link-stack/apps/metamigo-worker/tasks/import-leafcutter.ts

165 lines
4.7 KiB
TypeScript
Raw Normal View History

/* eslint-disable camelcase */
import fetch from "node-fetch";
import { URLSearchParams } from "url";
import { withDb, AppDatabase } from "../db";
import { loadConfig } from "@digiresilience/metamigo-config";
type LabelStudioTicket = {
id: string
is_labeled: boolean
annotations: Record<string, unknown>[]
data: Record<string, unknown>
updated_at: string;
}
type LeafcutterTicket = {
id: string
incident: string[]
technology: string[]
targeted_group: string[]
country: string[]
region: string[]
continent: string[]
date: Date
origin: string
origin_id: string
source_created_at: string
source_updated_at: string
}
const getLabelStudioTickets = async (page: number): Promise<LabelStudioTicket[]> => {
const {
leafcutter: {
labelStudioApiUrl,
labelStudioApiKey,
}
} = await loadConfig();
const headers = {
Authorization: `Token ${labelStudioApiKey}`,
Accept: "application/json",
};
const ticketsQuery = new URLSearchParams({
page_size: "50",
page: `${page}`,
});
console.log({ url: `${labelStudioApiUrl}/projects/1/tasks?${ticketsQuery}` })
const res = await fetch(`${labelStudioApiUrl}/projects/1/tasks?${ticketsQuery}`,
{ headers });
console.log({ res })
const tasksResult = await res.json();
console.log({ tasksResult });
return tasksResult;
}
const fetchFromLabelStudio = async (minUpdatedTimestamp: Date): Promise<LabelStudioTicket[]> => {
const pages = [...Array.from({ length: 10000 }).keys()];
const allDocs: LabelStudioTicket[] = [];
for await (const page of pages) {
const docs = await getLabelStudioTickets(page + 1);
console.log({ page, docs })
if (docs && docs.length > 0) {
for (const doc of docs) {
const updatedAt = new Date(doc.updated_at);
console.log({ updatedAt, minUpdatedTimestamp });
if (updatedAt > minUpdatedTimestamp) {
console.log(`Adding doc`, { doc })
allDocs.push(doc)
}
}
} else {
break;
}
}
console.log({ allDocs })
return allDocs;
}
const sendToLeafcutter = async (tickets: LabelStudioTicket[]) => {
const {
leafcutter: {
contributorId,
opensearchApiUrl,
opensearchUsername,
opensearchPassword
}
} = await loadConfig();
console.log({ tickets })
const filteredTickets = tickets.filter((ticket) => ticket.is_labeled);
console.log({ filteredTickets })
const finalTickets: LeafcutterTicket[] = filteredTickets.map((ticket) => {
const {
id,
annotations,
data: {
source_id,
source_created_at,
source_updated_at
}
} = ticket;
const getTags = (tags: Record<string, any>[], name: string) =>
tags
.filter((tag) => tag.from_name === name)
.map((tag) => tag.value.choices)
.flat();
const allTags = annotations.map(({ result }) => result).flat();
const incident = getTags(allTags, "incidentType tag");
const technology = getTags(allTags, "platform tag");
const country = getTags(allTags, "country tag");
const targetedGroup = getTags(allTags, "targetedGroup tag");
return {
id,
incident,
technology,
targeted_group: targetedGroup,
country,
region: [],
continent: [],
date: new Date(source_created_at as string),
origin: contributorId,
origin_id: source_id as string,
source_created_at: source_created_at as string,
source_updated_at: source_updated_at as string
};
});
console.log("Sending to Leafcutter");
console.log({ finalTickets })
const result = await fetch(opensearchApiUrl, {
method: "POST",
headers: {
"Content-Type": "application/json",
Authorization: `Basic ${Buffer.from(`${opensearchUsername}:${opensearchPassword}`).toString("base64")}`,
},
body: JSON.stringify({ tickets: finalTickets }),
});
console.log({ result });
};
const importLeafcutterTask = async (): Promise<void> => {
withDb(async (db: AppDatabase) => {
const { leafcutter: { contributorName } } = await loadConfig();
const settingName = `${contributorName}ImportLeafcutterTask`;
const res: any = await db.settings.findByName(settingName);
const startTimestamp = res?.value?.minUpdatedTimestamp ? new Date(res.value.minUpdatedTimestamp as string) : new Date("2023-03-01");
const newLastTimestamp = new Date();
console.log({ contributorName, settingName, res, startTimestamp, newLastTimestamp });
const tickets = await fetchFromLabelStudio(startTimestamp);
console.log({ tickets })
await sendToLeafcutter(tickets);
await db.settings.upsert(settingName, { minUpdatedTimestamp: newLastTimestamp })
});
};
export default importLeafcutterTask;