From 3f6e8ffe64b57b0eaba6812208e94500422ca40c Mon Sep 17 00:00:00 2001
From: Jeffrey Yasskin <jyasskin@gmail.com>
Date: Sun, 25 Jul 2010 00:36:03 -0700
Subject: [PATCH] Implement _str.len() to return the number of bytes, rename it
 to byte_len(), and add a test.

---
 src/Makefile                 |  1 +
 src/lib/_str.rs              | 15 ++++++++++-----
 src/rt/rust_builtin.cpp      |  6 ++++++
 src/test/run-pass/str-lib.rs | 16 ++++++++++++++++
 4 files changed, 33 insertions(+), 5 deletions(-)
 create mode 100644 src/test/run-pass/str-lib.rs

diff --git a/src/Makefile b/src/Makefile
index 792eeec0a3b..1d79a46700a 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -506,6 +506,7 @@ TEST_XFAILS_LLVM := $(TASK_XFAILS) \
                       str-append.rs \
                       str-concat.rs \
                       str-idx.rs \
+                      str-lib.rs \
                       tag.rs \
                       tail-cps.rs \
                       tail-direct.rs \
diff --git a/src/lib/_str.rs b/src/lib/_str.rs
index 7d1a2dbdcd7..a607c7d5df5 100644
--- a/src/lib/_str.rs
+++ b/src/lib/_str.rs
@@ -3,7 +3,7 @@ import rustrt.sbuf;
 native "rust" mod rustrt {
   type sbuf;
   fn str_buf(str s) -> sbuf;
-  fn str_len(str s) -> uint;
+  fn str_byte_len(str s) -> uint;
   fn str_alloc(uint n_bytes) -> str;
   fn refcount[T](str s) -> uint;
 }
@@ -13,7 +13,7 @@ fn is_utf8(vec[u8] v) -> bool {
 }
 
 fn is_ascii(str s) -> bool {
-  let uint i = len(s);
+  let uint i = byte_len(s);
   while (i > 0u) {
     i -= 1u;
     if ((s.(i) & 0x80u8) != 0u8) {
@@ -27,8 +27,13 @@ fn alloc(uint n_bytes) -> str {
   ret rustrt.str_alloc(n_bytes);
 }
 
-fn len(str s) -> uint {
-  ret rustrt.str_len(s);
+// Returns the number of bytes (a.k.a. UTF-8 code units) in s.
+// Contrast with a function that would return the number of code
+// points (char's), combining character sequences, words, etc.  See
+// http://icu-project.org/apiref/icu4c/classBreakIterator.html for a
+// way to implement those.
+fn byte_len(str s) -> uint {
+  ret rustrt.str_byte_len(s);
 }
 
 fn buf(str s) -> sbuf {
@@ -39,5 +44,5 @@ fn bytes(&str s) -> vec[u8] {
   fn ith(str s, uint i) -> u8 {
     ret s.(i);
   }
-  ret _vec.init_fn[u8](bind ith(s, _), _str.len(s));
+  ret _vec.init_fn[u8](bind ith(s, _), _str.byte_len(s));
 }
diff --git a/src/rt/rust_builtin.cpp b/src/rt/rust_builtin.cpp
index 657109c6df2..d8d9b8d6eb4 100644
--- a/src/rt/rust_builtin.cpp
+++ b/src/rt/rust_builtin.cpp
@@ -115,6 +115,12 @@ str_buf(rust_task *task, rust_str *s)
     return (char const *)&s->data[0];
 }
 
+extern "C" CDECL size_t
+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)
 {
diff --git a/src/test/run-pass/str-lib.rs b/src/test/run-pass/str-lib.rs
new file mode 100644
index 00000000000..585f9b8de5f
--- /dev/null
+++ b/src/test/run-pass/str-lib.rs
@@ -0,0 +1,16 @@
+use std;
+import std._str;
+
+fn test_bytes_len() {
+  check (_str.byte_len("") == 0u);
+  check (_str.byte_len("hello world") == 11u);
+  check (_str.byte_len("\x63") == 1u);
+  check (_str.byte_len("\xa2") == 2u);
+  check (_str.byte_len("\u03c0") == 2u);
+  check (_str.byte_len("\u2620") == 3u);
+  check (_str.byte_len("\U0001d11e") == 4u);
+}
+
+fn main() {
+  test_bytes_len();
+}