eduvia/workavia-drive
11
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

🩹 back: when indexing documents, call unoconv directly to ignore stderr output. And do not try indexing multiple ways.

Browse Source
This commit is contained in:
Eric Doughty-Papassideris
2024-04-23 15:48:58 +02:00
committed by ericlinagora
parent 79fc8f2ad8
commit a8de167547
1 changed files with 4 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tdrive/backend/node/src/services/documents/utils.ts
+4 -8
View File
@@ -1,8 +1,8 @@
import { spawnCheckingExitCode } from "../../utils/exec";
import archiver from "archiver"; import archiver from "archiver";
import { merge } from "lodash"; import { merge } from "lodash";
import PdfParse from "pdf-parse"; import PdfParse from "pdf-parse";
import { Readable } from "stream"; import { Readable } from "stream";
import unoconv from "unoconv-promise";
import Repository from "../../core/platform/services/database/services/orm/repository/repository"; import Repository from "../../core/platform/services/database/services/orm/repository/repository";
import { import {
cleanFiles, cleanFiles,
@@ -387,11 +387,7 @@ export const officeFileToString = async (file: Readable, extension: string): Pro
const outputPath = getTmpFile(".pdf"); const outputPath = getTmpFile(".pdf");
try { try {
await unoconv.run({ await spawnCheckingExitCode("unoconv", [`-o${outputPath}`, officeFilePath]);
file: officeFilePath,
output: outputPath,
});
cleanFiles([officeFilePath]); cleanFiles([officeFilePath]);
return await pdfFileToString(outputPath); return await pdfFileToString(outputPath);
@@ -633,11 +629,11 @@ export const getKeywordsOfFile = async (
logger.info(`Processing text file: ${filename}`); logger.info(`Processing text file: ${filename}`);
content_strings = await readableToString(file); content_strings = await readableToString(file);
} }
if (isFileType(mime, filename, pdfExtensions)) { if ((content_strings ?? "").trim().length == 0 && isFileType(mime, filename, pdfExtensions)) {
logger.info(`Processing PDF file: ${filename}`); logger.info(`Processing PDF file: ${filename}`);
content_strings = await pdfFileToString(file); content_strings = await pdfFileToString(file);
} }
if (isFileType(mime, filename, officeExtensions)) { if ((content_strings ?? "").trim().length == 0 && isFileType(mime, filename, officeExtensions)) {
logger.info(`Processing office file: ${filename}`); logger.info(`Processing office file: ${filename}`);
content_strings = await officeFileToString(file, extension); content_strings = await officeFileToString(file, extension);
} }