GitHub-Proxy/jina_reader
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/jina-ai/reader.git synced 2025-08-19 18:29:09 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: expect failure while loading pdf

Browse Source
This commit is contained in:
Yanlong Wang 2025-03-08 20:25:18 +08:00
parent 4bc6394692
commit 440ff4d729
No known key found for this signature in database
GPG Key ID: C0A623C0BADF9F37
1 changed files with 16 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
src/api/crawler.ts
View File

@ -1010,10 +1010,23 @@ export class CrawlerHost extends RPCHost {
return this.snapshotFormatter.formatSnapshot(mode, snapshotCopy, nominalUrl, urlValidMs);
}
const r = await this.curlControl.sideLoad(new URL(pdfUrl), scrappingOptions);
if (r.file) {
snapshotCopy.pdfs[0] = pathToFileURL(await r.file.filePath).href;
const r = await this.curlControl.sideLoad(new URL(pdfUrl), scrappingOptions).catch((err) => {
if (err instanceof ServiceBadAttemptError) {
return Promise.reject(new AssertionFailureError(`Failed to load PDF(${pdfUrl}): ${err.message}`));
}
return Promise.reject(err);
});
if (r.status !== 200) {
throw new AssertionFailureError(`Failed to load PDF(${pdfUrl}): Server responded status ${r.status}`);
}
if (!r.contentType.includes('application/pdf')) {
throw new AssertionFailureError(`Failed to load PDF(${pdfUrl}): Server responded with wrong content type ${r.contentType}`);
}
if (!r.file) {
throw new AssertionFailureError(`Failed to load PDF(${pdfUrl}): Server did not return a body`);
}
snapshotCopy.pdfs[0] = pathToFileURL(await r.file.filePath).href;
}
}