Harden NWS key message parsing

2026-08-03 00:19:48 +00:00
parent 9a09454621
commit 6943a5ebc9
2 changed files with 259 additions and 5 deletions
--- a/internal/providers/nws/forecast_discussion_test.go
+++ b/internal/providers/nws/forecast_discussion_test.go
@@ -740,6 +740,178 @@ func TestParseForecastDiscussionTextSectionKeepsArbitraryDashedProse(t *testing.
 	}
 }

+func TestIsForecastDiscussionKeyMessageMetadataLine(t *testing.T) {
+	tests := []struct {
+		name string
+		line string
+		want bool
+	}{
+		{
+			name: "issued at",
+			line: "Issued at 300 PM CDT Sat Mar 28 2026",
+			want: true,
+		},
+		{
+			name: "updated at ASCII case insensitive",
+			line: "uPdAtEd At 300 PM CDT Sat Mar 28 2026",
+			want: true,
+		},
+		{
+			name: "label without whitespace boundary",
+			line: "Updated at300 PM CDT Sat Mar 28 2026",
+			want: false,
+		},
+		{
+			name: "invalid timestamp",
+			line: "Updated at not a timestamp",
+			want: false,
+		},
+		{
+			name: "nonmetadata prose",
+			line: "Updated atmospheric conditions remain unsettled.",
+			want: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := isForecastDiscussionKeyMessageMetadataLine(tt.line); got != tt.want {
+				t.Fatalf("isForecastDiscussionKeyMessageMetadataLine(%q) = %t, want %t", tt.line, got, tt.want)
+			}
+		})
+	}
+}
+
+func TestStripForecastDiscussionKeyMessageMarker(t *testing.T) {
+	tests := []struct {
+		name string
+		line string
+		want string
+		ok   bool
+	}{
+		{name: "hyphen", line: "- First message.", want: "First message.", ok: true},
+		{name: "hyphen without whitespace", line: "-First message.", want: "First message.", ok: true},
+		{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: "composite hyphen", line: "- 1. First message.", want: "First message.", ok: true},
+		{name: "composite asterisk", 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: "overflow numeric marker", line: "999999999999999999999999999999. Too big.", 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},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got, ok := stripForecastDiscussionKeyMessageMarker(tt.line)
+			if ok != tt.ok || got != tt.want {
+				t.Fatalf("stripForecastDiscussionKeyMessageMarker(%q) = (%q, %t), want (%q, %t)", tt.line, got, ok, tt.want, tt.ok)
+			}
+		})
+	}
+}
+
+func TestParseForecastDiscussionKeyMessagesClassifiesListsAndMetadata(t *testing.T) {
+	tests := []struct {
+		name string
+		body []string
+		want []string
+	}{
+		{
+			name: "numbered list with continuation",
+			body: []string{
+				"1) First numbered message.",
+				"Wrapped continuation.",
+				"2. Second numbered message.",
+			},
+			want: []string{"First numbered message. Wrapped continuation.", "Second numbered message."},
+		},
+		{
+			name: "asterisk list",
+			body: []string{
+				"* First asterisk message.",
+				"*Second asterisk message.",
+			},
+			want: []string{"First asterisk message.", "Second asterisk message."},
+		},
+		{
+			name: "unmarked paragraphs",
+			body: []string{
+				"First paragraph.",
+				"Wrapped continuation.",
+				"",
+				"",
+				"Second paragraph.",
+			},
+			want: []string{"First paragraph. Wrapped continuation.", "Second paragraph."},
+		},
+		{
+			name: "introductory prose before marked items",
+			body: []string{
+				"Introductory prose.",
+				"",
+				"- First marked message.",
+				"Wrapped continuation.",
+				"- Second marked message.",
+			},
+			want: []string{"Introductory prose.", "First marked message. Wrapped continuation.", "Second marked message."},
+		},
+		{
+			name: "leading issued metadata",
+			body: []string{
+				"iSsUeD At 300 PM CDT Sat Mar 28 2026",
+				"- First message.",
+			},
+			want: []string{"First message."},
+		},
+		{
+			name: "leading updated metadata",
+			body: []string{
+				"Updated at 300 PM CDT Sat Mar 28 2026",
+				"- First message.",
+			},
+			want: []string{"First message."},
+		},
+		{
+			name: "updated atmospheric prose remains content",
+			body: []string{
+				"Updated atmospheric conditions remain unsettled.",
+				"Additional detail.",
+			},
+			want: []string{"Updated atmospheric conditions remain unsettled. Additional detail."},
+		},
+		{
+			name: "invalid updated metadata remains content",
+			body: []string{
+				"Updated at not a timestamp",
+				"Additional detail.",
+			},
+			want: []string{"Updated at not a timestamp Additional detail."},
+		},
+		{
+			name: "later valid metadata remains content",
+			body: []string{
+				"- First message.",
+				"Updated at 300 PM CDT Sat Mar 28 2026",
+			},
+			want: []string{"First message. Updated at 300 PM CDT Sat Mar 28 2026"},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := parseForecastDiscussionKeyMessages(tt.body); !reflect.DeepEqual(got, tt.want) {
+				t.Fatalf("KeyMessages = %#v, want %#v", got, tt.want)
+			}
+		})
+	}
+}
+
 func TestParseForecastDiscussionHTMLParsesCrossOfficeLayout(t *testing.T) {
 	got, err := ParseForecastDiscussionHTML(loadForecastDiscussionBOUSampleHTML(t))
 	if err != nil {