1 files changed, 21 insertions, 15 deletions

diff --git a/jim-regexp.c b/jim-regexp.c
index 8eb457d..b0411f8 100644
--- a/jim-regexp.c
+++ b/jim-regexp.c
@@ -55,13 +55,18 @@
     #include <regex.h>
 #endif
 #include "jim.h"
+#include "utf8.h"
 
 static void FreeRegexpInternalRep(Jim_Interp *interp, Jim_Obj *objPtr)
 {
-    regfree(objPtr->internalRep.regexpValue.compre);
-    Jim_Free(objPtr->internalRep.regexpValue.compre);
+    regfree(objPtr->internalRep.ptrIntValue.ptr);
+    Jim_Free(objPtr->internalRep.ptrIntValue.ptr);
 }
 
+/* internal rep is stored in ptrIntvalue
+ *  ptr = compiled regex
+ *  int1 = flags
+ */
 static const Jim_ObjType regexpObjType = {
     "regexp",
     FreeRegexpInternalRep,
@@ -78,9 +83,9 @@ static regex_t *SetRegexpFromAny(Jim_Interp *interp, Jim_Obj *objPtr, unsigned f
 
     /* Check if the object is already an uptodate variable */
     if (objPtr->typePtr == &regexpObjType &&
-        objPtr->internalRep.regexpValue.compre && objPtr->internalRep.regexpValue.flags == flags) {
+        objPtr->internalRep.ptrIntValue.ptr && objPtr->internalRep.ptrIntValue.int1 == flags) {
         /* nothing to do */
-        return objPtr->internalRep.regexpValue.compre;
+        return objPtr->internalRep.ptrIntValue.ptr;
     }
 
     /* Not a regexp or the flags do not match */
@@ -102,8 +107,8 @@ static regex_t *SetRegexpFromAny(Jim_Interp *interp, Jim_Obj *objPtr, unsigned f
     Jim_FreeIntRep(interp, objPtr);
 
     objPtr->typePtr = &regexpObjType;
-    objPtr->internalRep.regexpValue.flags = flags;
-    objPtr->internalRep.regexpValue.compre = compre;
+    objPtr->internalRep.ptrIntValue.int1 = flags;
+    objPtr->internalRep.ptrIntValue.ptr = compre;
 
     return compre;
 }
@@ -222,7 +227,7 @@ int Jim_RegexpCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
             source_str += source_len;
         }
         else if (offset > 0) {
-            source_str += offset;
+            source_str += utf8_index(source_str, offset);
         }
         eflags |= REG_NOTBOL;
     }
@@ -276,16 +281,15 @@ int Jim_RegexpCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
             }
         }
         else {
-            int len = pmatch[j].rm_eo - pmatch[j].rm_so;
-
             if (opt_indices) {
-                Jim_ListAppendElement(interp, resultObj, Jim_NewIntObj(interp,
-                        offset + pmatch[j].rm_so));
-                Jim_ListAppendElement(interp, resultObj, Jim_NewIntObj(interp,
-                        offset + pmatch[j].rm_so + len - 1));
+                /* rm_so and rm_eo are byte offsets. We need char offsets */
+                int so = utf8_strlen(source_str, pmatch[j].rm_so);
+                int eo = utf8_strlen(source_str, pmatch[j].rm_eo);
+                Jim_ListAppendElement(interp, resultObj, Jim_NewIntObj(interp, offset + so));
+                Jim_ListAppendElement(interp, resultObj, Jim_NewIntObj(interp, offset + eo - 1));
             }
             else {
-                Jim_AppendString(interp, resultObj, source_str + pmatch[j].rm_so, len);
+                Jim_AppendString(interp, resultObj, source_str + pmatch[j].rm_so, pmatch[j].rm_eo - pmatch[j].rm_so);
             }
         }
 
@@ -306,7 +310,7 @@ int Jim_RegexpCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
   try_next_match:
     if (opt_all && (pattern[0] != '^' || (regcomp_flags & REG_NEWLINE)) && *source_str) {
         if (pmatch[0].rm_eo) {
-            offset += pmatch[0].rm_eo;
+            offset += utf8_strlen(source_str, pmatch[0].rm_eo);
             source_str += pmatch[0].rm_eo;
         }
         else {
@@ -437,6 +441,8 @@ int Jim_RegsubCmd(Jim_Interp *interp, int argc, Jim_Obj *const *argv)
             offset = 0;
         }
     }
+    /* Convert from character offset to byte offset */
+    offset = utf8_index(source_str, offset);
 
     /* Copy the part before -start */
     Jim_AppendString(interp, resultObj, source_str, offset);