Handle additional NWS forecast discussion formats

internal/normalizers/nws/forecast_discussion_test.go

@@ -215,7 +215,7 @@ func TestForecastDiscussionNormalizerSupportsCrossOfficeLayout(t *testing.T) {
 	}
 }
 
-func TestForecastDiscussionNormalizerSupportsNumberedAndKeyPointsFixtures(t *testing.T) {
+func TestForecastDiscussionNormalizerSupportsCrossOfficeKeyMessageFixtures(t *testing.T) {
 	tests := []struct {
 		name        string
 		filename    string
@@ -249,6 +249,42 @@ func TestForecastDiscussionNormalizerSupportsNumberedAndKeyPointsFixtures(t *tes
 				"Inland valleys remain dry through Saturday.",
 			},
 		},
+		{
+			name:        "as of preamble",
+			filename:    "forecast_discussion_rah_as_of_sample.html",
+			id:          "evt-discussion-rah",
+			source:      "nws-discussion-rah-test",
+			emittedAt:   time.Date(2026, 8, 2, 16, 36, 0, 0, time.UTC),
+			effectiveAt: time.Date(2026, 8, 2, 16, 35, 0, 0, time.UTC),
+			messages: []string{
+				"Scattered storms may produce locally heavy rain this afternoon.",
+				"Drier weather arrives Monday.",
+			},
+		},
+		{
+			name:        "parenthesized numeric markers",
+			filename:    "forecast_discussion_lwx_parenthesized_number_sample.html",
+			id:          "evt-discussion-lwx",
+			source:      "nws-discussion-lwx-test",
+			emittedAt:   time.Date(2026, 8, 2, 18, 1, 0, 0, time.UTC),
+			effectiveAt: time.Date(2026, 8, 2, 18, 0, 0, 0, time.UTC),
+			messages: []string{
+				"Thunderstorms remain possible near the Blue Ridge this evening.",
+				"Seasonably warm conditions continue Monday.",
+			},
+		},
+		{
+			name:        "embedded key messages in previous discussion",
+			filename:    "forecast_discussion_mfr_prev_discussion_sample.html",
+			id:          "evt-discussion-mfr-previous",
+			source:      "nws-discussion-mfr-previous-test",
+			emittedAt:   time.Date(2026, 8, 2, 22, 20, 0, 0, time.UTC),
+			effectiveAt: time.Date(2026, 8, 2, 22, 19, 0, 0, time.UTC),
+			messages: []string{
+				"Heat returns to inland valleys Monday.",
+				"Gusty afternoon winds develop east of the Cascades.",
+			},
+		},
 	}
 
 	for _, tt := range tests {

internal/providers/nws/forecast_discussion.go

@@ -550,6 +550,7 @@ func isForecastDiscussionHorizontalWhitespace(b byte) bool {
 func parseForecastDiscussionSectionBlocks(lines []string) []forecastDiscussionSectionBlock {
 	var blocks []forecastDiscussionSectionBlock
 	var active *forecastDiscussionSectionBlock
+	embeddedHeadings := false
 
 	finish := func() {
 		if active == nil {
@@ -567,6 +568,7 @@ func parseForecastDiscussionSectionBlocks(lines []string) []forecastDiscussionSe
 		}
 		if line == "&&" || strings.Contains(line, "WATCHES/WARNINGS/ADVISORIES") {
 			finish()
+			embeddedHeadings = false
 			continue
 		}
 
@@ -574,8 +576,17 @@ func parseForecastDiscussionSectionBlocks(lines []string) []forecastDiscussionSe
 		if ok {
 			finish()
 			active = &forecastDiscussionSectionBlock{heading: heading}
+			embeddedHeadings = isForecastDiscussionEmbeddedSectionWrapper(heading.section)
 			continue
 		}
+		if embeddedHeadings {
+			heading, ok = parseForecastDiscussionEmbeddedSectionHeading(raw)
+			if ok {
+				finish()
+				active = &forecastDiscussionSectionBlock{heading: heading}
+				continue
+			}
+		}
 		if active != nil {
 			active.body = append(active.body, raw)
 		}
@@ -585,6 +596,18 @@ func parseForecastDiscussionSectionBlocks(lines []string) []forecastDiscussionSe
 	return blocks
 }
 
+func isForecastDiscussionEmbeddedSectionWrapper(section string) bool {
+	return section == "PREV DISCUSSION"
+}
+
+func parseForecastDiscussionEmbeddedSectionHeading(line string) (forecastDiscussionSectionHeading, bool) {
+	line = strings.TrimSpace(line)
+	if line == "" || line[0] == '.' {
+		return forecastDiscussionSectionHeading{}, false
+	}
+	return parseForecastDiscussionSectionHeading("." + line)
+}
+
 func parseForecastDiscussionKeyMessages(body []string) []string {
 	body = removeForecastDiscussionPresentationMarkers(body)
 	body = trimBlankLines(body)
@@ -716,7 +739,32 @@ func isForecastDiscussionKeyMessageMetadataLine(line string) bool {
 			return true
 		}
 	}
-	return false
+	return isForecastDiscussionKeyMessageAsOfLine(line)
+}
+
+func isForecastDiscussionKeyMessageAsOfLine(line string) bool {
+	const label = "As of"
+
+	if !hasForecastDiscussionASCIIPrefix(line, label) || len(line) == len(label) || !isForecastDiscussionHorizontalWhitespace(line[len(label)]) {
+		return false
+	}
+	remainder := strings.TrimSpace(line[len(label):])
+	if !strings.HasSuffix(remainder, "...") {
+		return false
+	}
+	fields := strings.Fields(strings.TrimSpace(strings.TrimSuffix(remainder, "...")))
+	if len(fields) != 3 {
+		return false
+	}
+	if _, _, err := parseForecastDiscussionClock(fields[0], fields[1]); err != nil {
+		return false
+	}
+	switch strings.ToLower(fields[2]) {
+	case "monday", "tuesday", "wednesday", "thursday", "friday", "saturday", "sunday":
+		return true
+	default:
+		return false
+	}
 }
 
 func stripForecastDiscussionKeyMessageMarker(line string) (string, bool) {
@@ -740,11 +788,23 @@ func stripForecastDiscussionKeyMessageMarker(line string) (string, bool) {
 }
 
 func stripForecastDiscussionKeyMessageNumericMarker(line string) (string, bool) {
+	digitStart := 0
 	digitEnd := 0
+	parenthesized := len(line) > 0 && line[0] == '('
+	if parenthesized {
+		digitStart = 1
+		digitEnd = 1
+	}
 	for digitEnd < len(line) && line[digitEnd] >= '0' && line[digitEnd] <= '9' {
 		digitEnd++
 	}
-	if digitEnd == 0 || digitEnd == len(line) || (line[digitEnd] != ')' && line[digitEnd] != '.') {
+	if digitEnd == digitStart || digitEnd == len(line) {
+		return "", false
+	}
+	if parenthesized && line[digitEnd] != ')' {
+		return "", false
+	}
+	if !parenthesized && line[digitEnd] != ')' && line[digitEnd] != '.' {
 		return "", false
 	}
 
@@ -752,7 +812,7 @@ func stripForecastDiscussionKeyMessageNumericMarker(line string) (string, bool)
 	if markerEnd < len(line) && !isForecastDiscussionHorizontalWhitespace(line[markerEnd]) {
 		return "", false
 	}
-	value, err := strconv.ParseUint(line[:digitEnd], 10, 0)
+	value, err := strconv.ParseUint(line[digitStart:digitEnd], 10, 0)
 	if err != nil || value == 0 {
 		return "", false
 	}

internal/providers/nws/forecast_discussion_test.go

@@ -391,6 +391,58 @@ func TestParseForecastDiscussionSectionBlocksUsesGenericHeadingsAsBoundaries(t *
 	}
 }
 
+func TestParseForecastDiscussionSectionBlocksRecognizesEmbeddedHeadingsInPreviousDiscussion(t *testing.T) {
+	got := parseForecastDiscussionSectionBlocks([]string{
+		".PREV DISCUSSION... /Issued 319 PM PDT Sun Aug 2 2026/",
+		"KEY MESSAGES...",
+		"* First embedded message.",
+		"DISCUSSION...",
+		"Discussion prose.",
+		"&&",
+	})
+	want := []forecastDiscussionSectionBlock{
+		{
+			heading: forecastDiscussionSectionHeading{
+				section:   "PREV DISCUSSION",
+				qualifier: "/Issued 319 PM PDT Sun Aug 2 2026/",
+			},
+		},
+		{
+			heading: forecastDiscussionSectionHeading{section: "KEY MESSAGES"},
+			body:    []string{"* First embedded message."},
+		},
+		{
+			heading: forecastDiscussionSectionHeading{section: "DISCUSSION"},
+			body:    []string{"Discussion prose."},
+		},
+	}
+	if !reflect.DeepEqual(got, want) {
+		t.Fatalf("blocks = %#v, want %#v", got, want)
+	}
+}
+
+func TestParseForecastDiscussionSectionBlocksKeepsUndottedHeadingLikeLinesOutsideWrappers(t *testing.T) {
+	got := parseForecastDiscussionSectionBlocks([]string{
+		".SHORT TERM...",
+		"Short-term prose.",
+		"KEY MESSAGES...",
+		"Still short-term prose.",
+	})
+	want := []forecastDiscussionSectionBlock{
+		{
+			heading: forecastDiscussionSectionHeading{section: "SHORT TERM"},
+			body: []string{
+				"Short-term prose.",
+				"KEY MESSAGES...",
+				"Still short-term prose.",
+			},
+		},
+	}
+	if !reflect.DeepEqual(got, want) {
+		t.Fatalf("blocks = %#v, want %#v", got, want)
+	}
+}
+
 func TestParseForecastDiscussionSectionBlocksUsesParenthesizedTerminalHeadingAsBoundary(t *testing.T) {
 	got := parseForecastDiscussionSectionBlocks([]string{
 		".SHORT TERM... (Tonight)",
@@ -756,6 +808,16 @@ func TestIsForecastDiscussionKeyMessageMetadataLine(t *testing.T) {
 			line: "uPdAtEd At 300 PM CDT Sat Mar 28 2026",
 			want: true,
 		},
+		{
+			name: "as of weekday preamble",
+			line: "As of 1235 PM Sunday...",
+			want: true,
+		},
+		{
+			name: "as of ASCII case insensitive",
+			line: "aS oF 1235 pm SuNdAy...",
+			want: true,
+		},
 		{
 			name: "label without whitespace boundary",
 			line: "Updated at300 PM CDT Sat Mar 28 2026",
@@ -766,6 +828,26 @@ func TestIsForecastDiscussionKeyMessageMetadataLine(t *testing.T) {
 			line: "Updated at not a timestamp",
 			want: false,
 		},
+		{
+			name: "as of invalid clock",
+			line: "As of 2535 PM Sunday...",
+			want: false,
+		},
+		{
+			name: "as of invalid weekday",
+			line: "As of 1235 PM Someday...",
+			want: false,
+		},
+		{
+			name: "as of without terminal ellipsis",
+			line: "As of 1235 PM Sunday",
+			want: false,
+		},
+		{
+			name: "as of prefix collision",
+			line: "As often happens, storms weaken overnight.",
+			want: false,
+		},
 		{
 			name: "nonmetadata prose",
 			line: "Updated atmospheric conditions remain unsettled.",
@@ -794,13 +876,18 @@ func TestStripForecastDiscussionKeyMessageMarker(t *testing.T) {
 		{name: "asterisk", line: "* First message.", want: "First message.", ok: true},
 		{name: "numeric parenthesis", line: "1) First message.", want: "First message.", ok: true},
 		{name: "numeric dot", line: "2. Second message.", want: "Second message.", ok: true},
+		{name: "parenthesized numeric", line: "(1) First message.", want: "First message.", ok: true},
 		{name: "composite hyphen", line: "- 1. First message.", want: "First message.", ok: true},
 		{name: "composite asterisk", line: "* 1) First message.", want: "First message.", ok: true},
+		{name: "composite parenthesized numeric", line: "- (1) First message.", want: "First message.", ok: true},
+		{name: "composite asterisk parenthesized numeric", line: "* (1) First message.", want: "First message.", ok: true},
 		{name: "multi digit numeric", line: "12. Twelfth message.", want: "Twelfth message.", ok: true},
 		{name: "hyphen only", line: "-", want: "", ok: true},
 		{name: "numeric marker only", line: "1)", want: "", ok: true},
 		{name: "zero numeric marker", line: "0) Not a marker.", want: "", ok: false},
+		{name: "zero parenthesized numeric marker", line: "(0) Not a marker.", want: "", ok: false},
 		{name: "overflow numeric marker", line: "999999999999999999999999999999. Too big.", want: "", ok: false},
+		{name: "parenthesized numeric marker without boundary", line: "(1)Not a marker.", want: "", ok: false},
 		{name: "alphanumeric numeric prefix", line: "1x Not a marker.", want: "", ok: false},
 		{name: "numeric marker without boundary", line: "1)Not a marker.", want: "", ok: false},
 		{name: "numeric marker with unsupported punctuation", line: "1, Not a marker.", want: "", ok: false},
@@ -877,6 +964,14 @@ func TestParseForecastDiscussionKeyMessagesClassifiesListsAndMetadata(t *testing
 			},
 			want: []string{"First message."},
 		},
+		{
+			name: "leading as of metadata",
+			body: []string{
+				"As of 1235 PM Sunday...",
+				"- First message.",
+			},
+			want: []string{"First message."},
+		},
 		{
 			name: "updated atmospheric prose remains content",
 			body: []string{
@@ -901,6 +996,14 @@ func TestParseForecastDiscussionKeyMessagesClassifiesListsAndMetadata(t *testing
 			},
 			want: []string{"First message. Updated at 300 PM CDT Sat Mar 28 2026"},
 		},
+		{
+			name: "later as of metadata remains content",
+			body: []string{
+				"- First message.",
+				"As of 1235 PM Sunday...",
+			},
+			want: []string{"First message. As of 1235 PM Sunday..."},
+		},
 	}
 
 	for _, tt := range tests {
@@ -1089,7 +1192,7 @@ func TestParseForecastDiscussionHTMLSupportsMixedHeadingFormats(t *testing.T) {
 	}
 }
 
-func TestParseForecastDiscussionHTMLParsesNumberedAndKeyPointsFixtures(t *testing.T) {
+func TestParseForecastDiscussionHTMLParsesCrossOfficeKeyMessageFixtures(t *testing.T) {
 	tests := []struct {
 		name       string
 		filename   string
@@ -1127,6 +1230,48 @@ func TestParseForecastDiscussionHTMLParsesNumberedAndKeyPointsFixtures(t *testin
 				"*", "DISCUSSION", "(Today through Thursday)", "Discussion details must not be included with key points.",
 			},
 		},
+		{
+			name:       "as of preamble",
+			filename:   "forecast_discussion_rah_as_of_sample.html",
+			officeID:   "RAH",
+			officeName: "National Weather Service Raleigh NC",
+			issuedAt:   time.Date(2026, 8, 2, 16, 35, 0, 0, time.UTC),
+			messages: []string{
+				"Scattered storms may produce locally heavy rain this afternoon.",
+				"Drier weather arrives Monday.",
+			},
+			unwanted: []string{
+				"As of", "1)", "2)", "DISCUSSION", "Discussion details remain boundary-only content.",
+			},
+		},
+		{
+			name:       "parenthesized numeric markers",
+			filename:   "forecast_discussion_lwx_parenthesized_number_sample.html",
+			officeID:   "LWX",
+			officeName: "National Weather Service Baltimore MD/Washington DC",
+			issuedAt:   time.Date(2026, 8, 2, 18, 0, 0, 0, time.UTC),
+			messages: []string{
+				"Thunderstorms remain possible near the Blue Ridge this evening.",
+				"Seasonably warm conditions continue Monday.",
+			},
+			unwanted: []string{
+				"(1)", "(2)", "AVIATION", "Aviation details remain boundary-only content.",
+			},
+		},
+		{
+			name:       "embedded key messages in previous discussion",
+			filename:   "forecast_discussion_mfr_prev_discussion_sample.html",
+			officeID:   "MFR",
+			officeName: "National Weather Service Medford OR",
+			issuedAt:   time.Date(2026, 8, 2, 22, 19, 0, 0, time.UTC),
+			messages: []string{
+				"Heat returns to inland valleys Monday.",
+				"Gusty afternoon winds develop east of the Cascades.",
+			},
+			unwanted: []string{
+				"PREV DISCUSSION", "Issued 319 PM", "*", "DISCUSSION", "Discussion details must not be included with key messages.",
+			},
+		},
 	}
 
 	for _, tt := range tests {

internal/providers/nws/testdata/forecast_discussion_lwx_parenthesized_number_sample.html

@@ -0,0 +1,27 @@
+<!DOCTYPE html>
+<!-- Representative LWX-style layout; prose is concise edited test data, not an archived product. -->
+<html>
+  <body>
+    <pre class="glossaryProduct">
+FXUS61 KLWX 021800
+AFDLWX
+
+Area Forecast Discussion
+National Weather Service Baltimore MD/Washington DC
+200 PM EDT Sun Aug 2 2026
+
+.KEY MESSAGES...
+- (1) Thunderstorms remain possible near the Blue Ridge this evening.
+- (2) Seasonably warm conditions continue Monday.
+
+&&
+
+.AVIATION...
+Aviation details remain boundary-only content.
+
+$$
+
+WFO LWX
+    </pre>
+  </body>
+</html>

internal/providers/nws/testdata/forecast_discussion_mfr_prev_discussion_sample.html

@@ -0,0 +1,33 @@
+<!DOCTYPE html>
+<!-- Representative current MFR previous-discussion wrapper layout; prose is concise edited test data, not an archived product. -->
+<html>
+  <body>
+    <pre class="glossaryProduct">
+FXUS66 KMFR 022219
+AFDMFR
+
+Area Forecast Discussion
+National Weather Service Medford OR
+319 PM PDT Sun Aug 2 2026
+
+.PREV DISCUSSION... /Issued 319 PM PDT Sun Aug 2 2026/
+
+KEY MESSAGES...
+
+* Heat returns to inland valleys Monday.
+* Gusty afternoon winds develop east of the Cascades.
+
+DISCUSSION...
+Discussion details must not be included with key messages.
+
+&&
+
+.MFR WATCHES/WARNINGS/ADVISORIES...
+None.
+
+$$
+
+WFO MFR
+    </pre>
+  </body>
+</html>

internal/providers/nws/testdata/forecast_discussion_rah_as_of_sample.html

@@ -0,0 +1,29 @@
+<!DOCTYPE html>
+<!-- Representative RAH-style layout; prose is concise edited test data, not an archived product. -->
+<html>
+  <body>
+    <pre class="glossaryProduct">
+FXUS62 KRAH 021635
+AFDRAH
+
+Area Forecast Discussion
+National Weather Service Raleigh NC
+1235 PM EDT Sun Aug 2 2026
+
+.KEY MESSAGES...
+As of 1235 PM Sunday...
+
+1) Scattered storms may produce locally heavy rain this afternoon.
+2) Drier weather arrives Monday.
+
+&&
+
+.DISCUSSION...
+Discussion details remain boundary-only content.
+
+$$
+
+WFO RAH
+    </pre>
+  </body>
+</html>