fix(decode): keep the targeted parse on the tree path's contract

Assisted-by: GLM 5.3
2026-09-22 21:15:00 +02:00
parent f7e427ae3f
commit b45f4d65da
2 changed files with 719 additions and 191 deletions
@@ -4,9 +4,12 @@
 package interpres

 import (
+	"errors"
+	"maps"
 	"net"
 	"reflect"
 	"strings"
+	"sync/atomic"
 	"testing"
 	"time"
 )
@@ -522,3 +525,350 @@ func TestTargetedMapTableShapes(t *testing.T) {
 		})
 	}
 }
+
+// TestTargetedNestedMapDescents pins the descents into a map of maps that
+// meet entries the document built earlier: a dotted key twice through the
+// same sub-table, a header into a dotted-built sub-table, and a typed array
+// under a map key. Each shape once panicked on a reflect Elem of a map.
+func TestTargetedNestedMapDescents(t *testing.T) {
+	t.Run("dotted key through one sub-table twice", func(t *testing.T) {
+		var cfg struct {
+			M map[string]map[string]any `toml:"m"`
+		}
+		err := Unmarshal([]byte("m.a.b = 1\nm.a.c = 2\n"), &cfg)
+		if err != nil {
+			t.Fatalf("unmarshal: %v", err)
+		}
+		if cfg.M["a"]["b"] != int64(1) || cfg.M["a"]["c"] != int64(2) {
+			t.Errorf("m = %#v", cfg.M)
+		}
+	})
+	t.Run("header under a dotted-built sub-table", func(t *testing.T) {
+		var cfg struct {
+			M map[string]map[string]any `toml:"m"`
+		}
+		err := Unmarshal([]byte("m.a.b = 1\n[m.a.deep]\nx = 2\n"), &cfg)
+		if err != nil {
+			t.Fatalf("unmarshal: %v", err)
+		}
+		if cfg.M["a"]["b"] != int64(1) || cfg.M["a"]["deep"].(map[string]any)["x"] != int64(2) {
+			t.Errorf("m = %#v", cfg.M)
+		}
+	})
+	t.Run("typed array under a map key", func(t *testing.T) {
+		var cfg struct {
+			M map[string][]map[string]any `toml:"m"`
+		}
+		err := Unmarshal([]byte("[[m.arr]]\nx = 1\n\n[[m.arr]]\ny = 2\n"), &cfg)
+		if err != nil {
+			t.Fatalf("unmarshal: %v", err)
+		}
+		if len(cfg.M["arr"]) != 2 || cfg.M["arr"][1]["y"] != int64(2) {
+			t.Errorf("m = %#v", cfg.M)
+		}
+	})
+}
+
+// TestTargetedPointerElementSlice pins that an array of tables over a slice
+// of pointer elements fills the pointed-to structs.
+func TestTargetedPointerElementSlice(t *testing.T) {
+	type item struct {
+		N int `toml:"n"`
+	}
+	var cfg struct {
+		Items []*item `toml:"items"`
+	}
+	err := Unmarshal([]byte("[[items]]\nn = 1\n\n[[items]]\nn = 2\n"), &cfg)
+	if err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if len(cfg.Items) != 2 || cfg.Items[0] == nil || cfg.Items[1].N != 2 {
+		t.Errorf("items = %#v", cfg.Items)
+	}
+}
+
+// TestTargetedArrayScopeResets pins that a new element of an array of tables
+// starts a fresh definition scope, the contract the changelog documents.
+func TestTargetedArrayScopeResets(t *testing.T) {
+	doc := "[[a]]\nb.c = 1\n\n[[a]]\n\n[a.b]\nx = 1\n"
+	var ref, tgt targetCfg
+	refErr := treeDecodeInto([]byte(doc), &ref)
+	if refErr != nil {
+		t.Fatalf("tree decode: %v", refErr)
+	}
+	if err := Unmarshal([]byte(doc), &tgt); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if !reflect.DeepEqual(ref, tgt) {
+		t.Errorf("targeted = %#v, tree = %#v", tgt, ref)
+	}
+}
+
+// TestTargetedUnknownArrayElements pins that every element of an unknown
+// array of tables is a fresh namespace, and a sub-table header reaches the
+// last element the way the tree parser's does.
+func TestTargetedUnknownArrayElements(t *testing.T) {
+	doc := "[[zz]]\nk = 1\n\n[[zz]]\nk = 2\n\n[zz.sub]\nx = 3\n"
+	var ref, tgt targetCfg
+	refErr := treeDecodeInto([]byte(doc), &ref)
+	tgtErr := Unmarshal([]byte(doc), &tgt)
+	if (refErr == nil) != (tgtErr == nil) {
+		t.Fatalf("error presence disagrees: tree %v, targeted %v", refErr, tgtErr)
+	}
+	if refErr != nil {
+		return
+	}
+	if !reflect.DeepEqual(ref, tgt) {
+		t.Errorf("targeted = %#v, tree = %#v", tgt, ref)
+	}
+	// A dotted key may not enter the array: the tree's own rule.
+	var dotted targetCfg
+	dErr := Unmarshal([]byte("[[zz]]\nk = 1\nzz.x = 2\n"), &dotted)
+	refDotted := treeDecodeInto([]byte("[[zz]]\nk = 1\nzz.x = 2\n"), &dotted)
+	if (dErr == nil) != (refDotted == nil) {
+		t.Errorf("dotted into an array: targeted %v, tree %v", dErr, refDotted)
+	}
+}
+
+// TestTargetedFixedArrayUnderFill pins that a fixed-size array the document
+// under-fills is the length mismatch the tree decode raises, with the
+// field's path.
+func TestTargetedFixedArrayUnderFill(t *testing.T) {
+	type item struct {
+		N int `toml:"n"`
+	}
+	var cfg struct {
+		Items [2]item `toml:"items"`
+	}
+	err := Unmarshal([]byte("[[items]]\nn = 1\n"), &cfg)
+	if err == nil {
+		t.Fatal("unmarshal accepted an under-filled array")
+	}
+	want := `interpres: items: cannot assign 1 elements to [2]interpres.item`
+	if err.Error() != want {
+		t.Errorf("err = %v\nwant %q", err, want)
+	}
+}
+
+// TestTargetedPrefilledSliceReplaced pins that a prefilled slice is replaced
+// by the document's elements on both paths, not appended to.
+func TestTargetedPrefilledSliceReplaced(t *testing.T) {
+	type item struct {
+		N int `toml:"n"`
+	}
+	doc := []byte("[[items]]\nn = 1\n")
+	var ref struct {
+		Items []item `toml:"items"`
+	}
+	ref.Items = []item{{N: 9}}
+	if err := treeDecodeInto(doc, &ref); err != nil {
+		t.Fatalf("tree decode: %v", err)
+	}
+	var tgt struct {
+		Items []item `toml:"items"`
+	}
+	tgt.Items = []item{{N: 9}}
+	if err := Unmarshal(doc, &tgt); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if !reflect.DeepEqual(ref, tgt) {
+		t.Errorf("targeted = %#v, tree = %#v", tgt, ref)
+	}
+	if len(tgt.Items) != 1 || tgt.Items[0].N != 1 {
+		t.Errorf("items = %#v, want the prefilled element replaced", tgt.Items)
+	}
+}
+
+// TestTargetedHeaderOverValueArrayKeepsCase pins that a value array assigned
+// under a differently cased key than the field's name still blocks the
+// array-of-tables header over it, the tree parse error.
+func TestTargetedHeaderOverValueArrayKeepsCase(t *testing.T) {
+	var cfg struct {
+		Arr []targetNested `toml:"arr"`
+	}
+	err := Unmarshal([]byte("Arr = [{x = 1}]\n[[Arr]]\nx = 2\n"), &cfg)
+	if err == nil || err.Error() != `interpres: line 2: key "Arr" is not an array of tables` {
+		t.Errorf("err = %v, want the parse error over the assigned field", err)
+	}
+}
+
+// TestTargetedDottedInlineFreezePath pins that an inline table assigned by a
+// dotted key freezes the whole path the statement wrote: a later header
+// under that path is the extension error, and a key outside it stays free.
+func TestTargetedDottedInlineFreezePath(t *testing.T) {
+	var cfg targetCfg
+	err := Unmarshal([]byte("m.a.b = {x = 1}\nb.y = 2\n"), &cfg)
+	if err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	err = Unmarshal([]byte("m.a.b = {x = 1}\n[m.a.b]\ny = 2\n"), &cfg)
+	want := `interpres: line 2: cannot extend inline table "m.a.b"`
+	if err == nil || err.Error() != want {
+		t.Errorf("err = %v\nwant %q", err, want)
+	}
+}
+
+// TestTargetedStrictThroughDottedKeys pins that strict and required findings
+// survive the transient tables a dotted descent builds.
+func TestTargetedStrictThroughDottedKeys(t *testing.T) {
+	var cfg targetCfg
+	err := Unmarshal([]byte("tab.zz = 1\n"), &cfg, RejectUnknownFields(true))
+	if err == nil || !strings.Contains(err.Error(), `unknown field "zz"`) {
+		t.Errorf("err = %v, want the strict failure through the dotted key", err)
+	}
+	if err == nil || !strings.HasPrefix(err.Error(), "interpres: tab:") {
+		t.Errorf("err = %v, want the path through the dotted key", err)
+	}
+}
+
+// TestTargetedRequiredThroughDottedKeys pins that a required tag is honoured
+// when the table is reached only through dotted keys.
+func TestTargetedRequiredThroughDottedKeys(t *testing.T) {
+	type nested struct {
+		X int `toml:"x,required"`
+		Y int `toml:"y"`
+	}
+	var cfg struct {
+		Tab nested `toml:"tab"`
+	}
+	err := Unmarshal([]byte("tab.y = 1\n"), &cfg)
+	if err == nil || !strings.Contains(err.Error(), `missing required key "x"`) {
+		t.Errorf("err = %v, want the missing required key through the dotted key", err)
+	}
+}
+
+// TestTargetedOrderedMapSliceFallsBack pins that a slice of OrderedMap
+// elements takes the tree path, whose fill keeps the written order.
+func TestTargetedOrderedMapSliceFallsBack(t *testing.T) {
+	var cfg struct {
+		Items []OrderedMap `toml:"items"`
+	}
+	err := Unmarshal([]byte("[[items]]\nk = \"v\"\n"), &cfg)
+	if err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if len(cfg.Items) != 1 || cfg.Items[0].Keys()[0] != "k" {
+		t.Errorf("items = %#v, want the element filled in written order", cfg.Items)
+	}
+}
+
+// hookMap is a named map type whose decode hook counts its calls.
+type hookMap map[string]any
+
+var hookMapCalls atomic.Int32
+
+func (h *hookMap) UnmarshalTOML(data any) error {
+	hookMapCalls.Add(1)
+	m, _ := data.(map[string]any)
+	if *h == nil {
+		*h = hookMap{}
+	}
+	maps.Copy((*h), m)
+	return nil
+}
+
+// TestTargetedMapFieldHookGetsWholeTable pins that a named map field with a
+// decode hook receives the whole parsed table, even in its header form.
+func TestTargetedMapFieldHookGetsWholeTable(t *testing.T) {
+	type cfg struct {
+		M hookMap `toml:"m"`
+	}
+	var c cfg
+	hookMapCalls.Store(0)
+	err := Unmarshal([]byte("[m]\na = 1\nb = 2\n"), &c)
+	if err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if hookMapCalls.Load() != 1 {
+		t.Errorf("hook calls = %d, want exactly one with the whole table", hookMapCalls.Load())
+	}
+	if c.M["a"] != int64(1) || c.M["b"] != int64(2) {
+		t.Errorf("m = %#v", c.M)
+	}
+}
+
+// errHook fails every decode with a fixed error and counts its calls.
+type errHook struct{ calls *int }
+
+func (e *errHook) UnmarshalTOML(any) error {
+	if e.calls != nil {
+		*e.calls++
+	}
+	return errors.New("boom")
+}
+
+// TestTargetedHookErrorRunsOnce pins that a failing hook's error is the
+// tree path's own, wrapped with the key, and that the hook is not run a
+// second time by a fallback.
+func TestTargetedHookErrorRunsOnce(t *testing.T) {
+	calls := 0
+	cfg := struct {
+		F errHook `toml:"f"`
+	}{F: errHook{calls: &calls}}
+	err := Unmarshal([]byte("f = 1\n"), &cfg)
+	if err == nil || err.Error() != "interpres: f: unmarshal: boom" {
+		t.Errorf("err = %v, want the wrapped hook failure", err)
+	}
+	if calls != 1 {
+		t.Errorf("hook calls = %d, want one", calls)
+	}
+}
+
+// TestTargetedUnknownBeforeRequired pins the report order the tree decode
+// produces: an unknown key wins over a missing required one.
+func TestTargetedUnknownBeforeRequired(t *testing.T) {
+	type inner struct {
+		X int `toml:"x,required"`
+	}
+	var cfg struct {
+		Tab inner `toml:"tab"`
+	}
+	err := Unmarshal([]byte("[tab]\nzz = 1\n"), &cfg, RejectUnknownFields(true))
+	if err == nil || !strings.Contains(err.Error(), `unknown field "zz"`) {
+		t.Errorf("err = %v, want the unknown key reported before the required one", err)
+	}
+}
+
+// TestTargetedStrictPathStableAcrossHeaders pins that the path a strict
+// finding wraps does not alias the parser's key buffer: the table that owns
+// the unknown key keeps its name after a later header.
+func TestTargetedStrictPathStableAcrossHeaders(t *testing.T) {
+	var cfg targetCfg
+	err := Unmarshal([]byte("[tab]\nzz = 1\n\n[lims]\nx = 1\n"), &cfg, RejectUnknownFields(true))
+	if err == nil || !strings.HasPrefix(err.Error(), "interpres: tab:") {
+		t.Errorf("err = %v, want the finding on tab, not the later header", err)
+	}
+}
+
+// TestTargetedPrefilledMapFieldMergesUnderHeader pins that a prefilled map
+// field merges the document's header-form table into it on both paths, the
+// rule the root map has always followed.
+func TestTargetedPrefilledMapFieldMergesUnderHeader(t *testing.T) {
+	doc := []byte("[lims]\nnew = 3\n")
+	var ref, tgt targetCfg
+	ref.Lims = map[string]any{"keep": "yes"}
+	if err := treeDecodeInto(doc, &ref); err != nil {
+		t.Fatalf("tree decode: %v", err)
+	}
+	tgt.Lims = map[string]any{"keep": "yes"}
+	if err := Unmarshal(doc, &tgt); err != nil {
+		t.Fatalf("unmarshal: %v", err)
+	}
+	if !reflect.DeepEqual(ref, tgt) {
+		t.Errorf("targeted = %#v, tree = %#v", tgt, ref)
+	}
+	if tgt.Lims["keep"] != "yes" || tgt.Lims["new"] != int64(3) {
+		t.Errorf("lims = %#v, want the merge", tgt.Lims)
+	}
+}
+
+// TestTargetedNumberTokenValidatesUTF8 pins that the token route the
+// targeted parse takes reports invalid UTF-8 with the scanner's own message
+// and position.
+func TestTargetedNumberTokenValidatesUTF8(t *testing.T) {
+	var cfg targetCfg
+	err := Unmarshal([]byte("num = 12\xff\n"), &cfg)
+	if err == nil || !strings.Contains(err.Error(), "invalid UTF-8 in value at byte offset 8") {
+		t.Errorf("err = %v, want the UTF-8 complaint on the invalid byte", err)
+	}
+}