Improve submenu handling and linkless fixture parsing

internal/site/league_parser.go

@@ -161,43 +161,93 @@ func parseFixturesTable(table *goquery.Selection) []Fixture {
 	fixtures := make([]Fixture, 0, 16)
 
 	table.Find("tr").Each(func(_ int, row *goquery.Selection) {
-		matchLinks := row.Find("a[href*='mecz.php']")
-		if matchLinks.Length() != 1 {
+		fixture, ok := parseFixtureRow(row)
+		if !ok {
 			return
 		}
 
+		fixtures = append(fixtures, fixture)
+	})
+
+	return fixtures
+}
+
+func parseFixtureRow(row *goquery.Selection) (Fixture, bool) {
+	tds := row.Find("td")
+	if tds.Length() < 3 {
+		return Fixture{}, false
+	}
+
+	scoreCell, scoreIdx, matchLink, ok := fixtureScoreCell(row, tds)
+	if !ok {
+		return Fixture{}, false
+	}
+
+	home, _ := nearestTeamCellText(tds, scoreIdx-1, -1)
+	away, awayIdx := nearestTeamCellText(tds, scoreIdx+1, 1)
+	if home == "" || away == "" || isScoreLikeText(home) || isScoreLikeText(away) {
+		return Fixture{}, false
+	}
+
+	score := normalizeWhitespace(scoreCell.Text())
+	if !isFixtureScoreText(score) {
+		return Fixture{}, false
+	}
+
+	return Fixture{
+		Home:     home,
+		Away:     away,
+		Score:    score,
+		WhenInfo: joinNonEmptyCells(tds, awayIdx+1),
+		MatchURL: matchLink,
+		MatchID:  extractMatchID(matchLink),
+	}, true
+}
+
+func fixtureScoreCell(row *goquery.Selection, tds *goquery.Selection) (*goquery.Selection, int, string, bool) {
+	matchLinks := row.Find("a[href*='mecz.php']")
+	if matchLinks.Length() > 1 {
+		return nil, -1, "", false
+	}
+	if matchLinks.Length() == 1 {
 		scoreCell := matchLinks.First().Closest("td")
 		if scoreCell.Length() == 0 {
-			return
+			return nil, -1, "", false
 		}
 
-		tds := row.Find("td")
-		scoreIdx := scoreCell.Index()
-		home, _ := nearestTeamCellText(tds, scoreIdx-1, -1)
-		away, awayIdx := nearestTeamCellText(tds, scoreIdx+1, 1)
-		if home == "" || away == "" || isScoreLikeText(home) || isScoreLikeText(away) {
-			return
+		matchLink := strings.TrimSpace(matchLinks.First().AttrOr("href", ""))
+		if matchLink == "" {
+			return nil, -1, "", false
 		}
 
-		matchLink := strings.TrimSpace(scoreCell.Find("a[href*='mecz.php']").First().AttrOr("href", ""))
-		score := normalizeWhitespace(scoreCell.Text())
-		whenInfo := joinNonEmptyCells(tds, awayIdx+1)
+		return scoreCell, scoreCell.Index(), matchLink, true
+	}
 
-		if score == "" || matchLink == "" {
-			return
+	scoreIdx := -1
+	for idx := 0; idx < tds.Length(); idx++ {
+		if !isFixtureScoreText(tds.Eq(idx).Text()) {
+			continue
 		}
+		home, _ := nearestTeamCellText(tds, idx-1, -1)
+		away, _ := nearestTeamCellText(tds, idx+1, 1)
+		if home == "" || away == "" || isScoreLikeText(home) || isScoreLikeText(away) {
+			continue
+		}
+		if scoreIdx >= 0 {
+			return nil, -1, "", false
+		}
+		scoreIdx = idx
+	}
+	if scoreIdx < 0 {
+		return nil, -1, "", false
+	}
 
-		fixtures = append(fixtures, Fixture{
-			Home:     home,
-			Away:     away,
-			Score:    score,
-			WhenInfo: whenInfo,
-			MatchURL: matchLink,
-			MatchID:  extractMatchID(matchLink),
-		})
-	})
+	return tds.Eq(scoreIdx), scoreIdx, "", true
+}
 
-	return fixtures
+func isFixtureScoreText(text string) bool {
+	cleaned := normalizeWhitespace(text)
+	return cleaned == "-" || isScoreLikeText(cleaned)
 }
 
 func parseStandings(doc *goquery.Document) []StandingRow {

internal/site/models.go

@@ -13,6 +13,12 @@ type Competition struct {
 	LeagueKey string
 }
 
+type CompetitionMenu struct {
+	Title string
+	URL   string
+	Items []Competition
+}
+
 type Fixture struct {
 	Home     string
 	Away     string

internal/site/parser_archive_test.go

@@ -83,17 +83,29 @@ func TestParseSeasonsAndCompetitionsFromArchiveFixtures(t *testing.T) {
 				ekstraklasaURL := c.Resolve("/liga/1/liga11233.html")
 				iLigaURL := c.Resolve("/liga/1/liga11234.html")
 				iiLigaURL := c.Resolve("/liga/1/liga11235.html")
+				iiiLigaSelectorURL := c.Resolve("/ligireg.php?poziom=4&id_sezon=97")
+				regionalneURL := c.Resolve("/ligireg.php?id_sezon=97")
+				regionalCupsURL := c.Resolve("/polcups.php?id_sezon=97")
 
 				ekstraklasaIdx := competitionIndexByURL(competitions, ekstraklasaURL)
 				iLigaIdx := competitionIndexByURL(competitions, iLigaURL)
 				iiLigaIdx := competitionIndexByURL(competitions, iiLigaURL)
+				iiiLigaSelectorIdx := competitionIndexByURL(competitions, iiiLigaSelectorURL)
+				regionalneIdx := competitionIndexByURL(competitions, regionalneURL)
+				regionalCupsIdx := competitionIndexByURL(competitions, regionalCupsURL)
 
 				if ekstraklasaIdx < 0 || iLigaIdx < 0 || iiLigaIdx < 0 {
 					t.Fatalf("missing expected league links in 2020/21 archive")
 				}
 				if !(ekstraklasaIdx < iLigaIdx && iLigaIdx < iiLigaIdx) {
 					t.Fatalf("competition order broken: Ekstraklasa=%d I liga=%d II liga=%d", ekstraklasaIdx, iLigaIdx, iiLigaIdx)
 				}
+				if iiiLigaSelectorIdx < 0 || regionalneIdx < 0 {
+					t.Fatalf("missing III liga or ligi regionalne links in 2020/21 archive")
+				}
+				if regionalCupsIdx < 0 {
+					t.Fatalf("missing regional cups link in 2020/21 archive")
+				}
 			}
 
 			for _, season := range seasons {
@@ -117,3 +129,86 @@ func TestParseSeasonsAndCompetitionsFromArchiveFixtures(t *testing.T) {
 		t.Fatalf("expected at least one decoded Polish diacritic in archive fixtures")
 	}
 }
+
+func TestParseCompetitionMenuForIIILigaSelector(t *testing.T) {
+	html := `<html><body><table class="main"><tr><td valign="top"><p align="center"><b>III liga 2025/26</b></p><table class="main"><tr><td width="100"></td><td><a href="/liga/1/liga14154.html" class="main">I</a></td></tr><tr><td width="100"></td><td><a href="/liga/1/liga14155.html" class="main">II</a></td></tr></table></td></tr></table></body></html>`
+	doc, err := decodeAndParse([]byte(html), "text/html; charset=utf-8")
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	menu := parseCompetitionMenu(doc, "http://www.90minut.pl/ligireg.php?poziom=4&id_sezon=107", NewClient())
+	if menu == nil {
+		t.Fatalf("expected III liga submenu")
+	}
+	if menu.Title != "III liga 2025/26" {
+		t.Fatalf("unexpected menu title: %q", menu.Title)
+	}
+	if len(menu.Items) != 2 || menu.Items[0].Name != "III liga 2025/26, gr. I" || menu.Items[1].Name != "III liga 2025/26, gr. II" {
+		t.Fatalf("unexpected III liga submenu items: %+v", menu.Items)
+	}
+}
+
+func TestParseCompetitionMenuForRegionalRoot(t *testing.T) {
+	html := `<html><body><table class="main"><tr><td valign="top"><p align="center"><b>Ligi regionalne 2025/26</b></p><a href="/ligireg-16.html" class="main">Dolnośląski ZPN</a><a href="/mecze_okreg.php?id_okreg=16" class="main">Dziś grają</a><a href="/ligireg-1.html" class="main">Kujawsko-Pomorski ZPN</a></td></tr></table></body></html>`
+	doc, err := decodeAndParse([]byte(html), "text/html; charset=utf-8")
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	menu := parseCompetitionMenu(doc, "http://www.90minut.pl/ligireg.php?id_sezon=107", NewClient())
+	if menu == nil {
+		t.Fatalf("expected regional submenu")
+	}
+	if len(menu.Items) != 2 || menu.Items[0].Name != "Dolnośląski ZPN" || menu.Items[1].Name != "Kujawsko-Pomorski ZPN" {
+		t.Fatalf("unexpected regional submenu items: %+v", menu.Items)
+	}
+}
+
+func TestParseCompetitionMenuForRegionalRootWithAssociationQueryLinks(t *testing.T) {
+	html := `<html><body><table class="main"><tr><td valign="top"><p align="center"><b>Ligi regionalne 2025/26</b></p><a href="/ligireg.php?id_okreg=16&id_sezon=107" class="main">Dolnośląski ZPN</a><a href="/ligireg.php?id_okreg=8&id_sezon=107" class="main">Lubuski ZPN</a></td></tr></table></body></html>`
+	doc, err := decodeAndParse([]byte(html), "text/html; charset=utf-8")
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	menu := parseCompetitionMenu(doc, "http://www.90minut.pl/ligireg.php?id_sezon=107", NewClient())
+	if menu == nil {
+		t.Fatalf("expected regional submenu")
+	}
+	if len(menu.Items) != 2 {
+		t.Fatalf("unexpected regional submenu items: %+v", menu.Items)
+	}
+}
+
+func TestParseCompetitionMenuForRegionalAssociationPage(t *testing.T) {
+	html := `<html><body><table class="main"><tr><td valign="top"><p align="center"><b>Ligi regionalne 2025/26 - Dolnośląski ZPN</b></p><a href="/liga/1/liga14169.html" class="main">IV liga 2025/2026, grupa: dolnośląska</a><a href="/liga/1/liga14204.html" class="main">Klasa okręgowa 2025/2026, grupa: Jelenia Góra</a></td></tr></table></body></html>`
+	doc, err := decodeAndParse([]byte(html), "text/html; charset=utf-8")
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	menu := parseCompetitionMenu(doc, "http://www.90minut.pl/ligireg-16.html", NewClient())
+	if menu == nil {
+		t.Fatalf("expected regional association submenu")
+	}
+	if len(menu.Items) != 2 {
+		t.Fatalf("unexpected regional association item count: %d", len(menu.Items))
+	}
+}
+
+func TestParseCompetitionMenuForRegionalCupsPage(t *testing.T) {
+	html := `<html><body><table class="main"><tr><td valign="top"><p align="center"><b>Puchary krajowe 2025/26</b></p><a href="/liga/1/liga14076.html" class="main">Puchar Polski</a><a href="/liga/1/liga14075.html" class="main">Superpuchar Polski</a><a href="/liga/1/liga14636.html" class="main">Puchar Polski 2025/2026, grupa: Lubuski ZPN</a><a href="/liga/1/liga14069.html" class="main">Puchar Polski 2025/2026, grupa: Lubuski ZPN - Gorzów Wielkopolski</a></td></tr></table></body></html>`
+	doc, err := decodeAndParse([]byte(html), "text/html; charset=utf-8")
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	menu := parseCompetitionMenu(doc, "http://www.90minut.pl/polcups.php?id_sezon=107", NewClient())
+	if menu == nil {
+		t.Fatalf("expected regional cups submenu")
+	}
+	if len(menu.Items) != 2 {
+		t.Fatalf("expected regional cups only, got %+v", menu.Items)
+	}
+}

internal/site/parser_league_test.go

@@ -17,15 +17,18 @@ func TestParseLeagueFixturesFromCorpus(t *testing.T) {
 		"league_14073": {firstRoundName: "Kolejka 1 - 19-20 lipca", firstRoundFixtures: 9},
 	}
 	leagues := fixturesByKind(m, "league")
-	if len(leagues) < 6 {
-		t.Fatalf("expected at least 6 league fixtures, got %d", len(leagues))
+	if len(leagues) < 7 {
+		t.Fatalf("expected at least 7 league fixtures, got %d", len(leagues))
 	}
 	if !containsFixtureName(leagues, "league_14072") {
 		t.Fatalf("expected league fixture for liga14072")
 	}
 	if !containsFixtureName(leagues, "league_14073") {
 		t.Fatalf("expected league fixture for liga14073")
 	}
+	if !containsFixtureName(leagues, "league_14141") {
+		t.Fatalf("expected league fixture for liga14141")
+	}
 
 	for _, fixture := range leagues {
 		fixture := fixture
@@ -61,6 +64,12 @@ func TestParseLeagueFixturesFromCorpus(t *testing.T) {
 					if isScoreLikeText(match.Home) || isScoreLikeText(match.Away) {
 						t.Fatalf("fixture side parsed as score token in %s: home=%q away=%q", fixture.Name, match.Home, match.Away)
 					}
+					if match.MatchURL == "" {
+						if match.MatchID != "" {
+							t.Fatalf("fixture without match url should keep empty match id in %s: %q", fixture.Name, match.MatchID)
+						}
+						continue
+					}
 					if !strings.Contains(match.MatchURL, "mecz.php") {
 						t.Fatalf("fixture match url is not a match link in %s: %q", fixture.Name, match.MatchURL)
 					}
@@ -102,6 +111,37 @@ func TestParseLeagueFixturesFromCorpus(t *testing.T) {
 	}
 }
 
+func TestParseLeagueFixturesWithoutMatchLinksFromCorpus(t *testing.T) {
+	doc, _ := fixtureDoc(t, "fixtures/league_14141.html")
+	page := parseLeaguePage(doc, "http://www.90minut.pl/liga/1/liga14141.html")
+	if page == nil {
+		t.Fatalf("expected league page")
+	}
+	if len(page.Rounds) == 0 {
+		t.Fatalf("expected rounds for linkless fixture league")
+	}
+
+	linklessFixtures := 0
+	for _, round := range page.Rounds {
+		for _, fixture := range round.Fixtures {
+			if fixture.MatchURL != "" {
+				continue
+			}
+			linklessFixtures++
+			if fixture.MatchID != "" {
+				t.Fatalf("expected empty match id for linkless fixture, got %q", fixture.MatchID)
+			}
+			if fixture.Home == "" || fixture.Away == "" || fixture.Score == "" {
+				t.Fatalf("expected sides and score for linkless fixture, got %+v", fixture)
+			}
+		}
+	}
+
+	if linklessFixtures == 0 {
+		t.Fatalf("expected at least one linkless fixture in liga14141")
+	}
+}
+
 func assertFixturesSortedByDate(t *testing.T, fixtureName string, round Round) {
 	t.Helper()
 

internal/site/parser_regression_test.go

@@ -192,6 +192,60 @@ func TestParseFixturesTableSkipsRowsWithMultipleMatchLinks(t *testing.T) {
 	}
 }
 
+func TestParseFixturesTableFallsBackToPlainTextScoresWithoutMatchLinks(t *testing.T) {
+	html := `
+	<table>
+	<tr><td>Slask Wroclaw</td><td>-</td><td>Pogon Tczew</td><td>16 maja, 12:00</td></tr>
+	<tr><td colspan="4">w pierwotnym terminie odwolany</td></tr>
+	<tr><td><b>Gornik Leczna</b></td><td><b>3-0</b></td><td><b>UKS SMS Lodz</b></td><td>25 marca, 16:00</td></tr>
+	</table>`
+
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		t.Fatalf("parse synthetic HTML: %v", err)
+	}
+
+	fixtures := parseFixturesTable(doc.Find("table").First())
+	if len(fixtures) != 2 {
+		t.Fatalf("expected 2 fixtures, got %d", len(fixtures))
+	}
+	if fixtures[0].Home != "Slask Wroclaw" || fixtures[0].Away != "Pogon Tczew" || fixtures[0].Score != "-" || fixtures[0].WhenInfo != "16 maja, 12:00" {
+		t.Fatalf("unexpected plain-text fixture: %+v", fixtures[0])
+	}
+	if fixtures[0].MatchURL != "" || fixtures[0].MatchID != "" {
+		t.Fatalf("expected empty match details for plain-text score fixture: %+v", fixtures[0])
+	}
+	if fixtures[1].Home != "Gornik Leczna" || fixtures[1].Away != "UKS SMS Lodz" || fixtures[1].Score != "3-0" {
+		t.Fatalf("unexpected second plain-text fixture: %+v", fixtures[1])
+	}
+}
+
+func TestParseLeaguePageHandlesSavedAmbiguousLinklessFixture(t *testing.T) {
+	doc, _ := fixtureDoc(t, "fixtures/league_ambiguous_linkless.html")
+
+	page := parseLeaguePage(doc, "http://www.90minut.pl/liga/1/liga99998.html")
+	if page == nil {
+		t.Fatalf("expected league page")
+	}
+	if len(page.Rounds) != 1 {
+		t.Fatalf("expected 1 round, got %d", len(page.Rounds))
+	}
+	if len(page.Rounds[0].Fixtures) != 2 {
+		t.Fatalf("expected 2 fixtures, got %d", len(page.Rounds[0].Fixtures))
+	}
+
+	first := page.Rounds[0].Fixtures[0]
+	if first.Home != "Team A" || first.Away != "Team B" || first.Score != "1-0" {
+		t.Fatalf("unexpected first fixture: %+v", first)
+	}
+	if first.WhenInfo != "walkower 3-0 24 lipca, 18:00" {
+		t.Fatalf("unexpected first fixture metadata: %+v", first)
+	}
+	if first.MatchURL != "" || first.MatchID != "" {
+		t.Fatalf("expected linkless fixture, got %+v", first)
+	}
+}
+
 func TestRoundNameFromTableSkipsNavigationBlocks(t *testing.T) {
 	html := `
 	<table>