Added mimetype verification option to media source to ensure server returned a plausible file. Added additional fallbacks to Jules Jordan poster scraper for Amateur Allure.

This commit is contained in:
DebaucheryLibrarian
2020-09-17 02:30:58 +02:00
parent 6d1f83bc40
commit 1a8de4fcf6
9 changed files with 29 additions and 6 deletions

View File

@@ -99,6 +99,9 @@ function toBaseSource(rawSource) {
baseSource.stream = rawSource.stream;
}
// reject source if response mimetype does not match specified type
if (rawSource.verifyType) baseSource.verifyType = rawSource.verifyType;
if (rawSource.referer) baseSource.referer = rawSource.referer;
if (rawSource.host) baseSource.host = rawSource.host;
if (rawSource.attempts) baseSource.attempts = rawSource.attempts;
@@ -441,6 +444,8 @@ streamQueue.define('fetchStreamSource', async ({ source, tempFileTarget, hashStr
});
async function fetchSource(source, baseMedia) {
const maxAttempts = source.attempts || 3;
logger.silly(`Fetching media from ${source.src}`);
// attempts
@@ -470,6 +475,10 @@ async function fetchSource(source, baseMedia) {
const [type, subtype] = mimetype.split('/');
const extension = mime.getExtension(mimetype);
if (source.verifyType && source.verifyType !== type) {
throw Object.assign(new Error(`Type '${type}' does not match type '${source.verifyType}' specified by source`), { code: 'VERIFY_TYPE' });
}
return {
...source,
file: {
@@ -486,14 +495,15 @@ async function fetchSource(source, baseMedia) {
};
} catch (error) {
hasher.end();
const maxAttempts = source.attempts || 3;
logger.warn(`Failed attempt ${attempts}/${maxAttempts} to fetch ${source.src}: ${error.message}`);
if (error.code !== 'VERIFY_TYPE') {
logger.warn(`Failed attempt ${attempts}/${maxAttempts} to fetch ${source.src}: ${error.message}`);
if (attempts < maxAttempts) {
await Promise.delay(1000);
if (attempts < maxAttempts) {
await Promise.delay(1000);
return attempt(attempts + 1);
return attempt(attempts + 1);
}
}
throw new Error(`Failed to fetch ${source.src}: ${error.message}`);

View File

@@ -160,10 +160,22 @@ function scrapeAll(scenes, site, entryIdFromTitle) {
{
src: prefixedSrc.replace(/.jpg$/, '-full.jpg'),
referer: site.url,
verifyType: 'image', // sometimes returns 200 OK with text/html instead of 403
},
{
src: prefixedSrc.replace(/-1x.jpg$/, '-4x.jpg'),
referer: site.url,
verifyType: 'image',
},
{
src: prefixedSrc.replace(/-1x.jpg$/, '-2x.jpg'),
referer: site.url,
verifyType: 'image',
},
{
src: prefixedSrc,
referer: site.url,
verifyType: 'image',
},
];
}