feat(ee): PDF import (#2142)

* feat: replace pdfjs-dist with firecrawl-pdf-inspector

* use modified firecrawl-pdf-inspector

* feat: pdf import

* increase single file upload size limit

* use npm package

* sync

* update package
This commit is contained in:
Philip Okugbe
2026-05-01 14:56:39 +01:00
committed by GitHub
parent 641ce142df
commit c247d4c1e3
7 changed files with 93 additions and 142 deletions
+1
View File
@@ -8,6 +8,7 @@ export const Feature = {
AI: 'ai',
CONFLUENCE_IMPORT: 'import:confluence',
DOCX_IMPORT: 'import:docx',
PDF_IMPORT: 'import:pdf',
ATTACHMENT_INDEXING: 'attachment:indexing',
SECURITY_SETTINGS: 'security:settings',
MCP: 'mcp',
@@ -12,6 +12,7 @@ import {
IconCheck,
IconFileCode,
IconFileTypeDocx,
IconFileTypePdf,
IconFileTypeZip,
IconMarkdown,
IconX,
@@ -90,12 +91,14 @@ function ImportFormatSelection({ spaceId, onClose }: ImportFormatSelection) {
const markdownFileRef = useRef<() => void>(null);
const htmlFileRef = useRef<() => void>(null);
const docxFileRef = useRef<() => void>(null);
const pdfFileRef = useRef<() => void>(null);
const notionFileRef = useRef<() => void>(null);
const confluenceFileRef = useRef<() => void>(null);
const zipFileRef = useRef<() => void>(null);
const canUseConfluence = useHasFeature(Feature.CONFLUENCE_IMPORT);
const canUseDocx = useHasFeature(Feature.DOCX_IMPORT);
const canUsePdf = useHasFeature(Feature.PDF_IMPORT);
const upgradeLabel = useUpgradeLabel();
const handleZipUpload = async (selectedFile: File, source: string) => {
@@ -244,7 +247,7 @@ function ImportFormatSelection({ spaceId, onClose }: ImportFormatSelection) {
}, 3000);
}, [fileTaskId]);
const maxSingleFileSize = bytes("20mb");
const maxSingleFileSize = bytes("30mb");
const handleFileUpload = async (selectedFiles: File[]) => {
if (!selectedFiles) {
@@ -298,6 +301,7 @@ function ImportFormatSelection({ spaceId, onClose }: ImportFormatSelection) {
if (markdownFileRef.current) markdownFileRef.current();
if (htmlFileRef.current) htmlFileRef.current();
if (docxFileRef.current) docxFileRef.current();
if (pdfFileRef.current) pdfFileRef.current();
const pageCountText =
pageCount === 1 ? `1 ${t("page")}` : `${pageCount} ${t("pages")}`;
@@ -378,6 +382,30 @@ function ImportFormatSelection({ spaceId, onClose }: ImportFormatSelection) {
)}
</FileButton>
<FileButton
onChange={handleFileUpload}
accept=".pdf"
multiple
resetRef={pdfFileRef}
>
{(props) => (
<Tooltip
label={upgradeLabel}
disabled={canUsePdf}
>
<Button
disabled={!canUsePdf}
justify="start"
variant="default"
leftSection={<IconFileTypePdf size={18} />}
{...props}
>
PDF
</Button>
</Tooltip>
)}
</FileButton>
<FileButton
onChange={(file) => handleZipUpload(file, "notion")}
accept="application/zip"