Fix Circa strikeout supplemental parsing

src/market-scanner.js

@@ -1087,6 +1087,84 @@ function extractNoisyStrikeoutEntries(text, section) {
   return entries;
 }
 
+function extractStrikeoutLaneEntriesFromSnippet(text, section) {
+  const rawLines = String(text ?? '')
+    .replace(/\r\n/g, '\n')
+    .split('\n')
+    .map((line) => line.replace(/\s+$/g, ''));
+
+  const headerIndex = rawLines.findIndex((line) => /TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i.test(line));
+  if (headerIndex === -1) {
+    return [];
+  }
+
+  const headerSegments = splitCircaLayoutColumns(rawLines[headerIndex]);
+  const strikeoutLaneIndex = headerSegments.findIndex((segment) => /TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i.test(segment.text));
+  if (strikeoutLaneIndex === -1) {
+    return [];
+  }
+
+  const states = new Map();
+  const entries = [];
+
+  for (const rawLine of rawLines.slice(headerIndex + 1)) {
+    if (!rawLine.trim()) {
+      continue;
+    }
+
+    if (detectCircaSection(rawLine) && !/TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i.test(rawLine)) {
+      break;
+    }
+
+    const segments = splitCircaLayoutColumns(rawLine);
+    if (segments.length === 0) {
+      continue;
+    }
+
+    const preferredSegment = segments.find((segment) => segment.index === strikeoutLaneIndex)
+      ?? segments.at(-1);
+    if (!preferredSegment?.text) {
+      continue;
+    }
+
+    const stateKey = `strikeout-snippet-${preferredSegment.index}`;
+    const state = states.get(stateKey) ?? {
+      section,
+      pendingPlayer: null,
+      pendingTeam: null,
+      pendingLineValue: null,
+      pendingOverOdds: null,
+      pendingUnderOdds: null,
+    };
+    state.section = section;
+    entries.push(...extractCircaOverUnderTableEntriesV2(preferredSegment.text, section, state));
+    states.set(stateKey, state);
+  }
+
+  for (const [stateKey, state] of states.entries()) {
+    entries.push(...extractCircaOverUnderTableEntriesV2('', section, state));
+    states.set(stateKey, state);
+  }
+
+  return entries;
+}
+
+export function extractSupplementalStrikeoutEntries(text) {
+  const section = detectCircaSection('TOTAL STRIKEOUTS');
+  const snippet = extractCircaSectionSnippet(text, /TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i, 40);
+  if (!snippet || !section) {
+    return [];
+  }
+
+  return dedupeBy(
+    [
+      ...extractStrikeoutLaneEntriesFromSnippet(snippet, section),
+      ...extractNoisyStrikeoutEntries(snippet, section),
+    ].filter(isLikelyValidCircaEntry),
+    (entry) => `${entry.marketKey}|${entry.book}`,
+  );
+}
+
 function normalizeCombinedBinaryLaneOdds(oddsTokens, section) {
   if (!Array.isArray(oddsTokens) || oddsTokens.length < 2) {
     return { yesOdds: null, noOdds: null };
@@ -1696,12 +1774,8 @@ export function parseCircaOcrText(text) {
     }
   }
 
-  const hasStrikeouts = entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic');
-  if (!hasStrikeouts && /TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i.test(String(text ?? ''))) {
-    const strikeoutsSnippet = extractCircaSectionSnippet(text, /TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i, 8);
-    if (strikeoutsSnippet) {
-      entries.push(...extractNoisyStrikeoutEntries(strikeoutsSnippet, detectCircaSection('TOTAL STRIKEOUTS')));
-    }
+  if (/TOTAL STRIKEOUTS|PITCHER STRIKEOUTS/i.test(String(text ?? ''))) {
+    entries.push(...extractSupplementalStrikeoutEntries(text));
   }
 
   return dedupeBy(entries.filter(isLikelyValidCircaEntry), (entry) => `${entry.marketKey}|${entry.book}`);

test/market-scanner.test.js

@@ -6,6 +6,7 @@ import {
   buildMarketKey,
   extractCircaPdfFromArchiveBuffer,
   extractCircaFileCandidatesFromHtml,
+  extractSupplementalStrikeoutEntries,
   fetchOddsApiEntries,
   matchesCircaMlbFilename,
   normalizeOddsApiEntries,
@@ -252,6 +253,42 @@ test('recovers noisy live-style strikeout rows', () => {
   assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'LOGAN GILBERT' && entry.side === 'over' && entry.lineValue === 6.5));
 });
 
+test('extracts strikeout lane from mixed Team Totals header', () => {
+  const entries = extractSupplementalStrikeoutEntries([
+    'Team Totals      Total Strikeouts',
+    '86601 PADRES 3½ -130      85001 GERMAN MARQUEZ (SD) 3½ +105',
+    '86602 3½ +110      85002 3½ -125',
+    '86603 PIRATES 4½ +105      85003 BUBBA CHANDLER (PIT) 5½ +135',
+    '86604 4½ -125      85004 5½ -155',
+    '86611 NATIONALS 3½ -140      85027 JOE RYAN (MIL) 6½ +105',
+    '86612 3½ +120      85028 6½ -125',
+  ].join('\n'));
+
+  assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'GERMAN MARQUEZ' && entry.side === 'over' && entry.lineValue === 3.5 && entry.oddsInput === '+105'));
+  assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'BUBBA CHANDLER' && entry.side === 'over' && entry.lineValue === 5.5 && entry.oddsInput === '+135'));
+  assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'JOE RYAN' && entry.side === 'over' && entry.lineValue === 6.5 && entry.oddsInput === '+105'));
+});
+
+test('supplements strikeouts even when generic parsing already found some rows', () => {
+  const entries = parseCircaOcrText([
+    'Team Totals Total Strikeouts',
+    'asst 0 [86627 "10 {ssn ns [85029 PE',
+    'asé02 PADRES 34150 [sez TVINS 3% 110 [saonz GERMAN MARUQUEZ. (SD) 3% "15% | saay LOGAN GILBERT (SEA) 6%',
+    '86603 ios [86679 "ro [65003 "ras [85031 E',
+    'sens PIRATES 4% 110% [Seen MARINERS 3% 110 |3aops BUBBA CHANDLER (PIT) 5% 1 5037 2ACOB DEGRON (TEX) 6% 15%',
+    '8665 "io [86st "is [esos "an (85033 E',
+    'ss606 REDS 3% in [sess RANGERS 3% ie [gaone BRANDON WILLIAMSON (CIN) 4% *1i0 (270 JAMESON TAILLON (CHC) 341%',
+    'ee MARLINS 4% 1 [4 cups 3 11055007 JANSON JUNK. (IA) ® = SHANE NCCLANAHAN (TB) 5% 110',
+    '86611 NATIONALS 3½ -140      85027 JOE RYAN (MIL) 6½ +105',
+    '86612 3½ +120      85028 6½ -125',
+  ].join('\n'));
+
+  const strikeoutOvers = entries.filter((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.side === 'over');
+  assert.ok(strikeoutOvers.some((entry) => entry.playerName === 'JOE RYAN' && entry.lineValue === 6.5 && entry.oddsInput === '+105'));
+  assert.ok(strikeoutOvers.some((entry) => entry.playerName.includes('GERMAN') && entry.lineValue === 3.5));
+  assert.ok(strikeoutOvers.length >= 6);
+});
+
 test('ranks discrepancy, width, and circa alerts', () => {
   const entries = [
     {