Add NWS forecast discussion heading regressions

2026-08-02 20:53:30 +00:00
parent f402e27542
commit 3ecf4c5b7f
2 changed files with 229 additions and 0 deletions
--- a/internal/normalizers/nws/forecast_discussion_test.go
+++ b/internal/normalizers/nws/forecast_discussion_test.go
@@ -71,6 +71,70 @@ func TestForecastDiscussionNormalizerProducesCanonicalSchema(t *testing.T) {
 	}
 }

+func TestForecastDiscussionNormalizerSupportsMixedHeadingFormats(t *testing.T) {
+	out, err := (ForecastDiscussionNormalizer{}).Normalize(nil, event.Event{
+		ID:        "evt-discussion-mixed-format",
+		Kind:      event.Kind(standards.KindForecastDiscussion),
+		Source:    "nws-discussion-test",
+		EmittedAt: time.Date(2026, 3, 28, 19, 25, 0, 0, time.UTC),
+		Schema:    standards.SchemaRawNWSForecastDiscussionV1,
+		Payload:   loadMixedFormatForecastDiscussionSampleHTML(t),
+	})
+	if err != nil {
+		t.Fatalf("Normalize() error = %v", err)
+	}
+	if out == nil {
+		t.Fatalf("Normalize() returned nil output")
+	}
+	if out.Kind != event.Kind(standards.KindForecastDiscussion) {
+		t.Fatalf("Kind = %q, want forecast_discussion", out.Kind)
+	}
+	if out.Schema != standards.SchemaWeatherForecastDiscussionV1 {
+		t.Fatalf("Schema = %q, want %q", out.Schema, standards.SchemaWeatherForecastDiscussionV1)
+	}
+	wantEffectiveAt := time.Date(2026, 3, 28, 19, 24, 0, 0, time.UTC)
+	if out.EffectiveAt == nil || !out.EffectiveAt.Equal(wantEffectiveAt) {
+		t.Fatalf("EffectiveAt = %v, want %s", out.EffectiveAt, wantEffectiveAt.Format(time.RFC3339))
+	}
+
+	payload, ok := out.Payload.(model.WeatherForecastDiscussion)
+	if !ok {
+		t.Fatalf("Payload type = %T, want model.WeatherForecastDiscussion", out.Payload)
+	}
+	if payload.ShortTerm == nil || payload.LongTerm == nil {
+		t.Fatalf("ShortTerm=%v LongTerm=%v, want both populated", payload.ShortTerm, payload.LongTerm)
+	}
+	if payload.ShortTerm.Qualifier != "Through Late Sunday Night" {
+		t.Fatalf("ShortTerm.Qualifier = %q", payload.ShortTerm.Qualifier)
+	}
+	if !strings.Contains(payload.ShortTerm.Text, "After a chilly morning") {
+		t.Fatalf("ShortTerm.Text missing expected prose: %q", payload.ShortTerm.Text)
+	}
+	if payload.LongTerm.Qualifier != "Monday through Next Saturday" {
+		t.Fatalf("LongTerm.Qualifier = %q", payload.LongTerm.Qualifier)
+	}
+	if !strings.Contains(payload.LongTerm.Text, "The peak of the warmth arrives Monday and Tuesday") {
+		t.Fatalf("LongTerm.Text missing expected prose: %q", payload.LongTerm.Text)
+	}
+	if strings.Contains(payload.LongTerm.Text, "AVIATION") || strings.Contains(payload.LongTerm.Text, "VFR conditions are expected") {
+		t.Fatalf("LongTerm.Text includes aviation content: %q", payload.LongTerm.Text)
+	}
+
+	b, err := json.Marshal(out.Payload)
+	if err != nil {
+		t.Fatalf("json.Marshal(payload) error = %v", err)
+	}
+	var fields map[string]any
+	if err := json.Unmarshal(b, &fields); err != nil {
+		t.Fatalf("json.Unmarshal(payload) error = %v", err)
+	}
+	for _, key := range []string{"aviation", "sections"} {
+		if _, ok := fields[key]; ok {
+			t.Fatalf("unexpected key %q in canonical payload", key)
+		}
+	}
+}
+
 func TestForecastDiscussionNormalizerRejectsMissingIssueTime(t *testing.T) {
 	_, err := (ForecastDiscussionNormalizer{}).Normalize(nil, event.Event{
 		ID:        "evt-discussion-bad",
@@ -128,3 +192,34 @@ func loadForecastDiscussionSampleHTML(t *testing.T) string {
 	}
 	return string(b)
 }
+
+func loadMixedFormatForecastDiscussionSampleHTML(t *testing.T) string {
+	t.Helper()
+
+	raw := loadForecastDiscussionSampleHTML(t)
+	replacements := []struct {
+		original    string
+		replacement string
+	}{
+		{
+			original:    ".SHORT TERM...  (Through Late Sunday Night)",
+			replacement: ".SHORT TERM /Through Late Sunday Night/...",
+		},
+		{
+			original:    ".LONG TERM...  (Monday through Next Saturday)",
+			replacement: ".LONG TERM /Monday through Next Saturday/...",
+		},
+		{
+			original:    ".AVIATION...  (For the 18z TAFs through 18z Sunday Afternoon)",
+			replacement: ".AVIATION /For the 18z TAFs through 18z Sunday Afternoon/...",
+		},
+	}
+	for _, replacement := range replacements {
+		if !strings.Contains(raw, replacement.original) {
+			t.Fatalf("fixture missing heading %q", replacement.original)
+		}
+		raw = strings.Replace(raw, replacement.original, replacement.replacement, 1)
+	}
+
+	return raw
+}
--- a/internal/providers/nws/forecast_discussion_test.go
+++ b/internal/providers/nws/forecast_discussion_test.go
@@ -3,6 +3,7 @@ package nws
 import (
 	"os"
 	"path/filepath"
+	"reflect"
 	"strings"
 	"testing"
 	"time"
@@ -137,6 +138,63 @@ func TestParseForecastDiscussionSectionHeading(t *testing.T) {
 	}
 }

+func TestExtractForecastDiscussionSectionStopsAtSlashQualifiedHeading(t *testing.T) {
+	block, ok := extractForecastDiscussionSection([]string{
+		".LONG TERM /Monday through Next Saturday/...",
+		"Long-term prose.",
+		".AVIATION /18Z TAFS/...",
+		"Aviation prose.",
+	}, forecastDiscussionSectionLongTerm)
+	if !ok {
+		t.Fatalf("extractForecastDiscussionSection() found no LONG TERM block")
+	}
+	if block.heading.qualifier != "Monday through Next Saturday" {
+		t.Fatalf("qualifier = %q, want normalized slash qualifier", block.heading.qualifier)
+	}
+	wantBody := []string{"Long-term prose."}
+	if !reflect.DeepEqual(block.body, wantBody) {
+		t.Fatalf("body = %#v, want %#v", block.body, wantBody)
+	}
+}
+
+func TestExtractForecastDiscussionSectionAllowsEmptyBodyBeforeHeading(t *testing.T) {
+	block, ok := extractForecastDiscussionSection([]string{
+		".SHORT TERM... (Tonight)",
+		".LONG TERM... (Tomorrow)",
+		"Long-term prose.",
+	}, forecastDiscussionSectionShortTerm)
+	if !ok {
+		t.Fatalf("extractForecastDiscussionSection() found no SHORT TERM block")
+	}
+	if len(block.body) != 0 {
+		t.Fatalf("body = %#v, want empty body", block.body)
+	}
+}
+
+func TestExtractForecastDiscussionSectionKeepsUnrecognizedHeadingLikeLines(t *testing.T) {
+	block, ok := extractForecastDiscussionSection([]string{
+		".SHORT TERM... (Tonight)",
+		".SYNOPSIS... This unknown section stays in the body.",
+		".LONG TERM /Tomorrow/..",
+		".LONG TERM /Tomorrow/Next Week/...",
+		"Expected short-term prose.",
+		".LONG TERM... (Tomorrow)",
+		"Long-term prose.",
+	}, forecastDiscussionSectionShortTerm)
+	if !ok {
+		t.Fatalf("extractForecastDiscussionSection() found no SHORT TERM block")
+	}
+	wantBody := []string{
+		".SYNOPSIS... This unknown section stays in the body.",
+		".LONG TERM /Tomorrow/..",
+		".LONG TERM /Tomorrow/Next Week/...",
+		"Expected short-term prose.",
+	}
+	if !reflect.DeepEqual(block.body, wantBody) {
+		t.Fatalf("body = %#v, want %#v", block.body, wantBody)
+	}
+}
+
 func TestParseForecastDiscussionHTMLParsesExpectedFields(t *testing.T) {
 	raw := loadForecastDiscussionSampleHTML(t)

@@ -215,6 +273,51 @@ func TestParseForecastDiscussionHTMLParsesExpectedFields(t *testing.T) {
 	}
 }

+func TestParseForecastDiscussionHTMLSupportsMixedHeadingFormats(t *testing.T) {
+	legacy, err := ParseForecastDiscussionHTML(loadForecastDiscussionSampleHTML(t))
+	if err != nil {
+		t.Fatalf("ParseForecastDiscussionHTML() legacy error = %v", err)
+	}
+
+	got, err := ParseForecastDiscussionHTML(loadMixedFormatForecastDiscussionSampleHTML(t))
+	if err != nil {
+		t.Fatalf("ParseForecastDiscussionHTML() mixed-format error = %v", err)
+	}
+	if !reflect.DeepEqual(got.KeyMessages, legacy.KeyMessages) {
+		t.Fatalf("KeyMessages = %#v, want unchanged %#v", got.KeyMessages, legacy.KeyMessages)
+	}
+
+	if got.ShortTerm == nil {
+		t.Fatalf("ShortTerm is nil")
+	}
+	if got.ShortTerm.Qualifier != "Through Late Sunday Night" {
+		t.Fatalf("ShortTerm.Qualifier = %q", got.ShortTerm.Qualifier)
+	}
+	wantIssuedAt := time.Date(2026, 3, 28, 19, 19, 0, 0, time.UTC)
+	if got.ShortTerm.IssuedAt == nil || !got.ShortTerm.IssuedAt.Equal(wantIssuedAt) {
+		t.Fatalf("ShortTerm.IssuedAt = %v, want %s", got.ShortTerm.IssuedAt, wantIssuedAt.Format(time.RFC3339))
+	}
+	if !strings.Contains(got.ShortTerm.Text, "After a chilly morning") {
+		t.Fatalf("ShortTerm.Text missing expected prose: %q", got.ShortTerm.Text)
+	}
+
+	if got.LongTerm == nil {
+		t.Fatalf("LongTerm is nil")
+	}
+	if got.LongTerm.Qualifier != "Monday through Next Saturday" {
+		t.Fatalf("LongTerm.Qualifier = %q", got.LongTerm.Qualifier)
+	}
+	if got.LongTerm.IssuedAt == nil || !got.LongTerm.IssuedAt.Equal(wantIssuedAt) {
+		t.Fatalf("LongTerm.IssuedAt = %v, want %s", got.LongTerm.IssuedAt, wantIssuedAt.Format(time.RFC3339))
+	}
+	if !strings.Contains(got.LongTerm.Text, "The peak of the warmth arrives Monday and Tuesday") {
+		t.Fatalf("LongTerm.Text missing expected prose: %q", got.LongTerm.Text)
+	}
+	if strings.Contains(got.LongTerm.Text, "AVIATION") || strings.Contains(got.LongTerm.Text, "VFR conditions are expected") {
+		t.Fatalf("LongTerm.Text includes aviation content: %q", got.LongTerm.Text)
+	}
+}
+
 func TestParseForecastDiscussionHTMLMissingPreBlock(t *testing.T) {
 	_, err := ParseForecastDiscussionHTML("<html><body><div>no pre block</div></body></html>")
 	if err == nil {
@@ -245,3 +348,34 @@ func loadForecastDiscussionSampleHTML(t *testing.T) string {
 	}
 	return string(b)
 }
+
+func loadMixedFormatForecastDiscussionSampleHTML(t *testing.T) string {
+	t.Helper()
+
+	raw := loadForecastDiscussionSampleHTML(t)
+	replacements := []struct {
+		original    string
+		replacement string
+	}{
+		{
+			original:    ".SHORT TERM...  (Through Late Sunday Night)",
+			replacement: ".SHORT TERM /Through Late Sunday Night/...",
+		},
+		{
+			original:    ".LONG TERM...  (Monday through Next Saturday)",
+			replacement: ".LONG TERM /Monday through Next Saturday/...",
+		},
+		{
+			original:    ".AVIATION...  (For the 18z TAFs through 18z Sunday Afternoon)",
+			replacement: ".AVIATION /For the 18z TAFs through 18z Sunday Afternoon/...",
+		},
+	}
+	for _, replacement := range replacements {
+		if !strings.Contains(raw, replacement.original) {
+			t.Fatalf("fixture missing heading %q", replacement.original)
+		}
+		raw = strings.Replace(raw, replacement.original, replacement.replacement, 1)
+	}
+
+	return raw
+}