Merge pull request #1544 from notro/ure_sub

extmod/ure: Extend functionality

4 files changed, 43 insertions, 0 deletions

diff --git a/py/mpconfig.h b/py/mpconfig.h
index a846f6a25..bb7952a8b 100755
--- a/py/mpconfig.h
+++ b/py/mpconfig.h
@@ -1164,6 +1164,18 @@ typedef double mp_float_t;
 #define MICROPY_PY_URE (0)
 #endif
 
+#ifndef MICROPY_PY_URE_MATCH_GROUPS
+#define MICROPY_PY_URE_MATCH_GROUPS (0)
+#endif
+
+#ifndef MICROPY_PY_URE_MATCH_SPAN_START_END
+#define MICROPY_PY_URE_MATCH_SPAN_START_END (0)
+#endif
+
+#ifndef MICROPY_PY_URE_SUB
+#define MICROPY_PY_URE_SUB (0)
+#endif
+
 #ifndef MICROPY_PY_UHEAPQ
 #define MICROPY_PY_UHEAPQ (0)
 #endif
diff --git a/py/objstr.c b/py/objstr.c
index 7236d9772..ebd11d5cc 100644
--- a/py/objstr.c
+++ b/py/objstr.c
@@ -408,6 +408,15 @@ mp_obj_t mp_obj_str_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_i
 
 #if !MICROPY_PY_BUILTINS_STR_UNICODE
 // objstrunicode defines own version
+size_t str_offset_to_index(const mp_obj_type_t *type, const byte *self_data, size_t self_len,
+                           size_t offset) {
+    if (offset > self_len) {
+        mp_raise_ValueError(translate("offset out of bounds"));
+    }
+
+    return offset;
+}
+
 const byte *str_index_to_ptr(const mp_obj_type_t *type, const byte *self_data, size_t self_len,
                              mp_obj_t index, bool is_slice) {
     size_t index_val = mp_get_index(type, self_len, index, is_slice);
diff --git a/py/objstr.h b/py/objstr.h
index 895130446..0efe62a80 100644
--- a/py/objstr.h
+++ b/py/objstr.h
@@ -71,6 +71,8 @@ mp_obj_t mp_obj_new_str_of_type(const mp_obj_type_t *type, const byte* data, siz
 mp_obj_t mp_obj_str_binary_op(mp_binary_op_t op, mp_obj_t lhs_in, mp_obj_t rhs_in);
 mp_int_t mp_obj_str_get_buffer(mp_obj_t self_in, mp_buffer_info_t *bufinfo, mp_uint_t flags);
 
+size_t str_offset_to_index(const mp_obj_type_t *type, const byte *self_data, size_t self_len,
+                           size_t offset);
 const byte *str_index_to_ptr(const mp_obj_type_t *type, const byte *self_data, size_t self_len,
                              mp_obj_t index, bool is_slice);
 const byte *find_subbytes(const byte *haystack, size_t hlen, const byte *needle, size_t nlen, int direction);
diff --git a/py/objstrunicode.c b/py/objstrunicode.c
index 03106f987..30000a51e 100644
--- a/py/objstrunicode.c
+++ b/py/objstrunicode.c
@@ -112,6 +112,26 @@ STATIC mp_obj_t uni_unary_op(mp_unary_op_t op, mp_obj_t self_in) {
     }
 }
 
+size_t str_offset_to_index(const mp_obj_type_t *type, const byte *self_data, size_t self_len,
+                           size_t offset) {
+    if (offset > self_len) {
+        mp_raise_ValueError(translate("offset out of bounds"));
+    }
+
+    if (type == &mp_type_bytes) {
+        return offset;
+    }
+
+    size_t index_val = 0;
+    const byte *s = self_data;
+    for (size_t i = 0; i < offset; i++, s++) {
+        if (!UTF8_IS_CONT(*s)) {
+            ++index_val;
+        }
+    }
+    return index_val;
+}
+
 // Convert an index into a pointer to its lead byte. Out of bounds indexing will raise IndexError or
 // be capped to the first/last character of the string, depending on is_slice.
 const byte *str_index_to_ptr(const mp_obj_type_t *type, const byte *self_data, size_t self_len,