├── README.md
└── scrape.sh


/README.md:
--------------------------------------------------------------------------------
 1 | # pastebin-scraper
 2 | 
 3 | A tool that leverages the API of https://psbdmp.ws/ to find emails and domains dumped in pastebin which could lead to finding some juicy information.
 4 | 
 5 | ## Usage
 6 | 
 7 | ```
 8 | $ ./scrape.sh 
 9 | Usage:
10 | Search for domains - ./script.sh -d domain.com
11 | Search for emails  - ./script.sh -e foo@bar.com
12 | General search	   - ./script.sh -g foobar
13 | ```
14 | 
15 | ## Example 
16 | 
17 | ``` 
18 | $ ./scrape.sh -d facebook.com
19 | Searching pastebin...
20 | Task completed. Output present in ./output/domain/facebook.com/
21 | ```
22 | 
23 | ## Installation
24 | 
25 | ```
26 | $ git clone https://github.com/streaak/pastebin-scraper.git
27 | $ cd pastebin-scraper
28 | $ sudo chmod +x ./scrape.sh
29 | $ sudo apt-get install jq
30 | $ mkdir -p output/domain/ output/email/ output/general/
31 | ```
32 | 
33 | ## Output
34 | 
35 | 
36 | Outputs will either be stored in `./output/domain`, `./output/email` or `./output/general depending` depending on what you search for. The output will contain 2 files, `output.json` and `urls.txt`. `output.json` will contain the original json returned by the API in a beautified format and `urls.txt` will contain the URLs returned by the API in text format.
37 | 
38 | ## TODO
39 | 
40 | * Add more features
41 | 
42 | ## Acknowledgments
43 | 
44 | * [Tomnomnom](https://twitter.com/tomnomnom)
45 | * [Ed](https://twitter.com/edoverflow)
46 | * [CaptainFreak](https://twitter.com/0xcaptainfreak)
47 | 


--------------------------------------------------------------------------------
/scrape.sh:
--------------------------------------------------------------------------------
 1 |  echo -e "$(tput setaf 1)        ____  ____          _____   __  ____    ____  ____   ___  ____  
 2 |         |    \|    \        / ___/  /  ]|    \  /    ||    \ /  _]|    \ 
 3 |         |  o  )  o  ) _____(   \_  /  / |  D  )|  o  ||  o  )  [_ |  D  )
 4 |         |   _/|     ||     |\__  |/  /  |    / |     ||   _/    _]|    / 
 5 |         |  |  |  O  ||_____|/  \ /   \_ |    \ |  _  ||  | |   [_ |    \ 
 6 |         |  |  |     |       \    \     ||  .  \|  |  ||  | |     ||  .  \ 
 7 |         |__|  |_____|        \___|\____||__|\_||__|__||__| |_____||__|\_|$(tput sgr0)"
 8 | 
 9 | echo -e "\n\t\t\t\t By @Streaak"
10 | if [ $# -eq 0 ] || [ $1 == '-h' ]; then
11 |   echo -e "$(tput setaf 2)\nUsage:$(tput sgr0)" 
12 |   echo "Search for domains - ./script.sh -d domain.com"
13 |   echo "Search for emails  - ./script.sh -e foo@bar.com"
14 |   echo "General search     - ./script.sh -g foobar"
15 |   exit 0
16 | fi
17 | 
18 | function scrape() {
19 |         echo "Searching pastebin..."
20 |         curl -s -X GET "$url" | python -m json.tool > ./output/$dir/output.json 
21 |         cat ./output/$dir/output.json | jq -r '.data[] | .id' | awk '{print "https://psbdmp.ws/" $1 }' > ./output/$dir/urls.txt
22 |         echo "Task completed. Output present in ./output/$dir/$2"
23 |         exit 0
24 |         return 0
25 | }
26 | 
27 | if [[ $1 == '-g' ]] && [[ $2 != '' ]] ; then
28 | dir=general/$2
29 | url=https://psbdmp.ws/api/search/$2
30 | mkdir -p ./output/$dir/
31 | scrape
32 | elif [[ $2 == '' ]]; then
33 |         echo "Missing Value. Try to run ./script.sh -g <Term here>"
34 |         exit 0
35 | fi
36 | 
37 | if [[ $1 == '-e' ]] && [[ $2 != '' ]] ; then
38 | dir=email/$2
39 | url=https://psbdmp.ws/api/search/email/$2
40 | mkdir -p ./output/$dir/
41 | scrape
42 | elif [[ $2 == '' ]]; then
43 |         echo "Missing Value. Try to run ./script.sh -e <Term here>"
44 |         exit 0
45 | fi
46 | 
47 | if [[ $1 == '-d' ]] && [[ $2 != '' ]] ; then
48 | dir=domain/$2
49 | url=https://psbdmp.ws/api/search/domain/$2
50 | mkdir -p ./output/$dir/
51 | scrape
52 | elif [[ $2 == '' ]]; then
53 |         echo "Missing Value. Try to run ./script.sh -d <Term here>"
54 |         exit 0
55 | fi
56 | 


--------------------------------------------------------------------------------