Fix bug #1 - avoid broken UTF-8 at line breaks

diff --git a/src/writer.c b/src/writer.c
index 5448d6c0b805d1e5b26e7e9f10081663eec42be3..5f95ba2fdb3659f09ba1fe8c24e3a508bf89c423 100644 (file)
--- a/src/writer.c
+++ b/src/writer.c
@@ -54,12 +54,18 @@ void out_char(const char *chunk) {
        } else if (bufpos>wrap_margin) {
                char *q=outputbuffer,*p=outputbuffer+wrap_margin;
                
-               while (p>outputbuffer&&!isspace(*p)) p--;
+               while (p>outputbuffer&&*p!=' '&& *p!='\t') p--;
                if (p==outputbuffer) {
                        /*worst case - nowhere to wrap. Will use brute force */
-                       fwrite(outputbuffer,wrap_margin,1,stdout);
+                       int i = wrap_margin;
+                       if (from_unicode == to_utf8) {
+                               /* go back to start of nearest utf-8 character */
+                               while(i>0 && (outputbuffer[i] & 0xC0) == 0x80) i--;
+                       }                       
+                       fwrite(outputbuffer,i,1,stdout);
+
                        fputc('\n',stdout);
-                       p=outputbuffer+wrap_margin;
+                       p=outputbuffer+i;
                } else {
                        *p=0;p++;
                        fputs(outputbuffer,stdout);