Add native vec[u8] to str converter. Put in workaround for leak in str to vec[u8] converter. Add testcase exercising both. Drive-by fix a potential array-out-of-bounds write on rust_str buffers.

src/Makefile

@@ -477,6 +477,7 @@ TEST_XFAILS_LLVM := $(TASK_XFAILS) \
                       lib-deque.rs \
                       lib-map.rs \
                       lib-rand.rs \
+                      lib-vec-str-conversions.rs \
                       linear-for-loop.rs \
                       list.rs \
                       many.rs \

src/lib/_str.rs

@@ -1,10 +1,13 @@
 import rustrt.sbuf;
 
+import std._vec.rustrt.vbuf;
+
 native "rust" mod rustrt {
   type sbuf;
   fn str_buf(str s) -> sbuf;
   fn str_byte_len(str s) -> uint;
   fn str_alloc(uint n_bytes) -> str;
+  fn str_from_vec(vec[u8] b) -> str;
   fn refcount[T](str s) -> uint;
 }
 
@@ -40,9 +43,33 @@ fn buf(str s) -> sbuf {
   ret rustrt.str_buf(s);
 }
 
-fn bytes(&str s) -> vec[u8] {
-  fn ith(str s, uint i) -> u8 {
-    ret s.(i);
+fn bytes(str s) -> vec[u8] {
+  /* FIXME (issue #58):
+   * Should be...
+   *
+   *  fn ith(str s, uint i) -> u8 {
+   *      ret s.(i);
+   *  }
+   *  ret _vec.init_fn[u8](bind ith(s, _), byte_len(s));
+   *
+   * but we do not correctly decrement refcount of s when
+   * the binding dies, so we have to do this manually.
+   */
+  let uint n = _str.byte_len(s);
+  let vec[u8] v = _vec.alloc[u8](n);
+  let uint i = 0u;
+  while (i < n) {
+    v += vec(s.(i));
+    i += 1u;
   }
-  ret _vec.init_fn[u8](bind ith(s, _), _str.byte_len(s));
+  ret v;
+}
+
+fn from_bytes(vec[u8] v) : is_utf8(v) -> str {
+  ret rustrt.str_from_vec(v);
+}
+
+fn refcount(str s) -> uint {
+  // -1 because calling this function incremented the refcount.
+  ret rustrt.refcount[u8](s) - 1u;
 }

src/rt/rust_builtin.cpp

@@ -2,19 +2,6 @@
 #include "rust_internal.h"
 
 /* Native builtins. */
-extern "C" CDECL rust_str*
-str_alloc(rust_task *task, size_t n_bytes)
-{
-    rust_dom *dom = task->dom;
-    size_t alloc = next_power_of_two(sizeof(rust_str) + n_bytes);
-    void *mem = dom->malloc(alloc);
-    if (!mem) {
-        task->fail(2);
-        return NULL;
-    }
-    rust_str *st = new (mem) rust_str(dom, alloc, 1, (uint8_t const *)"");
-    return st;
-}
 
 extern "C" CDECL rust_str*
 last_os_error(rust_task *task) {
@@ -109,6 +96,48 @@ vec_alloc(rust_task *task, type_desc *t, type_desc *elem_t, size_t n_elts)
     return vec;
 }
 
+extern "C" CDECL void *
+vec_buf(rust_task *task, type_desc *ty, rust_vec *v, size_t offset)
+{
+    return (void *)&v->data[ty->size * offset];
+}
+
+extern "C" CDECL size_t
+vec_len(rust_task *task, type_desc *ty, rust_vec *v)
+{
+    return v->fill / ty->size;
+}
+
+/* Helper for str_alloc and str_from_vec.  Returns NULL as failure. */
+static rust_str *
+str_alloc_with_data(rust_task *task,
+                    size_t n_bytes,
+                    size_t fill,
+                    uint8_t const *d)
+{
+    rust_dom *dom = task->dom;
+    size_t alloc = next_power_of_two(sizeof(rust_str) + n_bytes);
+    void *mem = dom->malloc(alloc);
+    if (!mem)
+        return NULL;
+    rust_str *st = new (mem) rust_str(dom, alloc, fill, d);
+    return st;
+}
+
+extern "C" CDECL rust_str*
+str_alloc(rust_task *task, size_t n_bytes)
+{
+    rust_str *st = str_alloc_with_data(task,
+                                       n_bytes + 1,  // +1 to fit at least ""
+                                       1,
+                                       (uint8_t const *)"");
+    if (!st) {
+        task->fail(2);
+        return NULL;
+    }
+    return st;
+}
+
 extern "C" CDECL char const *
 str_buf(rust_task *task, rust_str *s)
 {
@@ -121,17 +150,37 @@ str_byte_len(rust_task *task, rust_str *s)
     return s->fill - 1;  // -1 for the '\0' terminator.
 }
 
-extern "C" CDECL void *
-vec_buf(rust_task *task, type_desc *ty, rust_vec *v, size_t offset)
+extern "C" CDECL rust_str *
+str_from_vec(rust_task *task, rust_vec *v)
 {
-    return (void *)&v->data[ty->size * offset];
+    rust_str *st =
+        str_alloc_with_data(task,
+                            v->fill + 1,  // +1 to fit at least '\0'
+                            v->fill,
+                            v->fill ? (uint8_t const *)v->data : NULL);
+    if (!st) {
+        task->fail(2);
+        return NULL;
+    }
+    st->data[st->fill++] = '\0';
+    return st;
 }
 
-extern "C" CDECL size_t
-vec_len(rust_task *task, type_desc *ty, rust_vec *v)
+/*
+extern "C" CDECL rust_str*
+str_alloc(rust_task *task, size_t n_bytes)
 {
-    return v->fill / ty->size;
+    rust_dom *dom = task->dom;
+    size_t alloc = next_power_of_two(sizeof(rust_str) + n_bytes);
+    void *mem = dom->malloc(alloc);
+    if (!mem) {
+        task->fail(2);
+        return NULL;
+    }
+    rust_str *st = new (mem) rust_str(dom, alloc, 1, (uint8_t const *)"");
+    return st;
 }
+*/
 
 extern "C" CDECL void *
 rand_new(rust_task *task)

src/test/run-pass/lib-vec-str-conversions.rs

@@ -0,0 +1,41 @@
+// -*- rust -*-
+
+use std;
+import std._str;
+import std._vec;
+
+fn test_simple() {
+  let str s1 = "All mimsy were the borogoves";
+
+  /*
+   * FIXME from_bytes(vec[u8] v) has constraint is_utf(v), which is
+   * unimplemented and thereby just fails.  This doesn't stop us from
+   * using from_bytes for now since the constraint system isn't fully
+   * working, but we should implement is_utf8 before that happens.
+   */
+
+  let vec[u8] v = _str.bytes(s1);
+  let str s2 = _str.from_bytes(v);
+
+  let uint i = 0u;
+  let uint n1 = _str.byte_len(s1);
+  let uint n2 = _vec.len[u8](v);
+
+  check (n1 == n2);
+
+  while (i < n1) {
+    let u8 a = s1.(i);
+    let u8 b = s2.(i);
+    log a;
+    log b;
+    check (a == b);
+    i += 1u;
+  }
+
+  log "refcnt is";
+  log _str.refcount(s1);
+}
+
+fn main() {
+  test_simple();
+}