Contents
Short description
Nudecrawler is a spider that searches for naked women on telegra.ph, an anonymous content-posting platform. Developed by a programmer, it uses a built-in sensor to find NSFW images and videos on the site. Nudecrawler keeps a log of what it finds and where, and saves passwords for various resources. The spider can also connect to detectors like tensorflow or keras to improve its accuracy. Based on the detector’s results, the spider sorts content into “interesting” or “not interesting” categories. However, the spider comes with a disclaimer and warnings about its potential offensiveness and disturbing content.
A naked creeper on women on the telegraph
Nudecrawler is a spider for finding naked women on telegra.ph.
As a programmer, I tend to automate everything. As a man, I tend to Cherchez la femme. Also, as someone who’s been online longer than pornhub has been around, I’ve seen so many tits and porn that I’m not always sure if I’m interested.
NudeCrawler is the pinnacle of client-side technology for a girl lover: it not only searches for porn for you, but even watches for you so you don’t have to! However, he keeps a log of what and where he watched and how he liked it, so that you can “spy on the robot, how he spies on a not-so-publicly posted picture, where the photographer spies how a girl changes clothes.” (In general, it’s sex in a square, a cube, and a hyper-cube).
The final result of the program looks like this (just in case, I removed the numbers to specifically (thank God and for the Glory of Satan, here it is the same thing) spoil the link – but now you can easily find it yourself):
INTERESTING (16 video) https://telegra.ph/Alina-XX-XX-XX
Nude: 129 non-nude: 11
Total video: 16
INTERESTING (30 videos) https://telegra.ph/Alisa-XX-XX-XX
Nude: 52 non-nude: 11
Total videos: 30
Ineffective intriguing warning
It doesn’t matter what gender you are, how old you are, what your orientation is, how wide your sexual horizons are, and what perversions you’re most interested in – you’ll still find a lot to dislike with NudeCrawler.
Trust me, I’ve seen some shit. Literally.
For example, once NudeCrawler brought me this page (The criterion of interest was either how many NSFW pictures or any video.) I warn you: the video has bright colors, flashes, if you have epilepsy – it’s not worth it. And in any case, it’s not worth it.
What is interesting about telegra.ph and what can be found?
It was not for nothing that V. I. Lenin called to capture the telegraph first.
Telegraph allows you to post content (and has its own hosting of pictures and short videos) anonymously without moderation. In addition, it is very easy. The content is not encrypted, it is open to everyone, so it is unlikely that anyone keeps any completely confidential things there. But on the other hand, in the former era (before the appearance of nudecrawler), it was impossible to find any interesting page there. The gerbil is securely hidden in a pile of sand. But with the advent of nudecrawler, finding interesting things has become easier.
What I (no pun intended, just testing nudecrawler) found on telegra.ph:
-
Nude photos, comics, photosets and “plums” (amateur and from OnlyFans/Patreon). And in a truncated form (like “join our paid channel to see the shower in full”) and in a fairly complete form (more than 100 pictures. A record for 3 days – 163 pictures in one find).
-
Collections of various “gloomy” videos (fights, [роскомнадзор] in the subway under the train, etc.)
-
Description of several fraudulent schemes, ready-made guides, take and use. And tips and links where you need them. By the way, I liked some of them, they were witty. I didn’t know about them before (I have known about calls from the “security service of the Savings Bank” and the “investigative department” for 10 years).
-
Huge lists of dozens of links to various redirectors and hosts of “dubious reputation” like mega.nz. On part of the links there are pictures and videos of the model Amouranth, part of which has been deleted, and part of it “for gross violation of the terms of use of the service”.
-
Personal letters (someone communicates like this, probably writes an “unsealed” letter in the telegraph and sends a link. Honestly: I stopped reading when I realized what I was reading). It is not necessary to do so, the telegraph is not suitable for everything personal.
-
Just galleries in the genre “here I am walking in the park”, “here I made repairs in the country”
-
Advertisement for the sale of rental real estate and motor vehicles
-
Military and patriotic articles and photos (drafts maybe?) of different sides and in different languages.
-
A huge number of pages in the “porn video – porn story – second porn video” format. It seems that someone created them with a script. Moreover, in a story on a gay theme (I didn’t take into account! honestly!) porn videos are not in the theme of the story, but normal, with universal values, where several European African-Americans enjoy one white woman without apparent respect.
-
Yes, there are also enough passwords for various resources. Many Google a query like:
site:telegra.ph password
. Of course, you don’t have to do that either.
Installation
pip3 install nudecrawler
(better in virtualenv) or, directly from the repository (who does not take risks, does not drink antibiotics):
pip3 install git+https://github.com/yaroslaff/nudecrawler
It will come with an integrated (and slow) nudepy sensor. To “play” and not drag a bunch of dependencies with you – it’s quite normal, but if you need to use others, it’s better to use them via docker or download the necessary packages for using NudeNet (more details – on the project page)
By the way, if filtering is not needed at all, you can just use the flag -a
.
How to use?
Let’s start with the fastest (otherwise I can’t wait!):
nudecrawler -a Сиськи
INTERESTING (ALL) https://telegra.ph/Siski-XX-XX
Total images: 8
Total video: 1
INTERESTING (ALL) https://telegra.ph/Siski-XX-XX
Total images: 4
INTERESTING (ALL) https://telegra.ph/Siski-XX-XX
Total images: 0
We indicated -a
i.e. nudecrawler did not filter at all and showed all found pages. (even without pictures).
Let’s look for beautiful people in the hope of something beautiful
nudecrawler Маша Наташа "Анатолий Евгеньевич"
Now it will work more slowly, because each picture will be run through the built-in Nudepy filter. In addition, the default parameters are used – there must be 5+ pictures on the page (otherwise – it is immediately rejected), then for the page to count as interesting and appear, it must contain at least 1 NSFW, or any video. This can be tweaked through the settings --nude
, --total
and --video
. And you can add -v
to see how it works. By default, pages are searched for dates from the last 30 days (but the years can be any, that’s how telegraph is already arranged). This can be tweaked through -d
.
If we want to search for many words, we use the word list through -w file
. A wordlist can consist of both words and base URLs. The base URL is a URL without a date and number, for example https://telegra.ph/secret – there is no page at this address, but when we add a date and an index to it, there will be some.
From the repository, you can download wordlist.txt for 3000 English words, or urls.txt 11,200 base URLs of telegraph – these are already filtered URLs (through bulk-http-check), there are pages for each of them on telegra.ph (and here, for example, there were no pages for the word “lampshades”, so it was not included in urls.txt).
We are connecting the detectors!
It is difficult for me to answer which detector is more accurate, more accurate, because the very boundaries of the concept of “voices” are rather vague (on the same test page, different detectors found from 10 to 20 NSFW pictures), but in terms of speed, nudepy is definitely 6 times slower than the others .
Detectors usually work through tensorflow / keras, entail a huge bunch of dependencies, so I recommend using them through containers. Example:
sudo docker run --rm --name nsfw-api -d -p 3000:3000 ghcr.io/arnidan/nsfw-api:latest
Now you can tell nudecrawler: --detect nsfwapi
and he will turn to him. Everything will work faster. About other options, so as not to overload the article, on the project page. This method is quite good.
These detectors have a memory leak (adult-image-detector also has a memory leak). For a very long time, it will simply eat up all the available memory and crash. To prevent this from happening:
--stop 200 --refresh PATH/TO/refresh-nsfw-api.sh
Now after every 200 pictures we restart the container (but it is possible after much more, the main thing is that at least sometimes).
Not only for finding naked women in the Telegraph!
First, not only in the telegraph:
nudecrawler --detect nsfwapi --url1 https://oboobs.ru
....
INTERESTING (17 nudes) https://oboobs.ru
Nude: 17 non-nude: 1
It turns out that oboobs.ru, well-known to Khabrov residents, also has tits! You can take a list of all domains, for example, run each one through nudecrawler and now you have a ready database of “adult” sites that can be used in parental filters.
Secondly, you can connect any of your detectors. In fact, the detector is a simple script, any program on the disk that passes either the name of the downloaded file or its URL. Connects or through --detect-image PATH
or --detect-url PATH
. You can though /bin/true
use as such a “concerned” detector that sees obscenity in everything.
For example, AWS Rekognition can detect weapons in photos. Plug it in and you can find pages with pictures of weapons. Or you can simply search for your own photos on the Internet.
And if you run it with a key -a
and search for the word “Passwords”, you will be surprised how many people save passwords on open pages in the Telegraph… Including quite important resources such as hosting. (Here I must say that using other people’s passwords is illegal and unethical)
Wish?
I will be glad for any feedback and wishes. There may be tips on interesting detectors and other ideas, which chips should be added, how to better filter boring pages (Like a million of those pages, where there are two videos and a story between them).