buf: stream: switch approach for unget_char.

We switch to the method used in string streams for
ungetting characters, whereby we UTF-8 encode the pushed
back character and push back the subsequent bytes,
thereby unifying character and byte pushback.

* buf.c (struct buf_strm): Remove member unget_c.
(buf_strm_mark): Remove reference to unget_c.
(strm_get_char): Remove code for obtaining previously
pushed back character from s->unget_c stack.
(buf_strm_unget_char): Rewrite using the approach
of using utf8_encode to write the multi-byte representation
of the character into utf8_tiny_buf, and then
pushing back the bytes.
(make_buf_stream): Don't initialize removed unget_c.

* tests/018/streams.tl: New tests.

2 files changed, 47 insertions, 18 deletions

diff --git a/buf.c b/buf.c
index 360d5bec..d5782dc2 100644
--- a/buf.c
+++ b/buf.c
@@ -1086,7 +1086,6 @@ struct buf_strm {
   int is_byte_oriented;
   val buf;
   val pos;
-  val unget_c;
 };
 
 static void buf_strm_mark(val stream)
@@ -1095,7 +1094,6 @@ static void buf_strm_mark(val stream)
   strm_base_mark(&b->a);
   gc_mark(b->buf);
   gc_mark(b->pos);
-  gc_mark(b->unget_c);
 }
 
 static int buf_strm_put_byte_callback(int b, mem_t *ctx)
@@ -1147,23 +1145,18 @@ static int buf_strm_get_byte_callback(mem_t *ctx)
 static val buf_strm_get_char(val stream)
 {
   struct buf_strm *s = coerce(struct buf_strm *, stream->co.handle);
+  wint_t ch;
 
-  if (s->unget_c) {
-    return rcyc_pop(&s->unget_c);
+  if (s->is_byte_oriented) {
+    ch = buf_strm_get_byte_callback(coerce(mem_t *, s));
+    if (ch == 0)
+      ch = 0xDC00;
   } else {
-    wint_t ch;
-
-    if (s->is_byte_oriented) {
-      ch = buf_strm_get_byte_callback(coerce(mem_t *, s));
-      if (ch == 0)
-        ch = 0xDC00;
-    } else {
-      ch = utf8_decode(&s->ud, buf_strm_get_byte_callback,
-                       coerce(mem_t *, s));
-    }
-
-    return (ch != WEOF) ? chr(ch) : nil;
+    ch = utf8_decode(&s->ud, buf_strm_get_byte_callback,
+                     coerce(mem_t *, s));
   }
+
+  return (ch != WEOF) ? chr(ch) : nil;
 }
 
 static val buf_strm_get_byte(val stream)
@@ -1175,8 +1168,28 @@ static val buf_strm_get_byte(val stream)
 
 static val buf_strm_unget_char(val stream, val ch)
 {
+  val self = lit("unget-char");
   struct buf_strm *s = coerce(struct buf_strm *, stream->co.handle);
-  mpush(ch, mkloc(s->unget_c, stream));
+  struct buf *b = us_buf_handle(s->buf);
+  struct utf8_tiny_buf bu;
+  unsigned char *bend = bu.buf + sizeof bu.buf;
+  ucnum index = c_unum(s->pos, self);
+
+  bu.ptr = bend;
+
+  (void) utf8_encode(c_chr(ch), utf8_tiny_buf_putc, coerce(mem_t *, &bu));
+
+  if (convert(size_t, bend - bu.ptr) > index) {
+    uw_throwf(file_error_s,
+              lit("~a: cannot push back past start of stream ~s"),
+              stream, self, nao);
+  }
+
+  while (bu.ptr < bend)
+    b->data[--index] = *bu.ptr++;
+
+  s->pos = unum(index);
+
   return ch;
 }
 
@@ -1367,7 +1380,6 @@ val make_buf_stream(val buf_opt)
   s->buf = nil;
   s->pos = zero;
   s->is_byte_oriented = 0;
-  s->unget_c = nil;
   stream = cobj(coerce(mem_t *, s), stream_cls, &buf_strm_ops.cobj_ops);
   s->buf = buf;
 
diff --git a/tests/018/streams.tl b/tests/018/streams.tl
index f82bc06e..84f1b89b 100644
--- a/tests/018/streams.tl
+++ b/tests/018/streams.tl
@@ -249,3 +249,20 @@
       (seek-stream s -3 :from-end) t
       (fill-buf-adjust b 0 s) 3
       b #b'DDEEFF')))
+
+(with-in-buf-stream (s (buf-str "ABCD"))
+  (mtest
+    (seek-stream s 0 :from-start) t
+    (unget-byte 3 s) :error
+    (get-char s) #\A
+    (get-byte s) 66
+    (get-char s) #\C
+    (seek-stream s 0 :from-current) 3
+    (unget-char #\x3042 s) #\x3042
+    (seek-stream s 0 :from-current) 0
+    (get-char s) #\x3042
+    (unget-char #\x3042 s) #\x3042
+    (get-byte s) #xe3
+    (get-char s) #\xdc81
+    (seek-stream s 0 :from-current) 2
+    (unget-char #\x3042 s) :error))