GitHub-Proxy/jina_reader
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/jina-ai/reader.git synced 2025-08-18 17:45:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: compressed response from curl

Browse Source
This commit is contained in:
yanlong.wang 2025-02-05 16:09:54 +08:00
parent 6a58de590c
commit 0f36fe81a6
No known key found for this signature in database
GPG Key ID: C0A623C0BADF9F37
1 changed files with 40 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
backend/functions/src/services/curl.ts
View File

@ -10,6 +10,7 @@ import { AssertionFailureError, FancyFile } from 'civkit';
import { TempFileManager } from '../shared'; import { TempFileManager } from '../shared';
import { readFile } from 'fs/promises'; import { readFile } from 'fs/promises';
import { pathToFileURL } from 'url'; import { pathToFileURL } from 'url';
import { createBrotliDecompress, createInflate, createGunzip } from 'zlib';
@singleton() @singleton()
export class CurlControl extends AsyncService { export class CurlControl extends AsyncService {
@ -59,6 +60,7 @@ export class CurlControl extends AsyncService {
text: '', text: '',
} as PageSnapshot; } as PageSnapshot;
let contentType = '';
const result = await new Promise<{ const result = await new Promise<{
statusCode: number, statusCode: number,
data?: FancyFile, data?: FancyFile,
@ -102,14 +104,20 @@ export class CurlControl extends AsyncService {
}); });
curl.setOpt(Curl.option.MAXFILESIZE, 1024 * 1024 * 1024); // 1GB curl.setOpt(Curl.option.MAXFILESIZE, 1024 * 1024 * 1024); // 1GB
let status = -1; let status = -1;
let contentType = ''; let contentEncoding = '';
curl.on('stream', (stream, statusCode, headers) => { curl.on('stream', (stream, statusCode, headers) => {
status = statusCode; status = statusCode;
outerLoop: outerLoop:
for (const headerVec of headers) { for (const headerVec of headers) {
for (const [k, v] of Object.entries(headerVec)) { for (const [k, v] of Object.entries(headerVec)) {
if (k.toLowerCase() === 'content-type') { const kl = k.toLowerCase();
if (kl === 'content-type') {
contentType = v.toLowerCase(); contentType = v.toLowerCase();
}
if (kl === 'content-encoding') {
contentEncoding = v.toLowerCase();
}
if (contentType && contentEncoding) {
break outerLoop; break outerLoop;
} }
} }
@ -130,6 +138,30 @@ export class CurlControl extends AsyncService {
return; return;
} }
switch (contentEncoding) {
case 'gzip': {
const decompressed = createGunzip();
stream.pipe(decompressed);
stream = decompressed;
break;
}
case 'deflate': {
const decompressed = createInflate();
stream.pipe(decompressed);
stream = decompressed;
break;
}
case 'br': {
const decompressed = createBrotliDecompress();
stream.pipe(decompressed);
stream = decompressed;
break;
}
default: {
break;
}
}
const fpath = this.tempFileManager.alloc(); const fpath = this.tempFileManager.alloc();
const fancyFile = FancyFile.auto(stream, fpath); const fancyFile = FancyFile.auto(stream, fpath);
this.tempFileManager.bindPathTo(fancyFile, fpath); this.tempFileManager.bindPathTo(fancyFile, fpath);
@ -147,8 +179,13 @@ export class CurlControl extends AsyncService {
throw new AssertionFailureError(`Failed to access ${urlToCrawl}: HTTP ${result.statusCode}`); throw new AssertionFailureError(`Failed to access ${urlToCrawl}: HTTP ${result.statusCode}`);
} }
if (contentType === 'application/octet-stream') {
// Content declared as binary is same as unknown.
contentType = '';
}
if (result.data) { if (result.data) {
const mimeType: string = await result.data.mimeType; const mimeType: string = contentType || await result.data.mimeType;
if (mimeType.startsWith('text/html')) { if (mimeType.startsWith('text/html')) {
if ((await result.data.size) > 1024 * 1024 * 32) { if ((await result.data.size) > 1024 * 1024 * 32) {
throw new AssertionFailureError(`Failed to access ${urlToCrawl}: file too large`); throw new AssertionFailureError(`Failed to access ${urlToCrawl}: file too large`);