Handle additional NWS forecast discussion formats

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 {