Fix Circa strikeouts table parsing

src/market-scanner.js

@@ -758,6 +758,31 @@ function normalizeCircaHalfLine(value) {
   return Number.isFinite(numeric) ? numeric : null;
 }
 
+function extractCircaTableLineValue(segment) {
+  const normalized = normalizeWhitespace(segment)
+    .toUpperCase()
+    .replace(/I/g, '1')
+    .replace(/O/g, '0');
+
+  const matches = [...normalized.matchAll(/(\d+)([^0-9\s]{0,3})?(?=\s+[+\-~]?[0-9BOSI]{2,4}\b)/g)];
+  if (matches.length === 0) {
+    return null;
+  }
+
+  const match = matches[matches.length - 1];
+  const whole = Number(match[1]);
+  if (!Number.isFinite(whole)) {
+    return null;
+  }
+
+  const suffix = String(match[2] ?? '');
+  if (suffix && suffix !== '.0' && suffix !== '.00') {
+    return whole + 0.5;
+  }
+
+  return whole;
+}
+
 function isReasonableCircaOdds(oddsInput, section) {
   if (!oddsInput) {
     return false;
@@ -1278,6 +1303,93 @@ function extractCircaOverUnderTableEntries(segment, section, state) {
   return [];
 }
 
+function extractCircaOverUnderTableEntriesV2(segment, section, state) {
+  const normalizedSegment = normalizeWhitespace(segment)
+    .toUpperCase()
+    .replace(/[~]/g, '-')
+    .replace(/\s+/g, ' ');
+  const playerMatch = normalizedSegment.match(/([A-Z][A-Z.'\- ]{2,40}\([A-Z]{2,3}\))/);
+  const lineValue = extractCircaTableLineValue(normalizedSegment);
+  const oddsTokens = extractOddsTokensFromSegment(normalizedSegment);
+
+  const flushPending = () => {
+    if (!state.pendingPlayer || !state.pendingTeam || !Number.isFinite(state.pendingLineValue)) {
+      state.pendingPlayer = null;
+      state.pendingTeam = null;
+      state.pendingLineValue = null;
+      state.pendingOverOdds = null;
+      state.pendingUnderOdds = null;
+      return [];
+    }
+
+    const built = [];
+    for (const sideEntry of [
+      { side: 'over', oddsInput: state.pendingOverOdds },
+      { side: 'under', oddsInput: state.pendingUnderOdds ?? null },
+    ]) {
+      if (!sideEntry.oddsInput || !isReasonableCircaOdds(sideEntry.oddsInput, { mode: 'over_under' })) {
+        continue;
+      }
+
+      const impliedProbability = americanToImpliedProbability(sideEntry.oddsInput);
+      if (impliedProbability === null) {
+        continue;
+      }
+
+      const entry = {
+        source: 'circa',
+        book: 'Circa',
+        eventName: 'Circa MLB',
+        eventId: null,
+        team: state.pendingTeam,
+        playerName: state.pendingPlayer,
+        playerKey: normalizePlayerName(state.pendingPlayer),
+        marketType: section.type,
+        marketLabel: section.label,
+        side: sideEntry.side,
+        lineValue: state.pendingLineValue,
+        oddsInput: sideEntry.oddsInput,
+        impliedProbability,
+        rawLabel: normalizedSegment,
+      };
+      entry.marketKey = buildMarketKey(entry);
+      built.push(entry);
+    }
+
+    state.pendingPlayer = null;
+    state.pendingTeam = null;
+    state.pendingLineValue = null;
+    state.pendingOverOdds = null;
+    state.pendingUnderOdds = null;
+    return built;
+  };
+
+  if (!normalizedSegment) {
+    return flushPending();
+  }
+
+  if (playerMatch) {
+    const flushed = flushPending();
+    const { playerName, team } = extractPlayerAndTeam(playerMatch[1]);
+    state.pendingPlayer = playerName;
+    state.pendingTeam = team;
+    state.pendingLineValue = lineValue;
+    state.pendingOverOdds = oddsTokens[0] ? normalizeOcrOddsTokenForSide(oddsTokens[0], 'over') : null;
+    state.pendingUnderOdds = oddsTokens[1] ? normalizeOcrOddsTokenForSide(oddsTokens[1], 'under') : null;
+    if (state.pendingUnderOdds) {
+      return [...flushed, ...flushPending()];
+    }
+    return flushed;
+  }
+
+  if (state.pendingPlayer && Number.isFinite(lineValue) && oddsTokens.length > 0) {
+    state.pendingUnderOdds = normalizeOcrOddsTokenForSide(oddsTokens[0], 'under');
+    return flushPending();
+  }
+
+  return [];
+}
+
 export function parseCircaOcrText(text) {
   const entries = [];
   const rawLines = String(text ?? '')
@@ -1344,7 +1456,10 @@ export function parseCircaOcrText(text) {
     if (activeSections.length === 1) {
       const section = activeSections[0];
       for (const segment of segments) {
-        entries.push(...parseCircaSegmentWithSection(segment.text, section, overUnderStates, `single-${section.type}`));
+        const stateKey = section.mode === 'over_under_table'
+          ? `single-${section.type}-col-${segment.index}`
+          : `single-${section.type}`;
+        entries.push(...parseCircaSegmentWithSection(segment.text, section, overUnderStates, stateKey));
       }
       continue;
     }
@@ -1478,7 +1593,7 @@ function parseCircaSegmentWithSection(segment, section, overUnderStates, stateKe
       pendingUnderOdds: null,
     };
     existingState.section = section;
-    const parsed = extractCircaOverUnderTableEntries(segment, section, existingState);
+    const parsed = extractCircaOverUnderTableEntriesV2(segment, section, existingState);
     overUnderStates.set(stateKey, existingState);
     return parsed;
   }

test/market-scanner.test.js

@@ -209,6 +209,21 @@ test('parses table-style total bases and strikeouts sections', () => {
   assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'BUBBA CHANDLER' && entry.side === 'under' && entry.lineValue === 5.5 && entry.oddsInput === '-155'));
 });
 
+test('parses single-header two-column strikeouts table', () => {
+  const entries = parseCircaOcrText([
+    'Total Strikeouts',
+    '85001 GERMAN MARQUEZ (SD) 3½ +105      85029 LOGAN GILBERT (SEA) 6½ +105',
+    '3½ -125      6½ -125',
+    '85003 BUBBA CHANDLER (PIT) 5½ +135      85031 JACOB DEGROM (TEX) 6½ -145',
+    '5½ -155      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 === 'LOGAN GILBERT' && entry.side === 'under' && entry.lineValue === 6.5 && entry.oddsInput === '-125'));
+  assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'BUBBA CHANDLER' && entry.side === 'under' && entry.lineValue === 5.5 && entry.oddsInput === '-155'));
+  assert.ok(entries.some((entry) => entry.marketType === 'pitcher_strikeouts_generic' && entry.playerName === 'JACOB DEGROM' && entry.side === 'over' && entry.lineValue === 6.5 && entry.oddsInput === '-145'));
+});
+
 test('ranks discrepancy, width, and circa alerts', () => {
   const entries = [
     {