QSTR_heap_unlock), MP_ROM_PTR(&mp_micropython_heap_unlock_obj) },
     #if MICROPY_PY_MICROPYTHON_HEAP_LOCKED
     { MP_ROM_QSTR(MP_QSTR_heap_locked), MP_ROM_PTR(&mp_micropython_heap_locked_obj) },
     #endif
     #endif
-    #if MICROPY_KBD_EXCEPTION
+    #if CIRCUITPY_MICROPYTHON_ADVANCED && MICROPY_KBD_EXCEPTION
     { MP_ROM_QSTR(MP_QSTR_kbd_intr), MP_ROM_PTR(&mp_micropython_kbd_intr_obj) },
     #endif
     #if MICROPY_ENABLE_SCHEDULER

py/objstrunicode.c

@@ -41,6 +41,13 @@ STATIC mp_obj_t mp_obj_new_str_iterator(mp_obj_t str, mp_obj_iter_buf_t *iter_bu
 /******************************************************************************/
 /* str                                                                        */
 
+
+// These settings approximate CPython's printability. It is not
+// exhaustive and may print "unprintable" characters. All ASCII control codes
+// are escaped along with variable space widths and paragraph designators.
+// Unlike CPython, we do not escape private use codes or reserved characters.
+// We assume that the unicode is well formed.
+// CPython policy is documented here: https://github.com/python/cpython/blob/bb3e0c240bc60fe08d332ff5955d54197f79751c/Objects/unicodectype.c#L147-L159
 STATIC void uni_print_quoted(const mp_print_t *print, const byte *str_data, uint str_len) {
     // this escapes characters, but it will be very slow to print (calling print many times)
     bool has_single_quote = false;
@@ -61,25 +68,26 @@ STATIC void uni_print_quoted(const mp_print_t *print, const byte *str_data, uint
     while (s < top) {
         unichar ch;
         ch = utf8_get_char(s);
+        const byte *start = s;
         s = utf8_next_char(s);
         if (ch == quote_char) {
             mp_printf(print, "\\%c", quote_char);
         } else if (ch == '\\') {
             mp_print_str(print, "\\\\");
-        } else if (32 <= ch && ch <= 126) {
-            mp_printf(print, "%c", ch);
         } else if (ch == '\n') {
             mp_print_str(print, "\\n");
         } else if (ch == '\r') {
             mp_print_str(print, "\\r");
         } else if (ch == '\t') {
             mp_print_str(print, "\\t");
-        } else if (ch < 0x100) {
+        } else if (ch <= 0x1f || (0x7f <= ch && ch <= 0xa0) || ch == 0xad) {
             mp_printf(print, "\\x%02x", ch);
-        } else if (ch < 0x10000) {
+        } else if ((0x2000 <= ch && ch <= 0x200f) || ch == 0x2028 || ch == 0x2029) {
             mp_printf(print, "\\u%04x", ch);
         } else {
-            mp_printf(print, "\\U%08x", ch);
+            // Print the full character out.
+            int width = s - start;
+            mp_print_strn(print, (const char *)start, width, 0, ' ', width);
         }
     }
     mp_printf(print, "%c", quote_char);

tests/unicode/unicode.py

@@ -19,8 +19,7 @@ print(enc, enc.decode() == s)
 
 # printing of unicode chars using repr
 # NOTE: for some characters (eg \u10ff) we differ to CPython
-print(repr("a\uffff"))
-print(repr("a\U0001ffff"))
+print(repr("a\u2000"))
 
 # test invalid escape code
 try:

tests/unicode/unicode_repr.py

@@ -0,0 +1,8 @@
+# ¥ is 1 byte wide
+# Πis 2 bytes wide
+# 😅 is 4 bytes wide
+
+a = "hello¥Œ😅.txt\n\r\t'\"\\"
+
+print(a)
+print(repr(a))