more cleanup

Author: ktalebian

src/metrics/chat-processor.ts

@@ -3,12 +3,16 @@ import path from 'path';
 
 import { logger } from '../utils';
 import {
-  ApiHistoryEntry,
-  CONTROL_MARKER,
-  MCP_MARKER,
-  TaskSegment,
-  UIMessage,
-} from './types';
+  createTimestampBoundaries,
+  extractConversationHistoryIndex,
+  filterRelevantMessages,
+  findTaskBoundaries,
+  getLastValidMessageTimestamp,
+  getTimestampFromApiEntry,
+  identifyTestType,
+  validateTaskBoundaries,
+} from './metrics-utils';
+import { ApiHistoryEntry, TaskSegment, UIMessage } from './types';
 
 class ChatProcessor {
   private chatDir: string;
@@ -45,7 +49,7 @@ class ChatProcessor {
       }
 
       // Identify test type with validation
-      const testType = this.identifyTestType();
+      const testType = this.getTestType();
       if (!testType) {
         logger.warn(
           `Could not identify test type for ${this.directoryId}, skipping processing`,
@@ -95,7 +99,7 @@ class ChatProcessor {
    * Identify the test type (control or mcp) from the chat data
    * @returns {string|undefined} Test type or undefined if not identifiable
    */
-  identifyTestType(): string | undefined {
+  getTestType(): string | undefined {
     if (
       !this.initialized ||
       !this.uiMessages?.length ||
@@ -104,64 +108,7 @@ class ChatProcessor {
       return undefined;
     }
 
-    // Check UI messages first
-    for (const message of this.uiMessages) {
-      if (message.type === 'say' && message.say === 'text' && message.text) {
-        if (
-          message.text.includes(
-            "'agent-instructions/control_instructions.md'",
-          ) ||
-          message.text.includes(
-            '"agent-instructions/control_instructions.md"',
-          ) ||
-          message.text.includes(CONTROL_MARKER)
-        ) {
-          return 'control';
-        }
-
-        if (
-          message.text.includes("'agent-instructions/mcp_instructions.md'") ||
-          message.text.includes('"agent-instructions/mcp_instructions.md"') ||
-          message.text.includes(MCP_MARKER)
-        ) {
-          return 'mcp';
-        }
-      }
-
-      // Check for MCP usage
-      if (message.type === 'say' && message.say === 'use_mcp_server') {
-        return 'mcp';
-      }
-    }
-
-    // Check API history as backup
-    for (const entry of this.apiHistory) {
-      if (
-        entry.role === 'user' &&
-        entry.content &&
-        Array.isArray(entry.content)
-      ) {
-        const content = entry.content.map((c) => c.text ?? '').join(' ');
-
-        if (
-          content.includes(CONTROL_MARKER) ||
-          content.includes("'agent-instructions/control_instructions.md'") ||
-          content.includes('"agent-instructions/control_instructions.md"')
-        ) {
-          return 'control';
-        }
-
-        if (
-          content.includes(MCP_MARKER) ||
-          content.includes("'agent-instructions/mcp_instructions.md'") ||
-          content.includes('"agent-instructions/mcp_instructions.md"')
-        ) {
-          return 'mcp';
-        }
-      }
-    }
-
-    return undefined;
+    return identifyTestType(this.uiMessages, this.apiHistory);
   }
 
   /**
@@ -175,52 +122,12 @@ class ChatProcessor {
       return Promise.resolve([]);
     }
 
-    const taskBoundaries: TaskSegment[] = [];
-    const taskNumbers = new Set<number>();
-
-    // Find task boundaries in UI messages
-    for (let i = 0; i < this.uiMessages.length; i++) {
-      const message = this.uiMessages[i];
-
-      if (message.type === 'say' && message.say === 'text' && message.text) {
-        let taskNumber: number | undefined;
-        let messageTestType = testType;
-
-        // Only look for task starts in specific instruction formats
-        const taskStartMatch = message.text.match(
-          /Complete Task (\d+) using the (?:commands|tools|functions)/i,
-        );
-        if (taskStartMatch?.[1]) {
-          taskNumber = parseInt(taskStartMatch[1], 10);
-
-          // Verify this is actually a task instruction by checking for instruction file references
-          if (message.text.includes('control_instructions.md')) {
-            messageTestType = 'control';
-          } else if (message.text.includes('mcp_instructions.md')) {
-            messageTestType = 'mcp';
-          } else {
-            // Not a real task instruction
-            taskNumber = undefined;
-          }
-        }
-
-        if (taskNumber && !taskNumbers.has(taskNumber)) {
-          taskNumbers.add(taskNumber);
-
-          taskBoundaries.push({
-            taskNumber,
-            directoryId: this.directoryId,
-            startIndex: i,
-            startTime: message.ts as number,
-            apiCalls: [],
-            userMessages: [],
-            endIndex: null,
-            endTime: null,
-            testType: messageTestType,
-          });
-        }
-      }
-    }
+    // Find task boundaries
+    const taskBoundaries = findTaskBoundaries(
+      this.uiMessages,
+      this.directoryId,
+      testType,
+    );
 
     // If no task boundaries found, return empty array
     if (taskBoundaries.length === 0) {
@@ -229,48 +136,18 @@ class ChatProcessor {
     }
 
     // Get the last valid message timestamp
-    let lastValidMessageTs: number | undefined;
-    for (let i = this.uiMessages.length - 1; i >= 0; i--) {
-      const message = this.uiMessages[i];
-      // Skip resume_completed_task messages as they can appear much later
-      if (
-        message?.ts &&
-        typeof message.ts === 'number' &&
-        !(message.type === 'ask' && message.ask === 'resume_completed_task')
-      ) {
-        lastValidMessageTs = message.ts;
-        break;
-      }
-    }
+    const lastValidMessageTs = getLastValidMessageTimestamp(this.uiMessages);
 
-    // Set end boundaries with validation
-    for (let i = 0; i < taskBoundaries.length; i++) {
-      if (i < taskBoundaries.length - 1) {
-        // For non-last tasks, end at the start of next task
-        taskBoundaries[i].endIndex = taskBoundaries[i + 1].startIndex - 1;
-        taskBoundaries[i].endTime = taskBoundaries[i + 1].startTime;
-      } else {
-        // For the last task, use the last valid message timestamp
-        taskBoundaries[i].endIndex = this.uiMessages.length - 1;
-        taskBoundaries[i].endTime = lastValidMessageTs as number;
-      }
-
-      // Validate endTime is after startTime and within reasonable bounds
-      if (
-        !taskBoundaries[i].endTime ||
-        taskBoundaries[i].endTime! < taskBoundaries[i].startTime ||
-        (lastValidMessageTs && taskBoundaries[i].endTime! > lastValidMessageTs)
-      ) {
-        logger.warn(
-          `Invalid endTime detected for task ${taskBoundaries[i].taskNumber}. Using last valid message timestamp.`,
-        );
-        taskBoundaries[i].endTime = lastValidMessageTs as number;
-      }
-    }
+    // Validate and set end boundaries
+    const validatedBoundaries = validateTaskBoundaries(
+      taskBoundaries,
+      lastValidMessageTs,
+      this.uiMessages.length,
+    );
 
     // Process task segments in parallel
     return Promise.all(
-      taskBoundaries.map((task) => this.processTaskSegment(task)),
+      validatedBoundaries.map((task) => this.processTaskSegment(task)),
     );
   }
 
@@ -300,78 +177,28 @@ class ChatProcessor {
       `Processing ${messages.length} messages for task ${task.taskNumber}`,
     );
 
-    const TASK_SEGMENT_TEXTS = [
-      'read_file',
-      'codebase_search',
-      'grep_search',
-      'file_search',
-      '<function_calls>',
-      '<fnr>',
-    ];
-
     // Filter messages to include only relevant ones
-    const relevantMessages = messages.filter((msg) => {
-      // Include API requests and MCP server requests
-      if (
-        msg.type === 'say' &&
-        (msg.say === 'api_req_started' ||
-          msg.say === 'mcp_server_request_started')
-      ) {
-        return true;
-      }
-
-      // Include tool usage messages
-      if (msg.type === 'say' && msg.say === 'text' && msg.text) {
-        return TASK_SEGMENT_TEXTS.some((text) => msg.text?.includes(text));
-      }
-
-      // Include user messages
-      if (
-        msg.type === 'say' &&
-        msg.say === 'text' &&
-        (!msg.from || msg.from === 'user')
-      ) {
-        return true;
-      }
-
-      // Include all other 'say' messages for completeness
-      if (msg.type === 'say') {
-        return true;
-      }
-
-      return false;
-    });
+    const relevantMessages = filterRelevantMessages(messages);
 
-    // Create a combined task timestamp for filtering API entries
-    const startBoundary = task.startTime - 60 * 1000; // 1 minute before task start
-    const endBoundary = (task.endTime as number) + 5 * 60 * 1000; // 5 minutes after task end
+    // Create timestamp boundaries for filtering API entries
+    const [startBoundary, endBoundary] = createTimestampBoundaries(
+      task.startTime,
+      task.endTime as number,
+    );
 
     // Filter API entries that fall within this task's time boundaries
     const apiEntries = this.apiHistory.filter((entry) => {
-      const timestamp = this.getTimestampFromApiEntry(entry);
+      const timestamp = getTimestampFromApiEntry(entry);
       return (
         timestamp && timestamp >= startBoundary && timestamp <= endBoundary
       );
     });
 
-    // Extract conversation history index from UI messages if available
+    // Extract conversation history index from UI messages
     for (const msg of relevantMessages) {
-      if (msg.type === 'say' && msg.text) {
-        try {
-          // First try to parse JSON to get index from structured data
-          const jsonData = JSON.parse(msg.text);
-          if (jsonData.conversationHistoryIndex !== undefined) {
-            msg.conversationHistoryIndex = jsonData.conversationHistoryIndex;
-          }
-        } catch (e) {
-          // If JSON parsing fails, try regex
-          const indexMatch = msg.text.match(
-            /conversationHistoryIndex["\s:]+(\d+)/i,
-          );
-          if (indexMatch?.[1]) {
-            msg.conversationHistoryIndex = parseInt(indexMatch[1], 10);
-          }
-        }
+      const index = extractConversationHistoryIndex(msg);
+      if (index !== undefined) {
+        msg.conversationHistoryIndex = index;
       }
     }
 
@@ -385,36 +212,6 @@ class ChatProcessor {
       messageCount: relevantMessages.length,
     };
   }
-
-  /**
-   * Extract timestamp from an API entry
-   * @param {ApiHistoryEntry} entry The API entry
-   * @returns {number|undefined} Timestamp in milliseconds or undefined if not found
-   */
-  getTimestampFromApiEntry(entry: ApiHistoryEntry): number | undefined {
-    if (!entry?.content || !Array.isArray(entry.content)) {
-      return undefined;
-    }
-
-    for (const content of entry.content) {
-      if (
-        content.type === 'text' &&
-        content.text &&
-        content.text.includes('environment_details') &&
-        content.text.includes('Current Time')
-      ) {
-        const match = content.text.match(/Current Time\s+([^<]+)/);
-        if (match?.[1]) {
-          const date = new Date(match[1].trim());
-          if (!Number.isNaN(date.getTime())) {
-            return date.getTime();
-          }
-        }
-      }
-    }
-
-    return undefined;
-  }
 }
 
 export default ChatProcessor;