py/objstr: Make string formatting 8-bit clean.

dpgeorge · dpgeorge · commit 51b9a0d0c4df · 2015-08-29T23:13:51.000+01:00
diff --git a/py/objstr.c b/py/objstr.c
@@ -851,7 +851,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t *args, mp_map_t *kwa
         if (*str == '}') {
             str++;
             if (str < top && *str == '}') {
-                vstr_add_char(&vstr, '}');
+                vstr_add_byte(&vstr, '}');
                 continue;
             }
             if (MICROPY_ERROR_REPORTING == MICROPY_ERROR_REPORTING_TERSE) {
@@ -862,13 +862,13 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t *args, mp_map_t *kwa
             }
         }
         if (*str != '{') {
-            vstr_add_char(&vstr, *str);
+            vstr_add_byte(&vstr, *str);
             continue;
         }
 
         str++;
         if (str < top && *str == '{') {
-            vstr_add_char(&vstr, '{');
+            vstr_add_byte(&vstr, '{');
             continue;
         }
 
@@ -881,7 +881,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t *args, mp_map_t *kwa
         if (str < top && *str != '}' && *str != '!' && *str != ':') {
             field_name = vstr_new();
             while (str < top && *str != '}' && *str != '!' && *str != ':') {
-                vstr_add_char(field_name, *str++);
+                vstr_add_byte(field_name, *str++);
             }
         }
 
@@ -911,7 +911,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t *args, mp_map_t *kwa
             if (*str != '}') {
                 format_spec = vstr_new();
                 while (str < top && *str != '}') {
-                    vstr_add_char(format_spec, *str++);
+                    vstr_add_byte(format_spec, *str++);
                 }
             }
         }
@@ -1290,14 +1290,14 @@ STATIC mp_obj_t str_modulo_format(mp_obj_t pattern, mp_uint_t n_args, const mp_o
     for (const byte *top = str + len; str < top; str++) {
         mp_obj_t arg = MP_OBJ_NULL;
         if (*str != '%') {
-            vstr_add_char(&vstr, *str);
+            vstr_add_byte(&vstr, *str);
             continue;
         }
         if (++str >= top) {
             break;
         }
         if (*str == '%') {
-            vstr_add_char(&vstr, '%');
+            vstr_add_byte(&vstr, '%');
             continue;
         }
 
diff --git a/tests/unicode/unicode_str_format.py b/tests/unicode/unicode_str_format.py
@@ -0,0 +1,4 @@
+# test handling of unicode chars in format strings
+
+print('α'.format())
+print('{α}'.format(α=1))
diff --git a/tests/unicode/unicode_str_modulo.py b/tests/unicode/unicode_str_modulo.py
@@ -0,0 +1,3 @@
+# test handling of unicode chars in string % formatting
+
+print('α' % ())

Original file line number	Diff line number	Diff line change
`@@ -851,7 +851,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t args, mp_map_t kwa`
`851`	`851`	`if (*str == '}') {`
`852`	`852`	`str++;`
`853`	`853`	`if (str < top && *str == '}') {`
`854`		`- vstr_add_char(&vstr, '}');`
	`854`	`+ vstr_add_byte(&vstr, '}');`
`855`	`855`	`continue;`
`856`	`856`	`}`
`857`	`857`	`if (MICROPY_ERROR_REPORTING == MICROPY_ERROR_REPORTING_TERSE) {`
`@@ -862,13 +862,13 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t args, mp_map_t kwa`
`862`	`862`	`}`
`863`	`863`	`}`
`864`	`864`	`if (*str != '{') {`
`865`		`- vstr_add_char(&vstr, *str);`
	`865`	`+ vstr_add_byte(&vstr, *str);`
`866`	`866`	`continue;`
`867`	`867`	`}`
`868`	`868`
`869`	`869`	`str++;`
`870`	`870`	`if (str < top && *str == '{') {`
`871`		`- vstr_add_char(&vstr, '{');`
	`871`	`+ vstr_add_byte(&vstr, '{');`
`872`	`872`	`continue;`
`873`	`873`	`}`
`874`	`874`
`@@ -881,7 +881,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t args, mp_map_t kwa`
`881`	`881`	`if (str < top && str != '}' && str != '!' && *str != ':') {`
`882`	`882`	`field_name = vstr_new();`
`883`	`883`	`while (str < top && str != '}' && str != '!' && *str != ':') {`
`884`		`- vstr_add_char(field_name, *str++);`
	`884`	`+ vstr_add_byte(field_name, *str++);`
`885`	`885`	`}`
`886`	`886`	`}`
`887`	`887`
`@@ -911,7 +911,7 @@ mp_obj_t mp_obj_str_format(mp_uint_t n_args, const mp_obj_t args, mp_map_t kwa`
`911`	`911`	`if (*str != '}') {`
`912`	`912`	`format_spec = vstr_new();`
`913`	`913`	`while (str < top && *str != '}') {`
`914`		`- vstr_add_char(format_spec, *str++);`
	`914`	`+ vstr_add_byte(format_spec, *str++);`
`915`	`915`	`}`
`916`	`916`	`}`
`917`	`917`	`}`
`@@ -1290,14 +1290,14 @@ STATIC mp_obj_t str_modulo_format(mp_obj_t pattern, mp_uint_t n_args, const mp_o`
`1290`	`1290`	`for (const byte *top = str + len; str < top; str++) {`
`1291`	`1291`	`mp_obj_t arg = MP_OBJ_NULL;`
`1292`	`1292`	`if (*str != '%') {`
`1293`		`- vstr_add_char(&vstr, *str);`
	`1293`	`+ vstr_add_byte(&vstr, *str);`
`1294`	`1294`	`continue;`
`1295`	`1295`	`}`
`1296`	`1296`	`if (++str >= top) {`
`1297`	`1297`	`break;`
`1298`	`1298`	`}`
`1299`	`1299`	`if (*str == '%') {`
`1300`		`- vstr_add_char(&vstr, '%');`
	`1300`	`+ vstr_add_byte(&vstr, '%');`
`1301`	`1301`	`continue;`
`1302`	`1302`	`}`
`1303`	`1303`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+# test handling of unicode chars in string % formatting`
	`2`	`+`
	`3`	`+print('α' % ())`