Files
parcer/server/src/api/record.ts

392 lines
11 KiB
TypeScript
Raw Normal View History

2024-09-26 22:59:37 +05:30
import { readFile, readFiles } from "../workflow-management/storage";
2024-09-26 22:39:39 +05:30
import { Router, Request, Response } from 'express';
import { chromium } from "playwright";
2024-09-26 22:41:46 +05:30
import { requireAPIKey } from "../middlewares/api";
2024-10-10 05:47:42 +05:30
import Robot from "../models/Robot";
import Run from "../models/Run";
const router = Router();
import { getDecryptedProxyConfig } from "../routes/proxy";
import { uuid } from "uuidv4";
import { createRemoteBrowserForRun, destroyRemoteBrowser } from "../browser-management/controller";
import logger from "../logger";
2024-10-12 15:51:03 +05:30
import { browserPool } from "../server";
2024-10-12 15:55:00 +05:30
import { io, Socket } from "socket.io-client";
2024-09-26 22:39:39 +05:30
const formatRecording = (recordingData: any) => {
const recordingMeta = recordingData.recording_meta;
const workflow = recordingData.recording.workflow || [];
const firstWorkflowStep = workflow[0]?.where?.url || '';
2024-09-26 22:39:57 +05:30
2024-09-26 22:39:39 +05:30
const inputParameters = [
2024-09-26 22:39:57 +05:30
{
type: "string",
name: "originUrl",
label: "Origin URL",
required: true,
defaultValue: firstWorkflowStep,
},
2024-09-26 22:39:39 +05:30
];
2024-09-26 22:39:57 +05:30
2024-09-26 22:39:39 +05:30
return {
2024-09-26 22:39:57 +05:30
id: recordingMeta.id,
name: recordingMeta.name,
2024-10-08 20:50:17 +05:30
createdAt: new Date(recordingMeta.createdAt).getTime(),
2024-09-26 22:39:57 +05:30
inputParameters,
2024-09-26 22:39:39 +05:30
};
2024-09-26 22:39:57 +05:30
};
router.get("/robots", requireAPIKey, async (req: Request, res: Response) => {
2024-09-26 22:39:39 +05:30
try {
2024-10-10 06:03:56 +05:30
const robots = await Robot.findAll({ raw: true });
2024-10-10 05:57:49 +05:30
const formattedRecordings = robots.map(formatRecording);
2024-09-26 22:39:57 +05:30
const response = {
statusCode: 200,
messageCode: "success",
robots: {
totalCount: formattedRecordings.length,
items: formattedRecordings,
},
};
res.status(200).json(response);
2024-09-26 22:39:39 +05:30
} catch (error) {
2024-09-26 22:39:57 +05:30
console.error("Error fetching recordings:", error);
res.status(500).json({
statusCode: 500,
messageCode: "error",
message: "Failed to retrieve recordings",
});
2024-09-26 22:39:39 +05:30
}
2024-09-26 22:39:57 +05:30
});
2024-09-26 22:56:44 +05:30
const formatRecordingById = (recordingData: any) => {
const recordingMeta = recordingData.recording_meta;
const workflow = recordingData.recording.workflow || [];
const firstWorkflowStep = workflow[0]?.where?.url || '';
2024-09-26 22:57:07 +05:30
2024-09-26 22:56:44 +05:30
const inputParameters = [
2024-09-26 22:57:07 +05:30
{
type: "string",
name: "originUrl",
label: "Origin URL",
required: true,
defaultValue: firstWorkflowStep,
},
2024-09-26 22:56:44 +05:30
];
2024-09-26 22:57:07 +05:30
2024-09-26 22:56:44 +05:30
return {
2024-09-26 22:57:07 +05:30
id: recordingMeta.id,
name: recordingMeta.name,
2024-10-08 20:50:17 +05:30
createdAt: new Date(recordingMeta.createdAt).getTime(),
2024-09-26 22:57:07 +05:30
inputParameters,
2024-09-26 22:56:44 +05:30
};
2024-09-26 22:57:07 +05:30
};
2024-09-26 22:56:44 +05:30
2024-10-10 06:08:11 +05:30
router.get("/robots/:id", requireAPIKey, async (req: Request, res: Response) => {
2024-09-26 22:56:44 +05:30
try {
2024-10-10 06:08:11 +05:30
const robot = await Robot.findOne({
where: {
2024-10-10 06:08:26 +05:30
'recording_meta.id': req.params.id
2024-10-10 06:08:11 +05:30
},
raw: true
2024-10-10 06:08:26 +05:30
});
2024-09-26 22:57:07 +05:30
2024-10-10 06:08:11 +05:30
const formattedRecording = formatRecordingById(robot);
2024-09-26 22:57:07 +05:30
const response = {
statusCode: 200,
messageCode: "success",
robot: formattedRecording,
};
res.status(200).json(response);
2024-09-26 22:56:44 +05:30
} catch (error) {
2024-09-26 22:57:07 +05:30
console.error("Error fetching recording:", error);
res.status(404).json({
statusCode: 404,
messageCode: "not_found",
message: `Recording with name "${req.params.fileName}" not found.`,
});
2024-09-26 22:56:44 +05:30
}
2024-09-26 22:57:07 +05:30
});
2024-10-11 17:58:27 +05:30
// TODO: Format runs to send more data formatted
router.get("/robots/:id/runs", requireAPIKey, async (req: Request, res: Response) => {
try {
const runs = await Run.findAll({
where: {
2024-10-10 06:18:42 +05:30
robotMetaId: req.params.id
},
raw: true
});
const response = {
statusCode: 200,
messageCode: "success",
runs: {
totalCount: runs.length,
items: runs,
},
};
res.status(200).json(response);
} catch (error) {
console.error("Error fetching runs:", error);
res.status(500).json({
statusCode: 500,
messageCode: "error",
message: "Failed to retrieve runs",
});
}
}
);
router.get("/robots/:id/runs/:runId", requireAPIKey, async (req: Request, res: Response) => {
try {
const run = await Run.findOne({
where: {
2024-10-10 06:22:53 +05:30
runId: req.params.runId,
robotMetaId: req.params.id,
},
raw: true
});
2024-10-12 15:55:43 +05:30
const response = {
statusCode: 200,
messageCode: "success",
run: run,
};
res.status(200).json(response);
} catch (error) {
console.error("Error fetching run:", error);
res.status(404).json({
statusCode: 404,
messageCode: "not_found",
2024-10-10 06:24:04 +05:30
message: `Run with id "${req.params.runId}" for robot with id "${req.params.id}" not found.`,
});
}
});
async function createWorkflowAndStoreMetadata(id: string, userId: string) {
try {
const recording = await Robot.findOne({
where: {
'recording_meta.id': id
},
raw: true
});
if (!recording || !recording.recording_meta || !recording.recording_meta.id) {
return {
success: false,
error: 'Recording not found'
};
}
const proxyConfig = await getDecryptedProxyConfig(userId);
let proxyOptions: any = {};
if (proxyConfig.proxy_url) {
proxyOptions = {
server: proxyConfig.proxy_url,
...(proxyConfig.proxy_username && proxyConfig.proxy_password && {
username: proxyConfig.proxy_username,
password: proxyConfig.proxy_password,
}),
};
}
const browserId = createRemoteBrowserForRun({
browser: chromium,
launchOptions: {
headless: true,
proxy: proxyOptions.server ? proxyOptions : undefined,
}
});
const runId = uuid();
const run = await Run.create({
status: 'Running',
name: recording.recording_meta.name,
robotId: recording.id,
robotMetaId: recording.recording_meta.id,
startedAt: new Date().toLocaleString(),
finishedAt: '',
browserId,
interpreterSettings: { maxConcurrency: 1, maxRepeats: 1, debug: true },
log: '',
runId,
serializableOutput: {},
binaryOutput: {},
});
const plainRun = run.toJSON();
return {
browserId,
runId: plainRun.runId,
}
} catch (e) {
const { message } = e as Error;
logger.log('info', `Error while scheduling a run with id: ${id}`);
console.log(message);
return {
success: false,
error: message,
};
}
}
2024-10-12 15:52:26 +05:30
async function readyForRunHandler(browserId: string, id: string) {
try {
const result = await executeRun(id);
2024-10-12 15:55:43 +05:30
if (result && result.success) {
2024-10-12 15:55:43 +05:30
logger.log('info', `Interpretation of ${id} succeeded`);
2024-10-12 22:19:38 +05:30
resetRecordingState(browserId, id);
return result.interpretationInfo;
2024-10-12 15:55:43 +05:30
} else {
logger.log('error', `Interpretation of ${id} failed`);
await destroyRemoteBrowser(browserId);
2024-10-12 22:19:38 +05:30
resetRecordingState(browserId, id);
return null;
2024-10-12 15:55:43 +05:30
}
2024-10-12 15:52:26 +05:30
} catch (error: any) {
2024-10-12 15:55:43 +05:30
logger.error(`Error during readyForRunHandler: ${error.message}`);
await destroyRemoteBrowser(browserId);
return null;
2024-10-12 15:52:26 +05:30
}
2024-10-12 15:55:43 +05:30
}
2024-10-12 15:52:26 +05:30
2024-10-12 15:55:43 +05:30
function resetRecordingState(browserId: string, id: string) {
2024-10-12 15:53:13 +05:30
browserId = '';
id = '';
2024-10-12 15:55:43 +05:30
}
2024-10-12 15:53:13 +05:30
async function executeRun(id: string) {
try {
2024-10-12 15:55:43 +05:30
const run = await Run.findOne({ where: { runId: id } });
if (!run) {
return {
success: false,
error: 'Run not found'
};
}
2024-10-12 15:55:43 +05:30
const plainRun = run.toJSON();
const recording = await Robot.findOne({ where: { 'recording_meta.id': plainRun.robotMetaId }, raw: true });
if (!recording) {
return {
success: false,
error: 'Recording not found'
};
2024-10-12 15:55:43 +05:30
}
plainRun.status = 'running';
const browser = browserPool.getRemoteBrowser(plainRun.browserId);
if (!browser) {
throw new Error('Could not access browser');
}
const currentPage = await browser.getCurrentPage();
if (!currentPage) {
throw new Error('Could not create a new page');
}
2024-10-12 15:55:43 +05:30
const interpretationInfo = await browser.interpreter.InterpretRecording(
recording.recording, currentPage, plainRun.interpreterSettings
);
2024-10-12 15:55:43 +05:30
await destroyRemoteBrowser(plainRun.browserId);
2024-10-12 22:16:23 +05:30
const updatedRun = await run.update({
2024-10-12 15:55:43 +05:30
...run,
status: 'success',
finishedAt: new Date().toLocaleString(),
browserId: plainRun.browserId,
log: interpretationInfo.log.join('\n'),
serializableOutput: interpretationInfo.serializableOutput,
binaryOutput: interpretationInfo.binaryOutput,
});
return {
success: true,
2024-10-12 22:16:23 +05:30
interpretationInfo: updatedRun.toJSON()
};
} catch (error: any) {
2024-10-12 15:55:43 +05:30
logger.log('info', `Error while running a recording with id: ${id} - ${error.message}`);
return {
success: false,
error: error.message,
};
}
2024-10-12 15:55:43 +05:30
}
2024-10-12 15:55:43 +05:30
export async function handleRunRecording(id: string, userId: string) {
2024-10-12 15:54:14 +05:30
try {
2024-10-12 15:55:43 +05:30
const result = await createWorkflowAndStoreMetadata(id, userId);
const { browserId, runId: newRunId } = result;
if (!browserId || !newRunId || !userId) {
throw new Error('browserId or runId or userId is undefined');
}
const socket = io(`http://localhost:8080/${browserId}`, {
transports: ['websocket'],
rejectUnauthorized: false
});
socket.on('ready-for-run', () => readyForRunHandler(browserId, newRunId));
logger.log('info', `Running recording: ${id}`);
socket.on('disconnect', () => {
cleanupSocketListeners(socket, browserId, newRunId);
});
2024-10-12 15:54:14 +05:30
} catch (error: any) {
2024-10-12 15:55:43 +05:30
logger.error('Error running recording:', error);
2024-10-12 15:54:14 +05:30
}
2024-10-12 15:55:43 +05:30
}
function cleanupSocketListeners(socket: Socket, browserId: string, id: string) {
2024-10-12 15:54:14 +05:30
socket.off('ready-for-run', () => readyForRunHandler(browserId, id));
logger.log('info', `Cleaned up listeners for browserId: ${browserId}, runId: ${id}`);
2024-10-12 15:55:43 +05:30
}
2024-10-12 15:54:14 +05:30
router.post("/robots/:id/runs", requireAPIKey, async (req: Request, res: Response) => {
try {
2024-10-12 22:14:33 +05:30
const interpretationInfo = await handleRunRecording(req.params.id, req.user.dataValues.id);
console.log(`Result`, interpretationInfo);
const response = {
statusCode: 200,
messageCode: "success",
2024-10-12 22:14:33 +05:30
run: interpretationInfo,
};
res.status(200).json(response);
} catch (error) {
console.error("Error running robot:", error);
res.status(500).json({
statusCode: 500,
messageCode: "error",
message: "Failed to run robot",
});
}
});
2024-10-12 22:14:33 +05:30
export default router;