ure: Convert byte offsets to unicode indices when necessary.

.. and add a test. Closes: adafruit#9202.
jepler · Sep 5, 2022 · 0605004 · 0605004
1 parent 0b26efe
commit 0605004
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 0 deletions.
diff --git a/extmod/modure.c b/extmod/modure.c
@@ -33,6 +33,10 @@
 #include "py/objstr.h"
 #include "py/stackctrl.h"
 
+#if MICROPY_PY_BUILTINS_STR_UNICODE
+#include "py/unicode.h"
+#endif
+
 #if MICROPY_PY_URE
 
 #define re1_5_stack_chk() MP_STACK_CHECK()
@@ -121,6 +125,18 @@ STATIC void match_span_helper(size_t n_args, const mp_obj_t *args, mp_obj_t span
         e = self->caps[no * 2 + 1] - begin;
     }
 
+#if MICROPY_PY_BUILTINS_STR_UNICODE
+    if(mp_obj_get_type(self->str) == &mp_type_str) {
+        const byte *begin = (const byte *)mp_obj_str_get_str(self->str);
+        if (s != -1) {
+            s = utf8_ptr_to_index(begin, begin+s);
+        }
+        if (e != -1) {
+            e = utf8_ptr_to_index(begin, begin+e);
+        }
+    }
+#endif
+
     span[0] = mp_obj_new_int(s);
     span[1] = mp_obj_new_int(e);
 }

diff --git a/tests/extmod/ure_span.py b/tests/extmod/ure_span.py
@@ -34,6 +34,9 @@ def print_spans(match):
 m = re.match(r"([0-9]*)(([a-z]*)([0-9]*))", "1234hello567")
 print_spans(m)
 
+m = re.match(r"([0-9]*)(([a-z]*)([0-9]*))", "1234\u2764567")
+print_spans(m)
+
 # optional span that matches
 print_spans(re.match(r"(a)?b(c)", "abc"))