From c1743b9a2a21ffcf6948ab738ab4cf4b14a18f70 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Stanis=C5=82aw=20Barzowski?= <stanislaw.barzowski@gmail.com>
Date: Thu, 22 Nov 2018 18:37:20 +0100
Subject: [PATCH] Add std.encodeUTF8 and std.decodeUTF8

---
 builtins.go                 | 34 ++++++++++++++++++++++++++++++
 testdata/decodeUTF8.golden  |  7 +++++++
 testdata/decodeUTF8.jsonnet |  7 +++++++
 testdata/encodeUTF8.golden  | 42 +++++++++++++++++++++++++++++++++++++
 testdata/encodeUTF8.jsonnet |  7 +++++++
 5 files changed, 97 insertions(+)
 create mode 100644 testdata/decodeUTF8.golden
 create mode 100644 testdata/decodeUTF8.jsonnet
 create mode 100644 testdata/encodeUTF8.golden
 create mode 100644 testdata/encodeUTF8.jsonnet

diff --git a/builtins.go b/builtins.go
index 0bcb1d2..cff418a 100644
--- a/builtins.go
+++ b/builtins.go
@@ -575,6 +575,38 @@ func builtinMd5(i *interpreter, trace TraceElement, x value) (value, error) {
 	return makeValueString(hex.EncodeToString(hash[:])), nil
 }
 
+func builtinEncodeUTF8(i *interpreter, trace TraceElement, x value) (value, error) {
+	str, err := i.getString(x, trace)
+	if err != nil {
+		return nil, err
+	}
+	s := str.getString()
+	elems := make([]*cachedThunk, 0, len(s)) // it will be longer if characters fall outside of ASCII
+	for _, c := range []byte(s) {
+		elems = append(elems, readyThunk(makeValueNumber(float64(c))))
+	}
+	return makeValueArray(elems), nil
+}
+
+func builtinDecodeUTF8(i *interpreter, trace TraceElement, x value) (value, error) {
+	arr, err := i.getArray(x, trace)
+	if err != nil {
+		return nil, err
+	}
+	bs := make([]byte, len(arr.elements)) // it will be longer if characters fall outside of ASCII
+	for pos := range arr.elements {
+		v, err := i.evaluateInt(arr.elements[pos], trace)
+		if err != nil {
+			return nil, err
+		}
+		if v < 0 || v > 255 {
+			return nil, i.Error(fmt.Sprintf("Bytes must be integers in range [0, 255], got %d", v), trace)
+		}
+		bs[pos] = byte(v)
+	}
+	return makeValueString(string(bs)), nil
+}
+
 // Maximum allowed unicode codepoint
 // https://en.wikipedia.org/wiki/Unicode#Architecture_and_terminology
 const codepointMax = 0x10FFFF
@@ -1016,6 +1048,8 @@ var funcBuiltins = buildBuiltinMap([]builtin{
 	&unaryBuiltin{name: "md5", function: builtinMd5, parameters: ast.Identifiers{"x"}},
 	&ternaryBuiltin{name: "strReplace", function: builtinStrReplace, parameters: ast.Identifiers{"str", "from", "to"}},
 	&unaryBuiltin{name: "parseJson", function: builtinParseJSON, parameters: ast.Identifiers{"str"}},
+	&unaryBuiltin{name: "encodeUTF8", function: builtinEncodeUTF8, parameters: ast.Identifiers{"str"}},
+	&unaryBuiltin{name: "decodeUTF8", function: builtinDecodeUTF8, parameters: ast.Identifiers{"arr"}},
 	&unaryBuiltin{name: "native", function: builtinNative, parameters: ast.Identifiers{"x"}},
 
 	// internal
diff --git a/testdata/decodeUTF8.golden b/testdata/decodeUTF8.golden
new file mode 100644
index 0000000..02d2b03
--- /dev/null
+++ b/testdata/decodeUTF8.golden
@@ -0,0 +1,7 @@
+[
+   "",
+   "A",
+   "AAA",
+   "§",
+   "zażółć geślą jaźń"
+]
diff --git a/testdata/decodeUTF8.jsonnet b/testdata/decodeUTF8.jsonnet
new file mode 100644
index 0000000..fa34866
--- /dev/null
+++ b/testdata/decodeUTF8.jsonnet
@@ -0,0 +1,7 @@
+[
+    std.decodeUTF8([]),
+    std.decodeUTF8([65]),
+    std.decodeUTF8([65,65,65]),
+    std.decodeUTF8([194,167]),
+    std.decodeUTF8([122,97,197,188,195,179,197,130,196,135,32,103,101,197,155,108,196,133,32,106,97,197,186,197,132]),
+]
\ No newline at end of file
diff --git a/testdata/encodeUTF8.golden b/testdata/encodeUTF8.golden
new file mode 100644
index 0000000..33b169d
--- /dev/null
+++ b/testdata/encodeUTF8.golden
@@ -0,0 +1,42 @@
+[
+   [ ],
+   [
+      65
+   ],
+   [
+      65,
+      65,
+      65
+   ],
+   [
+      194,
+      167
+   ],
+   [
+      122,
+      97,
+      197,
+      188,
+      195,
+      179,
+      197,
+      130,
+      196,
+      135,
+      32,
+      103,
+      101,
+      197,
+      155,
+      108,
+      196,
+      133,
+      32,
+      106,
+      97,
+      197,
+      186,
+      197,
+      132
+   ]
+]
diff --git a/testdata/encodeUTF8.jsonnet b/testdata/encodeUTF8.jsonnet
new file mode 100644
index 0000000..f42a334
--- /dev/null
+++ b/testdata/encodeUTF8.jsonnet
@@ -0,0 +1,7 @@
+[
+    std.encodeUTF8(''),
+    std.encodeUTF8('A'),
+    std.encodeUTF8('AAA'),
+    std.encodeUTF8('§'),
+    std.encodeUTF8('zażółć geślą jaźń'),
+]
\ No newline at end of file