Refine Circa packed-row odds extraction

src/market-scanner.js

@@ -508,6 +508,86 @@ function extractBinaryYesNoEntriesV2(line, section) {
   return entries;
 }
 
+function extractOddsTokensFromSegment(segment) {
+  const cleaned = normalizeWhitespace(segment)
+    .toUpperCase()
+    .replace(/\//g, ' ')
+    .replace(/\s+/g, ' ');
+  const tokens = [...cleaned.matchAll(/(?:\d{4,6}|[0-9BOSI]{4,6})\s+([+\-~]?[0-9BOSI]{2,4})/g)]
+    .map((match) => match[1]);
+
+  if (tokens.length > 0) {
+    return tokens;
+  }
+
+  return [...cleaned.matchAll(/([+\-~][0-9BOSI]{2,4})/g)].map((match) => match[1]);
+}
+
+function extractBinaryYesNoEntriesV3(line, section) {
+  const entries = [];
+  const normalizedLine = normalizeWhitespace(line)
+    .toUpperCase()
+    .replace(/\//g, ' ')
+    .replace(/\s+/g, ' ');
+  const playerPattern = /([A-Z][A-Z.'\- ]{2,40}\([A-Z]{2,3}\))/g;
+  const playerMatches = [...normalizedLine.matchAll(playerPattern)];
+
+  for (let index = 0; index < playerMatches.length; index += 1) {
+    const current = playerMatches[index];
+    const next = playerMatches[index + 1];
+    const segmentStart = current.index ?? 0;
+    const segmentEnd = next?.index ?? normalizedLine.length;
+    const segment = normalizedLine.slice(segmentStart, segmentEnd);
+    const { playerName, team } = extractPlayerAndTeam(current[1]);
+    if (!playerName || playerName.length < 3) {
+      continue;
+    }
+
+    const oddsTokens = extractOddsTokensFromSegment(segment);
+    if (oddsTokens.length < 2) {
+      continue;
+    }
+
+    const yesOdds = normalizeOcrOddsTokenForSide(oddsTokens[0], 'yes');
+    const noOdds = normalizeOcrOddsTokenForSide(oddsTokens[1], 'no');
+
+    for (const sideEntry of [
+      { side: section.side ?? 'yes', oddsInput: yesOdds },
+      { side: 'no', oddsInput: noOdds },
+    ]) {
+      if (!sideEntry.oddsInput || !isReasonableCircaOdds(sideEntry.oddsInput, section)) {
+        continue;
+      }
+
+      const impliedProbability = americanToImpliedProbability(sideEntry.oddsInput);
+      if (impliedProbability === null) {
+        continue;
+      }
+
+      const entry = {
+        source: 'circa',
+        book: 'Circa',
+        eventName: 'Circa MLB',
+        eventId: null,
+        team,
+        playerName,
+        playerKey: normalizePlayerName(playerName),
+        marketType: section.type,
+        marketLabel: section.label,
+        side: sideEntry.side,
+        lineValue: section.lineValue,
+        oddsInput: sideEntry.oddsInput,
+        impliedProbability,
+        rawLabel: segment,
+      };
+      entry.marketKey = buildMarketKey(entry);
+      entries.push(entry);
+    }
+  }
+
+  return entries;
+}
+
 function extractOverUnderEntries(line, section) {
   const entries = [];
   const normalizedLine = normalizeWhitespace(line).toUpperCase().replace(/[~]/g, '-');
@@ -633,7 +713,7 @@ export function parseCircaOcrText(text) {
     }
 
     if (currentSection?.mode === 'binary_yes_no') {
-      const binaryEntries = extractBinaryYesNoEntriesV2(line, currentSection);
+      const binaryEntries = extractBinaryYesNoEntriesV3(line, currentSection);
       if (binaryEntries.length > 0) {
         entries.push(...binaryEntries);
         continue;

test/market-scanner.test.js

@@ -82,6 +82,17 @@ test('drops unreasonable binary odds from circa OCR rows', () => {
   assert.equal(entries[0].oddsInput, '-770');
 });
 
+test('recovers multiple packed home run rows from one OCR line', () => {
+  const entries = parseCircaOcrText([
+    'WILL [PLAYER] HIT A HOME RUN?',
+    'SHOHEI OHTANI (LAD) 6381 +235 63682 -280 ADDISON BARGER (TOR) 60765 +470 60786 -620 GUNNAR HENDERSON (BAL) 63683 +525 63685 -710',
+  ].join('\n'));
+
+  assert.ok(entries.some((entry) => entry.playerName === 'SHOHEI OHTANI' && entry.oddsInput === '+235'));
+  assert.ok(entries.some((entry) => entry.playerName === 'ADDISON BARGER' && entry.oddsInput === '+470'));
+  assert.ok(entries.some((entry) => entry.playerName === 'GUNNAR HENDERSON' && entry.oddsInput === '+525'));
+});
+
 test('ranks discrepancy, width, and circa alerts', () => {
   const entries = [
     {