@codeyam/codeyam-cli 0.1.0-staging.ad31e3e → 0.1.0-staging.b8a55ba

package/analyzer-template/packages/ai/src/lib/dataStructure/equivalencyManagers/frameworks/ReactFrameworkManager.ts

@@ -17,6 +17,12 @@ interface UseStateInfo {
 export class ReactFrameworkManager implements EquivalencyManager {
   private useStateInfos: UseStateInfo[] = [];
   private useCallbackEquivalents: Record<string, string> = {};
+  // Track variables used as ref props - they should be typed as 'null'
+  // Format: { scopeNodeName -> Set of variable paths }
+  private refVariables: Map<string, Set<string>> = new Map();
+  // Track ref variable source paths for updating external function call schemas in finalize
+  // Format: { functionCallName -> Set of source paths (e.g., 'useAutoAnimate().functionCallReturnValue[]') }
+  private refVariableSources: Map<string, Set<string>> = new Map();
 
   postProcess(scopeNode: ScopeNode, scopeDataStructure: ScopeDataStructure) {
     this.handleUseState(scopeNode, scopeDataStructure);
@@ -139,7 +145,27 @@ export class ReactFrameworkManager implements EquivalencyManager {
     'main',
   ]);
 
-  finalize(_scopeNode: ScopeNode, _scopeDataStructure: ScopeDataStructure) {}
+  finalize(scopeNode: ScopeNode, scopeDataStructure: ScopeDataStructure) {
+    // Set all tracked ref variables to 'null' type
+    // This ensures mock data generation uses null instead of placeholder strings
+    const refVarsForScope = this.refVariables.get(scopeNode.name);
+    if (refVarsForScope) {
+      for (const refVar of refVarsForScope) {
+        scopeNode.schema[refVar] = 'null';
+      }
+    }
+
+    // Update external function call schemas with ref source paths
+    // This runs AFTER schema building is complete, so the null values won't be overwritten
+    for (const efc of scopeDataStructure.externalFunctionCalls) {
+      const refSourcePaths = this.refVariableSources.get(efc.name);
+      if (refSourcePaths) {
+        for (const sourcePath of refSourcePaths) {
+          efc.schema[sourcePath] = 'null';
+        }
+      }
+    }
+  }
 
   private findUseStateInfo(useStateCall: string) {
     return this.useStateInfos.find(
@@ -251,9 +277,9 @@ export class ReactFrameworkManager implements EquivalencyManager {
               equivalentPathParts[0].replace(existingFunctionName, ''),
             ...equivalentPathParts.slice(1),
           ]);
-          for (const equivalentValue of scopeNode.equivalencies[
-            equivalentPath
-          ]) {
+          const equivalenciesToReroute =
+            scopeNode.equivalencies[equivalentPath] ?? [];
+          for (const equivalentValue of equivalenciesToReroute) {
             scopeDataStructure.removeEquivalency(
               equivalentPath,
               equivalentValue,
@@ -273,7 +299,9 @@ export class ReactFrameworkManager implements EquivalencyManager {
           equivalentPath = newEquivalentPath;
         }
 
-        for (const equivalentValue of scopeNode.equivalencies[equivalentPath]) {
+        const equivalenciesToModify =
+          scopeNode.equivalencies[equivalentPath] ?? [];
+        for (const equivalentValue of equivalenciesToModify) {
           if (equivalentValue.scopeNodeName === useCallbackScopeInfoName) {
             modifiedEquivalencies = true;
             equivalentValue.scopeNodeName =
@@ -562,6 +590,102 @@ export class ReactFrameworkManager implements EquivalencyManager {
     for (const [equivalentPath, equivalentValues] of Object.entries(
       scopeNode.equivalencies,
     )) {
+      // Check if the equivalency KEY (e.g., "div().signature[0].ref") ends with .ref
+      // If so, the VALUE (e.g., "parent") should be marked as null
+      const equivalentPathParts = scopeDataStructure.splitPath(equivalentPath);
+      const pathLength = equivalentPathParts.length;
+
+      if (pathLength >= 2) {
+        const secondToLast = equivalentPathParts[pathLength - 2];
+        const last = equivalentPathParts[pathLength - 1];
+
+        if (secondToLast === 'signature[0]' && last === 'ref') {
+          // This is a ref prop usage like div().signature[0].ref -> parent
+          for (const equivalentValue of equivalentValues) {
+            propagated = true;
+
+            // Track this variable as a ref - we'll set its type to 'null' in finalize()
+            // after the schema is built (setting directly here would be overwritten)
+            // IMPORTANT: The ref variable might be defined in a parent scope, not the
+            // current scope. We need to find where the variable is actually defined by
+            // walking up the scope tree.
+            const varName = equivalentValue.schemaPath;
+            let targetScope: ScopeNode | null = scopeNode;
+
+            // Walk up the scope tree to find where this variable is defined
+            while (targetScope) {
+              // Check if this scope has the variable in its schema or equivalencies
+              const hasVariable =
+                varName in (targetScope.schema || {}) ||
+                varName in (targetScope.equivalencies || {});
+
+              if (hasVariable) {
+                if (!this.refVariables.has(targetScope.name)) {
+                  this.refVariables.set(targetScope.name, new Set());
+                }
+                this.refVariables.get(targetScope.name)!.add(varName);
+
+                // IMPORTANT: Also update the source of the ref variable in the external
+                // function call's schema. This ensures dataForMocks produces [null]
+                // instead of ["string"] for hooks like useAutoAnimate.
+                const varEquivalencies = targetScope.equivalencies?.[varName];
+                if (varEquivalencies) {
+                  for (const varEquivalency of varEquivalencies) {
+                    const sourcePath = varEquivalency.schemaPath;
+                    // Check if this comes from an external function call
+                    // (contains functionCallReturnValue which indicates it's a hook/function return)
+                    if (sourcePath.includes('functionCallReturnValue')) {
+                      // Store the source path for later - we'll update the external function
+                      // call's schema in finalize() which runs AFTER schema building is complete.
+                      // Setting it here would be overwritten by later schema building.
+                      for (const efc of scopeDataStructure.externalFunctionCalls) {
+                        if (sourcePath.startsWith(efc.name)) {
+                          if (!this.refVariableSources.has(efc.name)) {
+                            this.refVariableSources.set(efc.name, new Set());
+                          }
+                          // Store both the specific indexed path and the generic [] path
+                          this.refVariableSources
+                            .get(efc.name)!
+                            .add(sourcePath);
+                          const genericPath = sourcePath.replace(
+                            /\[\d+\]/g,
+                            '[]',
+                          );
+                          if (genericPath !== sourcePath) {
+                            this.refVariableSources
+                              .get(efc.name)!
+                              .add(genericPath);
+                          }
+                        }
+                      }
+                    }
+                  }
+                }
+              }
+
+              // Move up to parent scope
+              if (targetScope.tree && targetScope.tree.length > 0) {
+                const parentScopeName =
+                  targetScope.tree[targetScope.tree.length - 1];
+                targetScope = scopeDataStructure.getScopeOrFunctionCallInfo(
+                  parentScopeName,
+                ) as ScopeNode | null;
+              } else {
+                targetScope = null;
+              }
+            }
+
+            scopeDataStructure.removeEquivalency(
+              equivalentPath,
+              equivalentValue,
+              scopeNode,
+            );
+          }
+          continue;
+        }
+      }
+
+      // Original logic for checking the value side
       for (const equivalentValue of equivalentValues) {
         const equivalentValueParts = scopeDataStructure.splitPath(
           equivalentValue.schemaPath,

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/cleanKnownObjectFunctions.ts

@@ -64,6 +64,14 @@ const protoMethodSets: Record<PrimitiveKind, Set<string>> = {
   ),
 };
 
+/** Built-in properties (not methods) that should be removed from primitives.
+ * These are properties that exist on the prototype and shouldn't appear
+ * as schema entries (they cause malformed mock data structures). */
+const protoPropertySets: Partial<Record<PrimitiveKind, Set<string>>> = {
+  array: new Set(['length']),
+  string: new Set(['length']),
+};
+
 /**
  * Extract primitive kind from a type string, handling union types.
  * E.g., "number | undefined" -> "number", "string | null" -> "string"
@@ -118,11 +126,21 @@ function scrub(
       const rootKind = rootTable.get(prefix);
       if (!rootKind) continue;
 
-      const rawMethod = keyParts[cut].split('(')[0];
+      const rawToken = keyParts[cut];
+      const rawMethod = rawToken.split('(')[0];
+
+      // Check if it's a prototype method (e.g., map, filter, toString)
       if (protoMethodSets[rootKind].has(rawMethod)) {
         onRemove(keyPath);
         break;
       }
+
+      // Check if it's a built-in property (e.g., length on arrays/strings)
+      const propertySet = protoPropertySets[rootKind];
+      if (propertySet && propertySet.has(rawToken)) {
+        onRemove(keyPath);
+        break;
+      }
     }
   }
 }

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/cleanNonObjectFunctions.ts

@@ -30,11 +30,30 @@ export const nonTransformingArrayMethodsSet = new Set([
   'reverse',
   'entries',
 ]);
-const isolatingArrayMethodsSet = new Set(['find', 'at', 'pop', 'shift']);
+const isolatingArrayMethodsSet = new Set([
+  'find',
+  'findLast', // Returns element like find, but from end
+  'at',
+  'pop',
+  'shift',
+]);
 
 // Methods that are ambiguous - used on both arrays and non-arrays (e.g., router.push, history.pop)
 // Don't infer array type from these alone without additional evidence
 const ambiguousArrayMethodsSet = new Set(['push', 'pop']);
+
+// Methods that return primitives (boolean, number) - their return values are not mockable
+// These should have their .functionCallReturnValue paths removed from the schema
+const primitiveReturningMethodsSet = new Set([
+  'some', // Returns boolean
+  'every', // Returns boolean
+  'includes', // Returns boolean
+  'indexOf', // Returns number (index)
+  'lastIndexOf', // Returns number (index from end)
+  'findIndex', // Returns number (index)
+  'findLastIndex', // Returns number (index from end)
+]);
+
 export const transformingArrayMethodsSet = new Set([
   'map',
   'reduce',
@@ -42,6 +61,12 @@ export const transformingArrayMethodsSet = new Set([
   'concat',
   'join',
   'some',
+  'every', // Returns boolean
+  'findIndex', // Returns number (index)
+  'findLastIndex', // Returns number (index from end)
+  'indexOf', // Returns number (index)
+  'lastIndexOf', // Returns number (index from end)
+  'includes', // Returns boolean
 ]);
 
 export const arrayMethodsSet = new Set([
@@ -225,6 +250,54 @@ export function clearAttributesFromMapping(
   clearAttributes(mapping, knownBasePaths);
 }
 
+/**
+ * Remove duplicate function calls (same function with different arguments) that have no return data.
+ * Example: setSelectedFilter(arg1) and setSelectedFilter(arg2) where neither has functionCallReturnValue
+ * These create invalid mock code when spread together (orphan spread statements).
+ *
+ * This is a targeted fix that can be called independently without the full clearAttributes logic,
+ * which was disabled in constructMockCode.ts due to side effects on array method calls.
+ */
+export function removeDuplicateFunctionCalls(
+  mapping: Record<string, string>,
+): void {
+  const allKeys = Object.keys(mapping);
+  const functionCallGroups: { [baseName: string]: string[] } = {};
+
+  for (const key of allKeys) {
+    // Only process paths that:
+    // 1. Are marked as 'function'
+    // 2. END with a function call (parentheses) - not just contain one somewhere
+    // 3. Are on an object path (contain '.')
+    if (mapping[key] === 'function' && key.endsWith(')') && key.includes('.')) {
+      // Extract base path + function name (without the arguments)
+      // e.g., "returnValue.setSelectedFilter(arg)" -> "returnValue.setSelectedFilter"
+      const baseName = key.replace(/\([^)]*\)$/, '');
+      functionCallGroups[baseName] ||= [];
+      functionCallGroups[baseName].push(key);
+    }
+  }
+
+  // For groups with multiple calls (same function, different args), remove those without return data
+  for (const baseName in functionCallGroups) {
+    const calls = functionCallGroups[baseName];
+    if (calls.length > 1) {
+      // Multiple calls to the same function with different arguments
+      for (const callPath of calls) {
+        // Check if this call has return data
+        const hasReturnData = allKeys.some(
+          (k: string) =>
+            k.startsWith(callPath + '.functionCallReturnValue') ||
+            k.startsWith(callPath + '.signature['),
+        );
+        if (!hasReturnData) {
+          delete mapping[callPath];
+        }
+      }
+    }
+  }
+}
+
 function transformPath({
   oldPath,
   newPath,
@@ -642,13 +715,16 @@ function handleFunctionMapping(
     }
 
     // Don't add [] element for ambiguous methods without other array evidence
+    // Also don't add [] if the path was already identified as a string type
     const methodName = getMethodName(lastPart);
     const isAmbiguousMethod = ambiguousArrayMethodsSet.has(methodName);
+    const isStringType = keyMapping[previousPath] === 'string';
     if (
       !isValidFunctionCall(previousPart) &&
       arrayMethodsSet.has(methodName) &&
       !keyMapping[previousPath + '[]'] &&
-      !isAmbiguousMethod
+      !isAmbiguousMethod &&
+      !isStringType
     ) {
       keyMapping[previousPath + '[]'] = 'unknown';
     }
@@ -901,6 +977,9 @@ function clearAttributes(
     }
   }
 
+  // Remove duplicate function calls (same function with different arguments) that have no return data
+  removeDuplicateFunctionCalls(mapping);
+
   for (const key in mapping) {
     const value = mapping[key];
     if (knownTypes.has(value)) {
@@ -918,11 +997,20 @@ function clearAttributes(
         basePath
       )
         continue;
+      // Preserve functionCallReturnValue for non-string types, UNLESS
+      // the method is a primitive-returning method (like some, every, includes, indexOf)
+      // whose return values are not mockable
+      const methodPart = keyParts[basePathParts.length];
+      const methodName = methodPart?.split('(')[0];
+      const isPrimitiveReturning = primitiveReturningMethodsSet.has(
+        methodName ?? '',
+      );
       if (
         mapping[basePath] !== 'string' &&
         keyParts[basePathParts.length + 1]?.startsWith(
           'functionCallReturnValue',
-        )
+        ) &&
+        !isPrimitiveReturning
       )
         continue;
 

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/convertDotNotation.ts

@@ -6,6 +6,14 @@ const STANDALONE_INDEX_RE = /^\[(\d+)\]$/;
 const isExplicitIndex = (s?: string) => !!s && INDEXED_RE.test(s);
 const isStandaloneIndex = (s?: string) => !!s && STANDALONE_INDEX_RE.test(s);
 
+// Detect .length properties that come from dynamic iteration patterns
+// These aren't useful for mock data - arrays work without explicit length,
+// and objects shouldn't have length. We filter .length that appears:
+// 1. After array notation: metadata[].length, items[][].length
+// 2. On object paths that likely come from iteration: metadata.length (when metadata has other props)
+// The regex matches any path ending with .length that has [] somewhere before it
+const DYNAMIC_LENGTH_RE = /\[\].*\.length$/;
+
 // Treat these as structural placeholders (don't commit them as concrete leaves)
 function isSkippableLeafType(t: string) {
   // 'unknown' by itself is a placeholder (but 'boolean | unknown' is not)
@@ -23,10 +31,13 @@ export default function convertDotNotation(
   const result: JsonTypeDefinition = {};
 
   for (const [rawPath, typ] of Object.entries(schema)) {
+    // Skip paths with .length after array notation from dynamic iteration
+    // e.g., metadata[].length, metadata[][].length
+    if (DYNAMIC_LENGTH_RE.test(rawPath)) continue;
+
     const parts = splitOutsideParenthesesAndArrays(rawPath);
     if (!parts || parts.length === 0) continue;
 
-    // eslint-disable-next-line @typescript-eslint/no-explicit-any
     let cursor: any = result;
 
     for (let i = 0; i < parts.length; i++) {
@@ -181,12 +192,24 @@ export default function convertDotNotation(
           if (!isSkippableLeafType(typ)) {
             cursor[key] = [typ];
           } else {
-            // structural/placeholder subtype — ensure array exists but don't clobber
-            if (!Array.isArray(cursor[key])) cursor[key] = [];
+            // structural/placeholder subtype — ensure array exists but don't clobber existing values
+            // Only create an array if nothing exists yet; preserve concrete types like 'string'
+            if (cursor[key] === undefined) cursor[key] = [];
           }
           i++; // consume '[]'
         } else {
           // Pattern like: "items", "[]", "id" -> items: [ { id: ... } ]
+          // But if cursor[key] is already a non-array object (from paths like "metadata: object"),
+          // don't overwrite it with an array - the object type should take precedence.
+          // This handles spurious "metadata[]" paths from dynamic iteration (Object.keys, for...in).
+          if (
+            cursor[key] !== undefined &&
+            typeof cursor[key] === 'object' &&
+            !Array.isArray(cursor[key])
+          ) {
+            // Skip this path - it's trying to treat an object as an array
+            break;
+          }
           if (!Array.isArray(cursor[key])) cursor[key] = [];
 
           // Look ahead to see what comes after '[]'

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/fillInSchemaGapsAndUnknowns.ts

@@ -208,10 +208,12 @@ export default function fillInSchemaGapsAndUnknowns(
 export function fillInDirectSchemaGapsAndUnknowns({
   scopeName,
   schema,
+  mergedSchema,
   attempts = 0,
 }: {
   scopeName?: string;
   schema: Record<string, string>;
+  mergedSchema?: Record<string, string>;
   attempts?: number;
 }) {
   try {
@@ -245,7 +247,9 @@ export function fillInDirectSchemaGapsAndUnknowns({
     const guessTypeForPath = (path: string, defaultType = 'string') => {
       let knownType = checkIfKnownType(path, functionKeysMapping);
       if (!knownType) {
-        knownType = checkIfObjectOrFunction(path, schema);
+        // Use mergedSchema for type inference if available (has more complete nested paths)
+        // Fall back to schema if mergedSchema is not provided
+        knownType = checkIfObjectOrFunction(path, mergedSchema ?? schema);
       }
 
       if (knownType && schema[path] !== knownType) {
@@ -373,10 +377,39 @@ export function fillInDirectSchemaGapsAndUnknowns({
       fillInDirectSchemaGapsAndUnknowns({
         scopeName,
         schema,
+        mergedSchema,
         attempts: ++attempts,
       });
     }
 
+    // Remove .length paths when the parent is typed as 'array' or 'string' (or nullable variants).
+    // These are built-in properties that were used for type inference but shouldn't appear
+    // in the final schema (they create { length: N } objects instead of proper arrays).
+    // However, if the parent could be an 'object' or other type, keep .length as it may be
+    // a custom property.
+    for (const key of Object.keys(schema)) {
+      const parts = splitOutsideParenthesesAndArrays(key);
+      if (parts[parts.length - 1] === 'length') {
+        const parentPath = joinParenthesesAndArrays(parts.slice(0, -1));
+        const parentType = schema[parentPath];
+        if (parentType) {
+          // Split union types and filter out nullable parts
+          const typeParts = parentType.split(' | ').map((t) => t.trim());
+          const nonNullableTypes = typeParts.filter(
+            (t) => t !== 'undefined' && t !== 'null',
+          );
+          // Only remove if ALL non-nullable types are array or string
+          // This ensures we don't remove .length from potential custom objects
+          if (
+            nonNullableTypes.length > 0 &&
+            nonNullableTypes.every((t) => t === 'array' || t === 'string')
+          ) {
+            delete schema[key];
+          }
+        }
+      }
+    }
+
     return schema;
   } catch (error) {
     console.info(
@@ -454,6 +487,8 @@ function checkIfKnownType(
   let isString: boolean | undefined;
   let hasStrongArrayEvidence = false;
   let hasOnlyAmbiguousArrayMethods = true;
+  let hasLengthAccess = false;
+  let hasArrayStringAmbiguousMethod = false;
 
   for (const functionOrAttributeName of functionAndAttributesKeysMapping[key] ??
     []) {
@@ -467,11 +502,18 @@ function checkIfKnownType(
       knownMethodCalls(functionOrAttributeName).includes('string') ||
       functionOrAttributeName === 'length';
 
-    // Check if this is strong evidence of an array
-    if (
-      STRONG_ARRAY_METHODS.has(methodName) ||
-      functionOrAttributeName === 'length'
-    ) {
+    // Track .length access
+    if (functionOrAttributeName === 'length') {
+      hasLengthAccess = true;
+    }
+
+    // Track methods that exist on both arrays and strings (like includes, indexOf)
+    if (couldBeArray && couldBeString && functionOrAttributeName !== 'length') {
+      hasArrayStringAmbiguousMethod = true;
+    }
+
+    // Check if this is strong evidence of an array (unambiguous array methods)
+    if (STRONG_ARRAY_METHODS.has(methodName)) {
       hasStrongArrayEvidence = true;
     }
 
@@ -495,10 +537,24 @@ function checkIfKnownType(
     }
   }
 
+  // Strong array evidence (unambiguous methods like .map, .filter) -> definitely array
+  if (isArray && hasStrongArrayEvidence) {
+    return 'array';
+  }
+
+  // When we have BOTH .length AND an ambiguous method like .includes(),
+  // prefer array. This pattern (checking .length AND calling .includes)
+  // is very common with arrays (e.g., arr.length > 0 && arr.includes(x))
+  // but rare with strings (you'd typically just call str.includes(x))
+  if (isArray && hasLengthAccess && hasArrayStringAmbiguousMethod) {
+    return 'array';
+  }
+
   if (isString) return 'string';
+
   // Only infer array if we have strong evidence OR not only ambiguous methods
   // This prevents router.push() from being incorrectly inferred as an array
-  if (isArray && (hasStrongArrayEvidence || !hasOnlyAmbiguousArrayMethods)) {
+  if (isArray && !hasOnlyAmbiguousArrayMethods) {
     return 'array';
   }
 }

package/analyzer-template/packages/ai/src/lib/dataStructure/helpers/getFunctionCallRoot.ts

@@ -1,11 +1,41 @@
 import { splitOutsideParenthesesAndArrays } from '../../splitOutsideParentheses';
 
+/**
+ * Find the position of the first '(' that is NOT inside angle brackets.
+ * Returns -1 if not found.
+ */
+function findFirstParenOutsideTypeParams(signature: string): number {
+  let angleDepth = 0;
+
+  for (let i = 0; i < signature.length; i++) {
+    const char = signature[i];
+
+    if (char === '<') {
+      angleDepth++;
+    } else if (char === '>') {
+      angleDepth--;
+    } else if (char === '(' && angleDepth === 0) {
+      return i;
+    }
+  }
+
+  return -1;
+}
+
 export default function getFunctionCallRoot(functionCallSignature: string) {
   const functionCallSignatureParts = splitOutsideParenthesesAndArrays(
     functionCallSignature,
   );
 
-  return functionCallSignatureParts[0]
-    .replace(/(cyScope\d+)F/, '$1')
-    .split('(')[0];
+  const firstPart = functionCallSignatureParts[0];
+
+  // Find the first '(' that's not inside type parameters (angle brackets)
+  // This handles cases like "useFetcher<{ data: UserData | null }>()"
+  // which should return "useFetcher<{ data: UserData | null }>"
+  // (preserving type params for scope node matching)
+  const parenPos = findFirstParenOutsideTypeParams(firstPart);
+
+  const result = parenPos >= 0 ? firstPart.substring(0, parenPos) : firstPart;
+
+  return result.replace(/(cyScope\d+)F/, '$1');
 }