seaworthycode 1.2.16 → 1.2.18

package/dist/index.js

@@ -407,10 +407,11 @@ async function crawl(options) {
           const { isBinary, isGenerated } = await peekFileHead(fullPath);
           if (isBinary || isGenerated) continue;
           if (await shouldSkipMinifiedJs(fullPath, entry.name)) continue;
+          let cachedContent;
           files.push({
             path: fullPath,
             relativePath: relPath,
-            content: () => fs.readFile(fullPath, "utf-8"),
+            content: () => cachedContent ??= fs.readFile(fullPath, "utf-8"),
             language: inferLanguage(entry.name),
             size: stat2.size
           });
@@ -428,10 +429,11 @@ async function crawl(options) {
         const { isBinary, isGenerated } = await peekFileHead(fullPath);
         if (isBinary || isGenerated) continue;
         if (await shouldSkipMinifiedJs(fullPath, entry.name)) continue;
+        let cachedContent;
         files.push({
           path: fullPath,
           relativePath: relPath,
-          content: () => fs.readFile(fullPath, "utf-8"),
+          content: () => cachedContent ??= fs.readFile(fullPath, "utf-8"),
           language: inferLanguage(entry.name),
           size: stat2.size
         });
@@ -5985,6 +5987,14 @@ function getMaxFilesPerLlmCheck() {
   }
   return 150;
 }
+function getMechanicalMaxFiles() {
+  const env2 = process.env.SEAWORTHY_MECHANICAL_MAX_FILES;
+  if (env2 !== void 0) {
+    const n = parseInt(env2, 10);
+    if (!isNaN(n) && n > 0) return n;
+  }
+  return 2e3;
+}
 function withLLMTimeout(promise, ms) {
   return new Promise((resolve22, reject) => {
     const timerId = setTimeout(() => {
@@ -6028,14 +6038,27 @@ function downgradeSeverity(severity) {
   if (idx < 0 || idx >= SEVERITY_ORDER3.length - 1) return severity;
   return SEVERITY_ORDER3[idx + 1];
 }
-function filterFilesByBudget(files, budget) {
+function estimateTokens(text22) {
+  return Math.ceil(text22.length / 4);
+}
+async function filterFilesByBudget(files, budget) {
   if (budget === void 0 || budget === Infinity) {
     return files;
   }
-  return files.filter((file) => {
-    const tokens2 = file.size !== void 0 ? Math.ceil(file.size / 4) : budget;
-    return tokens2 <= budget;
-  });
+  const result = [];
+  for (const file of files) {
+    let tokens2;
+    if (file.size !== void 0) {
+      tokens2 = Math.ceil(file.size / 4);
+    } else {
+      const content = await file.content();
+      tokens2 = estimateTokens(content);
+    }
+    if (tokens2 <= budget) {
+      result.push(file);
+    }
+  }
+  return result;
 }
 async function buildCacheKey(ctx, template, files) {
   if (!ctx.llm) return void 0;
@@ -6086,7 +6109,9 @@ function capProjectScopedFindings(findings) {
 }
 async function runProjectScopedMechanical(eligibleFiles, template) {
   if (eligibleFiles.length === 0) return [];
-  const contents = await Promise.all(eligibleFiles.map((f) => f.content()));
+  const maxFiles = getMechanicalMaxFiles();
+  const scanFiles = eligibleFiles.length > maxFiles ? eligibleFiles.slice(0, maxFiles) : eligibleFiles;
+  const contents = await Promise.all(scanFiles.map((f) => f.content()));
   const combined = contents.join("\n");
   let shouldFlag = false;
   const pattern = template.fallbackPatterns[0];
@@ -6113,13 +6138,23 @@ async function runProjectAbsenceChecks(eligibleFiles, template) {
   if (absencePatterns.length === 0 || eligibleFiles.length === 0) {
     return [];
   }
-  const contents = await Promise.all(eligibleFiles.map((f) => f.content()));
+  const maxFiles = getMechanicalMaxFiles();
+  const scanFiles = eligibleFiles.length > maxFiles ? eligibleFiles.slice(0, maxFiles) : eligibleFiles;
   const anchorFile = pickProjectAnchorFile(eligibleFiles);
+  const foundIndices = /* @__PURE__ */ new Set();
+  for (const file of scanFiles) {
+    if (foundIndices.size === absencePatterns.length) break;
+    const content = await file.content();
+    for (let i = 0; i < absencePatterns.length; i++) {
+      if (!foundIndices.has(i) && absencePatterns[i].regex.test(content)) {
+        foundIndices.add(i);
+      }
+    }
+  }
   const results = [];
-  for (const pattern of absencePatterns) {
-    const foundAnywhere = contents.some((content) => pattern.regex.test(content));
-    if (!foundAnywhere) {
-      results.push(makeMechanicalFinding(template, pattern, anchorFile, 1));
+  for (let i = 0; i < absencePatterns.length; i++) {
+    if (!foundIndices.has(i)) {
+      results.push(makeMechanicalFinding(template, absencePatterns[i], anchorFile, 1));
     }
   }
   return results;
@@ -6131,8 +6166,16 @@ async function runMechanicalCheck(ctx, template) {
   }
   const perFilePatterns = template.fallbackPatterns.filter((p) => !p.absenceBased);
   const projectResults = await runProjectAbsenceChecks(eligibleFiles, template);
+  const maxMechanicalFiles = getMechanicalMaxFiles();
+  let cappedFiles = eligibleFiles;
+  if (cappedFiles.length > maxMechanicalFiles) {
+    console.error(
+      `[seaworthy] ${template.checkId}: ${cappedFiles.length} eligible files \u2014 capping mechanical scan to ${maxMechanicalFiles}`
+    );
+    cappedFiles = cappedFiles.slice(0, maxMechanicalFiles);
+  }
   const fileResults = perFilePatterns.length === 0 ? [] : await runWithConcurrencyCollect(
-    eligibleFiles,
+    cappedFiles,
     async (file) => {
       const content = await file.content();
       const lines = content.split("\n");
@@ -6212,7 +6255,7 @@ async function runLLMCheck(ctx, template) {
     return [];
   }
   const proEligibleFiles = ctx.files.filter((f) => isProCheckEligible(f, template));
-  let eligibleFiles = filterFilesByBudget(proEligibleFiles, template.tokenBudget);
+  let eligibleFiles = await filterFilesByBudget(proEligibleFiles, template.tokenBudget);
   const maxFiles = getMaxFilesPerLlmCheck();
   if (eligibleFiles.length > maxFiles) {
     console.error(