@codeyam/codeyam-cli 0.1.0-staging.1669d45 → 0.1.0-staging.2a88920

package/analyzer-template/project/constructMockCode.ts

@@ -294,6 +294,19 @@ function isValidKey(key: string) {
   return !/\s/.test(keyWithOutArguments);
 }
 
+/**
+ * Known hooks that return tuples [value, setter] instead of arrays.
+ * These should NOT use the .map() pattern even when the schema has generic array access ([]).
+ * Instead, they should return [data, () => {}] where data is from scenarios().
+ */
+const TUPLE_RETURNING_HOOKS = new Set([
+  'useAtom', // Jotai
+  'useState', // React
+  'useReducer', // React
+  'useRecoilState', // Recoil
+  'useImmerAtom', // Jotai with Immer
+]);
+
 export default function constructMockCode(
   mockName: string,
   dependencySchemas: DeepReadonly<DataStructure['dependencySchemas']>,
@@ -530,6 +543,99 @@ export default function constructMockCode(
     dataStructureValue === 'array' &&
     (dataStructurePath === 'returnValue' || pathDepth <= mockNameParts.length);
 
+  // OPTIMIZATION: Early return for tuple-returning hooks (useAtom, useState, etc.)
+  // These hooks have simple [value, setter] return patterns that don't need the full
+  // 9216-key schema processing. Check if this is a tuple-returning hook and generate
+  // the mock code directly without iterating over all schema keys.
+  if (TUPLE_RETURNING_HOOKS.has(baseMockName) && isFunction) {
+    // Check if schema has generic array pattern (indicates tuple return like [value, setter])
+    const schemaKeys = Object.keys(relevantReturnValueSchema ?? {});
+    const hasGenericArrayInSchema = schemaKeys.some(
+      (k) =>
+        k.includes('.functionCallReturnValue[]') ||
+        k === `${dataKey}.functionCallReturnValue[]` ||
+        k === 'returnValue[]',
+    );
+
+    // Check for differentiated tuple indices (e.g., functionCallReturnValue[2], [3]) which would NOT be a standard tuple
+    // We only check indices immediately after functionCallReturnValue, not nested indices like signature[2]
+    const tupleHasDifferentiatedIndices = schemaKeys.some((k) => {
+      // Look for .functionCallReturnValue[N] where N >= 2
+      const match = k.match(/\.functionCallReturnValue\[(\d+)\]/);
+      if (!match) return false;
+      const idx = parseInt(match[1], 10);
+      return idx >= 2;
+    });
+
+    const isTupleReturningHook =
+      hasGenericArrayInSchema && !tupleHasDifferentiatedIndices;
+
+    if (isTupleReturningHook) {
+      // Find all call patterns for this hook (e.g., useAtom(quoteFilterAtom), useAtom(supplierAtom))
+      const hookCallPatterns = Object.keys(relevantReturnValueSchema ?? {})
+        .filter((k) => {
+          const regex = new RegExp(`^${baseMockName}\\([^)]+\\)$`);
+          return regex.test(k);
+        })
+        .map((k) => {
+          const match = k.match(/\(([^)]+)\)/);
+          return { key: k, arg: match?.[1] ?? '' };
+        });
+
+      let tupleReturnCode: string;
+      if (hookCallPatterns.length > 1) {
+        // Multiple patterns - generate conditional dispatch
+        const conditions = hookCallPatterns
+          .map(
+            ({ key, arg }) =>
+              `if (argLabel === '${arg}' || argStr.includes('${arg}')) {\n    return [scenarios().data()?.["${key}"]?.[0] ?? [], () => {}];\n  }`,
+          )
+          .join('\n  ');
+        const fallbackKey = hookCallPatterns[0]?.key ?? dataKey;
+        tupleReturnCode = `(() => {
+  // Dynamic dispatch for tuple-returning hook with multiple argument patterns
+  const argLabel = args[0]?.debugLabel ?? '';
+  const argStr = args[0]?.toString?.() ?? String(args[0] ?? '');
+  ${conditions}
+  return [scenarios().data()?.["${fallbackKey}"]?.[0] ?? [], () => {}];
+})()`;
+      } else {
+        // Single or no patterns - use dynamic dispatch
+        const fallbackKey = hookCallPatterns[0]?.key ?? `${baseMockName}()`;
+        tupleReturnCode = `(() => {
+  // Dynamic dispatch for tuple-returning hook
+  const argLabel = args[0]?.debugLabel ?? '';
+  const argStr = args[0]?.toString?.() ?? '';
+  const allData = scenarios().data() ?? {};
+  if (argLabel) {
+    const labelKey = '${baseMockName}(' + argLabel + ')';
+    if (allData[labelKey]) {
+      return [allData[labelKey]?.[0] ?? [], () => {}];
+    }
+  }
+  const keys = Object.keys(allData).filter(k => k.startsWith('${baseMockName}('));
+  for (const key of keys) {
+    const keyArg = key.slice(${baseMockName.length + 1}, -1);
+    if (argStr.includes(keyArg)) {
+      return [allData[key]?.[0] ?? [], () => {}];
+    }
+  }
+  return [allData[keys[0] ?? '${fallbackKey}']?.[0] ?? [], () => {}];
+})()`;
+      }
+
+      const safeFunctionName = options?.uniqueFunctionSuffix
+        ? `${baseMockName}_${options.uniqueFunctionSuffix}`
+        : options?.keepOriginalFunctionName
+          ? baseMockName
+          : mockNameIsCallSignature && derivedFunctionName
+            ? derivedFunctionName
+            : baseMockName;
+
+      return `function ${safeFunctionName}(...args) {\n  return ${tupleReturnCode};\n}`;
+    }
+  }
+
   const returnValueParts: ReturnValuePart = {
     name: dataStructureName,
     isArray: isRootArray,
@@ -864,7 +970,123 @@ export default function constructMockCode(
         });
 
       let returnValueContents = '';
-      if (
+
+      // Check if this is a known tuple-returning hook (useAtom, useState, etc.)
+      // These should return [value, setter] tuples, not arrays or data paths
+      // Check isGenericArray from current context OR from schema for root level calls
+      // (at root level, isGenericArray might not be set yet but the schema contains [] pattern)
+      const hasGenericArrayInSchema =
+        root &&
+        TUPLE_RETURNING_HOOKS.has(baseMockName) &&
+        Object.keys(relevantReturnValueSchema ?? {}).some((k) =>
+          k.includes('.functionCallReturnValue[]'),
+        );
+      // Check if there are array indices beyond what a standard 2-element tuple would have
+      // For tuple-returning hooks, [0] and [1] are expected (value and setter)
+      // Only consider it "differentiated" if there are indices >= 2 (e.g., [2], [3])
+      const tupleHasDifferentiatedIndices = nested?.some((n) => {
+        const indexMatch = n.name.match(/^\[(\d+)\]$/);
+        if (!indexMatch) return false;
+        const index = parseInt(indexMatch[1], 10);
+        return index >= 2;
+      });
+      const isTupleReturningHook =
+        TUPLE_RETURNING_HOOKS.has(baseMockName) &&
+        (isGenericArray || hasGenericArrayInSchema) &&
+        !tupleHasDifferentiatedIndices;
+
+      // Debug logging for tuple-returning hooks
+      if (TUPLE_RETURNING_HOOKS.has(baseMockName) && root) {
+        const schemaKeys = Object.keys(relevantReturnValueSchema ?? {});
+        const hasArrayPattern = schemaKeys.some((k) =>
+          k.includes('.functionCallReturnValue[]'),
+        );
+        console.log(
+          `CodeYam: Tuple hook check for ${baseMockName} (root):`,
+          `hasGenericArrayInSchema=${hasGenericArrayInSchema}`,
+          `hasArrayPattern=${hasArrayPattern}`,
+          `tupleHasDifferentiatedIndices=${tupleHasDifferentiatedIndices}`,
+          `isTupleReturningHook=${isTupleReturningHook}`,
+          `schemaKeysSample=${schemaKeys.slice(0, 5).join(', ')}`,
+        );
+      }
+
+      if (isTupleReturningHook) {
+        // Tuple-returning hooks should return [value, setter] tuple
+        // The value is the first element from scenarios data, setter is a no-op
+        // Default to [] when data is undefined to prevent errors like ".includes is not a function"
+
+        // Check if there are multiple call patterns for this hook in the schema
+        // (e.g., useAtom(quoteFilterAtom) and useAtom(supplierAtom))
+        const hookCallPatterns = Object.keys(relevantReturnValueSchema ?? {})
+          .filter((k) => {
+            // Match patterns like "useAtom(someArg)" but not nested paths like "useAtom(x).foo"
+            const regex = new RegExp(`^${baseMockName}\\([^)]+\\)$`);
+            return regex.test(k);
+          })
+          .map((k) => {
+            // Extract the argument from the key like "useAtom(quoteFilterAtom)" -> "quoteFilterAtom"
+            const match = k.match(/\(([^)]+)\)/);
+            return { key: k, arg: match?.[1] ?? '' };
+          });
+
+        if (hookCallPatterns.length > 1) {
+          // Multiple patterns - generate conditional dispatch based on first argument
+          // For Jotai atoms, we use debugLabel; for others, we try to match the argument string
+          const conditions = hookCallPatterns
+            .map(
+              ({ key, arg }) =>
+                `if (argLabel === '${arg}' || argStr.includes('${arg}')) {\n    return [scenarios().data()?.["${key}"]?.[0] ?? [], () => {}];\n  }`,
+            )
+            .join('\n  ');
+
+          // Use the first pattern as fallback
+          const fallbackKey = hookCallPatterns[0]?.key ?? dataKey;
+
+          returnValueContents = `(() => {
+  // Dynamic dispatch for tuple-returning hook with multiple argument patterns
+  const argLabel = args[0]?.debugLabel ?? '';
+  const argStr = args[0]?.toString?.() ?? String(args[0] ?? '');
+  ${conditions}
+  // Fallback to first pattern
+  return [scenarios().data()?.["${fallbackKey}"]?.[0] ?? [], () => {}];
+})()`;
+        } else {
+          // Single pattern or no patterns - use dynamic dispatch to handle case where
+          // the mock is used with different atoms than what was captured in the schema.
+          // Use the first argument to construct the data key dynamically.
+          const fallbackKey = hookCallPatterns[0]?.key ?? `${baseMockName}()`;
+
+          returnValueContents = `(() => {
+  // Dynamic dispatch for tuple-returning hook
+  // Try to construct key from argument's debugLabel (Jotai atoms) or toString
+  const argLabel = args[0]?.debugLabel ?? '';
+  const argStr = args[0]?.toString?.() ?? '';
+  const allData = scenarios().data() ?? {};
+
+  // Try to find a matching key using debugLabel first
+  if (argLabel) {
+    const labelKey = '${baseMockName}(' + argLabel + ')';
+    if (allData[labelKey]) {
+      return [allData[labelKey]?.[0] ?? [], () => {}];
+    }
+  }
+
+  // Try to find any matching key that contains part of the argument string
+  const keys = Object.keys(allData).filter(k => k.startsWith('${baseMockName}('));
+  for (const key of keys) {
+    const keyArg = key.slice(${baseMockName.length + 1}, -1);
+    if (argStr.includes(keyArg)) {
+      return [allData[key]?.[0] ?? [], () => {}];
+    }
+  }
+
+  // Fallback to first matching key or default
+  const fallback = keys[0] ?? '${fallbackKey}';
+  return [allData[fallback]?.[0] ?? [], () => {}];
+})()`;
+        }
+      } else if (
         !returnsFunctionArgs &&
         nestedContent.length === 0 &&
         dataPaths.length === 1
@@ -967,11 +1189,20 @@ export default function constructMockCode(
                   // Valid patterns where arrow function is NOT bare:
                   // 1. Property value: "key: (arg) => ..." - ends with ':'
                   // 2. Function argument: ".map((arg) => ..." - ends with '('
+                  // 3. Method call: "?.map" followed directly by the arrow function
+                  //    In this case, the '(' is consumed by the arrow function regex match,
+                  //    so beforeTrimmed ends with the method name (e.g., 'map'), not '('.
+                  //    We detect this by checking if beforeTrimmed ends with an identifier
+                  //    that could be a method name (preceded by '.' or '?.').
                   // NOTE: We don't include ',' because "{ prop, () => {} }" is invalid
                   // (can't distinguish function argument from object property context)
                   const isPropertyValue = beforeTrimmed.endsWith(':');
                   const isFunctionArg = beforeTrimmed.endsWith('(');
-                  const hasPropertyName = isPropertyValue || isFunctionArg;
+                  // Check if before ends with a method call pattern like ".map" or "?.map"
+                  // The '(' after the method name is consumed by the arrow function regex
+                  const isMethodCallArg = /\??\.\w+$/.test(beforeTrimmed);
+                  const hasPropertyName =
+                    isPropertyValue || isFunctionArg || isMethodCallArg;
 
                   if (!hasPropertyName) {
                     // This is a bare arrow function - filter it out
@@ -1254,7 +1485,38 @@ export default function constructMockCode(
             returnValueContents = `[\n${indent(levelContents)}\n]`;
           }
         } else {
-          returnValueContents = `{\n${indent(levelContents)}\n}`;
+          // When we have a single data path and nested content that creates an object structure,
+          // and we're NOT at the root level, we need to handle the case where the parent data
+          // value is null or undefined. Without this check, `{ ...null, prop: null?.["prop"] }`
+          // creates `{ prop: undefined }` instead of `null`, causing errors like
+          // "Cannot read properties of undefined (reading 'some')" when code does
+          // data?.prop.some(...) because data is an object with prop: undefined, not null.
+          // We only apply this to non-root cases because root-level mocks are expected to exist.
+          // We also skip structural elements (like [0] inside arrays) because the null check
+          // syntax doesn't work inside .map() callbacks where structural elements are used.
+          // We also skip array index elements ([0], [1], etc.) because they represent tuple/array
+          // elements, not properties that could be null.
+          // We also only apply this when we're inside a function return value context - i.e.,
+          // when the data path contains a function call pattern like ?.["someFunction(...)"].
+          // This prevents adding null checks to intermediate objects in chains like supabase.auth.
+          const hasNestedContent = nestedContent.filter(Boolean).length > 0;
+          const isArrayIndexElement = name.match(/^\[\d*\]$/);
+          // Check if data path contains a function call pattern, indicating we're inside a function return value
+          const isInsideFunctionReturnValue =
+            dataPaths.length === 1 &&
+            dataPaths[0].match(/\?\.\["\w+\([^"]*\)"\]/);
+          if (
+            !root &&
+            !returnValue.isStructural &&
+            !isArrayIndexElement &&
+            isInsideFunctionReturnValue &&
+            hasNestedContent
+          ) {
+            // Wrap with null check: if parent is null/undefined, return it directly; otherwise create object
+            returnValueContents = `${dataPaths[0]} == null ? ${dataPaths[0]} : {\n${indent(levelContents)}\n}`;
+          } else {
+            returnValueContents = `{\n${indent(levelContents)}\n}`;
+          }
         }
       }
 
@@ -1390,34 +1652,91 @@ export default function constructMockCode(
   };
 
   // Create the return value structure
-  const sortedKeys = Object.keys(relevantReturnValueSchema ?? {}).sort(
-    (a: string, b: string) => {
-      const aParts = splitOutsideParenthesesAndArrays(a);
-      const bParts = splitOutsideParenthesesAndArrays(b);
-
-      const maxLength = Math.max(aParts.length, bParts.length);
-      for (let i = 0; i < maxLength; ++i) {
-        const aPart = aParts[i];
-        const bPart = bParts[i];
-
-        if (!aPart) return -1;
-        if (!bPart) return 1;
-
-        if (aPart === bPart) continue;
-
-        const aName = aPart.split('(')[0];
-        const bName = bPart.split('(')[0];
+  // OPTIMIZATION: Filter keys to only those starting with baseMockName before sorting.
+  // This dramatically reduces processing time for large schemas (e.g., 9216 keys -> ~100 relevant keys).
+  // Without this filter, the loop would call splitOutsideParenthesesAndArrays on every key
+  // even though most are filtered out later by the baseMockName check.
+  const allSchemaKeys = Object.keys(relevantReturnValueSchema ?? {});
+  const relevantKeys = allSchemaKeys.filter((key) => {
+    // Fast prefix check - key must start with baseMockName followed by ( or < or .
+    // This matches: "useAtom()", "useAtom<T>()", "useAtom.something", but not "useAtomValue()"
+    if (key === baseMockName) return true;
+    if (key.startsWith(baseMockName + '(')) return true;
+    if (key.startsWith(baseMockName + '<')) return true;
+    if (key.startsWith(baseMockName + '.')) return true;
+    // Also include 'returnValue' paths which are normalized later
+    if (
+      key === 'returnValue' ||
+      key.startsWith('returnValue.') ||
+      key.startsWith('returnValue[')
+    )
+      return true;
+    return false;
+  });
+
+  const schemaKeyCount = relevantKeys.length;
+  const sortedKeys = relevantKeys.sort((a: string, b: string) => {
+    const aParts = splitOutsideParenthesesAndArrays(a);
+    const bParts = splitOutsideParenthesesAndArrays(b);
+
+    const maxLength = Math.max(aParts.length, bParts.length);
+    for (let i = 0; i < maxLength; ++i) {
+      const aPart = aParts[i];
+      const bPart = bParts[i];
+
+      if (!aPart) return -1;
+      if (!bPart) return 1;
+
+      if (aPart === bPart) continue;
+
+      const aName = aPart.split('(')[0];
+      const bName = bPart.split('(')[0];
+
+      if (aName !== bName) {
+        return aName.localeCompare(bName);
+      } else {
+        return aPart.localeCompare(bPart);
+      }
+    }
 
-        if (aName !== bName) {
-          return aName.localeCompare(bName);
-        } else {
-          return aPart.localeCompare(bPart);
-        }
+    return 0;
+  });
+
+  // OPTIMIZATION: Pre-compute prefix indexes for O(1) lookups instead of O(n) scans.
+  // This reduces complexity from O(n²) to O(n) for large schemas (9k+ keys).
+  //
+  // 1. extendedReturnValuePrefixes: Set of all path prefixes that have a .functionCallReturnValue extension
+  //    Used by hasExtendedFunctionCallReturnValue check at line ~1754
+  // 2. functionCallsWithReturnValue: Set of function call paths where .functionCallReturnValue IMMEDIATELY follows
+  //    Used by hasProperFunctionCallPath check at line ~1787
+  //    IMPORTANT: Only includes paths where the function call is directly followed by .functionCallReturnValue
+  //    e.g., "a.b().functionCallReturnValue" -> adds "a.b()" but NOT "a" even if "a" ends with ")"
+  const extendedReturnValuePrefixes = new Set<string>();
+  const functionCallsWithReturnValue = new Set<string>();
+
+  for (const k of relevantKeys) {
+    const parts = splitOutsideParenthesesAndArrays(k);
+    const returnValueIndex = parts.findIndex((part) =>
+      part.startsWith(RETURN_VALUE),
+    );
+    if (returnValueIndex !== -1) {
+      // Add all prefixes of k up to (but not including) functionCallReturnValue
+      const prefix = joinParenthesesAndArrays(parts.slice(0, returnValueIndex));
+      extendedReturnValuePrefixes.add(prefix);
+
+      // ONLY add to functionCallsWithReturnValue if functionCallReturnValue IMMEDIATELY follows
+      if (prefix.endsWith(')')) {
+        functionCallsWithReturnValue.add(prefix);
       }
 
-      return 0;
-    },
-  );
+      // Also add intermediate prefixes for nested paths to extendedReturnValuePrefixes
+      // This helps hasExtendedFunctionCallReturnValue which checks key + '.'
+      for (let i = 1; i < returnValueIndex; i++) {
+        const partialPrefix = joinParenthesesAndArrays(parts.slice(0, i));
+        extendedReturnValuePrefixes.add(partialPrefix);
+      }
+    }
+  }
 
   for (const key of sortedKeys) {
     const value = relevantReturnValueSchema[key];
@@ -1475,9 +1794,10 @@ export default function constructMockCode(
     // nested inside (e.g., methods on array elements passed as arguments).
     if (hasSignaturePath) continue;
 
-    const hasExtendedFunctionCallReturnValue = Object.keys(
-      relevantReturnValueSchema,
-    ).some((k) => k.startsWith(key + '.') && k.includes(`.${RETURN_VALUE}`));
+    // OPTIMIZATION: Use pre-computed index instead of O(n) scan
+    // Old code: Object.keys(relevantReturnValueSchema).some((k) => k.startsWith(key + '.') && k.includes('.functionCallReturnValue'))
+    const hasExtendedFunctionCallReturnValue =
+      extendedReturnValuePrefixes.has(key);
 
     // Skip JSX components - they look like function calls (e.g., Context.Provider())
     // but they're React components used in JSX, not functions that need mocking
@@ -1506,11 +1826,10 @@ export default function constructMockCode(
           const functionCallPath = joinParenthesesAndArrays(
             parts.slice(0, i + 1),
           );
-          const hasProperFunctionCallPath = Object.keys(
-            relevantReturnValueSchema,
-          ).some((k) =>
-            k.startsWith(functionCallPath + '.functionCallReturnValue'),
-          );
+          // OPTIMIZATION: Use pre-computed index instead of O(n) scan
+          // Old code: Object.keys(relevantReturnValueSchema).some((k) => k.startsWith(functionCallPath + '.functionCallReturnValue'))
+          const hasProperFunctionCallPath =
+            functionCallsWithReturnValue.has(functionCallPath);
           if (hasProperFunctionCallPath) {
             // Skip this path - the .functionCallReturnValue path will handle it correctly
             shouldSkipKey = true;
@@ -2098,7 +2417,18 @@ export default function constructMockCode(
 
     // Inline the return value directly in the function to avoid module-level const
     // that would be evaluated before scenario context is ready
-    return `${isRootAsyncFunction ? 'async ' : ''}function ${safeFunctionName}(...args) {\n${indent(`return ${returnValue};`)}\n}`;
+    // Add fallback for simple data path returns to prevent undefined errors (e.g., createTheme)
+    // Only add fallback if returnValue is a simple data accessor (starts with scenarios().data())
+    // and doesn't already have nested structure (object literal, array, or method chains like .map())
+    const isSimpleDataPath =
+      returnValue.startsWith('scenarios().data()') &&
+      !returnValue.trim().startsWith('{') &&
+      !returnValue.trim().startsWith('[') &&
+      !returnValue.includes('.map('); // Exclude method chains
+    const safeReturnValue = isSimpleDataPath
+      ? `${returnValue} ?? {}`
+      : returnValue;
+    return `${isRootAsyncFunction ? 'async ' : ''}function ${safeFunctionName}(...args) {\n${indent(`return ${safeReturnValue};`)}\n}`;
   } else {
     // Generate safe const name:
     // 1. For call signatures: use derivedFunctionName

package/analyzer-template/project/createEntitiesAndSortFiles.ts

@@ -28,6 +28,14 @@ export interface SortedEntitiesAndMap {
   entityMap: { [key: string]: { [key: string]: EntityWithImportCount } };
 }
 
+// Timing stats for debugging
+let findOrCreateEntityTime = 0;
+let gatherEntityMapTime = 0;
+let recursiveCallTime = 0;
+let findOrCreateEntityCalls = 0;
+let gatherEntityMapCalls = 0;
+let recursiveCalls = 0;
+
 async function findAndGatherEntity({
   project,
   branch,
@@ -37,6 +45,7 @@ async function findAndGatherEntity({
   projectAnalyzer,
   context,
   fileStore,
+  depth = 0,
 }: {
   project: Project;
   branch: Branch;
@@ -46,6 +55,7 @@ async function findAndGatherEntity({
   projectAnalyzer: ProjectAnalyzer;
   context: AnalysisContext;
   fileStore?: LazyFileStore;
+  depth?: number;
 }) {
   let entityWithImportCount: EntityWithImportCount = context.getEntity(
     file.path,
@@ -53,6 +63,7 @@ async function findAndGatherEntity({
   ) as EntityWithImportCount;
   if (entityWithImportCount?.importCount === undefined) {
     try {
+      const t0 = performance.now();
       entityWithImportCount = await findOrCreateEntity({
         file,
         branchId: branch?.id,
@@ -61,6 +72,8 @@ async function findAndGatherEntity({
         projectAnalyzer,
         context,
       });
+      findOrCreateEntityTime += performance.now() - t0;
+      findOrCreateEntityCalls++;
     } catch (e) {
       console.log(
         'CodeYam Warning: Entity not found in create and sort entities',
@@ -72,12 +85,15 @@ async function findAndGatherEntity({
     // Add entity to context
     context.addEntity(entityWithImportCount);
 
+    const t1 = performance.now();
     await gatherEntityMap({
       entity: entityWithImportCount,
       branchId: branch?.id,
       context,
       projectAnalyzer,
     });
+    gatherEntityMapTime += performance.now() - t1;
+    gatherEntityMapCalls++;
 
     entityWithImportCount.importCount = 0;
     for (const importedEntity of entityWithImportCount.metadata
@@ -105,6 +121,8 @@ async function findAndGatherEntity({
         continue;
       }
 
+      const t2 = performance.now();
+      recursiveCalls++;
       const info = await findAndGatherEntity({
         project,
         branch,
@@ -114,7 +132,11 @@ async function findAndGatherEntity({
         projectAnalyzer,
         context,
         fileStore,
+        depth: depth + 1,
       });
+      if (depth === 0) {
+        recursiveCallTime += performance.now() - t2;
+      }
       entityWithImportCount.importCount +=
         info.entityWithImportCount.importCount;
     }
@@ -147,6 +169,14 @@ export default async function createEntitiesAndSortFiles({
   branchCommit, // just for progress updates
   fileStore,
 }: EntitiesAndSortFiles): Promise<SortedEntitiesAndMap> {
+  // Reset timing stats
+  findOrCreateEntityTime = 0;
+  gatherEntityMapTime = 0;
+  recursiveCallTime = 0;
+  findOrCreateEntityCalls = 0;
+  gatherEntityMapCalls = 0;
+  recursiveCalls = 0;
+
   const sortedEntities: Entity[] = [];
 
   // We slightly abuse the AnalysisContext here, initializing and later throwing it away
@@ -161,6 +191,16 @@ export default async function createEntitiesAndSortFiles({
   // Enable baseline mode to skip DB reads during entity creation
   context.setBaselineMode(true);
   let ii = 0;
+  let updateCommitMetadataTime = 0;
+  let getFileAnalyzerTime = 0;
+  let getAllEntityNodesTime = 0;
+  const loopStart = performance.now();
+  let lastLogTime = loopStart;
+
+  console.log(
+    `CodeYam: createEntitiesAndSortFiles starting, ${filePaths.length} files to process`,
+  );
+
   for (const currentFilePath of filePaths) {
     if (filePath && currentFilePath !== filePath) {
       ii++;
@@ -179,16 +219,23 @@ export default async function createEntitiesAndSortFiles({
     }
 
     if (branchCommit) {
+      const t0 = performance.now();
       branchCommit.metadata = await updateCommitMetadata({
         commitId: branchCommit.id,
         runStatusUpdate: {
           filesCompleted: ii + 1,
         },
       });
+      updateCommitMetadataTime += performance.now() - t0;
     }
 
+    const t1 = performance.now();
     const fileAnalyzer = projectAnalyzer.getFileAnalyzer(file);
+    getFileAnalyzerTime += performance.now() - t1;
+
+    const t2 = performance.now();
     const allEntityNodes = fileAnalyzer.getAllEntityNodesOnlyAliases();
+    getAllEntityNodesTime += performance.now() - t2;
 
     for (const entityName of Object.keys(allEntityNodes)) {
       const { entityWithImportCount: entity } = await findAndGatherEntity({
@@ -206,8 +253,44 @@ export default async function createEntitiesAndSortFiles({
     }
 
     ii++;
+
+    // Log timing stats every 5 seconds
+    const now = performance.now();
+    if (now - lastLogTime > 5000) {
+      lastLogTime = now;
+      const elapsed = ((now - loopStart) / 1000).toFixed(1);
+      console.log(`CodeYam: createEntitiesAndSortFiles progress`, {
+        filesProcessed: ii,
+        totalFiles: filePaths.length,
+        elapsedSec: elapsed,
+        timings: {
+          updateCommitMetadata: `${(updateCommitMetadataTime / 1000).toFixed(1)}s`,
+          getFileAnalyzer: `${(getFileAnalyzerTime / 1000).toFixed(1)}s`,
+          getAllEntityNodes: `${(getAllEntityNodesTime / 1000).toFixed(1)}s`,
+          findOrCreateEntity: `${(findOrCreateEntityTime / 1000).toFixed(1)}s (${findOrCreateEntityCalls} calls)`,
+          gatherEntityMap: `${(gatherEntityMapTime / 1000).toFixed(1)}s (${gatherEntityMapCalls} calls)`,
+          recursiveCalls: `${(recursiveCallTime / 1000).toFixed(1)}s (${recursiveCalls} calls)`,
+        },
+      });
+    }
   }
 
+  // Final timing log
+  const totalElapsed = ((performance.now() - loopStart) / 1000).toFixed(1);
+  console.log(`CodeYam: createEntitiesAndSortFiles loop complete`, {
+    filesProcessed: ii,
+    totalFiles: filePaths.length,
+    elapsedSec: totalElapsed,
+    timings: {
+      updateCommitMetadata: `${(updateCommitMetadataTime / 1000).toFixed(1)}s`,
+      getFileAnalyzer: `${(getFileAnalyzerTime / 1000).toFixed(1)}s`,
+      getAllEntityNodes: `${(getAllEntityNodesTime / 1000).toFixed(1)}s`,
+      findOrCreateEntity: `${(findOrCreateEntityTime / 1000).toFixed(1)}s (${findOrCreateEntityCalls} calls)`,
+      gatherEntityMap: `${(gatherEntityMapTime / 1000).toFixed(1)}s (${gatherEntityMapCalls} calls)`,
+      recursiveCalls: `${(recursiveCallTime / 1000).toFixed(1)}s (${recursiveCalls} calls)`,
+    },
+  });
+
   await setImportedExports({
     entities: Object.values(sortedEntities).flat(),
     context,