* Bug fix: Memory corruption when alpine searches for a string that is

     an incomplete utf8 string in a local folder. This could happen by
     chopping a string to make it fit a buffer without regard to its content.
     We fix the string so that chopping it does not damage it. Reported
     by Andrew.

2 files changed, 44 insertions, 0 deletions

diff --git a/pith/charconv/utf8.c b/pith/charconv/utf8.c
index 6b1e2e9d..ee5ab852 100644
--- a/pith/charconv/utf8.c
+++ b/pith/charconv/utf8.c
@@ -683,6 +683,49 @@ ucs4_to_utf8_cpystr_n(UCS *ucs4src, int ucs4src_len)
     return ((char *) ret);
 }
 
+/*
+ * Similar to above but copy what is possible to a
+ * string of a size at most the given retlen.
+ */
+char *
+ucs4_to_utf8_n_cpystr(UCS *ucs4src, int retlen)
+{
+    unsigned char *ret = NULL;
+    unsigned char *writeptr;
+    int            i, oldlen, len;
+
+    if(!ucs4src)
+      return NULL;
+
+    /*
+     * Over-allocate and then resize at the end.
+     */
+
+    /* count characters in source */
+    for(i = 0; ucs4src[i]; i++)
+      ;
+
+    ret = (unsigned char *) fs_get((6*i + 1) * sizeof(unsigned char));
+    memset(ret, 0, (6*i + 1) * sizeof(unsigned char));
+
+    writeptr = ret;
+    oldlen = len = 0;
+    for(i = 0; ucs4src[i] && (len < retlen); i++){
+      oldlen = len;
+      writeptr = utf8_put(writeptr, (unsigned long) ucs4src[i]);
+      len = strlen(ret);
+    }
+    if(len > retlen){
+      ret[oldlen] = '\0';
+      len = oldlen;
+    }
+
+    /* get rid of excess size */
+    fs_resize((void **) &ret, (len + 1) * sizeof(unsigned char));
+
+    return ((char *) ret);
+}
+
 
 #ifdef _WINDOWS
 /*
diff --git a/pith/charconv/utf8.h b/pith/charconv/utf8.h
index f9597cbf..954821cc 100644
--- a/pith/charconv/utf8.h
+++ b/pith/charconv/utf8.h
@@ -66,6 +66,7 @@ UCS           *ucs4_particular_width(UCS*, int);
 UCS           *utf8_to_ucs4_cpystr(char *);
 char          *ucs4_to_utf8_cpystr(UCS *);
 char          *ucs4_to_utf8_cpystr_n(UCS *, int);
+char	      *ucs4_to_utf8_n_cpystr(UCS *, int);
 #ifdef _WINDOWS
 LPTSTR         utf8_to_lptstr(LPSTR);
 LPSTR          lptstr_to_utf8(LPTSTR);