starlarkjson: a standard json module for Starlark

This is a sketch of a standard module for JSON encoding/decoding for Starlark.

It is intended to subsume, generalize, and eventually
replace the ill-conceived struct.to_json method.

See related issues:
bazelbuild/bazel#7896
https://buganizer.corp.google.com/issues/23962735
https://buganizer.corp.google.com/issues/70210417
bazelbuild/bazel#7879 (comment)
bazelbuild/bazel#5542

Change-Id: I297ffaee9349eedeeb52f5a88f40636a4095f997

starlark/eval_test.go

@@ -16,6 +16,8 @@ import (
 	"go.starlark.net/internal/chunkedfile"
 	"go.starlark.net/resolve"
 	"go.starlark.net/starlark"
+	"go.starlark.net/starlarkjson"
+	"go.starlark.net/starlarkstruct"
 	"go.starlark.net/starlarktest"
 	"go.starlark.net/syntax"
 )
@@ -112,6 +114,7 @@ func TestExecFile(t *testing.T) {
 		"testdata/float.star",
 		"testdata/function.star",
 		"testdata/int.star",
+		"testdata/json.star",
 		"testdata/list.star",
 		"testdata/misc.star",
 		"testdata/set.star",
@@ -125,6 +128,7 @@ func TestExecFile(t *testing.T) {
 			predeclared := starlark.StringDict{
 				"hasfields": starlark.NewBuiltin("hasfields", newHasFields),
 				"fibonacci": fib{},
+				"struct":    starlark.NewBuiltin("struct", starlarkstruct.Make),
 			}
 
 			setOptions(chunk.Source)
@@ -179,6 +183,9 @@ func load(thread *starlark.Thread, module string) (starlark.StringDict, error) {
 	if module == "assert.star" {
 		return starlarktest.LoadAssertModule()
 	}
+	if module == "json.star" {
+		return starlark.StringDict{"json": starlarkjson.Module}, nil
+	}
 
 	// TODO(adonovan): test load() using this execution path.
 	filename := filepath.Join(filepath.Dir(thread.Caller().Position().Filename()), module)

starlark/testdata/json.star

@@ -0,0 +1,68 @@
+# Tests of json module.
+# option:float
+
+load("assert.star", "assert")
+load("json.star", "json")
+
+assert.eq(dir(json), ["decode", "encode", "indent"])
+
+## json.encode
+
+assert.eq(json.encode(None), "null")
+assert.eq(json.encode(True), "true")
+assert.eq(json.encode(-123), "-123")
+assert.eq(json.encode(12345*12345*12345*12345*12345*12345), "3539537889086624823140625")
+assert.eq(json.encode(12.345e67), "1.2345e+68")
+assert.eq(json.encode("hello"), '"hello"')
+assert.eq(json.encode([1, 2, 3]), "[1,2,3]")
+assert.eq(json.encode((1, 2, 3)), "[1,2,3]")
+assert.eq(json.encode(range(3)), "[0,1,2]") # a built-in iterable
+assert.eq(json.encode(dict(x = 1, y = "two")), '{"x":1,"y":"two"}')
+assert.eq(json.encode(struct(x = 1, y = "two")), '{"x":1,"y":"two"}')  # a user-defined HasAttrs
+
+# errors
+assert.fails(lambda: json.encode(float("NaN")), "cannot encode non-finite float NaN")
+assert.fails(lambda: json.encode({1: "two"}), "dict has int key, want string")
+assert.fails(lambda: json.encode(len), "cannot encode builtin_function_or_method as JSON")
+assert.fails(lambda: json.encode(struct(x=[1, {"x": len}])), # nested failure
+             'in field .x: at list index 1: in dict key "x": cannot encode...')
+
+## json.decode
+
+assert.eq(json.decode("null"), None)
+assert.eq(json.decode("true"), True)
+assert.eq(json.decode("-123"), -123)
+assert.eq(json.decode("3539537889086624823140625"), float(3539537889086624823140625))
+assert.eq(json.decode('[]'), ())
+assert.eq(json.decode('[1]'), (1,))
+assert.eq(json.decode('[1,2,3]'), (1, 2, 3))
+assert.eq(json.decode('{"one": 1, "two": 2}'), dict(one=1, two=2))
+
+# Exercise JSON string coding by round-tripping a string with every 16-bit code point.
+def codec(x):
+  return json.decode(json.encode(x))
+codepoints = ''.join(['%c' %c for c in range(65536)])
+assert.eq(codec(codepoints), codepoints)
+
+## json.indent
+
+s = json.encode(dict(x = 1, y = ["one", "two"]))
+
+assert.eq(json.indent(s), '''{
+	"x": 1,
+	"y": [
+		"one",
+		"two"
+	]
+}''')
+
+assert.eq(json.indent(s, prefix='¶', indent='–––'), '''{
+¶–––"x": 1,
+¶–––"y": [
+¶––––––"one",
+¶––––––"two"
+¶–––]
+¶}''')
+
+assert.fails(lambda: json.indent("!@#$%^& this is not json"), 'invalid character')
+---

starlark/testdata/paths.star

@@ -20,6 +20,7 @@ with backslash separators or drive letters.
 """
 
 # This file is in the Bazel build language dialect of Starlark,
+
 # so declarations of 'fail' and 'struct' are required to make
 # it compile in the core language.
 def fail(msg):