fix: 统一文档解析文件格式校验
- 统一知识库和工作流支持格式并增加前后端上传拦截 - 拒绝 XLS 与伪装 XLSX,避免空内容解析成功
This commit is contained in:
@@ -0,0 +1,83 @@
|
||||
import { $t } from '#/locales';
|
||||
|
||||
const OLE2_SIGNATURE = [208, 207, 17, 224, 161, 177, 26, 225];
|
||||
|
||||
const supportedExtensions = Object.freeze([
|
||||
'txt',
|
||||
'pdf',
|
||||
'docx',
|
||||
'md',
|
||||
'pptx',
|
||||
'xlsx',
|
||||
'csv',
|
||||
]);
|
||||
const supportedExtensionSet = new Set(supportedExtensions);
|
||||
const supportedTypeLabel = supportedExtensions
|
||||
.map((extension) => extension.toUpperCase())
|
||||
.join('、');
|
||||
|
||||
function extensionOf(fileName: string) {
|
||||
const normalizedName = String(fileName || '')
|
||||
.trim()
|
||||
.toLowerCase();
|
||||
const dotIndex = normalizedName.lastIndexOf('.');
|
||||
return dotIndex === -1 ? '' : normalizedName.slice(dotIndex + 1);
|
||||
}
|
||||
|
||||
function supports(fileName: string) {
|
||||
return supportedExtensionSet.has(extensionOf(fileName));
|
||||
}
|
||||
|
||||
async function validateFiles(files: File[]) {
|
||||
for (const file of files) {
|
||||
const extension = extensionOf(file.name);
|
||||
if (!supportedExtensionSet.has(extension)) {
|
||||
throw new Error(
|
||||
$t('message.upload.unsupportedFileType', {
|
||||
fileName: file.name,
|
||||
types: supportedTypeLabel,
|
||||
}),
|
||||
);
|
||||
}
|
||||
if (extension !== 'xlsx') {
|
||||
continue;
|
||||
}
|
||||
const prefix = new Uint8Array(await file.slice(0, 8).arrayBuffer());
|
||||
if (hasZipSignature(prefix)) {
|
||||
continue;
|
||||
}
|
||||
const messageKey = startsWith(prefix, OLE2_SIGNATURE)
|
||||
? 'message.upload.legacyExcelAsXlsx'
|
||||
: 'message.upload.invalidXlsx';
|
||||
throw new Error($t(messageKey, { fileName: file.name }));
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 知识库导入与工作流文档解析共用的文件格式策略。
|
||||
*/
|
||||
export const DocumentParseFilePolicy = Object.freeze({
|
||||
accept: supportedExtensions.map((extension) => `.${extension}`).join(','),
|
||||
supportedExtensions,
|
||||
supportedTypeLabel,
|
||||
supports,
|
||||
validateFiles,
|
||||
});
|
||||
|
||||
function hasZipSignature(prefix: Uint8Array) {
|
||||
return (
|
||||
prefix.length >= 4 &&
|
||||
prefix[0] === 80 &&
|
||||
prefix[1] === 75 &&
|
||||
((prefix[2] === 3 && prefix[3] === 4) ||
|
||||
(prefix[2] === 5 && prefix[3] === 6) ||
|
||||
(prefix[2] === 7 && prefix[3] === 8))
|
||||
);
|
||||
}
|
||||
|
||||
function startsWith(prefix: Uint8Array, signature: number[]) {
|
||||
return (
|
||||
prefix.length >= signature.length &&
|
||||
signature.every((value, index) => prefix[index] === value)
|
||||
);
|
||||
}
|
||||
Reference in New Issue
Block a user