tomltestgen: add toml-test unit test generation command (#610)

Tests are hidden behind a "testsuite" build tag for now since many tests
are failing.  Use `go test -tags testsuite` to activate.

Use `go generate` to regenerate toml_testgen_test.go.

Co-authored-by: Thomas Pelletier <thomas@pelletier.codes>

cmd/tomltestgen/main.go

@@ -0,0 +1,224 @@
+// tomltestgen retrieves a given version of the language-agnostic TOML test suite in
+// https://github.com/BurntSushi/toml-test and generates go-toml unit tests.
+//
+// Within the go-toml package, run `go generate`.  Otherwise, use:
+//
+//     go run github.com/pelletier/go-toml/cmd/tomltestgen -o toml_testgen_test.go
+package main
+
+import (
+	"archive/zip"
+	"bytes"
+	"flag"
+	"fmt"
+	"go/format"
+	"io"
+	"io/ioutil"
+	"log"
+	"net/http"
+	"os"
+	"regexp"
+	"strconv"
+	"strings"
+	"text/template"
+	"time"
+)
+
+type invalid struct {
+	Name  string
+	Input string
+}
+
+type valid struct {
+	Name    string
+	Input   string
+	JsonRef string
+}
+
+type testsCollection struct {
+	Ref       string
+	Timestamp string
+	Invalid   []invalid
+	Valid     []valid
+	Count     int
+}
+
+const srcTemplate = "// +build testsuite\n\n" +
+	"// Generated by tomltestgen for toml-test ref {{.Ref}} on {{.Timestamp}}\n" +
+	"package toml_test\n" +
+	" import (\n" +
+	"	\"testing\"\n" +
+	")\n" +
+
+	"{{range .Invalid}}\n" +
+	"func TestTOMLTest_Invalid_{{.Name}}(t *testing.T) {\n" +
+	"	input := {{.Input|gostr}}\n" +
+	"	testgenInvalid(t, input)\n" +
+	"}\n" +
+	"{{end}}\n" +
+	"\n" +
+	"{{range .Valid}}\n" +
+	"func TestTOMLTest_Valid_{{.Name}}(t *testing.T) {\n" +
+	"   input := {{.Input|gostr}}\n" +
+	"   jsonRef := {{.JsonRef|gostr}}\n" +
+	"   testgenValid(t, input, jsonRef)\n" +
+	"}\n" +
+	"{{end}}\n"
+
+func downloadTmpFile(url string) string {
+	log.Println("starting to download file from", url)
+	resp, err := http.Get(url)
+	if err != nil {
+		panic(err)
+	}
+	defer resp.Body.Close()
+
+	tmpfile, err := ioutil.TempFile("", "toml-test-*.zip")
+	if err != nil {
+		panic(err)
+	}
+	defer tmpfile.Close()
+
+	copiedLen, err := io.Copy(tmpfile, resp.Body)
+	if err != nil {
+		panic(err)
+	}
+	if resp.ContentLength > 0 && copiedLen != resp.ContentLength {
+		panic(fmt.Errorf("copied %d bytes, request body had %d", copiedLen, resp.ContentLength))
+	}
+	return tmpfile.Name()
+}
+
+func kebabToCamel(kebab string) string {
+	camel := ""
+	nextUpper := true
+	for _, c := range kebab {
+		if nextUpper {
+			camel += strings.ToUpper(string(c))
+			nextUpper = false
+		} else if c == '-' {
+			nextUpper = true
+		} else if c == '/' {
+			nextUpper = true
+			camel += "_"
+		} else {
+			camel += string(c)
+		}
+	}
+	return camel
+}
+
+func readFileFromZip(f *zip.File) string {
+	reader, err := f.Open()
+	if err != nil {
+		panic(err)
+	}
+	defer reader.Close()
+	bytes, err := ioutil.ReadAll(reader)
+	if err != nil {
+		panic(err)
+	}
+	return string(bytes)
+}
+
+func templateGoStr(input string) string {
+	return strconv.Quote(input)
+}
+
+var (
+	ref = flag.String("r", "master", "git reference")
+	out = flag.String("o", "", "output file")
+)
+
+func usage() {
+	_, _ = fmt.Fprintf(os.Stderr, "usage: tomltestgen [flags]\n")
+	flag.PrintDefaults()
+}
+
+func main() {
+	flag.Usage = usage
+	flag.Parse()
+
+	url := "https://codeload.github.com/BurntSushi/toml-test/zip/" + *ref
+	resultFile := downloadTmpFile(url)
+	defer os.Remove(resultFile)
+	log.Println("file written to", resultFile)
+
+	zipReader, err := zip.OpenReader(resultFile)
+	if err != nil {
+		panic(err)
+	}
+	defer zipReader.Close()
+
+	collection := testsCollection{
+		Ref:       *ref,
+		Timestamp: time.Now().Format(time.RFC3339),
+	}
+
+	zipFilesMap := map[string]*zip.File{}
+
+	for _, f := range zipReader.File {
+		zipFilesMap[f.Name] = f
+	}
+
+	testFileRegexp := regexp.MustCompile(`([^/]+/tests/(valid|invalid)/(.+))\.(toml)`)
+	for _, f := range zipReader.File {
+		groups := testFileRegexp.FindStringSubmatch(f.Name)
+		if len(groups) > 0 {
+			name := kebabToCamel(groups[3])
+			testType := groups[2]
+
+			log.Printf("> [%s] %s\n", testType, name)
+
+			tomlContent := readFileFromZip(f)
+
+			switch testType {
+			case "invalid":
+				collection.Invalid = append(collection.Invalid, invalid{
+					Name:  name,
+					Input: tomlContent,
+				})
+				collection.Count++
+			case "valid":
+				baseFilePath := groups[1]
+				jsonFilePath := baseFilePath + ".json"
+				jsonContent := readFileFromZip(zipFilesMap[jsonFilePath])
+
+				collection.Valid = append(collection.Valid, valid{
+					Name:    name,
+					Input:   tomlContent,
+					JsonRef: jsonContent,
+				})
+				collection.Count++
+			default:
+				panic(fmt.Sprintf("unknown test type: %s", testType))
+			}
+		}
+	}
+
+	log.Printf("Collected %d tests from toml-test\n", collection.Count)
+
+	funcMap := template.FuncMap{
+		"gostr": templateGoStr,
+	}
+	t := template.Must(template.New("src").Funcs(funcMap).Parse(srcTemplate))
+	buf := new(bytes.Buffer)
+	err = t.Execute(buf, collection)
+	if err != nil {
+		panic(err)
+	}
+	outputBytes, err := format.Source(buf.Bytes())
+	if err != nil {
+		panic(err)
+	}
+
+	if *out == "" {
+		fmt.Println(string(outputBytes))
+		return
+	}
+
+	err = os.WriteFile(*out, outputBytes, 0644)
+	if err != nil {
+		panic(err)
+	}
+}

testsuite/add.go

@@ -0,0 +1,74 @@
+package testsuite
+
+import (
+	"fmt"
+	"math"
+	"time"
+
+	"github.com/pelletier/go-toml/v2"
+)
+
+// addTag adds JSON tags to a data structure as expected by toml-test.
+func addTag(key string, tomlData interface{}) interface{} {
+	// Switch on the data type.
+	switch orig := tomlData.(type) {
+	default:
+		//return map[string]interface{}{}
+		panic(fmt.Sprintf("Unknown type: %T", tomlData))
+
+	// A table: we don't need to add any tags, just recurse for every table
+	// entry.
+	case map[string]interface{}:
+		typed := make(map[string]interface{}, len(orig))
+		for k, v := range orig {
+			typed[k] = addTag(k, v)
+		}
+		return typed
+
+	// An array: we don't need to add any tags, just recurse for every table
+	// entry.
+	case []map[string]interface{}:
+		typed := make([]map[string]interface{}, len(orig))
+		for i, v := range orig {
+			typed[i] = addTag("", v).(map[string]interface{})
+		}
+		return typed
+	case []interface{}:
+		typed := make([]interface{}, len(orig))
+		for i, v := range orig {
+			typed[i] = addTag("", v)
+		}
+		return typed
+
+	// Datetime: tag as datetime.
+	case toml.LocalTime:
+		return tag("time-local", orig.String())
+	case toml.LocalDate:
+		return tag("date-local", orig.String())
+	case toml.LocalDateTime:
+		return tag("datetime-local", orig.String())
+	case time.Time:
+		return tag("datetime", orig.Format("2006-01-02T15:04:05.999999999Z07:00"))
+
+	// Tag primitive values: bool, string, int, and float64.
+	case bool:
+		return tag("bool", fmt.Sprintf("%v", orig))
+	case string:
+		return tag("string", orig)
+	case int64:
+		return tag("integer", fmt.Sprintf("%d", orig))
+	case float64:
+		// Special case for nan since NaN == NaN is false.
+		if math.IsNaN(orig) {
+			return tag("float", "nan")
+		}
+		return tag("float", fmt.Sprintf("%v", orig))
+	}
+}
+
+func tag(typeName string, data interface{}) map[string]interface{} {
+	return map[string]interface{}{
+		"type":  typeName,
+		"value": data,
+	}
+}

testsuite/parser.go

@@ -0,0 +1,69 @@
+package testsuite
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+
+	"github.com/pelletier/go-toml/v2"
+)
+
+type parser struct{}
+
+func (p parser) Decode(input string) (output string, outputIsError bool, retErr error) {
+	defer func() {
+		if r := recover(); r != nil {
+			switch rr := r.(type) {
+			case error:
+				retErr = rr
+			default:
+				retErr = fmt.Errorf("%s", rr)
+			}
+		}
+	}()
+
+	var v interface{}
+
+	if err := toml.Unmarshal([]byte(input), &v); err != nil {
+		return err.Error(), true, nil
+	}
+
+	j, err := json.MarshalIndent(addTag("", v), "", "  ")
+	if err != nil {
+		return "", false, retErr
+	}
+
+	return string(j), false, retErr
+}
+
+func (p parser) Encode(input string) (output string, outputIsError bool, retErr error) {
+	defer func() {
+		if r := recover(); r != nil {
+			switch rr := r.(type) {
+			case error:
+				retErr = rr
+			default:
+				retErr = fmt.Errorf("%s", rr)
+			}
+		}
+	}()
+
+	var tmp interface{}
+	err := json.Unmarshal([]byte(input), &tmp)
+	if err != nil {
+		return "", false, err
+	}
+
+	rm, err := rmTag(tmp)
+	if err != nil {
+		return err.Error(), true, retErr
+	}
+
+	buf := new(bytes.Buffer)
+	err = toml.NewEncoder(buf).Encode(rm)
+	if err != nil {
+		return err.Error(), true, retErr
+	}
+
+	return buf.String(), false, retErr
+}

testsuite/rm.go

@@ -0,0 +1,110 @@
+package testsuite
+
+import (
+	"fmt"
+	"strconv"
+	"time"
+)
+
+// Remove JSON tags to a data structure as returned by toml-test.
+func rmTag(typedJson interface{}) (interface{}, error) {
+	// Check if key is in the table m.
+	in := func(key string, m map[string]interface{}) bool {
+		_, ok := m[key]
+		return ok
+	}
+
+	// Switch on the data type.
+	switch v := typedJson.(type) {
+
+	// Object: this can either be a TOML table or a primitive with tags.
+	case map[string]interface{}:
+		// This value represents a primitive: remove the tags and return just
+		// the primitive value.
+		if len(v) == 2 && in("type", v) && in("value", v) {
+			ut, err := untag(v)
+			if err != nil {
+				return ut, fmt.Errorf("tag.Remove: %w", err)
+			}
+			return ut, nil
+		}
+
+		// Table: remove tags on all children.
+		m := make(map[string]interface{}, len(v))
+		for k, v2 := range v {
+			var err error
+			m[k], err = rmTag(v2)
+			if err != nil {
+				return nil, err
+			}
+		}
+		return m, nil
+
+	// Array: remove tags from all itenm.
+	case []interface{}:
+		a := make([]interface{}, len(v))
+		for i := range v {
+			var err error
+			a[i], err = rmTag(v[i])
+			if err != nil {
+				return nil, err
+			}
+		}
+		return a, nil
+	}
+
+	// The top level must be an object or array.
+	return nil, fmt.Errorf("unrecognized JSON format '%T'", typedJson)
+}
+
+// Return a primitive: read the "type" and convert the "value" to that.
+func untag(typed map[string]interface{}) (interface{}, error) {
+	t := typed["type"].(string)
+	v := typed["value"].(string)
+	switch t {
+	case "string":
+		return v, nil
+	case "integer":
+		n, err := strconv.ParseInt(v, 10, 64)
+		if err != nil {
+			return nil, fmt.Errorf("untag: %w", err)
+		}
+		return n, nil
+	case "float":
+		f, err := strconv.ParseFloat(v, 64)
+		if err != nil {
+			return nil, fmt.Errorf("untag: %w", err)
+		}
+		return f, nil
+	case "datetime":
+		return parseTime(v, "2006-01-02T15:04:05.999999999Z07:00", false)
+	case "datetime-local":
+		return parseTime(v, "2006-01-02T15:04:05.999999999", true)
+	case "date-local":
+		return parseTime(v, "2006-01-02", true)
+	case "time-local":
+		return parseTime(v, "15:04:05.999999999", true)
+	case "bool":
+		switch v {
+		case "true":
+			return true, nil
+		case "false":
+			return false, nil
+		}
+		return nil, fmt.Errorf("untag: could not parse %q as a boolean", v)
+	}
+
+	return nil, fmt.Errorf("untag: unrecognized tag type %q", t)
+}
+
+func parseTime(v, format string, local bool) (t time.Time, err error) {
+	if local {
+		t, err = time.ParseInLocation(format, v, time.Local)
+	} else {
+		t, err = time.Parse(format, v)
+	}
+	if err != nil {
+		return time.Time{}, fmt.Errorf("Could not parse %q as a datetime: %w", v, err)
+	}
+	return t, nil
+}

testsuite/testsuite.go

@@ -0,0 +1,48 @@
+// Package testsuite provides helper functions for interoperating with the
+// language-agnostic TOML test suite at github.com/BurntSushi/toml-test.
+package testsuite
+
+import (
+	"encoding/json"
+	"log"
+	"os"
+
+	"github.com/pelletier/go-toml/v2"
+)
+
+// Marshal is a helpfer function for calling toml.Marshal
+//
+// Only needed to avoid package import loops.
+func Marshal(v interface{}) ([]byte, error) {
+	return toml.Marshal(v)
+}
+
+// Unmarshal is a helper function for calling toml.Unmarshal.
+//
+// Only needed to avoid package import loops.
+func Unmarshal(data []byte, v interface{}) error {
+	return toml.Unmarshal(data, v)
+}
+
+// ValueToTaggedJSON takes a data structure and returns the tagged JSON
+// representation.
+func ValueToTaggedJSON(doc interface{}) ([]byte, error) {
+	return json.MarshalIndent(addTag("", doc), "", "  ")
+}
+
+// DecodeStdin is a helper function for the toml-test binary interface.  TOML input
+// is read from STDIN and a resulting tagged JSON representation is written to
+// STDOUT.
+func DecodeStdin() {
+	var decoded map[string]interface{}
+
+	if err := toml.NewDecoder(os.Stdin).Decode(&decoded); err != nil {
+		log.Fatalf("Error decoding TOML: %s", err)
+	}
+
+	j := json.NewEncoder(os.Stdout)
+	j.SetIndent("", "  ")
+	if err := j.Encode(addTag("", decoded)); err != nil {
+		log.Fatalf("Error encoding JSON: %s", err)
+	}
+}

toml_testgen_support_test.go

@@ -1,14 +1,13 @@
+//go:generate go run ./cmd/tomltestgen/main.go -o toml_testgen_test.go
+
 // This is a support file for toml_testgen_test.go
 package toml_test
 
 import (
 	"encoding/json"
-	"fmt"
-	"strconv"
 	"testing"
-	"time"
 
-	"github.com/pelletier/go-toml/v2"
+	"github.com/pelletier/go-toml/v2/testsuite"
 	"github.com/stretchr/testify/require"
 )
 
@@ -17,10 +16,14 @@ func testgenInvalid(t *testing.T, input string) {
 	t.Logf("Input TOML:\n%s", input)
 
 	doc := map[string]interface{}{}
-	err := toml.Unmarshal([]byte(input), &doc)
+	err := testsuite.Unmarshal([]byte(input), &doc)
 
 	if err == nil {
-		t.Log(json.Marshal(doc))
+		out, err := json.Marshal(doc)
+		if err != nil {
+			panic("could not marshal map to json")
+		}
+		t.Log("JSON output from unmarshal:", string(out))
 		t.Fatalf("test did not fail")
 	}
 }
@@ -29,124 +32,14 @@ func testgenValid(t *testing.T, input string, jsonRef string) {
 	t.Helper()
 	t.Logf("Input TOML:\n%s", input)
 
-	doc := map[string]interface{}{}
+	// TODO: change this to interface{}
+	var doc map[string]interface{}
 
-	err := toml.Unmarshal([]byte(input), &doc)
+	err := testsuite.Unmarshal([]byte(input), &doc)
 	if err != nil {
 		t.Fatalf("failed parsing toml: %s", err)
 	}
-
+	j, err := testsuite.ValueToTaggedJSON(doc)
-	refDoc := testgenBuildRefDoc(jsonRef)
-
-	require.Equal(t, refDoc, doc)
-
-	out, err := toml.Marshal(doc)
 	require.NoError(t, err)
-
+	require.Equal(t, jsonRef, string(j)+"\n")
-	doc2 := map[string]interface{}{}
-	err = toml.Unmarshal(out, &doc2)
-	require.NoError(t, err)
-
-	require.Equal(t, refDoc, doc2)
-}
-
-func testgenBuildRefDoc(jsonRef string) map[string]interface{} {
-	descTree := map[string]interface{}{}
-
-	err := json.Unmarshal([]byte(jsonRef), &descTree)
-	if err != nil {
-		panic(fmt.Sprintf("reference doc should be valid JSON: %s", err))
-	}
-
-	doc := testGenTranslateDesc(descTree)
-	if doc == nil {
-		return map[string]interface{}{}
-	}
-
-	return doc.(map[string]interface{})
-}
-
-//nolint:funlen,gocognit,cyclop
-func testGenTranslateDesc(input interface{}) interface{} {
-	a, ok := input.([]interface{})
-	if ok {
-		xs := make([]interface{}, len(a))
-		for i, v := range a {
-			xs[i] = testGenTranslateDesc(v)
-		}
-
-		return xs
-	}
-
-	d, ok := input.(map[string]interface{})
-	if !ok {
-		panic(fmt.Sprintf("input should be valid map[string]: %v", input))
-	}
-
-	var (
-		dtype  string
-		dvalue interface{}
-	)
-
-	//nolint:nestif
-	if len(d) == 2 {
-		dtypeiface, ok := d["type"]
-		if ok {
-			dvalue, ok = d["value"]
-			if ok {
-				dtype = dtypeiface.(string)
-
-				switch dtype {
-				case "string":
-					return dvalue.(string)
-				case "float":
-					v, err := strconv.ParseFloat(dvalue.(string), 64)
-					if err != nil {
-						panic(fmt.Sprintf("invalid float '%s': %s", dvalue, err))
-					}
-
-					return v
-				case "integer":
-					v, err := strconv.ParseInt(dvalue.(string), 10, 64)
-					if err != nil {
-						panic(fmt.Sprintf("invalid int '%s': %s", dvalue, err))
-					}
-
-					return v
-				case "bool":
-					return dvalue.(string) == "true"
-				case "datetime":
-					dt, err := time.Parse("2006-01-02T15:04:05Z", dvalue.(string))
-					if err != nil {
-						panic(fmt.Sprintf("invalid datetime '%s': %s", dvalue, err))
-					}
-
-					return dt
-				case "array":
-					if dvalue == nil {
-						return nil
-					}
-
-					a := dvalue.([]interface{})
-
-					xs := make([]interface{}, len(a))
-
-					for i, v := range a {
-						xs[i] = testGenTranslateDesc(v)
-					}
-
-					return xs
-				}
-
-				panic(fmt.Sprintf("unknown type: %s", dtype))
-			}
-		}
-	}
-
-	dest := map[string]interface{}{}
-	for k, v := range d {
-		dest[k] = testGenTranslateDesc(v)
-	}
-
-	return dest
 }