docs-i18n: add behavior baseline fixtures

scripts/docs-i18n/behavior_baseline_test.go

@@ -0,0 +1,206 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+type behaviorReplacePair struct {
+	From string `json:"from"`
+	To   string `json:"to"`
+}
+
+type behaviorRule struct {
+	Method       string                `json:"method"`
+	MatchAll     []string              `json:"match_all"`
+	ResponseFile string                `json:"response_file,omitempty"`
+	ReplacePairs []behaviorReplacePair `json:"replace_pairs,omitempty"`
+}
+
+type behaviorFixture struct {
+	Name                      string         `json:"name"`
+	Mode                      string         `json:"mode"`
+	RelPath                   string         `json:"rel_path"`
+	SourceFile                string         `json:"source_file"`
+	ExpectedFile              string         `json:"expected_file,omitempty"`
+	ExpectedErrorContains     string         `json:"expected_error_contains,omitempty"`
+	ExpectedOutputContains    []string       `json:"expected_output_contains,omitempty"`
+	ExpectedOutputNotContains []string       `json:"expected_output_not_contains,omitempty"`
+	Rules                     []behaviorRule `json:"rules"`
+}
+
+type behaviorFixtureTranslator struct {
+	t     *testing.T
+	dir   string
+	rules []behaviorRule
+}
+
+func (t *behaviorFixtureTranslator) Translate(_ context.Context, text, _, _ string) (string, error) {
+	return t.run("masked", text), nil
+}
+
+func (t *behaviorFixtureTranslator) TranslateRaw(_ context.Context, text, _, _ string) (string, error) {
+	return t.run("raw", text), nil
+}
+
+func (t *behaviorFixtureTranslator) Close() {}
+
+func (t *behaviorFixtureTranslator) run(method, text string) string {
+	t.t.Helper()
+	for _, rule := range t.rules {
+		if rule.Method != method {
+			continue
+		}
+		if !matchesAll(text, rule.MatchAll) {
+			continue
+		}
+		switch {
+		case rule.ResponseFile != "":
+			return readFixtureText(t.t, filepath.Join(t.dir, rule.ResponseFile))
+		case len(rule.ReplacePairs) > 0:
+			out := text
+			for _, pair := range rule.ReplacePairs {
+				out = strings.ReplaceAll(out, pair.From, pair.To)
+			}
+			return out
+		default:
+			return text
+		}
+	}
+	return text
+}
+
+func matchesAll(text string, fragments []string) bool {
+	for _, fragment := range fragments {
+		if !strings.Contains(text, fragment) {
+			return false
+		}
+	}
+	return true
+}
+
+func TestDocsI18nBehaviorBaselines(t *testing.T) {
+	t.Parallel()
+
+	root := filepath.Join("testdata", "behavior")
+	entries, err := os.ReadDir(root)
+	if err != nil {
+		t.Fatalf("ReadDir(%q): %v", root, err)
+	}
+
+	found := false
+	for _, entry := range entries {
+		if !entry.IsDir() {
+			continue
+		}
+		found = true
+		dir := filepath.Join(root, entry.Name())
+		fixture := loadBehaviorFixture(t, dir)
+		name := fixture.Name
+		if name == "" {
+			name = entry.Name()
+		}
+		t.Run(name, func(t *testing.T) {
+			t.Parallel()
+			runBehaviorFixture(t, dir, fixture)
+		})
+	}
+
+	if !found {
+		t.Fatalf("no behavior fixtures found under %s", root)
+	}
+}
+
+func loadBehaviorFixture(t *testing.T, dir string) behaviorFixture {
+	t.Helper()
+	data, err := os.ReadFile(filepath.Join(dir, "case.json"))
+	if err != nil {
+		t.Fatalf("ReadFile(case.json): %v", err)
+	}
+	var fixture behaviorFixture
+	if err := json.Unmarshal(data, &fixture); err != nil {
+		t.Fatalf("Unmarshal(case.json): %v", err)
+	}
+	return fixture
+}
+
+func runBehaviorFixture(t *testing.T, dir string, fixture behaviorFixture) {
+	t.Helper()
+
+	source := readFixtureText(t, filepath.Join(dir, fixture.SourceFile))
+	translator := &behaviorFixtureTranslator{
+		t:     t,
+		dir:   dir,
+		rules: fixture.Rules,
+	}
+
+	var (
+		got string
+		err error
+	)
+
+	switch fixture.Mode {
+	case "doc_body_chunked":
+		got, err = translateDocBodyChunked(context.Background(), translator, fixture.RelPath, source, "en", "zh-CN")
+	case "frontmatter_scalar":
+		got, err = translateSnippet(
+			context.Background(),
+			translator,
+			&TranslationMemory{entries: map[string]TMEntry{}},
+			fixture.RelPath+":frontmatter:title",
+			source,
+			"en",
+			"zh-CN",
+		)
+	default:
+		t.Fatalf("unsupported fixture mode %q", fixture.Mode)
+	}
+
+	if fixture.ExpectedErrorContains != "" {
+		if err == nil {
+			t.Fatalf("expected error containing %q, got nil", fixture.ExpectedErrorContains)
+		}
+		if !strings.Contains(err.Error(), fixture.ExpectedErrorContains) {
+			t.Fatalf("expected error containing %q, got %v", fixture.ExpectedErrorContains, err)
+		}
+		return
+	}
+	if err != nil {
+		t.Fatalf("unexpected error: %v", err)
+	}
+
+	if fixture.ExpectedFile != "" {
+		want := readFixtureText(t, filepath.Join(dir, fixture.ExpectedFile))
+		if normalizeBehaviorText(got) != normalizeBehaviorText(want) {
+			t.Fatalf("unexpected output\nwant:\n%s\n\ngot:\n%s", want, got)
+		}
+	}
+
+	for _, fragment := range fixture.ExpectedOutputContains {
+		if !strings.Contains(got, fragment) {
+			t.Fatalf("expected output to contain %q\noutput:\n%s", fragment, got)
+		}
+	}
+	for _, fragment := range fixture.ExpectedOutputNotContains {
+		if strings.Contains(got, fragment) {
+			t.Fatalf("expected output to exclude %q\noutput:\n%s", fragment, got)
+		}
+	}
+}
+
+func readFixtureText(t *testing.T, path string) string {
+	t.Helper()
+	data, err := os.ReadFile(path)
+	if err != nil {
+		t.Fatalf("ReadFile(%q): %v", path, err)
+	}
+	return string(data)
+}
+
+func normalizeBehaviorText(value string) string {
+	return strings.TrimSpace(strings.ReplaceAll(value, "\r\n", "\n"))
+}

scripts/docs-i18n/testdata/behavior/fenced-singleton-retry/case.json

@@ -0,0 +1,24 @@
+{
+  "name": "fenced singleton retries after malformed raw output",
+  "mode": "doc_body_chunked",
+  "rel_path": "gateway/configuration-reference.md",
+  "source_file": "source.txt",
+  "expected_output_contains": [
+    "Translated line 01",
+    "Translated line 02",
+    "Translated line 03",
+    "Translated line 04"
+  ],
+  "expected_output_not_contains": ["Line 01", "Line 02", "Line 03", "Line 04"],
+  "rules": [
+    {
+      "method": "raw",
+      "match_all": ["Line 01", "Line 04"],
+      "response_file": "raw-malformed.txt"
+    },
+    {
+      "method": "raw",
+      "replace_pairs": [{ "from": "Line ", "to": "Translated line " }]
+    }
+  ]
+}

scripts/docs-i18n/testdata/behavior/fenced-singleton-retry/expected.txt

@@ -0,0 +1,7 @@
+```md
+Translated line 01
+Translated line 02
+Translated line 03
+Translated line 04
+```
+

scripts/docs-i18n/testdata/behavior/fenced-singleton-retry/raw-malformed.txt

@@ -0,0 +1,6 @@
+```md
+Line 01
+Line 02
+Line 03
+Line 04
+

scripts/docs-i18n/testdata/behavior/fenced-singleton-retry/source.txt

@@ -0,0 +1,7 @@
+```md
+Line 01
+Line 02
+Line 03
+Line 04
+```
+

scripts/docs-i18n/testdata/behavior/frontmatter-fallback/case.json

@@ -0,0 +1,14 @@
+{
+  "name": "frontmatter scalar falls back instead of keeping tagged wrapper",
+  "mode": "frontmatter_scalar",
+  "rel_path": "install/fly.md",
+  "source_file": "source.txt",
+  "expected_file": "expected.txt",
+  "rules": [
+    {
+      "method": "masked",
+      "match_all": ["Deploying OpenClaw on Fly.io"],
+      "response_file": "masked-tagged-wrapper.txt"
+    }
+  ]
+}

scripts/docs-i18n/testdata/behavior/frontmatter-fallback/expected.txt

@@ -0,0 +1 @@
+Deploying OpenClaw on Fly.io

scripts/docs-i18n/testdata/behavior/frontmatter-fallback/masked-tagged-wrapper.txt

@@ -0,0 +1,7 @@
+<frontmatter>
+title: Fly.io
+</frontmatter>
+
+<body>
+# Fly.io 部署
+</body>

scripts/docs-i18n/testdata/behavior/frontmatter-fallback/source.txt

@@ -0,0 +1 @@
+Deploying OpenClaw on Fly.io

scripts/docs-i18n/testdata/behavior/protocol-leak-split/case.json

@@ -0,0 +1,25 @@
+{
+  "name": "protocol leak retries on smaller chunks",
+  "mode": "doc_body_chunked",
+  "rel_path": "gateway/configuration-reference.md",
+  "source_file": "source.txt",
+  "expected_file": "expected.txt",
+  "expected_output_not_contains": ["<frontmatter>", "<body>", "[[[FM_"],
+  "rules": [
+    {
+      "method": "raw",
+      "match_all": ["First chunk", "Second chunk"],
+      "response_file": "raw-leaked.txt"
+    },
+    {
+      "method": "raw",
+      "match_all": ["First chunk"],
+      "response_file": "raw-first.txt"
+    },
+    {
+      "method": "raw",
+      "match_all": ["Second chunk"],
+      "response_file": "raw-second.txt"
+    }
+  ]
+}

scripts/docs-i18n/testdata/behavior/protocol-leak-split/expected.txt

@@ -0,0 +1,4 @@
+First translated
+
+Second translated
+

scripts/docs-i18n/testdata/behavior/protocol-leak-split/raw-first.txt

@@ -0,0 +1,2 @@
+First translated
+

scripts/docs-i18n/testdata/behavior/protocol-leak-split/raw-leaked.txt

@@ -0,0 +1,9 @@
+<frontmatter>
+title: leaked
+</frontmatter>
+
+<body>
+First translated
+
+Second translated
+</body>

scripts/docs-i18n/testdata/behavior/protocol-leak-split/raw-second.txt

@@ -0,0 +1,2 @@
+Second translated
+

scripts/docs-i18n/testdata/behavior/protocol-leak-split/source.txt

@@ -0,0 +1,4 @@
+First chunk
+
+Second chunk
+

scripts/docs-i18n/testdata/behavior/uppercase-wrapper-strip/case.json

@@ -0,0 +1,14 @@
+{
+  "name": "uppercase body wrapper is stripped",
+  "mode": "doc_body_chunked",
+  "rel_path": "help/testing.md",
+  "source_file": "source.txt",
+  "expected_file": "expected.txt",
+  "rules": [
+    {
+      "method": "raw",
+      "match_all": ["Regular paragraph."],
+      "response_file": "raw-uppercase-wrapper.txt"
+    }
+  ]
+}

scripts/docs-i18n/testdata/behavior/uppercase-wrapper-strip/expected.txt

@@ -0,0 +1 @@
+Translated paragraph.

scripts/docs-i18n/testdata/behavior/uppercase-wrapper-strip/raw-uppercase-wrapper.txt

@@ -0,0 +1,3 @@
+<BODY>
+Translated paragraph.
+</BODY>

scripts/docs-i18n/testdata/behavior/uppercase-wrapper-strip/source.txt

@@ -0,0 +1 @@
+Regular paragraph.