el.xwx.moe/lib/api/archiveHandler.ts

214 lines
6.6 KiB
TypeScript
Raw Normal View History

2024-06-28 11:12:52 -05:00
import { LaunchOptions, chromium, devices } from "playwright";
import { prisma } from "./db";
2024-03-26 00:38:08 -05:00
import sendToWayback from "./preservationScheme/sendToWayback";
import { Collection, Link, User } from "@prisma/client";
2024-06-29 16:18:38 -05:00
import fetchHeaders from "./fetchHeaders";
import createFolder from "./storage/createFolder";
2024-04-08 18:35:06 -05:00
import { removeFiles } from "./manageLinkFiles";
2024-06-27 11:39:03 -05:00
import handleMonolith from "./preservationScheme/handleMonolith";
import handleReadablility from "./preservationScheme/handleReadablility";
import handleArchivePreview from "./preservationScheme/handleArchivePreview";
import handleScreenshotAndPdf from "./preservationScheme/handleScreenshotAndPdf";
2024-06-28 11:20:56 -05:00
import imageHandler from "./preservationScheme/imageHandler";
import pdfHandler from "./preservationScheme/pdfHandler";
type LinksAndCollectionAndOwner = Link & {
collection: Collection & {
owner: User;
};
};
const BROWSER_TIMEOUT = Number(process.env.BROWSER_TIMEOUT) || 5;
export default async function archiveHandler(link: LinksAndCollectionAndOwner) {
2024-11-17 15:06:05 -06:00
if (process.env.DISABLE_PRESERVATION === "true")
return await prisma.link.update({
where: { id: link.id },
data: {
lastPreserved: new Date().toISOString(),
readable: "unavailable",
image: "unavailable",
monolith: "unavailable",
pdf: "unavailable",
preview: "unavailable",
},
});
2024-03-26 00:38:08 -05:00
const timeoutPromise = new Promise((_, reject) => {
setTimeout(
() =>
reject(
new Error(
`Browser has been open for more than ${BROWSER_TIMEOUT} minutes.`
)
),
BROWSER_TIMEOUT * 60000
);
});
// allow user to configure a proxy
let browserOptions: LaunchOptions = {};
2024-02-19 14:38:36 -06:00
if (process.env.PROXY) {
browserOptions.proxy = {
2024-02-19 14:38:36 -06:00
server: process.env.PROXY,
bypass: process.env.PROXY_BYPASS,
username: process.env.PROXY_USERNAME,
password: process.env.PROXY_PASSWORD,
};
}
if (process.env.PLAYWRIGHT_LAUNCH_OPTIONS_EXECUTABLE_PATH) {
2024-11-17 15:06:05 -06:00
browserOptions.executablePath =
process.env.PLAYWRIGHT_LAUNCH_OPTIONS_EXECUTABLE_PATH;
}
const browser = await chromium.launch(browserOptions);
const context = await browser.newContext({
...devices["Desktop Chrome"],
2024-02-10 18:34:25 -06:00
ignoreHTTPSErrors: process.env.IGNORE_HTTPS_ERRORS === "true",
});
2024-02-18 16:02:35 -06:00
const page = await context.newPage();
2024-04-08 18:35:06 -05:00
createFolder({
filePath: `archives/preview/${link.collectionId}`,
});
createFolder({
filePath: `archives/${link.collectionId}`,
});
try {
await Promise.race([
(async () => {
2024-03-26 00:38:08 -05:00
const user = link.collection?.owner;
2024-06-29 16:18:38 -05:00
const header = link.url ? await fetchHeaders(link.url) : undefined;
2024-06-29 16:18:38 -05:00
const contentType = header?.get("content-type");
let linkType = "url";
let imageExtension = "png";
if (!link.url) linkType = link.type;
2024-01-01 09:37:20 -06:00
else if (contentType?.includes("application/pdf")) linkType = "pdf";
else if (contentType?.startsWith("image")) {
linkType = "image";
2024-01-01 09:37:20 -06:00
if (contentType.includes("image/jpeg")) imageExtension = "jpeg";
else if (contentType.includes("image/png")) imageExtension = "png";
}
2024-03-26 00:38:08 -05:00
await prisma.link.update({
where: { id: link.id },
data: {
type: linkType,
image:
user.archiveAsScreenshot && !link.image?.startsWith("archive")
? "pending"
2024-07-25 12:43:55 -05:00
: undefined,
pdf:
user.archiveAsPDF && !link.pdf?.startsWith("archive")
? "pending"
2024-07-25 12:43:55 -05:00
: undefined,
monolith:
user.archiveAsMonolith && !link.monolith?.startsWith("archive")
? "pending"
: undefined,
readable: !link.readable?.startsWith("archive")
? "pending"
: undefined,
preview: !link.readable?.startsWith("archive")
? "pending"
: undefined,
lastPreserved: new Date().toISOString(),
},
});
2023-12-23 11:11:47 -06:00
2024-03-26 00:38:08 -05:00
// send to archive.org
if (user.archiveAsWaybackMachine && link.url) sendToWayback(link.url);
if (linkType === "image" && !link.image?.startsWith("archive")) {
await imageHandler(link, imageExtension); // archive image (jpeg/png)
return;
} else if (linkType === "pdf" && !link.pdf?.startsWith("archive")) {
await pdfHandler(link); // archive pdf
return;
} else if (link.url) {
// archive url
2023-12-23 11:11:47 -06:00
await page.goto(link.url, { waitUntil: "domcontentloaded" });
const content = await page.content();
2023-12-23 11:11:47 -06:00
2024-03-26 00:38:08 -05:00
// Preview
2024-06-26 12:54:03 -05:00
if (
!link.preview?.startsWith("archives") &&
!link.preview?.startsWith("unavailable")
)
2024-06-27 11:39:03 -05:00
await handleArchivePreview(link, page);
// Readability
if (
!link.readable?.startsWith("archives") &&
!link.readable?.startsWith("unavailable")
)
await handleReadablility(content, link);
2023-12-23 11:11:47 -06:00
2024-06-18 11:27:29 -05:00
// Screenshot/PDF
2024-06-26 12:54:03 -05:00
if (
(!link.image?.startsWith("archives") &&
!link.image?.startsWith("unavailable")) ||
(!link.pdf?.startsWith("archives") &&
!link.pdf?.startsWith("unavailable"))
)
2024-06-27 11:39:03 -05:00
await handleScreenshotAndPdf(link, page, user);
2024-06-27 20:58:07 -05:00
// Monolith
2024-06-27 11:39:03 -05:00
if (
2024-06-27 20:58:07 -05:00
!link.monolith?.startsWith("archive") &&
!link.monolith?.startsWith("unavailable") &&
user.archiveAsMonolith &&
2024-06-27 11:39:03 -05:00
link.url
)
await handleMonolith(link, content);
}
})(),
timeoutPromise,
]);
} catch (err) {
console.log(err);
console.log("Failed Link details:", link);
throw err;
} finally {
const finalLink = await prisma.link.findUnique({
where: { id: link.id },
});
if (finalLink)
await prisma.link.update({
where: { id: link.id },
data: {
2023-12-23 11:11:47 -06:00
lastPreserved: new Date().toISOString(),
readable: !finalLink.readable?.startsWith("archives")
2023-12-15 14:47:08 -06:00
? "unavailable"
: undefined,
image: !finalLink.image?.startsWith("archives")
2023-12-15 14:47:08 -06:00
? "unavailable"
: undefined,
2024-06-27 20:58:07 -05:00
monolith: !finalLink.monolith?.startsWith("archives")
2024-03-15 13:41:41 -05:00
? "unavailable"
: undefined,
pdf: !finalLink.pdf?.startsWith("archives")
2023-12-15 14:47:08 -06:00
? "unavailable"
: undefined,
2023-12-23 11:11:47 -06:00
preview: !finalLink.preview?.startsWith("archives")
? "unavailable"
: undefined,
},
});
2023-12-19 16:20:09 -06:00
else {
2024-04-08 18:35:06 -05:00
await removeFiles(link.id, link.collectionId);
2023-12-19 16:20:09 -06:00
}
await browser.close();
}
}