@codeyam/codeyam-cli 0.1.0-staging.036391e → 0.1.0-staging.07d8f7b

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

@@ -397,6 +397,7 @@ const SILENTLY_IGNORED_EQUIVALENCY_REASONS = new Set([
   'transformed non-object function equivalency - implicit parent equivalency - rerouted via useCallback',
   'transformed non-object function equivalency - Array.from() equivalency',
   'Spread operator equivalency key update: Explicit array deconstruction equivalency value',
+  // 'transformed non-object function equivalency - Explicit array deconstruction equivalency value',
 ]);
 
 export class ScopeDataStructure {
@@ -1107,6 +1108,33 @@ export class ScopeDataStructure {
       return;
     }
 
+    // Case 3: Circular reference through scope-suffixed names (____cyScope pattern)
+    // When a named arrow function is defined inside a scope (e.g., useEffect callback):
+    //   const identifyUser = async () => { ... };
+    //   identifyUser();
+    // This creates a variable "identifyUser" and a scope "identifyUser____cyScope9F".
+    // Mutual equivalencies between these cause infinite loops in Phase 2 because
+    // processing one triggers addToSchema → followEquivalencies → addEquivalency
+    // on the reverse, which repeats indefinitely.
+    // Only block when the REVERSE direction already exists (creating a cycle).
+    // The initial one-directional equivalency is necessary for scope resolution.
+    if (
+      path &&
+      equivalentPath &&
+      (equivalentPath.startsWith(path + '____') ||
+        path.startsWith(equivalentPath + '____'))
+    ) {
+      // Check if the reverse equivalency already exists
+      const reverseEquivalencies =
+        scopeNode.equivalencies[equivalentPath] || [];
+      const reverseExists = reverseEquivalencies.some(
+        (v) => v.schemaPath === path,
+      );
+      if (reverseExists) {
+        return;
+      }
+    }
+
     if (!equivalentScopeName) {
       console.error(
         'CodeYam Error: Missing equivalent scope name - FULL CONTEXT:',
@@ -2797,6 +2825,8 @@ export class ScopeDataStructure {
             usageEquivalency.scopeNodeName,
           ) as ScopeNode;
 
+          if (!usageScopeNode) continue;
+
           // Guard against infinite recursion by tracking which paths we've already
           // added from addComplexSourcePathVariables
           if (
@@ -2876,6 +2906,8 @@ export class ScopeDataStructure {
             usageEquivalency.scopeNodeName,
           ) as ScopeNode;
 
+          if (!usageScopeNode) continue;
+
           // This is put in place to avoid propagating array functions like 'filter' through complex equivalencies
           // but may cause problems if the funtion call is not on a known object (e.g. string or array)
           if (
@@ -3355,6 +3387,22 @@ export class ScopeDataStructure {
    * ensure all sub-paths of that variable are reflected under signature[N].
    */
   private propagateParameterToSignaturePaths(scopeNode: ScopeNode) {
+    // Helper: check if a type is a concrete scalar that cannot have sub-properties.
+    const SCALAR_TYPES = new Set([
+      'string',
+      'number',
+      'boolean',
+      'bigint',
+      'symbol',
+      'void',
+      'never',
+    ]);
+    const isDefinitelyScalar = (type: string): boolean => {
+      const parts = type.split('|').map((s) => s.trim());
+      const base = parts.filter((s) => s !== 'undefined' && s !== 'null');
+      return base.length > 0 && base.every((b) => SCALAR_TYPES.has(b));
+    };
+
     // Find variable → signature[N] equivalencies
     for (const [varName, equivalencies] of Object.entries(
       scopeNode.equivalencies,
@@ -3385,7 +3433,38 @@ export class ScopeDataStructure {
 
               // Only add if the signature path doesn't already exist
               if (!scopeNode.schema[sigKey]) {
-                scopeNode.schema[sigKey] = scopeNode.schema[key];
+                // Check if this path represents variable conflation:
+                // When a standalone variable (e.g., showWorkoutForm from useState)
+                // appears as a sub-property of a scalar-typed ancestor (e.g.,
+                // activity_type = "string"), it's from scope conflation, not real
+                // property access. Block these while allowing legitimate built-in
+                // accesses like string.length or string.slice.
+                let isConflatedPath = false;
+                let checkPos = signaturePath.length;
+                while (true) {
+                  checkPos = sigKey.indexOf('.', checkPos + 1);
+                  if (checkPos === -1) break;
+                  const ancestorPath = sigKey.substring(0, checkPos);
+                  const ancestorType = scopeNode.schema[ancestorPath];
+                  if (ancestorType && isDefinitelyScalar(ancestorType)) {
+                    // Ancestor is scalar — check if the immediate sub-property
+                    // is also a standalone variable (indicating conflation)
+                    const afterDot = sigKey.substring(checkPos + 1);
+                    const nextSep = afterDot.search(/[.\[]/);
+                    const subPropName =
+                      nextSep === -1
+                        ? afterDot
+                        : afterDot.substring(0, nextSep);
+                    if (scopeNode.schema[subPropName] !== undefined) {
+                      isConflatedPath = true;
+                      break;
+                    }
+                  }
+                }
+
+                if (!isConflatedPath) {
+                  scopeNode.schema[sigKey] = scopeNode.schema[key];
+                }
               }
             }
           }
@@ -3898,25 +3977,116 @@ export class ScopeDataStructure {
       return [source];
     };
 
-    return entries.reduce(
-      (acc, entry) => {
-        if (entry.sourceCandidates.length === 0) return acc;
+    const acc = entries.reduce(
+      (result, entry) => {
+        if (entry.sourceCandidates.length === 0) return result;
         const usages = entry.usages.filter(usageMatchesScope);
         for (const usage of usages) {
-          acc[usage.schemaPath] ||= [];
+          result[usage.schemaPath] ||= [];
           // Resolve each source candidate through the equivalency chain
           for (const source of entry.sourceCandidates) {
             const resolvedSources = resolveToSignature(source, new Set());
-            acc[usage.schemaPath].push(...resolvedSources);
+            result[usage.schemaPath].push(...resolvedSources);
           }
         }
-        return acc;
+        return result;
       },
       {} as Record<
         string,
         Pick<ScopeVariable, 'scopeNodeName' | 'schemaPath'>[]
       >,
     );
+
+    // Post-processing: enrich useState-backed sources with co-located external
+    // function calls. When a useState value resolves to a setter variable that
+    // lives in the same scope as a fetch/API call, that fetch is a data source.
+    this.enrichUseStateSourcesWithCoLocatedCalls(acc);
+
+    return acc;
+  }
+
+  /**
+   * For each source that ends at a useState path, check if the setter was called
+   * from a scope that also contains external function calls (like fetch).
+   * If so, add those external calls as additional source candidates.
+   */
+  private enrichUseStateSourcesWithCoLocatedCalls(
+    acc: Record<string, Pick<ScopeVariable, 'scopeNodeName' | 'schemaPath'>[]>,
+  ) {
+    const rootScopeName = this.scopeTreeManager.getRootName();
+    const rootScope = this.scopeNodes[rootScopeName];
+    if (!rootScope) return;
+
+    // Collect all descendants for each scope node
+    const getAllDescendants = (
+      node: import('./helpers/ScopeTreeManager').ScopeTreeNode,
+    ): Set<string> => {
+      const names = new Set<string>([node.name]);
+      for (const child of node.children) {
+        for (const name of getAllDescendants(child)) {
+          names.add(name);
+        }
+      }
+      return names;
+    };
+
+    for (const [usagePath, sources] of Object.entries(acc)) {
+      const additionalSources: Pick<
+        ScopeVariable,
+        'scopeNodeName' | 'schemaPath'
+      >[] = [];
+
+      for (const source of sources) {
+        // Check if this source is a useState-related terminal path
+        // (e.g., useState(X).functionCallReturnValue[1] or useState(X).signature[0])
+        if (!source.schemaPath.match(/^useState\([^)]*\)\./)) continue;
+
+        // Find the useState call from the source path
+        const useStateCallMatch = source.schemaPath.match(
+          /^(useState\([^)]*\))\./,
+        );
+        if (!useStateCallMatch) continue;
+        const useStateCall = useStateCallMatch[1];
+
+        // Look in the root scope for the useState value equivalency
+        // which tells us where the setter was called from
+        const valuePath = `${useStateCall}.functionCallReturnValue[0]`;
+        const valueEquivs = rootScope.equivalencies[valuePath];
+        if (!valueEquivs) continue;
+
+        for (const equiv of valueEquivs) {
+          // Find the scope where the setter was called
+          const setterScopeName = equiv.scopeNodeName;
+          const setterScopeTree =
+            this.scopeTreeManager.findNode(setterScopeName);
+          if (!setterScopeTree) continue;
+
+          // Get all descendant scope names from the setter scope
+          const relatedScopes = getAllDescendants(setterScopeTree);
+
+          // Find external function calls in those scopes whose return values
+          // are actually consumed (assigned to a variable). This excludes
+          // fire-and-forget calls like analytics.track() or console.log().
+          const coLocatedCalls = this.externalFunctionCalls.filter(
+            (efc) =>
+              relatedScopes.has(efc.callScope) &&
+              efc.receivingVariableNames &&
+              efc.receivingVariableNames.length > 0,
+          );
+
+          for (const call of coLocatedCalls) {
+            additionalSources.push({
+              scopeNodeName: call.callScope,
+              schemaPath: `${call.callSignature}.functionCallReturnValue`,
+            });
+          }
+        }
+      }
+
+      if (additionalSources.length > 0) {
+        acc[usagePath].push(...additionalSources);
+      }
+    }
   }
 
   getUsageEquivalencies(functionName?: string) {
@@ -4069,6 +4239,24 @@ export class ScopeDataStructure {
       }
     }
 
+    // Helper: check if a type is a concrete scalar that cannot have sub-properties.
+    // e.g., "string", "number | undefined", "boolean | null" are scalar.
+    // "object", "array", "function", "unknown", "Workout", etc. are NOT scalar.
+    const SCALAR_TYPES = new Set([
+      'string',
+      'number',
+      'boolean',
+      'bigint',
+      'symbol',
+      'void',
+      'never',
+    ]);
+    const isDefinitelyScalarType = (type: string): boolean => {
+      const parts = type.split('|').map((s) => s.trim());
+      const base = parts.filter((s) => s !== 'undefined' && s !== 'null');
+      return base.length > 0 && base.every((b) => SCALAR_TYPES.has(b));
+    };
+
     // Propagate nested paths from variables to their signature equivalents
     // e.g., if workouts = signature[0].workouts, then workouts[].title becomes
     // signature[0].workouts[].title
@@ -4093,7 +4281,69 @@ export class ScopeDataStructure {
 
           // Add to schema if not already present
           if (!tempScopeNode.schema[signatureKey]) {
-            tempScopeNode.schema[signatureKey] = schema[schemaKey];
+            // Check if this path represents variable conflation:
+            // When a standalone variable (e.g., showWorkoutForm from useState)
+            // appears as a sub-property of a scalar-typed ancestor (e.g.,
+            // activity_type = "string"), it's from scope conflation, not real
+            // property access. Block these while allowing legitimate built-in
+            // accesses like string.length or string.slice.
+            let isConflatedPath = false;
+            let checkPos = signaturePath.length;
+            while (true) {
+              checkPos = signatureKey.indexOf('.', checkPos + 1);
+              if (checkPos === -1) break;
+              const ancestorPath = signatureKey.substring(0, checkPos);
+              const ancestorType = tempScopeNode.schema[ancestorPath];
+              if (ancestorType && isDefinitelyScalarType(ancestorType)) {
+                // Ancestor is scalar — check if the immediate sub-property
+                // is also a standalone variable (indicating conflation)
+                const afterDot = signatureKey.substring(checkPos + 1);
+                const nextSep = afterDot.search(/[.\[]/);
+                const subPropName =
+                  nextSep === -1 ? afterDot : afterDot.substring(0, nextSep);
+                if (schema[subPropName] !== undefined) {
+                  isConflatedPath = true;
+                  break;
+                }
+              }
+            }
+
+            if (!isConflatedPath) {
+              tempScopeNode.schema[signatureKey] = schema[schemaKey];
+            }
+          }
+        }
+      }
+    }
+
+    // Post-process: filter out conflated signature paths.
+    // During phase 2 scope analysis, useState(false) conflation can create
+    // bad paths like signature[0].mockWorkouts[].activity_type.showWorkoutForm
+    // directly in scopeNode.schema. These flow through signatureInSchema into
+    // tempScopeNode.schema without any guard. Filter them out here by checking:
+    //   1. An ancestor in the path has a concrete scalar type (string, number, boolean, etc.)
+    //   2. The immediate sub-property of that scalar ancestor is also a standalone
+    //      variable in the schema (indicating conflation, not a real property access)
+    for (const key of Object.keys(tempScopeNode.schema)) {
+      if (!key.startsWith('signature[')) continue;
+
+      // Walk through the path looking for scalar-typed ancestors
+      let pos = 0;
+      while (true) {
+        pos = key.indexOf('.', pos + 1);
+        if (pos === -1) break;
+        const ancestorPath = key.substring(0, pos);
+        const ancestorType = tempScopeNode.schema[ancestorPath];
+        if (ancestorType && isDefinitelyScalarType(ancestorType)) {
+          // Found a scalar ancestor — check if the sub-property name
+          // is a standalone variable in the getSchema() result
+          const afterDot = key.substring(pos + 1);
+          const nextSep = afterDot.search(/[.\[]/);
+          const subPropName =
+            nextSep === -1 ? afterDot : afterDot.substring(0, nextSep);
+          if (schema[subPropName] !== undefined) {
+            delete tempScopeNode.schema[key];
+            break;
           }
         }
       }
@@ -4423,6 +4673,15 @@ export class ScopeDataStructure {
           !equivalentValue.schemaPath.startsWith('signature[') && // not a signature path
           !equivalentValue.schemaPath.endsWith('.functionCallReturnValue') // not already handled above
         ) {
+          // Skip bare "returnValue" from child scopes — this is the child's return value,
+          // not a meaningful data source path in the parent scope
+          if (
+            equivalentValue.schemaPath === 'returnValue' &&
+            equivalentValue.scopeNodeName !==
+              this.scopeTreeManager.getRootName()
+          ) {
+            continue;
+          }
           // Add equivalency (will accumulate if multiple values for OR expressions)
           addEquivalency(path, equivalentValue.schemaPath);
         }
@@ -4824,9 +5083,109 @@ export class ScopeDataStructure {
     // Replace cyScope placeholders in all external function call data
     // This ensures call signatures and schema paths use actual callback text
     // instead of internal cyScope names, preventing mock data merge conflicts.
-    return this.externalFunctionCalls.map((efc) =>
-      this.cleanCyScopeFromFunctionCallInfo(efc),
-    );
+    const rootScopeName = this.scopeTreeManager.getRootName();
+    const rootSchema = this.scopeNodes[rootScopeName]?.schema ?? {};
+
+    return this.externalFunctionCalls.map((efc) => {
+      const cleaned = this.cleanCyScopeFromFunctionCallInfo(efc);
+      return this.filterConflatedExternalPaths(cleaned, rootSchema);
+    });
+  }
+
+  /**
+   * Filters out conflated paths from external function call schemas.
+   *
+   * When multiple useState(false) calls create equivalency conflation during
+   * Phase 1 analysis, standalone boolean state variables (like showWorkoutForm,
+   * showGoalForm) can bleed into external function call schemas as sub-properties
+   * of unrelated data fields (like data[].activity_type.showWorkoutForm).
+   *
+   * Detection: group sub-properties by parent path. If 2+ sub-properties of
+   * the same parent all match standalone root scope variable names, treat them
+   * as conflation artifacts and remove them.
+   */
+  private filterConflatedExternalPaths(
+    efc: FunctionCallInfo,
+    rootSchema: Record<string, string>,
+  ): FunctionCallInfo {
+    // Build a set of top-level root scope variable names (simple names, no dots/brackets)
+    const topLevelRootVars = new Set<string>();
+    for (const key of Object.keys(rootSchema)) {
+      if (!key.includes('.') && !key.includes('[')) {
+        topLevelRootVars.add(key);
+      }
+    }
+
+    if (topLevelRootVars.size === 0) return efc;
+
+    // Group sub-property matches by their parent path.
+    // For a path like "...data[].activity_type.showWorkoutForm",
+    // parent = "...data[].activity_type", child = "showWorkoutForm"
+    const parentToConflatedKeys = new Map<string, string[]>();
+
+    for (const key of Object.keys(efc.schema)) {
+      const lastDot = key.lastIndexOf('.');
+      if (lastDot === -1) continue;
+
+      const parent = key.substring(0, lastDot);
+      const child = key.substring(lastDot + 1);
+
+      // Skip array access or function call patterns
+      if (child.includes('[') || child.includes('(')) continue;
+
+      // Only consider paths inside array element chains (contains []).
+      // Direct children of functionCallReturnValue are legitimate destructured
+      // return values, not conflation. Conflation happens deeper in the chain
+      // when array element fields get corrupted sub-properties.
+      if (!parent.includes('[')) continue;
+
+      if (topLevelRootVars.has(child)) {
+        if (!parentToConflatedKeys.has(parent)) {
+          parentToConflatedKeys.set(parent, []);
+        }
+        parentToConflatedKeys.get(parent)!.push(key);
+      }
+    }
+
+    // Only filter when 2+ sub-properties of the same parent match root scope vars.
+    // This threshold avoids false positives from coincidental name matches.
+    const keysToRemove = new Set<string>();
+    const parentsToRestore = new Set<string>();
+
+    for (const [parent, conflatedKeys] of parentToConflatedKeys) {
+      if (conflatedKeys.length >= 2) {
+        for (const key of conflatedKeys) {
+          keysToRemove.add(key);
+        }
+        parentsToRestore.add(parent);
+      }
+    }
+
+    if (keysToRemove.size === 0) return efc;
+
+    // Create a new schema without the conflated paths
+    const newSchema: Record<string, string> = {};
+    for (const [key, value] of Object.entries(efc.schema)) {
+      if (keysToRemove.has(key)) continue;
+
+      // Restore parent type: if it was changed to "object" because of conflated
+      // sub-properties, and now all those sub-properties are removed, change it
+      // back to "unknown" (we don't know the original type)
+      if (parentsToRestore.has(key) && value === 'object') {
+        // Check if there are any remaining sub-properties
+        const hasRemainingSubProps = Object.keys(efc.schema).some(
+          (k) =>
+            !keysToRemove.has(k) &&
+            k !== key &&
+            (k.startsWith(key + '.') || k.startsWith(key + '[')),
+        );
+        newSchema[key] = hasRemainingSubProps ? value : 'unknown';
+      } else {
+        newSchema[key] = value;
+      }
+    }
+
+    return { ...efc, schema: newSchema };
   }
 
   /**
@@ -5119,6 +5478,10 @@ export class ScopeDataStructure {
   getEnrichedConditionalUsages(): Record<string, EnrichedConditionalUsage[]> {
     const enriched: Record<string, EnrichedConditionalUsage[]> = {};
 
+    console.log(
+      `[getEnrichedConditionalUsages] Processing ${Object.keys(this.rawConditionalUsages).length} conditional paths: [${Object.keys(this.rawConditionalUsages).join(', ')}]`,
+    );
+
     for (const [path, usages] of Object.entries(this.rawConditionalUsages)) {
       // Try to trace this path back to a data source
       // First, try the root scope
@@ -5127,10 +5490,69 @@ export class ScopeDataStructure {
 
       let sourceDataPath: string | undefined;
       if (explanation.source) {
-        // Build the full data path: scopeName.path
-        sourceDataPath = `${explanation.source.scope}.${explanation.source.path}`;
+        const { scope, path: sourcePath } = explanation.source;
+
+        // Build initial path — avoid redundant prefix when path already contains the scope call
+        let fullPath: string;
+        if (sourcePath.startsWith(`${scope}(`)) {
+          fullPath = sourcePath;
+        } else {
+          fullPath = `${scope}.${sourcePath}`;
+        }
+
+        sourceDataPath = fullPath;
+        console.log(
+          `[getEnrichedConditionalUsages] "${path}" explainPath → scope="${scope}", sourcePath="${sourcePath}" → sourceDataPath="${sourceDataPath}"`,
+        );
+      } else {
+        console.log(
+          `[getEnrichedConditionalUsages] "${path}" explainPath → no source found`,
+        );
       }
 
+      // If explainPath didn't find a useful external source (e.g., it traced to
+      // useState or just to the component scope itself), check sourceEquivalencies
+      // for an external function call source like a fetch call
+      const hasExternalSource = sourceDataPath?.includes(
+        '.functionCallReturnValue',
+      );
+      if (!hasExternalSource) {
+        console.log(
+          `[getEnrichedConditionalUsages] "${path}" no external source (sourceDataPath="${sourceDataPath}"), checking sourceEquivalencies fallback...`,
+        );
+        const sourceEquiv = this.getSourceEquivalencies();
+        const returnValueKey = `returnValue.${path}`;
+        const sources = sourceEquiv[returnValueKey];
+        if (sources) {
+          console.log(
+            `[getEnrichedConditionalUsages] "${path}" sourceEquivalencies["${returnValueKey}"] has ${sources.length} sources: [${sources.map((s: { schemaPath: string }) => s.schemaPath).join(', ')}]`,
+          );
+          const externalSource = sources.find(
+            (s: { schemaPath: string }) =>
+              s.schemaPath.includes('.functionCallReturnValue') &&
+              !s.schemaPath.startsWith('useState('),
+          );
+          if (externalSource) {
+            console.log(
+              `[getEnrichedConditionalUsages] "${path}" sourceEquivalencies fallback found external source: "${externalSource.schemaPath}"`,
+            );
+            sourceDataPath = externalSource.schemaPath;
+          } else {
+            console.log(
+              `[getEnrichedConditionalUsages] "${path}" sourceEquivalencies fallback found no external function call source`,
+            );
+          }
+        } else {
+          console.log(
+            `[getEnrichedConditionalUsages] "${path}" sourceEquivalencies["${returnValueKey}"] not found`,
+          );
+        }
+      }
+
+      console.log(
+        `[getEnrichedConditionalUsages] "${path}" FINAL sourceDataPath="${sourceDataPath ?? '(none)'}" (${usages.length} usages)`,
+      );
+
       enriched[path] = usages.map((usage) => ({
         ...usage,
         sourceDataPath,

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

@@ -0,0 +1,62 @@
+/**
+ * Coerces primitive values to empty arrays when the schema expects an array.
+ *
+ * The LLM sometimes generates a primitive (boolean, number, string) where the
+ * schema expects an array. For example:
+ *   Schema: { webapps: [{ name: "string" }] }
+ *   LLM output: { webapps: true }
+ *
+ * This causes runtime errors like `TypeError: config.webapps?.forEach is not a function`
+ * because the code expects an array but gets a boolean.
+ *
+ * Coercion rule:
+ * - primitive → array: replace with []
+ *
+ * Null and undefined values are preserved (they may be intentionally nullable).
+ *
+ * @param data The mock data object to coerce (mutated in place)
+ * @param schema The schema describing expected types
+ */
+export default function coercePrimitivesToArraysBySchema<T>(
+  data: T,
+  schema: unknown,
+): T {
+  if (data === null || data === undefined) {
+    return data;
+  }
+
+  if (typeof data !== 'object') {
+    return data;
+  }
+
+  if (Array.isArray(data)) {
+    const itemSchema = Array.isArray(schema) ? schema[0] : undefined;
+    for (let i = 0; i < data.length; i++) {
+      data[i] = coercePrimitivesToArraysBySchema(data[i], itemSchema);
+    }
+    return data;
+  }
+
+  const dataObj = data as Record<string, unknown>;
+  const schemaObj = schema as Record<string, unknown> | undefined;
+
+  for (const key of Object.keys(dataObj)) {
+    const value = dataObj[key];
+    const fieldSchema = schemaObj?.[key];
+
+    if (value === null || value === undefined) {
+      continue;
+    }
+
+    if (typeof value !== 'object' && Array.isArray(fieldSchema)) {
+      // Schema expects an array but the value is a primitive.
+      // Replace with an empty array.
+      dataObj[key] = [];
+    } else if (typeof value === 'object' && value !== null) {
+      // Both value and schema are objects — recurse
+      coercePrimitivesToArraysBySchema(value, fieldSchema);
+    }
+  }
+
+  return data;
+}