import { logger, MCPTool } from "mcp-framework";
import {getLogForPipelineStep, getPipelineResult} from "./helpers/woodpeckerCI.js";
import {pipelineDetailSchema} from "../Schema/pipelineDetail.js";
import { SimpleCache } from "./helpers/cache.js";
interface WoodpeckerCiPipelineReportGeneratorInput {
repoId: string;
pipelineNumber: string;
}
const pipelineLogCache = new SimpleCache<any>(2 * 60 * 60 * 1000); // 2 hours cache
class WoodpeckerCiPipelineReportGeneratorTool extends MCPTool<WoodpeckerCiPipelineReportGeneratorInput> {
name = "woodpecker-ci-pipeline-report-generator";
description = "This tool is used to extract all the different machine logs for the given pipeline"
schema = pipelineDetailSchema;
async execute(input: WoodpeckerCiPipelineReportGeneratorInput) {
const {pipelineNumber, repoId} = input;
const cacheKey = `${repoId}:${pipelineNumber}`;
// check cache first
if (pipelineLogCache.has(cacheKey)) {
logger.info('Returning cached pipeline log for key: ' + cacheKey);
return pipelineLogCache.get(cacheKey);
}
// check failure first
const pipelineResult = await getPipelineResult(repoId, pipelineNumber);
const responseToAI = {
content: [{
type: "text",
details: pipelineResult.isSuccess ? `Pipeline ${pipelineNumber} for PR ${pipelineResult.pullRequestUrl} completed successfully. No failed steps detected.`
: pipelineResult.error,
moreDetails: [] as any[]
}],
}
if (pipelineResult.isSuccess || pipelineResult.error?.includes("Failed to fetch")) {
pipelineLogCache.set(cacheKey, responseToAI);
return responseToAI;
}
// If the pipeline failed, we need to fetch the logs for each failed step
const stepDetails = pipelineResult.failedStepDetails;
const response = await getLogForPipelineStep(repoId, pipelineNumber, stepDetails);
responseToAI.content[0].details = `Pipeline ${pipelineNumber} for PR ${pipelineResult.pullRequestUrl} has failed.`
responseToAI.content[0].moreDetails = response;
pipelineLogCache.set(cacheKey, responseToAI);
return responseToAI
}
}
export default WoodpeckerCiPipelineReportGeneratorTool;
