Generalize NWS forecast discussion heading parsing

2026-08-02 22:53:00 +00:00
parent 29c65971eb
commit 2b19a121fa
4 changed files with 618 additions and 331 deletions
--- a/internal/providers/nws/forecast_discussion.go
+++ b/internal/providers/nws/forecast_discussion.go
@@ -27,13 +27,12 @@ type ForecastDiscussionSection struct {
 	Text      string
 }

-const (
-	forecastDiscussionSectionKeyMessages = "KEY MESSAGES"
-	forecastDiscussionSectionShortTerm   = "SHORT TERM"
-	forecastDiscussionSectionLongTerm    = "LONG TERM"
-	forecastDiscussionSectionAviation    = "AVIATION"
+type forecastDiscussionSectionRole uint8

-	forecastDiscussionSectionIdentityPattern = "KEY MESSAGES|SHORT TERM|LONG TERM|AVIATION"
+const (
+	forecastDiscussionSectionRoleKeyMessages forecastDiscussionSectionRole = iota
+	forecastDiscussionSectionRoleShortTerm
+	forecastDiscussionSectionRoleLongTerm
 )

 type forecastDiscussionSectionHeading struct {
@@ -47,11 +46,14 @@ type forecastDiscussionSectionBlock struct {
 }

 var (
-	forecastDiscussionEllipsisHeadingRE = regexp.MustCompile(`^\.(` + forecastDiscussionSectionIdentityPattern + `)\.\.\.(.*)$`)
-	forecastDiscussionSlashHeadingRE    = regexp.MustCompile(`^\.(` + forecastDiscussionSectionIdentityPattern + `)[ \t]+/([^/]+)/\.\.\.$`)
-	forecastDiscussionAFDRE             = regexp.MustCompile(`^AFD([A-Z]{3})$`)
-	forecastDiscussionWMORE             = regexp.MustCompile(`\bK([A-Z]{3})\b`)
-	forecastDiscussionSigRE             = regexp.MustCompile(`^[A-Z]{2,6}$`)
+	forecastDiscussionSectionRoles = map[string]forecastDiscussionSectionRole{
+		"KEY MESSAGES": forecastDiscussionSectionRoleKeyMessages,
+		"SHORT TERM":   forecastDiscussionSectionRoleShortTerm,
+		"LONG TERM":    forecastDiscussionSectionRoleLongTerm,
+	}
+	forecastDiscussionAFDRE = regexp.MustCompile(`^AFD([A-Z]{3})$`)
+	forecastDiscussionWMORE = regexp.MustCompile(`\bK([A-Z]{3})\b`)
+	forecastDiscussionSigRE = regexp.MustCompile(`^[A-Z]{2,6}$`)
 )

 func ParseForecastDiscussionHTML(raw string) (ForecastDiscussion, error) {
@@ -119,20 +121,20 @@ func ParseForecastDiscussionText(text string) (ForecastDiscussion, error) {
 		IssuedAt:   issuedAt.UTC(),
 	}

-	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionKeyMessages); ok {
+	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionForRole(forecastDiscussionSectionRoleKeyMessages)); ok {
 		out.KeyMessages = parseForecastDiscussionKeyMessages(block.body)
 	}
-	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionShortTerm); ok {
+	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionForRole(forecastDiscussionSectionRoleShortTerm)); ok {
 		section, err := parseForecastDiscussionTextSection(block)
 		if err != nil {
-			return ForecastDiscussion{}, fmt.Errorf("parse %s: %w", forecastDiscussionSectionShortTerm, err)
+			return ForecastDiscussion{}, fmt.Errorf("parse %s: %w", block.heading.section, err)
 		}
 		out.ShortTerm = &section
 	}
-	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionLongTerm); ok {
+	if block, ok := extractForecastDiscussionSection(lines, forecastDiscussionSectionForRole(forecastDiscussionSectionRoleLongTerm)); ok {
 		section, err := parseForecastDiscussionTextSection(block)
 		if err != nil {
-			return ForecastDiscussion{}, fmt.Errorf("parse %s: %w", forecastDiscussionSectionLongTerm, err)
+			return ForecastDiscussion{}, fmt.Errorf("parse %s: %w", block.heading.section, err)
 		}
 		out.LongTerm = &section
 	}
@@ -408,23 +410,103 @@ func forecastDiscussionLocation(abbrev string) (*time.Location, error) {

 func parseForecastDiscussionSectionHeading(line string) (forecastDiscussionSectionHeading, bool) {
 	line = strings.TrimSpace(line)
-	if m := forecastDiscussionEllipsisHeadingRE.FindStringSubmatch(line); len(m) == 3 {
-		return forecastDiscussionSectionHeading{
-			section:   m[1],
-			qualifier: strings.TrimSpace(m[2]),
-		}, true
+	if len(line) < 2 || line[0] != '.' {
+		return forecastDiscussionSectionHeading{}, false
 	}
-	if m := forecastDiscussionSlashHeadingRE.FindStringSubmatch(line); len(m) == 3 {
-		qualifier := strings.TrimSpace(m[2])
-		if qualifier == "" {
-			return forecastDiscussionSectionHeading{}, false
+
+	if strings.HasSuffix(line, "/...") {
+		return parseForecastDiscussionSlashQualifiedHeading(line)
+	}
+	return parseForecastDiscussionEllipsisHeading(line)
+}
+
+func parseForecastDiscussionSlashQualifiedHeading(line string) (forecastDiscussionSectionHeading, bool) {
+	content := strings.TrimSuffix(line[1:], "/...")
+	separator := -1
+	for i := 1; i < len(content); i++ {
+		if content[i] == '/' && isForecastDiscussionHorizontalWhitespace(content[i-1]) {
+			separator = i
+			break
 		}
-		return forecastDiscussionSectionHeading{
-			section:   m[1],
-			qualifier: qualifier,
-		}, true
 	}
-	return forecastDiscussionSectionHeading{}, false
+	if separator < 0 {
+		return forecastDiscussionSectionHeading{}, false
+	}
+
+	section, ok := normalizeForecastDiscussionSectionIdentity(content[:separator])
+	if !ok {
+		return forecastDiscussionSectionHeading{}, false
+	}
+	qualifier := strings.TrimSpace(content[separator+1:])
+	if qualifier == "" {
+		return forecastDiscussionSectionHeading{}, false
+	}
+	return forecastDiscussionSectionHeading{section: section, qualifier: qualifier}, true
+}
+
+func parseForecastDiscussionEllipsisHeading(line string) (forecastDiscussionSectionHeading, bool) {
+	content := line[1:]
+	delimiter := strings.Index(content, "...")
+	if delimiter < 0 {
+		return forecastDiscussionSectionHeading{}, false
+	}
+
+	section, ok := normalizeForecastDiscussionSectionIdentity(content[:delimiter])
+	if !ok {
+		return forecastDiscussionSectionHeading{}, false
+	}
+	return forecastDiscussionSectionHeading{
+		section:   section,
+		qualifier: strings.TrimSpace(content[delimiter+3:]),
+	}, true
+}
+
+func normalizeForecastDiscussionSectionIdentity(raw string) (string, bool) {
+	var normalized strings.Builder
+	pendingSpace := false
+	hasLetterOrDigit := false
+
+	for i := 0; i < len(raw); i++ {
+		b := raw[i]
+		switch {
+		case isForecastDiscussionIdentityLetterOrDigit(b):
+			hasLetterOrDigit = true
+		case b == ' ' || b == '\t':
+			pendingSpace = normalized.Len() > 0
+			continue
+		case b == '/' && i > 0 && isForecastDiscussionHorizontalWhitespace(raw[i-1]):
+			return "", false
+		case b != '/' && b != '&' && b != '\'' && b != '-':
+			return "", false
+		}
+
+		if pendingSpace {
+			normalized.WriteByte(' ')
+			pendingSpace = false
+		}
+		normalized.WriteByte(b)
+	}
+	if !hasLetterOrDigit {
+		return "", false
+	}
+	return normalized.String(), true
+}
+
+func isForecastDiscussionIdentityLetterOrDigit(b byte) bool {
+	return b >= 'A' && b <= 'Z' || b >= '0' && b <= '9'
+}
+
+func isForecastDiscussionHorizontalWhitespace(b byte) bool {
+	return b == ' ' || b == '\t'
+}
+
+func forecastDiscussionSectionForRole(role forecastDiscussionSectionRole) string {
+	for section, registeredRole := range forecastDiscussionSectionRoles {
+		if registeredRole == role {
+			return section
+		}
+	}
+	return ""
 }

 func extractForecastDiscussionSection(lines []string, section string) (forecastDiscussionSectionBlock, bool) {