From 71bbc40ffa4bb16ef0fc31f8015f61709fac36ce Mon Sep 17 00:00:00 2001
From: Samuel Williams <samuel.williams@oriontransfer.co.nz>
Date: Mon, 20 Dec 2021 12:11:21 +1300
Subject: Rename `to_str` -> `get_string` and add support for encodings.

---
 io_buffer.c                 | 62 +++++++++++++++++++++++----------------------
 test/ruby/test_io_buffer.rb | 19 +++++++++++---
 2 files changed, 48 insertions(+), 33 deletions(-)

diff --git a/io_buffer.c b/io_buffer.c
index cc022a7c9f..075a9bfb78 100644
--- a/io_buffer.c
+++ b/io_buffer.c
@@ -717,35 +717,6 @@ rb_io_buffer_slice(VALUE self, VALUE _offset, VALUE _length)
     return instance;
 }
 
-VALUE
-rb_io_buffer_to_str(int argc, VALUE *argv, VALUE self)
-{
-    struct rb_io_buffer *data = NULL;
-    TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, data);
-
-    size_t offset = 0;
-    size_t length = data->size;
-
-    if (argc == 0) {
-        // Defaults.
-    }
-    else if (argc == 1) {
-        offset = NUM2SIZET(argv[0]);
-        length = data->size - offset;
-    }
-    else if (argc == 2) {
-        offset = NUM2SIZET(argv[0]);
-        length = NUM2SIZET(argv[1]);
-    }
-    else {
-        rb_error_arity(argc, 0, 2);
-    }
-
-    rb_io_buffer_validate(data, offset, length);
-
-    return rb_usascii_str_new((char*)data->base + offset, length);
-}
-
 void
 rb_io_buffer_get_mutable(VALUE self, void **base, size_t *size)
 {
@@ -1144,6 +1115,37 @@ io_buffer_set(VALUE self, VALUE type, VALUE _offset, VALUE value)
     return SIZET2NUM(offset);
 }
 
+static VALUE
+io_buffer_get_string(int argc, VALUE *argv, VALUE self)
+{
+    if (argc > 3) rb_error_arity(argc, 0, 3);
+
+    struct rb_io_buffer *data = NULL;
+    TypedData_Get_Struct(self, struct rb_io_buffer, &rb_io_buffer_type, data);
+
+    size_t offset = 0;
+    size_t length = data->size;
+    rb_encoding *encoding = NULL;
+
+    if (argc >= 1) {
+        offset = NUM2SIZET(argv[0]);
+    }
+
+    if (argc >= 2 && !RB_NIL_P(argv[1])) {
+        length = NUM2SIZET(argv[1]);
+    } else {
+        length = data->size - offset;
+    }
+
+    if (argc >= 3) {
+        encoding = rb_find_encoding(argv[2]);
+    }
+
+    rb_io_buffer_validate(data, offset, length);
+
+    return rb_enc_str_new((char*)data->base + offset, length, encoding);
+}
+
 void
 rb_io_buffer_clear(VALUE self, uint8_t value, size_t offset, size_t length)
 {
@@ -1278,7 +1280,6 @@ Init_IO_Buffer(void)
 
     // Manipulation:
     rb_define_method(rb_cIOBuffer, "slice", rb_io_buffer_slice, 2);
-    rb_define_method(rb_cIOBuffer, "to_str", rb_io_buffer_to_str, -1);
     rb_define_method(rb_cIOBuffer, "copy", io_buffer_copy, 2);
     rb_define_method(rb_cIOBuffer, "<=>", rb_io_buffer_compare, 1);
     rb_define_method(rb_cIOBuffer, "resize", io_buffer_resize, 1);
@@ -1298,4 +1299,5 @@ Init_IO_Buffer(void)
     // Data access:
     rb_define_method(rb_cIOBuffer, "get", io_buffer_get, 2);
     rb_define_method(rb_cIOBuffer, "set", io_buffer_set, 3);
+    rb_define_method(rb_cIOBuffer, "get_string", io_buffer_get_string, -1);
 }
diff --git a/test/ruby/test_io_buffer.rb b/test/ruby/test_io_buffer.rb
index 9c989f5cfe..45aab318f9 100644
--- a/test/ruby/test_io_buffer.rb
+++ b/test/ruby/test_io_buffer.rb
@@ -71,7 +71,7 @@ class TestIOBuffer < Test::Unit::TestCase
 
   def test_file_mapped
     buffer = File.open(__FILE__) {|file| IO::Buffer.map(file)}
-    assert_include buffer.to_str, "Hello World"
+    assert_include buffer.get_string, "Hello World"
   end
 
   def test_file_mapped_invalid
@@ -130,7 +130,7 @@ class TestIOBuffer < Test::Unit::TestCase
     buffer = IO::Buffer.new(1024)
     buffer.copy(message, 0)
     buffer.resize(2048)
-    assert_equal message, buffer.to_str(0, message.bytesize)
+    assert_equal message, buffer.get_string(0, message.bytesize)
   end
 
   def test_compare_same_size
@@ -157,7 +157,7 @@ class TestIOBuffer < Test::Unit::TestCase
     buffer = IO::Buffer.new(128)
     slice = buffer.slice(8, 32)
     slice.copy("Hello World", 0)
-    assert_equal("Hello World", buffer.to_str(8, 11))
+    assert_equal("Hello World", buffer.get_string(8, 11))
   end
 
   def test_slice_bounds
@@ -188,6 +188,19 @@ class TestIOBuffer < Test::Unit::TestCase
     assert_equal 128, buffer.size
   end
 
+  def test_get_string
+    message = "Hello World 🤓"
+
+    buffer = IO::Buffer.new(128)
+    buffer.copy(message, 0)
+
+    chunk = buffer.get_string(0, message.bytesize, Encoding::UTF_8)
+    assert_equal message, chunk
+    assert_equal Encoding::UTF_8, chunk.encoding
+
+    chunk = buffer.get_string(0, message.bytesize, Encoding::BINARY)
+    assert_equal Encoding::BINARY, chunk.encoding
+  end
   def test_invalidation
     input, output = IO.pipe
 
-- 
cgit v1.2.1