Automatically generate and set a title for documents with a specific tag in paperless.ngx using ollama

autotitle.js

#! bun run
/**
 * The author disclaims copyright to this source code.
 */

const PAPERLESS_URL = process.env.PAPERLESS_URL ?? "";
const PAPERLESS_TOKEN = process.env.PAPERLESS_TOKEN ?? "";
const PAPERLESS_TAG_ID = process.env.PAPERLESS_TAG_ID ?? 0;
const OLLAMA_URL = process.env.OLLAMA_URL ?? "";

if (PAPERLESS_URL.length < 1) throw new Error("PAPERLESS_URL must be set.");
if (PAPERLESS_TOKEN.length < 1) throw new Error("PAPERLESS_TOKEN must be set.");
if (PAPERLESS_TAG_ID < 1) throw new Error("PAPERLESS_TAG_ID must be set.");
if (OLLAMA_URL.length < 1) throw new Error("OLLAMA_URL must be set.");

async function autotitle(offset = 0) {
  const docs =
    (
      await fetch(
        `${PAPERLESS_URL}/api/documents/?page=${offset + 1}&page_size=1&tags__id__all=${PAPERLESS_TAG_ID}`,
        {
          headers: {
            Authorization: `Token ${PAPERLESS_TOKEN}`,
          },
          redirect: "error",
        },
      ).then((res) => res.json())
    ).results ?? [];

  if (docs.length < 1) {
    console.log("Nothing to do.");
    return;
  }

  for (const doc of docs) {
    const content = Array.from(new Set(doc.content.split("\n")));
    if (
      content.length == 1 &&
      content[0] == "Scanned by VueScan - get a free trial at www.hamrick.com"
    ) {
      console.log(`Document ${doc.id} only contains the scanner's watermark.`);
      continue;
    }

    if (doc.content.trim().length < 1) {
      console.log(`Document ${doc.id} doesn't have any content.`);
      continue;
    }

    const title = (
      await fetch(`${OLLAMA_URL}/api/generate`, {
        headers: {
          "Content-Type": "application/json",
        },
        method: "POST",
        body: JSON.stringify({
          model: "llama3.1:latest",
          prompt: `${doc.content}\n\nWhat's a plausible title for the above document? Only return the title; no comments; no quotation marks.`,
          stream: false,
        }),
      }).then((res) => res.json())
    ).response;

    console.log(`Document ${doc.id}'s new title is "${title}".`);
    await fetch(`${PAPERLESS_URL}/api/documents/${doc.id}/`, {
      method: "PUT",
      headers: {
        Authorization: `Token ${PAPERLESS_TOKEN}`,
        "Content-Type": "application/json",
      },
      body: JSON.stringify({
        correspondent: doc.correspondent,
        document_type: doc.document_type,
        storage_path: doc.storage_path,
        tags: doc.tags.filter((tagId) => tagId != PAPERLESS_TAG_ID),
        title: title,
      }),
    });
  }

  await autotitle(offset + 1);
}

while (true) {
  await autotitle(0);
  await new Promise((resolve) => setTimeout(resolve, 2000)); // wait for 2 seconds
}