Improve Circa odds cleanup and sign inference

src/market-scanner.js

@@ -351,7 +351,10 @@ function normalizeOcrOddsToken(token) {
   const normalized = normalizeWhitespace(token)
     .replace(/[~−–—]/g, '-')
     .replace(/B/g, '8')
-    .replace(/O/g, '0');
+    .replace(/O/g, '0')
+    .replace(/S/g, '5')
+    .replace(/I/g, '1')
+    .replace(/[)\],.;:]/g, '');
 
   if (/^[+-]\d{2,4}$/i.test(normalized)) {
     return normalized;
@@ -360,6 +363,29 @@ function normalizeOcrOddsToken(token) {
   return null;
 }
 
+function normalizeOcrOddsTokenForSide(token, side) {
+  const direct = normalizeOcrOddsToken(token);
+  if (direct) {
+    return direct;
+  }
+
+  const cleaned = normalizeWhitespace(token)
+    .replace(/[~−–—]/g, '-')
+    .replace(/B/g, '8')
+    .replace(/O/g, '0')
+    .replace(/S/g, '5')
+    .replace(/I/g, '1')
+    .replace(/[)\],.;:]/g, '');
+
+  if (/^\d{2,4}$/.test(cleaned)) {
+    return side === 'no' || side === 'under'
+      ? `-${cleaned}`
+      : `+${cleaned}`;
+  }
+
+  return null;
+}
+
 function isReasonableCircaOdds(oddsInput, section) {
   if (!oddsInput) {
     return false;
@@ -428,6 +454,60 @@ function extractBinaryYesNoEntries(line, section) {
   return entries;
 }
 
+function extractBinaryYesNoEntriesV2(line, section) {
+  const entries = [];
+  const normalizedLine = normalizeWhitespace(line)
+    .toUpperCase()
+    .replace(/\//g, ' ')
+    .replace(/\s+/g, ' ');
+  const pattern = /([A-Z][A-Z.'\- ]{2,40}\([A-Z]{2,3}\))\s+\d{4,6}\s+([+\-~]?[0-9BOSI]{2,4})\s+\d{4,6}\s+([+\-~]?[0-9BOSI]{2,4})/g;
+
+  for (const match of normalizedLine.matchAll(pattern)) {
+    const { playerName, team } = extractPlayerAndTeam(match[1]);
+    const yesOdds = normalizeOcrOddsTokenForSide(match[2], 'yes');
+    const noOdds = normalizeOcrOddsTokenForSide(match[3], 'no');
+
+    if (!playerName || playerName.length < 3) {
+      continue;
+    }
+
+    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: line,
+      };
+      entry.marketKey = buildMarketKey(entry);
+      entries.push(entry);
+    }
+  }
+
+  return entries;
+}
+
 function extractOverUnderEntries(line, section) {
   const entries = [];
   const normalizedLine = normalizeWhitespace(line).toUpperCase().replace(/[~]/g, '-');
@@ -480,6 +560,62 @@ function extractOverUnderEntries(line, section) {
   return entries;
 }
 
+function extractOverUnderEntriesV2(line, section) {
+  const entries = [];
+  const normalizedLine = normalizeWhitespace(line)
+    .toUpperCase()
+    .replace(/[~]/g, '-')
+    .replace(/\//g, ' ')
+    .replace(/\s+/g, ' ');
+  const pattern = /([A-Z][A-Z.'\- ]{2,40}\([A-Z]{2,3}\))\s+(?:O|OVER)\s*(\d+(?:\.\d+)?)\s+([+\-~]?[0-9BOSI]{2,4})\s+(?:U|UNDER)\s*\d+(?:\.\d+)?\s+([+\-~]?[0-9BOSI]{2,4})/g;
+
+  for (const match of normalizedLine.matchAll(pattern)) {
+    const { playerName, team } = extractPlayerAndTeam(match[1]);
+    const lineValue = Number(match[2]);
+    const overOdds = normalizeOcrOddsTokenForSide(match[3], 'over');
+    const underOdds = normalizeOcrOddsTokenForSide(match[4], 'under');
+
+    if (!playerName || playerName.length < 3 || !Number.isFinite(lineValue)) {
+      continue;
+    }
+
+    for (const sideEntry of [
+      { side: 'over', oddsInput: overOdds },
+      { side: 'under', oddsInput: underOdds },
+    ]) {
+      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,
+        oddsInput: sideEntry.oddsInput,
+        impliedProbability,
+        rawLabel: line,
+      };
+      entry.marketKey = buildMarketKey(entry);
+      entries.push(entry);
+    }
+  }
+
+  return entries;
+}
+
 export function parseCircaOcrText(text) {
   const entries = [];
   const lines = String(text ?? '')
@@ -497,7 +633,7 @@ export function parseCircaOcrText(text) {
     }
 
     if (currentSection?.mode === 'binary_yes_no') {
-      const binaryEntries = extractBinaryYesNoEntries(line, currentSection);
+      const binaryEntries = extractBinaryYesNoEntriesV2(line, currentSection);
       if (binaryEntries.length > 0) {
         entries.push(...binaryEntries);
         continue;
@@ -505,7 +641,7 @@ export function parseCircaOcrText(text) {
     }
 
     if (currentSection?.mode === 'over_under') {
-      const overUnderEntries = extractOverUnderEntries(line, currentSection);
+      const overUnderEntries = extractOverUnderEntriesV2(line, currentSection);
       if (overUnderEntries.length > 0) {
         entries.push(...overUnderEntries);
         continue;