pulse/pkg/reporting/pdf_codepage_test.go

package reporting

import (
	"strings"
	"testing"
	"time"
)

// Free-form report strings arrive as UTF-8 but fpdf core fonts read cp1252
// bytes. These tests render narratives full of typographic punctuation (the
// exact characters AI narrators emit) plus a non-ASCII resource name, then
// assert the content streams carry cp1252 encodings rather than raw UTF-8
// byte sequences, which a PDF viewer displays as mojibake (an em dash
// becomes "—", an é becomes "é").

func TestGenerate_NarrativeUTF8RendersWithoutMojibake(t *testing.T) {
	now := time.Now()
	data := &ReportData{
		Title:        "Monthly Review — “Production”",
		ResourceType: "vm",
		ResourceID:   "vm-7f8b2b6cd98c2089",
		Resource:     &ResourceInfo{Name: "café-web-01", Status: "running"},
		Start:        now.Add(-time.Hour),
		End:          now,
		GeneratedAt:  now,
		Metrics:      map[string][]MetricDataPoint{},
		Summary: MetricSummary{ByMetric: map[string]MetricStats{
			"cpu": {Avg: 10, Max: 20, Count: 60},
		}},
		TotalPoints: 60,
		Narrative: &Narrative{
			Source:           NarrativeSourceAI,
			HealthStatus:     "HEALTHY",
			HealthMessage:    "Quiet period — no incidents recorded",
			ExecutiveSummary: "Utilisation stayed “well within” limits — the host’s capacity is sufficient.",
			Observations:     []NarrativeBullet{{Text: "CPU averaged 10% — flat across the window", Severity: NarrativeSeverityOK}},
			Recommendations:  []string{"Keep monitoring — no changes required"},
			Disclaimer:       "Narrative generated by Pulse Assistant.",
		},
		Alerts: []AlertInfo{{
			Type:      "cpu",
			Level:     "warning",
			Message:   "CPU high — sustained above threshold",
			StartTime: now.Add(-30 * time.Minute),
		}},
	}
	assertNoMojibake(t, renderExecutiveSummaryText(t, data),
		[]string{"—", "“", "”", "’", "café-web-01"})
}

func TestGenerateMulti_FleetNarrativeUTF8RendersWithoutMojibake(t *testing.T) {
	now := time.Now()
	multi := &MultiReportData{
		Title:       "Fleet Review — “June”",
		Start:       now.Add(-time.Hour),
		End:         now,
		GeneratedAt: now,
		Resources: []*ReportData{{
			ResourceID:   "vm-a",
			ResourceType: "vm",
			Resource:     &ResourceInfo{Name: "café-web-01", Status: "running"},
			Summary:      MetricSummary{ByMetric: map[string]MetricStats{"cpu": {Avg: 10, Max: 15, Count: 60}}},
			TotalPoints:  60,
		}},
		FleetNarrative: &FleetNarrative{
			Source:        NarrativeSourceAI,
			HealthStatus:  "HEALTHY",
			HealthMessage: "Fleet stable — “no outliers” this period",
			Outliers: []FleetOutlier{{
				ResourceID:   "vm-a",
				ResourceName: "café-web-01",
				Reason:       "Memory averaging 91.2% — sustained pressure",
				Severity:     NarrativeSeverityWarning,
			}},
			Disclaimer: "Narrative generated by Pulse Assistant.",
		},
	}
	assertNoMojibake(t, renderFleetSummaryText(t, multi),
		[]string{"—", "“", "”", "café-web-01"})
}

// assertNoMojibake decodes the raw content-stream bytes the way a PDF
// viewer decodes core-font text (cp1252/WinAnsi) and asserts the
// typographic characters survived translation. Untranslated UTF-8 decodes
// to fragments led by "â€" (punctuation range) or "Ã" (latin-1 range).
func assertNoMojibake(t *testing.T, raw string, want []string) {
	t.Helper()
	rendered := decodeWinAnsi(raw)
	for _, frag := range []string{"â€", "é"} {
		if strings.Contains(rendered, frag) {
			t.Errorf("rendered PDF text contains mojibake %q:\n%s", frag, rendered)
		}
	}
	for _, w := range want {
		if !strings.Contains(rendered, w) {
			t.Errorf("rendered PDF text missing %q after cp1252 translation:\n%s", w, rendered)
		}
	}
}

// winAnsiSpecials covers the 0x80-0x9F block where cp1252 places the
// typographic punctuation that latin-1 lacks; every other byte maps to the
// identical Unicode code point.
var winAnsiSpecials = map[byte]rune{
	0x80: '€', 0x82: '‚', 0x83: 'ƒ', 0x84: '„', 0x85: '…', 0x86: '†', 0x87: '‡',
	0x88: 'ˆ', 0x89: '‰', 0x8A: 'Š', 0x8B: '‹', 0x8C: 'Œ', 0x8E: 'Ž',
	0x91: '‘', 0x92: '’', 0x93: '“', 0x94: '”', 0x95: '•', 0x96: '–', 0x97: '—',
	0x98: '˜', 0x99: '™', 0x9A: 'š', 0x9B: '›', 0x9C: 'œ', 0x9E: 'ž', 0x9F: 'Ÿ',
}

func decodeWinAnsi(s string) string {
	var b strings.Builder
	for i := 0; i < len(s); i++ {
		c := s[i]
		if c < 0x80 {
			b.WriteByte(c)
			continue
		}
		if r, ok := winAnsiSpecials[c]; ok {
			b.WriteRune(r)
			continue
		}
		b.WriteRune(rune(c))
	}
	return b.String()
}