@codeyam/codeyam-cli 0.1.0-staging.b8a55ba → 0.1.0-staging.c90f8c9

package/analyzer-template/project/serverOnlyModules.ts

@@ -213,6 +213,108 @@ export function applyServerOnlyMocks(fileContent: string): string {
   return applyServerOnlyMocksAst(fileContent);
 }
 
+/**
+ * Check if a file exports an "env" variable that's derived from a node module import.
+ * This is a generic pattern for env configuration packages (like @t3-oss/env-*, envalid, etc.)
+ * that create server-only environment objects.
+ *
+ * Pattern detected:
+ *   import { someFunc } from "some-package";
+ *   export const env = someFunc({...});
+ *
+ * Returns the name of the imported function if detected, null otherwise.
+ */
+function detectEnvStoreFromNodeModule(sourceFile: ts.SourceFile): {
+  importedFunctionName: string;
+  modulePath: string;
+} | null {
+  // First, collect all imported function names from node modules
+  const nodeModuleImports = new Map<string, string>(); // functionName -> modulePath
+
+  for (const statement of sourceFile.statements) {
+    if (!ts.isImportDeclaration(statement)) continue;
+
+    const moduleSpecifier = statement.moduleSpecifier;
+    if (!ts.isStringLiteral(moduleSpecifier)) continue;
+
+    const modulePath = moduleSpecifier.text;
+    // Skip relative imports - only look at node modules
+    if (modulePath.startsWith('.') || modulePath.startsWith('/')) continue;
+
+    if (
+      statement.importClause?.namedBindings &&
+      ts.isNamedImports(statement.importClause.namedBindings)
+    ) {
+      for (const element of statement.importClause.namedBindings.elements) {
+        nodeModuleImports.set(element.name.text, modulePath);
+      }
+    }
+    if (statement.importClause?.name) {
+      nodeModuleImports.set(statement.importClause.name.text, modulePath);
+    }
+  }
+
+  // Now look for: export const env = importedFunc(...)
+  for (const statement of sourceFile.statements) {
+    if (!ts.isVariableStatement(statement)) continue;
+
+    // Check if it's exported
+    const isExported = statement.modifiers?.some(
+      (m) => m.kind === ts.SyntaxKind.ExportKeyword,
+    );
+    if (!isExported) continue;
+
+    for (const decl of statement.declarationList.declarations) {
+      // Check if variable is named "env"
+      if (!ts.isIdentifier(decl.name) || decl.name.text !== 'env') continue;
+
+      // Check if initializer is a call expression
+      if (!decl.initializer || !ts.isCallExpression(decl.initializer)) continue;
+
+      // Get the function being called
+      const callExpr = decl.initializer.expression;
+      let funcName: string | null = null;
+
+      if (ts.isIdentifier(callExpr)) {
+        funcName = callExpr.text;
+      } else if (
+        ts.isPropertyAccessExpression(callExpr) &&
+        ts.isIdentifier(callExpr.name)
+      ) {
+        funcName = callExpr.name.text;
+      }
+
+      if (funcName && nodeModuleImports.has(funcName)) {
+        return {
+          importedFunctionName: funcName,
+          modulePath: nodeModuleImports.get(funcName)!,
+        };
+      }
+    }
+  }
+
+  return null;
+}
+
+/**
+ * Generate mock code for env store patterns.
+ * Replaces the imported function with one that returns a proxy.
+ */
+function generateEnvStoreMock(functionName: string): string {
+  return `// Mock for env store function (server-only env vars)
+const ${functionName} = (config: any) => {
+  // Return a proxy that returns undefined for any property access
+  // This allows the code to run on the client without throwing
+  return new Proxy({} as any, {
+    get: (target, prop) => {
+      if (prop === 'then' || prop === Symbol.toStringTag) return undefined;
+      return undefined;
+    },
+  });
+};
+`;
+}
+
 /**
  * AST-based implementation of applyServerOnlyMocks for large files.
  * Uses TypeScript parser instead of regex to avoid catastrophic backtracking.
@@ -229,6 +331,10 @@ function applyServerOnlyMocksAst(fileContent: string): string {
     const replacements: { start: number; end: number; replacement: string }[] =
       [];
 
+    // Check if this file exports an "env" variable from a node module function call
+    // This is a generic pattern for env configuration packages
+    const envStoreInfo = detectEnvStoreFromNodeModule(sourceFile);
+
     for (const statement of sourceFile.statements) {
       if (!ts.isImportDeclaration(statement)) continue;
 
@@ -236,8 +342,6 @@ function applyServerOnlyMocksAst(fileContent: string): string {
       if (!ts.isStringLiteral(moduleSpecifier)) continue;
 
       const modulePath = moduleSpecifier.text;
-      const config = findServerOnlyModule(modulePath);
-      if (!config) continue;
 
       // Extract import names
       const importNames: string[] = [];
@@ -265,6 +369,27 @@ function applyServerOnlyMocksAst(fileContent: string): string {
         }
       }
 
+      // Check if this import provides the function used to create the env store
+      if (
+        envStoreInfo &&
+        modulePath === envStoreInfo.modulePath &&
+        importNames.includes(envStoreInfo.importedFunctionName)
+      ) {
+        const mockCode = generateEnvStoreMock(
+          envStoreInfo.importedFunctionName,
+        );
+        replacements.push({
+          start: statement.getStart(sourceFile),
+          end: statement.getEnd(),
+          replacement: mockCode,
+        });
+        continue;
+      }
+
+      // Check for known server-only modules (from explicit list)
+      const config = findServerOnlyModule(modulePath);
+      if (!config) continue;
+
       const mockCode = config.getMockCode(importNames, defaultImportName);
       replacements.push({
         start: statement.getStart(sourceFile),

package/analyzer-template/project/start.ts

@@ -1,4 +1,5 @@
 import 'dotenv/config';
+import { once } from 'events';
 import yargs from 'yargs';
 import { hideBin } from 'yargs/helpers';
 import {
@@ -23,7 +24,7 @@ import {
   syncHeadBranches,
   syncPrimaryBranch,
 } from '~codeyam/github';
-import { installSignalHandlers } from '../process';
+import { installSignalHandlers } from '~codeyam/process';
 
 // Set process title for Activity Monitor visibility
 if (process.env.CODEYAM_PROCESS_NAME) {
@@ -207,7 +208,16 @@ async function main({
         while (!controller?.hasCompletedRepoCheckout()) {
           await new Promise((resolve) => setTimeout(resolve, 1000));
         }
-        awsLog('CodeYam: Repo checkout complete, starting orchestration...');
+        awsLog('CodeYam: Repo checkout complete.');
+
+        // Wait for analysis to enter producing phase (after prepareDataStructures)
+        // This prevents the 20-minute orchestration timeout from triggering before
+        // any analyses become ready to capture
+        awsLog('CodeYam: Waiting for analysis to start producing...');
+        if (!controller.hasStartedProducing()) {
+          await once(controller, 'analysisProducing');
+        }
+        awsLog('CodeYam: Analysis producing, starting orchestration...');
 
         // Create task runner based on orchestration mode
         let taskRunner;
@@ -260,8 +270,8 @@ async function main({
             }
           },
           project: sharedContext.project,
-          // In local-sequential mode, don't filter by commit since each analysis run
-          // creates a new fake commit. We want to capture all ready analyses for the project.
+          // In local-sequential mode, don't filter by commit (full=true handles that).
+          // For AWS mode, filter by the specific commit being analyzed.
           commit:
             orchestrateCapture === 'local-sequential'
               ? null
@@ -279,6 +289,8 @@ async function main({
               : envConfig.entityShas,
           full: orchestrateCapture === 'local-sequential',
           taskRunner,
+          // Always track capture progress on the commit, even when not filtering by it.
+          metadataCommit: sharedContext.commit ?? sharedContext.branchCommit,
         });
         awsLog('CodeYam Capture: runCaptureOrchestration() completed');
       };

package/analyzer-template/project/startScenarioCapture.ts

@@ -36,6 +36,7 @@ import {
 } from './utils/webappMapping';
 import mergeEnvVarsIntoStartCommand from './utils/mergeEnvVarsIntoStartCommand';
 import { LazyFileStore } from './LazyFileStore';
+import { DEFAULT_SCENARIO_NAME } from '../common/constants';
 
 export interface StartScenarioCaptureArgs {
   readyToBeCaptured: Analysis[];
@@ -226,6 +227,11 @@ export async function startScenarioCapture({
     }
 
     const sortedScenarios = (analysis.scenarios ?? []).sort((a, b) => {
+      // Default Scenario always goes first
+      if (a.name === DEFAULT_SCENARIO_NAME) return -1;
+      if (b.name === DEFAULT_SCENARIO_NAME) return 1;
+
+      // Then prioritize new or revised scenarios
       const newOrRevisedA =
         !a.previousVersionId || a.metadata?.revisedFromPreviousVersion;
       const newOrRevisedB =

package/analyzer-template/project/writeMockDataTsx.ts

@@ -43,6 +43,108 @@ function normalizeParamsMockData(
   return result;
 }
 
+/**
+ * Convert comma-separated string values to arrays when they look like array data.
+ * This handles cases where the LLM generates strings like "color,size" instead
+ * of arrays like ["color", "size"] due to schema type misdetection.
+ */
+function convertCommaSeparatedStringsToArrays(
+  mockData: Record<string, unknown>,
+): Record<string, unknown> {
+  const result: Record<string, unknown> = {};
+
+  for (const [key, value] of Object.entries(mockData)) {
+    if (typeof value === 'string') {
+      const asArray = parseCommaSeparatedStringAsArray(value, key);
+      if (asArray !== null) {
+        result[key] = asArray;
+        console.log(
+          `CodeYam: Converted comma-separated string to array for key "${key}": "${value}" -> [${asArray.map((s) => `"${s}"`).join(', ')}]`,
+        );
+      } else {
+        result[key] = value;
+      }
+    } else if (
+      value !== null &&
+      typeof value === 'object' &&
+      !Array.isArray(value)
+    ) {
+      // Recursively process nested objects
+      result[key] = convertCommaSeparatedStringsToArrays(
+        value as Record<string, unknown>,
+      );
+    } else if (Array.isArray(value)) {
+      // Recursively process arrays (each element could be an object)
+      result[key] = value.map((item) => {
+        if (item !== null && typeof item === 'object' && !Array.isArray(item)) {
+          return convertCommaSeparatedStringsToArrays(
+            item as Record<string, unknown>,
+          );
+        }
+        return item;
+      });
+    } else {
+      result[key] = value;
+    }
+  }
+
+  return result;
+}
+
+/**
+ * Detect if a string looks like a comma-separated list that should be an array.
+ */
+function parseCommaSeparatedStringAsArray(
+  value: string,
+  key: string,
+): string[] | null {
+  if (!value.includes(',')) {
+    return null;
+  }
+
+  // Heuristic: if the key name suggests it's an array field, convert it
+  // Check this FIRST before sentence detection, since "color, size" looks like
+  // a sentence but should be converted to ["color", "size"] for _attributes fields
+  const arrayFieldPatterns = [
+    /_attributes$/i,
+    /_ids$/i,
+    /_items$/i,
+    /_tags$/i,
+    /_values$/i,
+    /_types$/i,
+    /_names$/i,
+    /_keys$/i,
+    /^attributes$/i,
+    /^items$/i,
+    /^tags$/i,
+    /^values$/i,
+  ];
+
+  const looksLikeArrayField = arrayFieldPatterns.some((pattern) =>
+    pattern.test(key),
+  );
+
+  if (looksLikeArrayField) {
+    // For array-like fields, only reject if it contains newlines (likely prose)
+    if (value.includes('\n')) {
+      return null;
+    }
+    const parts = value.split(',').map((s) => s.trim());
+    return parts.filter((s) => s.length > 0);
+  }
+
+  // For non-array-named fields, apply stricter sentence detection
+  // Looks like a sentence: "Hello, world"
+  if (/,\s+[a-z]/.test(value)) {
+    return null;
+  }
+  if (value.includes('\n')) {
+    return null;
+  }
+
+  return null;
+}
+
 /**
  * Return true when `markup` contains 0 or >1 meaningful
  * top-level nodes (element *or* non-blank text), so React
@@ -81,12 +183,12 @@ const cleanOutTypes = (str: string) => {
 };
 
 const safeString = (s: string, level: number) => {
-  if (s?.endsWith && s.endsWith(')') && (level === 1 || level === 2)) {
-    // remove arguments from top level functions
-    s = s.split('(')[0] + '()';
-    // strip type parameters like <typeof loader>
-    s = cleanOutTypes(s);
-  }
+  // NOTE: We no longer strip arguments from top-level function keys (level 2)
+  // because normalizeMockData now handles merging duplicate keys while preserving
+  // original keys when there's only one call site. This ensures constructMockCode
+  // can look up data with the full key (e.g., "useCustomSizes(projectSlug)").
+  //
+  // Only escape quotes and newlines for safe string output.
   return s?.replace ? s.replace(/"/g, "'").replace(/\n/g, '\\n') : s;
 };
 
@@ -135,12 +237,14 @@ const shouldBeMap = (mockKey: string, mergedDataStructure: any): boolean => {
 };
 
 /**
- * Normalize a function call key by stripping arguments and type parameters,
- * but preserving variable qualifiers (::variableName suffix).
+ * Normalize a function call key by stripping arguments but PRESERVING type parameters.
+ * Type parameters like <typeof loader> are important for disambiguation.
+ * Variable qualifiers (::variableName suffix) are also preserved.
+ *
  * e.g., "useLastLogLine(projectSlug, !!simulatingEntitySha)" -> "useLastLogLine()"
- * e.g., "useLoaderData<typeof loader>()" -> "useLoaderData()"
+ * e.g., "useLoaderData<typeof loader>()" -> "useLoaderData<typeof loader>()" (PRESERVED)
  * e.g., "useFetcher()::entityDiffFetcher" -> "useFetcher()::entityDiffFetcher" (preserved)
- * e.g., "useFetcher<T>()::entityDiffFetcher" -> "useFetcher()::entityDiffFetcher"
+ * e.g., "useFetcher<T>()::entityDiffFetcher" -> "useFetcher<T>()::entityDiffFetcher" (PRESERVED)
  */
 const normalizeKey = (key: string): string => {
   // Check for variable qualifier (::variableName)
@@ -150,10 +254,34 @@ const normalizeKey = (key: string): string => {
 
   // Check if key ends with ) - indicating a function call (after removing qualifier)
   if (keyWithoutQualifier.endsWith(')') && keyWithoutQualifier.includes('(')) {
-    // Strip arguments: "foo(a, b)" -> "foo()"
-    const normalized = keyWithoutQualifier.split('(')[0] + '()';
-    // Strip type parameters: "foo<T>()" -> "foo()"
-    return cleanOutTypes(normalized) + qualifier;
+    // Find the opening paren that matches the closing paren (accounting for nested parens and generics)
+    // We want to strip arguments but preserve type parameters
+    // e.g., "useLoaderData<typeof loader>(arg1, arg2)" -> "useLoaderData<typeof loader>()"
+    // e.g., "foo(a, b)" -> "foo()"
+
+    // Find where type parameters end (if any) by finding matching < >
+    let typeParamEnd = 0;
+    if (keyWithoutQualifier.includes('<')) {
+      let depth = 0;
+      for (let i = 0; i < keyWithoutQualifier.length; i++) {
+        if (keyWithoutQualifier[i] === '<') depth++;
+        if (keyWithoutQualifier[i] === '>') {
+          depth--;
+          if (depth === 0) {
+            typeParamEnd = i + 1;
+            break;
+          }
+        }
+      }
+    }
+
+    // Find the opening paren for function arguments (after type params)
+    const argsStart = keyWithoutQualifier.indexOf('(', typeParamEnd);
+    if (argsStart !== -1) {
+      // Preserve everything up to and including the opening paren, then close it
+      const normalized = keyWithoutQualifier.slice(0, argsStart) + '()';
+      return normalized + qualifier;
+    }
   }
   return key;
 };
@@ -245,7 +373,11 @@ const normalizeMockData = (
   // Second pass: merge values for duplicate normalized keys
   for (const [normKey, originalKeys] of keyMapping) {
     if (originalKeys.length === 1) {
-      // No duplicates, use the original key (may be normalized or not based on safeString)
+      // No duplicates - PRESERVE the original key with arguments
+      // This is critical because constructMockCode generates code that accesses
+      // data with the FULL key (e.g., scenarios().data()?.["useCustomSizes(projectSlug)"])
+      // If we normalize to "useCustomSizes()" but code looks for "useCustomSizes(projectSlug)",
+      // the lookup fails and returns undefined, causing runtime errors.
       const value = mockData[originalKeys[0]];
       // Skip bare type markers like "function" or "async-function"
       if (isBareTypeMarker(value)) {
@@ -256,7 +388,7 @@ const normalizeMockData = (
       if (isEmptyObject(value)) {
         continue;
       }
-      normalized[normKey] = value;
+      normalized[originalKeys[0]] = value;
     } else {
       // Multiple keys normalize to the same value - merge them
       let mergedValue: unknown = {};
@@ -284,10 +416,12 @@ const normalizeMockData = (
         }
       }
       // Only add if we have actual data (not just filtered type markers)
+      // Note: typeof null === 'object' in JS, so we must check for null explicitly
       if (
         mergedValue !== undefined &&
         !isBareTypeMarker(mergedValue) &&
         !(
+          mergedValue !== null &&
           typeof mergedValue === 'object' &&
           Object.keys(mergedValue as object).length === 0
         )
@@ -415,7 +549,7 @@ const jsonToCode = (
               `${levelSpacing}"${safeString(key, level)}": ${arrayString}`,
             );
           }
-        } else if (typeof jsonObj[key] === 'object') {
+        } else if (jsonObj[key] !== null && typeof jsonObj[key] === 'object') {
           const objKeys = Object.keys(jsonObj[key] as object);
 
           // Check if this is a nested function signature (e.g., "postMessage(message, targetOrigin)": { signature: [...] })
@@ -716,14 +850,17 @@ export default async function writeMockDataTsx(
   const expectedKeys = extractAllKeys(dataForMocks);
 
   // Reconcile mock data keys to match expected keys from dataForMocks
-  // This handles cases where LLM generated literal values instead of variable references,
-  // and remaps old variable-based keys to canonical keys
-  const defaultMockData = reconcileMockDataKeys(
+  // This handles cases where LLM generated literal values instead of variable references
+  const reconciledMockData = reconcileMockDataKeys(
     rawDefaultMockData,
     expectedKeys,
-    analysis.entityName,
   );
 
+  // Convert comma-separated strings to arrays for fields that should be arrays
+  // This handles cases where LLM generates "color,size" instead of ["color", "size"]
+  const defaultMockData =
+    convertCommaSeparatedStringsToArrays(reconciledMockData);
+
   const mockDataComment = `// This file is auto-generated by CodeYam. Do not edit this file manually.
 // This file contains mock data required for:
 // Scenario: ${mockScenario.id} - ${mockScenario.name}
@@ -761,11 +898,14 @@ import React from 'react';
     };
 
     // Reconcile scenario-specific mock data keys as well
-    // Pass entityName for canonical key remapping
-    const scenarioMockData = reconcileMockDataKeys(
+    const reconciledScenarioMockData = reconcileMockDataKeys(
       rawScenarioMockData,
       expectedKeys,
-      analysis.entityName,
+    );
+
+    // Convert comma-separated strings to arrays for scenario data
+    const scenarioMockData = convertCommaSeparatedStringsToArrays(
+      reconciledScenarioMockData,
     );
 
     const mergedArguments = defaultArguments.map((defaultArg, index) => {

package/analyzer-template/project/writeScenarioClientWrapper.ts

@@ -0,0 +1,21 @@
+import { writeFile } from 'node:fs/promises';
+import { join } from 'node:path';
+import type { Scenario } from '~codeyam/types';
+import { generateScenarioClientWrapper } from '~codeyam/generate';
+
+/**
+ * Writes the ScenarioClientWrapper component to the mocks directory.
+ * This client component is used by Server Component pages to handle
+ * interactive features (iframe messaging, console interception, etc.)
+ */
+export default async function writeScenarioClientWrapper(
+  scenario: Scenario,
+  mocksDir: string,
+): Promise<string> {
+  const content = generateScenarioClientWrapper(scenario);
+  const filePath = join(mocksDir, 'ScenarioClientWrapper.tsx');
+
+  await writeFile(filePath, content);
+
+  return filePath;
+}