descrybe/apps/api/internal/feeds/specs_test.go

package feeds

import (
	"strings"
	"testing"
)

func TestParseSpecificationsA1BrokenClosers(t *testing.T) {
	raw := "<ul><li>Energijski razred: E</><li>Dimenzije: 605x95x395 </></ul>"
	pairs := ParseSpecifications(raw)
	if len(pairs) != 2 {
		t.Fatalf("got %d pairs: %#v", len(pairs), pairs)
	}
	if pairs[0].Label != "Energijski razred" || pairs[0].Value != "E" {
		t.Fatalf("first=%#v", pairs[0])
	}
	if pairs[1].Label != "Dimenzije" || pairs[1].Value != "605x95x395" {
		t.Fatalf("second=%#v", pairs[1])
	}
	m := specsToMap(pairs)
	if m["energy_class"] != "E" || m["dimenzije"] != "605x95x395" {
		t.Fatalf("map=%#v", m)
	}
}

func TestAttributeKeyFromLabel(t *testing.T) {
	if got := AttributeKeyFromLabel("Energijski razred"); got != "energijski-razred" {
		t.Fatalf("got %q", got)
	}
	if got := AttributeKeyFromLabel("Širina"); got != "sirina" {
		t.Fatalf("got %q", got)
	}
}

func TestCanonicalAttributeKey(t *testing.T) {
	cases := map[string]string{
		"Visina":       "net_height",
		"netheight":    "net_height",
		"net_height":   "net_height",
		"Širina":       "net_width",
		"Globina":      "net_depth",
		"netMass":      "net_mass",
		"Teža":         "net_mass",
		"Energijski razred": "energy_class",
		":":            "",
		"true":         "",
		"":             "",
	}
	for in, want := range cases {
		if got := CanonicalAttributeKey(in); got != want {
			t.Fatalf("%q → %q want %q", in, got, want)
		}
	}
}

func TestParseSpecificationsSkipsBareListItems(t *testing.T) {
	raw := `<ul><li>:</li><li>Color: Red</li><li>Waterproof</li></ul>`
	pairs := ParseSpecifications(raw)
	if len(pairs) != 1 || pairs[0].Label != "Color" || pairs[0].Value != "Red" {
		t.Fatalf("got %#v", pairs)
	}
	m := specsToMap(pairs)
	if _, ok := m[":"]; ok {
		t.Fatalf("junk key present: %#v", m)
	}
	if m["color"] != "Red" {
		t.Fatalf("map=%#v", m)
	}
}

func TestParseSpecificationsHTML(t *testing.T) {
	raw := `<ul><li>Color: Red</li><li>Size: Large</li><li>Material: Cotton</li></ul>`
	pairs := ParseSpecifications(raw)
	if len(pairs) != 3 {
		t.Fatalf("got %d pairs: %#v", len(pairs), pairs)
	}
	if pairs[0].Label != "Color" || pairs[0].Value != "Red" {
		t.Fatalf("first=%#v", pairs[0])
	}
}

func TestParseSpecificationsHTMLEmpty(t *testing.T) {
	for _, raw := range []string{"", "   ", "<ul></ul>", "<ul><li></li></ul>", "<![CDATA[]]>"} {
		if got := ParseSpecifications(raw); got != nil {
			t.Fatalf("raw=%q got %#v", raw, got)
		}
	}
}

func TestParseSpecificationsFlat(t *testing.T) {
	raw := "Color: Red; Size: L; Weight: 1.2 kg"
	pairs := ParseSpecifications(raw)
	if len(pairs) != 3 {
		t.Fatalf("got %d: %#v", len(pairs), pairs)
	}
	pipe := ParseSpecifications("Voltage: 230V | Frequency: 50Hz")
	if len(pipe) != 2 {
		t.Fatalf("pipe=%#v", pipe)
	}
	nl := ParseSpecifications("Width: 10\nHeight: 20")
	if len(nl) != 2 {
		t.Fatalf("nl=%#v", nl)
	}
}

func TestParseSpecificationsCSVLike(t *testing.T) {
	raw := `"Brand","Acme","Model","X1"`
	// Single chunk without strong separators — treat as one line; may not split.
	// Use semicolon CSV-ish pairs:
	raw = `"Brand","Acme"; "Model","X1"`
	pairs := ParseSpecifications(raw)
	if len(pairs) < 2 {
		t.Fatalf("got %#v", pairs)
	}
}

func TestExpandSpecificationFieldsNestedXML(t *testing.T) {
	xmlBody := `<?xml version="1.0"?>
<Export>
  <Item>
    <EAN>5901234123457</EAN>
    <name>Washer</name>
    <brand>Janus</brand>
    <specifications>
      <EnergyClass>A</EnergyClass>
      <Capacity>8 kg</Capacity>
    </specifications>
    <EPRELID></EPRELID>
    <netMass>72</netMass>
  </Item>
</Export>`
	var row feedRow
	n, err := parseXMLItems(strings.NewReader(xmlBody), "Item", func(r feedRow) error {
		row = r
		return nil
	})
	if err != nil {
		t.Fatal(err)
	}
	if n != 1 {
		t.Fatalf("n=%d", n)
	}
	if row["specifications/EnergyClass"] != "A" {
		t.Fatalf("nested energy=%q row=%v", row["specifications/EnergyClass"], row)
	}
	if row["specifications/Capacity"] != "8 kg" {
		t.Fatalf("capacity=%q", row["specifications/Capacity"])
	}
	v, ok := lookupRow(row, "specifications/EnergyClass")
	if !ok || v != "A" {
		t.Fatalf("lookup nested: ok=%v v=%q", ok, v)
	}
}

func TestExpandSpecificationFieldsCDATA(t *testing.T) {
	xmlBody := `<?xml version="1.0"?>
<products>
  <item>
    <gtin>111</gtin>
    <specifications><![CDATA[<ul><li>Color: Blue</li><li>Finish: Matte</li></ul>]]></specifications>
  </item>
</products>`
	var row feedRow
	_, err := parseXMLItems(strings.NewReader(xmlBody), "item", func(r feedRow) error {
		row = r
		return nil
	})
	if err != nil {
		t.Fatal(err)
	}
	if row["specifications/Color"] != "Blue" {
		t.Fatalf("color=%q row=%v", row["specifications/Color"], row)
	}
	if row["specifications/Finish"] != "Matte" {
		t.Fatalf("finish=%q", row["specifications/Finish"])
	}
}

func TestExpandSpecificationFieldsFlatAndEmpty(t *testing.T) {
	xmlBody := `<?xml version="1.0"?>
<products>
  <item>
    <ean>222</ean>
    <specifications>Width: 10; Height: 20</specifications>
  </item>
  <item>
    <ean>333</ean>
    <specifications><![CDATA[<ul></ul>]]></specifications>
  </item>
</products>`
	var rows []feedRow
	_, err := parseXMLItems(strings.NewReader(xmlBody), "item", func(r feedRow) error {
		cp := make(feedRow, len(r))
		for k, v := range r {
			cp[k] = v
		}
		rows = append(rows, cp)
		return nil
	})
	if err != nil {
		t.Fatal(err)
	}
	if len(rows) != 2 {
		t.Fatalf("rows=%d", len(rows))
	}
	if rows[0]["specifications/Width"] != "10" {
		t.Fatalf("width=%q", rows[0]["specifications/Width"])
	}
	// Empty HTML must not invent children.
	for k := range rows[1] {
		if strings.HasPrefix(k, "specifications/") {
			t.Fatalf("unexpected child %q", k)
		}
	}
}

func TestApplyMappingsSpecificationsObject(t *testing.T) {
	row := feedRow{
		"EAN":                        "5901234123457",
		"specifications":             `<ul><li>Color: Red</li></ul>`,
		"specifications/Color":       "Red",
		"specifications/EnergyClass": "B",
	}
	expandSpecificationFields(row)
	mappings := parseMappings([]any{
		map[string]any{"source": "EAN", "target": "gtin"},
		map[string]any{"source": "specifications", "target": "specifications"},
		map[string]any{"source": "specifications/EnergyClass", "target": "energy_class"},
	})
	mapped, gtin := applyMappings(row, mappings)
	if gtin != "5901234123457" {
		t.Fatalf("gtin=%q", gtin)
	}
	specs, ok := mapped["specifications"].(map[string]string)
	if !ok || specs["color"] != "Red" {
		t.Fatalf("specs=%#v", mapped["specifications"])
	}
	if mapped["energy_class"] != "B" {
		t.Fatalf("energy=%v", mapped["energy_class"])
	}
}

func TestExtractXMLSchemaNestedSpecs(t *testing.T) {
	data := []byte(`<?xml version="1.0"?>
<Export><Item>
  <EAN>1</EAN>
  <specifications><Color>Red</Color><Size>M</Size></specifications>
</Item></Export>`)
	fields, rows, err := extractXMLSchema(data, "Item")
	if err != nil {
		t.Fatal(err)
	}
	if rows != 1 {
		t.Fatalf("rows=%d", rows)
	}
	paths := map[string]bool{}
	for _, f := range fields {
		paths[f.Path] = true
	}
	if !paths["specifications/Color"] || !paths["specifications/Size"] {
		t.Fatalf("missing nested paths: %#v", paths)
	}
	// Bare Color leaf should be suppressed when nested exists.
	if paths["Color"] {
		t.Fatalf("bare Color should be dropped: %#v", paths)
	}
}

func TestLookupRowNestedPrefersPath(t *testing.T) {
	row := feedRow{
		"name":               "Product",
		"specifications":     "ignored",
		"specifications/foo": "nested",
		"other/foo":          "other",
	}
	v, ok := lookupRow(row, "specifications/foo")
	if !ok || v != "nested" {
		t.Fatalf("got ok=%v v=%q", ok, v)
	}
}