The Pappy (**P**roxy **A**ttack **P**roxy **P**rox**Y**) Proxy is an intercepting proxy for performing web application security testing. Its features are often similar, or straight up rippoffs from [Burp Suite](https://portswigger.net/burp/). However, Burp Suite is neither open source nor a command line tool, thus making a proxy like Pappy inevitable. The project is still in its early stages, so there are bugs and only the bare minimum features, but it can already do some cool stuff.
**I am taking any and all feature requests.** If you've used Burp and had any inconvenience with it, tell me about it and I'll do everything in my power to make sure Pappy doesn't have those issues. Or even better, if you want Burp to do something that it doesn't already, let me know so that I can ~~use it to stomp them into the dust~~ improve my project.
If you're brave and want to try and contribute code, please let me know. Right now the codebase is kind of rough and I have refactored it a few times already, but I would be more than happy to find a stable part of the codebase that you can contribute to.
Another option is to try writing a plugin. It might be a bit easier than contributing code and plugins are extremely easy to integrate as a core feature. So you can also contribute by writing a plugin and letting me know about it. You can find out more by looking at [the official plugin docs](https://roglew.github.io/pappy-proxy/pappyplugins.html).
Yes! If you don't want to go completely over to Pappy yet, you can configure Burp to use Pappy as an upstream proxy server. That way, traffic will go through both Burp and Pappy and you can use whichever you want to do your testing.
How to have Burp forward traffic through Pappy:
1. Open Burp
2. Go to `Options -> Connections -> Upstream Proxy Servers`
3. Click `Add`
4. Leave `Destination Host` blank, but put `127.0.0.1` in `Proxy Host` and `8000` into `Port` (assuming you're using the default listener)
Pappy supports OS X and Linux (sorry Windows). Installation requires `pip` or some other command that can handle a `setup.py` with requirements. Once the requirements are installed, you can check that it installed correctly by running `pappy -l` to start the proxy.
It is also possible (and encouraged!) to install pappy into a [virtual environment](http://docs.python-guide.org/en/latest/dev/virtualenvs/). After installing into a virtual environment, you will have to enter the virtual environment each time you want to run Pappy.
Pappy projects take up an entire directory. Any generated scripts, exported responses, plugin data, etc. will be placed in the current directory so it's good to give your project a directory of its own. To start a project, do something like:
And that's it! The proxy will by default be running on port 8000 and bound to localhost (to keep the hackers out). You can modify the port/interface in `config.json`. You can list all your intercepted requests with `ls`, view a full request with `vfq <reqid>` or view a full response with `vfs <reqid>`. Right now, the only command to delete requests is `filter_prune` which deletes all the requests that aren't in the current context (look at the sections on the context/filter strings for more information on that).
Here's everything you need to know to get the basics done:
* This quickstart assumes you've used Burp Suite
* Make a directory for your project and `cd` into it in the terminal. Type `pappy` into the terminal and hit enter
* Commands are entered into the prompt that appears
* The proxy starts listening on port 8000 once the program starts
* Use `ls` to look at recent requests, `ls a` to look at the entire history
* You will use the number in the `id` column to perform actions on that request
* Use `vfq <id>` and `vfs <id>` to view full requests/responses
* Use `ic` to modify requests with a text editor as they go through the proxy or `ic req rsp` to modify both requests and responses
* Use `rp <id>` to send a request to the repeater. In the repeater, use `<leader>f` to send the current buffer (you may need to configre a leader key in vim). Use `:qa!` to quit the repeater.
If you want to do more, I highly suggest reading the whole readme!
If you don't want to dirty up a directory, you can run Pappy in "lite" mode. Pappy will use the default configuration settings and will create a temporary data file in `/tmp` to use. When you quit, the file will be deleted. If you want to run Pappy in lite mode, run Pappy with either `-l` or `--lite`.
In order for Pappy to view data sent using HTTPS, you need to add a generated CA cert (`certificate.crt`) to your browser. Certificates are generated using the `gencerts` command and are by default stored in `~/.pappy/certs`. This allows Pappy to act as a CA and sign any HTTPS certificate it wants without the browser complaining. This allows Pappy to decrypt and modify HTTPS requests. The certificate installation instructions are different for each browser.
You can add the CA cert to Firefox by going to `Preferences -> Advanced -> View Certificates -> Authorities -> Import` and selecting the `certificate.crt` file in the `certs` directory.
### Chrome
You can add the CA cert to Chrome by going to `Settings -> Show advanced settings -> HTTPS/SSL -> Manage Certificates -> Authorities -> Import` and selecting the `certificate.crt` file in the `certs` directory.
### Safari
For Safari (on macs, obviously), you need to add the CA cert to your system keychain. You can do this by double clicking on the CA cert and following the prompts.
I didn't search too hard for instructions on this (since Pappy doesn't support windows) and I don't own a Windows machine to try this, so if you have trouble, I'm not the one to ask. According to Google you can double-click the cert to install it to the system, or you can do `Tools -> Content -> Certificates -> Trusted Root Certificates -> Import`.
Configuration for each project is done in the `config.json` file. The file is a JSON-formatted dictionary that contains settings for the proxy. The following fields can be used to configure the proxy:
| `data_file` | The file where requests and images will be stored |
| `debug_dir` (optional) | Where connection debug info should be stored. If not present, debug info is not saved to a file. |
| `cert_dir` | Where the CA cert and the private key for the CA cert are stored |
| `proxy_listeners` | A list of dicts which describe which ports the proxy will listen on. Each item is a dict with "port" and "interface" values which determine which port and interface to listen on. For example, if port=8000 and the interface is 127.0.0.1, the proxy will only accept connections from localhost on port 8000. To accept connections from anywhere, set the interface to 0.0.0.0. |
| `socks_proxy` | A dictionary with details on how to connect to an upstream SOCKS proxy to send all in-scope requests through. See the secion on upstream SOCKS proxies for more information. |
| `http_proxy` | A dictionary with details on how to connect to an upstream http proxy to send all in-scope requests through. See the section on upstream http proxies for more information. |
There are a few tricks you can use in general when using Pappy's console. Most of these are provided by the [cmd](https://docs.python.org/2/library/cmd.html) and [cmd2](https://pythonhosted.org/cmd2/index.html).
### Run a shell command
You can run a shell command with `!`:
```
pappy> ls
ID Verb Host Path S-Code Req Len Rsp Len Time Mngl
5 GET vitaly.sexy /netscape.gif 304 Not Modified 0 0 0.08 --
4 GET vitaly.sexy /esr1.jpg 304 Not Modified 0 0 0.07 --
3 GET vitaly.sexy /construction.gif 304 Not Modified 0 0 0.07 --
2 GET vitaly.sexy /vitaly2.jpg 0 N/A -- --
1 GET vitaly.sexy / 304 Not Modified 0 0 0.07 --
pappy> !ls
cmdhistory config.json data.db
pappy>
```
### Running Python Code
You can use the `py` command to either run python code or to drop down to a Python shell.
```
pappy> py print ':D '*10
:D :D :D :D :D :D :D :D :D :D
pappy> py
Python 2.7.6 (default, Jun 22 2015, 17:58:13)
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
(ProxyCmd)
py <command>: Executes a Python command.
py: Enters interactive Python mode.
End with ``Ctrl-D`` (Unix) / ``Ctrl-Z`` (Windows), ``quit()``, '`exit()``.
Non-python commands can be issued with ``cmd("your command")``.
Run python code from external files with ``run("filename.py")``
You can use `>` to direct output to a file. However, a number of commands use colored output. If you just redirect these to a file, there will be additional bytes which represent the ANSI color codes. To get around this, use the `nocolor` command to remove the color from the command output.
If you want to write the contents of a request or response to a file, don't use `nocolor` with `vfq` or `vfs`. Use just the `vbq` or `vbs` commands.
| Command | Description |
|:--------|:------------|
| `nocolor` | Run a command and print its output without ASCII escape codes. Intended for use when redirecting output to a file. Should only be used with text and not with binary data. |
In order to intercept and modify requests to sites that use HTTPS, you have to generate and install CA certs to your browser. You can do this by running the `gencerts` command in Pappy. By default, certs are stored `~/.pappy/certs`. This is also the default location that Pappy will look for certificates (unless you specify otherwise in `config.json`.) In addition, you can give the `gencerts` command an argument to have it put the generated certs in a different directory.
| `gencerts [/path/to/put/certs/in]` | Generate a CA cert that can be added to your browser to let Pappy decrypt HTTPS traffic. Also generates the private key for that cert in the same directory. If no path is given, the certs will be placed in the default certificate location. Overwrites any existing certs. |
| `ls [a|<num>]`| list, ls |List requests that are in the current context (see Context section). Has information like the host, target path, and status code. With no arguments, it will print the 25 most recent requests in the current context. If you pass 'a' or 'all' as an argument, it will print all the requests in the current context. If you pass a number "n" as an argument, it will print the n most recent requests in the current context. |
| `sm [p]` | sm, site_map | Print a tree showing the site map. It will display all requests in the current context that did not have a 404 response. This has to go through all of the requests in the current context so it may be slow. If the `p` option is given, it will print the paths as paths rather than as a tree. |
| `viq <id(s)>` | view_request_info, viq | View additional information about requests. Includes the target port, if SSL was used, applied tags, and other information. |
| `vfq <id(s)>` | view_full_request, vfq, kjq | [V]iew [F]ull Re[Q]uest, prints the full request including headers and data. If the request is part of a websocket handshake, it will also print the messages sent over the websocket. |
| `vbq <id(s)>` | view_request_bytes, vbq | [V]iew [B]ytes of Re[Q]uest, prints the full request including headers and data without coloring or additional newlines. Use this if you want to write a request to a file. |
| `vfs <id(s)>` | view_full_response, vfs, kjs |[V]iew [F]ull Re[S]ponse, prints the full response associated with a request including headers and data. |
| `vbs <id(s)>` | view_response_bytes, vbs | [V]iew [B]ytes of Re[S]ponse, prints the full response including headers and data without coloring or additional newlines. Use this if you want to write a response to a file. |
| `pps <format> <id(s)>` | pretty_print_response, pps | Pretty print a response with a specific format. See the table below for a list of formats. |
| `pprm <id(s)>` | print_params, pprm | Print a summary of the parameters submitted with the request. It will include URL params, POST params, and/or cookies |
| `pri [ct] [key(s)]` | param_info, pri | Print a summary of the parameters and values submitted by in-context requests. You can pass in keys to limit which values will be shown. If you also provide `ct` as the first argument, it will include any keys that are passed as arguments. |
| `urls <id(s)>` | urls | Search the full request and response of the given IDs for urls and prints them. Especially useful with a wildcard (`*`) to find URLs from all history. |
| ID | The request ID of that request. Used to identify the request for other commands. |
| Method | The method(/http verb) for the request |
| Host | The host that the request was sent to |
| Path | The path of the request |
| S-Code | The status code of the response |
| Req Len | The length of the data submitted |
| Rsp Len | The length of the data returned in the response |
| Time | The time in seconds it took to complete the request |
| Mngl | If the request or response were mangled with the interceptor. If the request was mangled, the column will show 'q'. If the response was mangled, the column will show 's'. If both were mangled, it will show 'q/s'. |
You can apply tags to a request and use filters to view specific tags. The following commands can be used to apply and remove tags to requests:
| Command | Aliases | Description |
|:--------|:--------|:------------|
| `tag <tag> [id(s)]` | tag | Apply a tag to the given requests. If no IDs are given, the tag will be applied to all in-context requests. |
| `untag <tag> [id(s)]` | untag | Remove a tag from the given ids. If no IDs are given, the tag is removed from every in-context request. |
| `clrtag <id(s)>` | clrtag | Removes all tags from the given ids. |
Request IDs
-----------
Request IDs are how you identify a request and every command that involves specifying a request will take one or more request IDs. You can see it when you run `ls`. In addition, you can prepend an ID with prefixes to get requests or responses associated with the request (for example if you modified the request or its response with the interceptor, you can get the unmangled versions.) Here are the valid prefixes:
| Prefix | Description |
|:-------|:------------|
| `u` | If the request was mangled, prefixing the ID with `u` will result in the unmangled version of the request. The resulting request will not have an associated response because it was never submitted to the server. |
| `s` | If the response was mangled, prefixing the request ID `s` will result in the same request but its associated response will be the unmangled version. |
I know it sounds kind of unintuitive. Here are some example commands that will hopefully make things clearer. Suppose request 1 had its request mangled, and request 2 had its response mangled.
*`vfq 1` Prints the mangled version of request 1
*`vfq u1` Prints the unmangled version of request 1
*`rp u1` Open the repeater with the unmangled version of request 1
*`vfs u1` Throws an error because the unmangled version was never submitted
*`vfs s1` Throws an error because the response for request 1 was never mangled
*`vfs 2` Prints the mangled response of request 2
*`vfs s2` Prints the unmangled response of request 2
*`vfq u2` Throws an error because request 2's request was never mangled
*`vfs u2` Throws an error because request 2's request was never mangled
### Passing Multiple Request IDs to a Command
Some arguments can take multiple IDs for an argument. To pass multiple IDs to a command, separate the IDs with commas **(no spaces!)**. A few examples:
*`viq 1,2,u3` View information about requests 1, 2, and the unmangled version of 3
*`gma foo 4,5,6` Generate a macro with definitions for requests 4, 5, and 6
The context is a set of filters that define which requests are considered "active". Only requests in the current context are displayed with `ls`. By default, the context includes every single request that passes through the proxy. You can limit down the current context by applying filters. Filters apply rules such as "the response code must equal 500" or "the host must contain google.com". Once you apply one or more filters, only requests/responses which pass every active filter will be a part of the current context.
| `f <filter string>` | filter, fl, f |Add a filter that limits which requests are included in the current context. See the Filter String section for how to create a filter string |
You are also able to save and load contexts. When saving a context you pass the command a name. The context can then be loaded with that name. Whenever you load a context, the current context is saved with the name `_` to make it easier to quickly load a context, view requests, then return to the original context.
| Command | Aliases | Description |
|:--------|:------------|:---|
| `sc <name>` | `sc`, `save_context` | Save the current filters with the provided name. |
| `lc <name>` | `lc`, `load_context` | Load a saved context by its name. |
| `dc <name>` | `dc`, `delete_context` | Delete a saved context by its name. |
| `cls` | `cls`, `list_contexts` | Show a list of saved contexts and the filters for each of them. |
Here is an example session demonstrating saving/loading contexts:
```
pappy> ls
ID Verb Host Path S-Code Req Len Rsp Len Time Mngl
16 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 10515 0.06 --
15 GET cdn.sstatic.net /Sites/stackoverflow/img/sprites.svg?... 200 OK 0 8131 0.05 --
14 GET i.stack.imgur.com /eoNf5.png 403 Forbidden 0 173 0.07 --
13 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 12919 0.07 --
12 GET cdn.sstatic.net /img/favicons-sprite16.png?v=4b071e01... 200 OK 0 66460 0.09 --
11 GET i.stack.imgur.com /xqoqk.png 403 Forbidden 0 173 0.07 --
10 GET i.stack.imgur.com /6HFc3.png 403 Forbidden 0 173 0.06 --
9 GET i.stack.imgur.com /tKsDb.png 403 Forbidden 0 173 0.06 --
8 GET i.stack.imgur.com /5d55j.png 403 Forbidden 0 173 0.08 --
7 GET cdn.sstatic.net /Js/full-anon.en.js?v=a65ef7e053bb 200 OK 0 116828 0.27 --
6 GET cdn.sstatic.net /img/share-sprite-new.svg?v=78be252218f3 200 OK 0 34771 0.93 --
5 GET cdn.sstatic.net /clc/clc.min.js?v=6f49b407ccbc 200 OK 0 6969 0.92 --
4 GET cdn.sstatic.net /Sites/stackoverflow/all.css?v=40629f... 200 OK 0 476855 0.07 --
3 GET cdn.sstatic.net /Js/stub.en.js?v=5cc84a62e045 200 OK 0 38661 0.08 --
2 GET ajax.googleapis.com /ajax/libs/jquery/1.12.4/jquery.min.js 200 OK 0 97163 0.09 --
1 GET stackoverflow.com / 200 OK 0 244280 0.43 --
pappy> fbi not_jscss
path nctr "(\.js$|\.css$)"
pappy> fbi not_image
path nctr "(\.png$|\.jpg$|\.gif$)"
pappy> sc clean
Filters saved to clean:
path nctr "(\.js$|\.css$)"
path nctr "(\.png$|\.jpg$|\.gif$)"
pappy> cls
Saved contexts:
clean
path nctr "(\.js$|\.css$)"
path nctr "(\.png$|\.jpg$|\.gif$)"
pappy> sr
pappy> fls
pappy> f host ct sstatic
pappy> ls
ID Verb Host Path S-Code Req Len Rsp Len Time Mngl
16 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 10515 0.06 --
15 GET cdn.sstatic.net /Sites/stackoverflow/img/sprites.svg?... 200 OK 0 8131 0.05 --
13 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 12919 0.07 --
12 GET cdn.sstatic.net /img/favicons-sprite16.png?v=4b071e01... 200 OK 0 66460 0.09 --
7 GET cdn.sstatic.net /Js/full-anon.en.js?v=a65ef7e053bb 200 OK 0 116828 0.27 --
6 GET cdn.sstatic.net /img/share-sprite-new.svg?v=78be252218f3 200 OK 0 34771 0.93 --
5 GET cdn.sstatic.net /clc/clc.min.js?v=6f49b407ccbc 200 OK 0 6969 0.92 --
4 GET cdn.sstatic.net /Sites/stackoverflow/all.css?v=40629f... 200 OK 0 476855 0.07 --
3 GET cdn.sstatic.net /Js/stub.en.js?v=5cc84a62e045 200 OK 0 38661 0.08 --
pappy> lc clean
Set the context to:
path nctr "(\.js$|\.css$)"
path nctr "(\.png$|\.jpg$|\.gif$)"
pappy> ls
ID Verb Host Path S-Code Req Len Rsp Len Time Mngl
16 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 10515 0.06 --
15 GET cdn.sstatic.net /Sites/stackoverflow/img/sprites.svg?... 200 OK 0 8131 0.05 --
13 GET cdn.sstatic.net /img/developer-story/announcement_ban... 200 OK 0 12919 0.07 --
6 GET cdn.sstatic.net /img/share-sprite-new.svg?v=78be252218f3 200 OK 0 34771 0.93 --
Where `<field>` is some part of the request/response, `<comparer>` is some comparison to `<value>`. For example, if you wanted a filter that only matches requests to `target.org`, you could use the following filter string:
Also **if you prefix a comparer with 'n' it turns it into a negation.** Using the previous example, the following will match any request except for ones where the host contains `target.org`:
For fields that are a list of key/value pairs (headers, get params, post params, and cookies) you can use the following format:
```
<field><comparer1><value1>[ <comparer2><value2>]
```
This is a little more complicated. If you don't give comparer2/value2, the filter will pass any pair where the key or the value matches comparer1 and value1. If you do give comparer2/value2, the key must match comparer1/value1 and the value must match comparer2/value2 For example:
```
Filter A:
cookie contains Session
Filter B:
cookie contains Session contains 456
Filter C:
cookie ncontains Ultra
Cookie: SuperSession=abc123
Matches A and C but not B
Cookie: UltraSession=abc123456
Matches both A and B but not C
```
### List of fields
| Field Name | Aliases | Description | Format |
|:--------|:------------|:-----|:------|
| all | all | The entire request represented as one string | String |
Scope is a set of rules to define whether Pappy should mess with a request. You define the scope by setting the context to what you want the scope to be and running `scope_save`. The scope is saved in the data file and is automatically restored when using the same project directory.
Any requests which don't match all the filters in the scope will be passed straight to the browser and will not be caught by the interceptor or recorded in the data file. This is useful to make sure you don't accidentally do something like log in to your email through the proxy and have your plaintext username/password stored.
| `scope_save` |`scope_save`| Set the current context to be the scope |
| `sr` |`scope_reset`, `sr`| Set the current context to the scope |
| `scope_delete` |`scope_delete`| Clear the scope (everything's in scope!) |
| `scope_list` |`scope_list`, `sls`| List all the filters that are applied to the scope |
### Built-In Filters
Pappy also includes some built in filters that you can apply. These are things that you may want to filter by but may be too tedius to type out. The `fbi` command also supports tab completion.
These features try to fill a similar role to Burp's decoder. Each command will automatically copy the results to the clipboard. In addition, if no string is given, the commands will encode/decode whatever is already in the clipboard. Here is an example of how to base64 encode/decode a string.
```
pappy> b64e "Hello World!"
SGVsbG8gV29ybGQh
pappy> b64d
Hello World!
pappy>
```
And if the result contains non-printable characters, a hexdump will be produced instead
```
pappy> b64d ImALittleTeapot=
0000 22 60 0b 8a db 65 79 37 9a a6 8b "`...ey7...
pappy>
```
The following commands can be used to encode/decode strings:
| Command | Aliases | Description |
|:--------|:--------|:------------|
|`base64_decode`|`base64_decode`, `b64d` | Base64 decode a string |
|`base64_encode`|`base64_encode`, `b64e` | Base64 encode a string |
|`asciihex_decode`|`asciihex_decode`, `ahd` | Decode an ASCII hex string |
|`asciihex_encode`|`asciihex_encode`, `ahe` | Encode an ASCII hex string |
|`url_decode`|`url_decode`, `urld` | Url decode a string |
|`url_encode`|`url_encode`, `urle` | Url encode a string |
|`gzip_decode`|`gzip_decode`, `gzd` | Gzip decompress a string. Probably won't work too well since there's not a great way to get binary data passed in as an argument. I'm working on this. |
|`gzip_encode`|`gzip_encode`, `gze` | Gzip compress a string. Result doesn't get copied to the clipboard. |
|`base64_decode_raw`|`base64_decode_raw`, `b64dr` | Same as `base64_decode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`base64_encode_raw`|`base64_encode_raw`, `b64er` | Same as `base64_encode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`asciihex_decode_raw`|`asciihex_decode_raw`, `ahdr` | Same as `asciihex_decode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`asciihex_encode_raw`|`asciihex_encode_raw`, `aher` | Same as `asciihex_encode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`html_decode_raw`|`html_decode_raw`, `htmldr` | Same as `html_decode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`html_encode_raw`|`html_encode_raw`, `htmler` | Same as `html_encode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`url_decode_raw`|`url_decode_raw`, `urldr` | Same as `url_decode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`url_encode_raw`|`url_encode_raw`, `urler` | Same as `url_encode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`gzip_decode_raw`|`gzip_decode_raw`, `gzdr` | Same as `gzip_decode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
|`gzip_encode_raw`|`gzip_encode_raw`, `gzer` | Same as `gzip_encode` but will not print a hexdump if it contains non-printable characters. It is suggested you use `>` to redirect the output to a file. |
This feature is like Burp's proxy with "Intercept Mode" turned on, except it's not turned on unless you explicitly turn it on. When the proxy gets a request while in intercept mode, it lets you edit it before forwarding it to the server. In addition, it can stop responses from the server and let you edit them before they get forwarded to the browser. When you run the command, you can pass `req` and/or `rsp` as arguments to say whether you would like to intercept requests and/or responses. Only in-scope requests/responses will be intercepted (see Scope section).
The interceptor will use your EDITOR variable to decide which editor to edit the request/response with. If no editor variable is set, it will default to `vi`.
| `ic <req,rsp>+` | `intercept`, `ic` | Begins interception mode. Press enter to leave interception mode and return to the command prompt. Pass in `request` to intercept requests, `response` to intercept responses, or both to intercept both. |
This feature is like Burp's repeater (yes, really). You choose a request and Pappy will open vim in a split window with your request on the left and the original response on the right. You can make changes to the request and then run ":RepeaterSubmitBuffer" to submit the modified request. The response will be displayed on the right. This command is bound to `<leader>f` by default, but you can bind it to something else too in your vimrc (I think, dunno if vim will complain if the function undefined which it will be for regular files). This command will submit whatever buffer your cursor is in, so make sure it's in the request buffer.
Macros are Pappy's version of Burp's intruder. You can use macros to make automated requests through the proxy and save them to the data file. A macro file is any python script file in the current directory that is in the form `macro_<name>.py`. An example project directory with macros would be:
```
$ ls -l
-rw-r--r-- 1 scaryhacker wheel 150 Nov 26 11:17 config.json
-rw------- 1 scaryhacker wheel 2639872 Nov 26 17:18 data.db
-rw-r--r-- 1 scaryhacker wheel 471 Nov 26 18:42 macro_blank.py
-rw-r--r-- 1 scaryhacker wheel 264 Nov 26 18:49 macro_hackthensa.py
-rw-r--r-- 1 scaryhacker wheel 1261 Nov 26 18:37 macro_testgen.py
-rw-r--r-- 1 scaryhacker wheel 241 Nov 26 17:18 macro_test.py
In this case we have a `blank`, `hackthensa`, `testgen`, and `test` macro. A macro script is any python script that defines a `run_macro(args)` function and a `MACRO_NAME` variable. To start with, we'll write a macro to iterate over a numbered image to try and find other images. We will take the following steps to do it:
If you enter in a value for `SHORT_NAME`, you can use it as a shortcut to run that macro. So if in a macro you set `SHORT_NAME='tm'` you can run it by running `pappy> rma tm`.
When you run the macro, any additional command line arguments will be passed to the run_macro function in the `args` argument. For example, if you run your macro using
```
pappy> rma foo thisis an "amazing argument"
```
The `args` argument of run_macro will be `["thisis", "an", "amazing argument"]`. If no arguments are give, `args` will be an empty list.
| `lma [dir]` | `load_macros`, `lma` | Load macros from a directory. If `dir` is not given, use the current directory (the project directory) |
| `rma <macro name>` | `run_macro`, `rma` | Run a macro with the given name. You can use the shortname, filename, or long name. |
| `gma <name> [id(s)]` | `generate_macro`, `gma` | Generate a macro with the given name. If request IDs are given, the macro will contain request objects that contain each request. |
| `rpy <id(s)>` | `rpy` | Print the Python object definitions for each of the given ids |
The main method of interacting with the proxy is through `Request` objects. You can submit a request with `req.sumbit()` and save it to the data file with `req.save()`. The objects also have attributes which can be used to modify the request in a high-level way. You can see the [full documentation](https://roglew.github.io/pappy-proxy/pappyproxy.html#module-pappyproxy.http) for more details on using these objects.
Dict-like objects are represented with a custom class called a `RepeatableDict`. Again, look at the docs for details. For the most part, you can interact with it like a normal dictionary, but don't be surprised if it's missing some methods you would expect.
| port | Yes | Integer | The port the request is/was sent to |
| post_params | Yes | RepeatableDict | Post parameters |
| raw_data | Yes | String | The data part of the request |
| raw_headers | No | String | The text of the headers section of the request |
| reqid | Yes | Integer | The ID of the request. If set when save() is called, it replaces the request with the same id in the database |
| response | Yes | Response | The associated response for the request |
| rsptime | No | Datetime Delta | The time it took to complete the request. Set when submit() is called |
| status_line | Yes | String | The status line of the request (ie 'GET / HTTP/1.1') |
| time_end | Yes | Datetime | The time when the request was completed |
| time_start | Yes | Datetime | The time when the request was started |
| unmangled | Yes | Request | If the request was mangled, the unmangled version of the request |
| url | Yes | String | The URL of the request (ie 'https://www.google.com') |
| url_params | Yes | RepeatableDict | The URL parameters of the request |
| verb | Yes | String | The verb used for the request (ie GET, POST, PATCH, HEAD, etc). Doesn't have to be a valid verb. |
| version | Yes | String | The version part of the status line (ie 'HTTP/1.1') |
Request methods:
| Function | Description |
|:--|:--|
| submit() | Submit the request through the proxy. Does not save the request to the data file |
| save() | Save the request, its unmangled version, its associated response, and the unmangled version of the response to the database |
And here is a quick list of attributes that you can use with `Response` objects:
| Attribute | Settable? | Data Type | Description |
|:--|:--|:--|:--|
| cookies | Yes | RepeatableDict | Cookies set by the response |
| headers | Yes | RepeatableDict | The headers of the response |
| response_code | Yes | Integer | The response code of the response |
| response_text | Yes | String | The text associated with the response code (ie OK, NOT FOUND)
| rspid | Yes | Integer | The response id of the response. If this is the same as another response in the database, calling save() on the associated request will replace that response in the database |
| unmangled | Yes | Response | If the response was mangled, this will refer to the unmangled version of the response. Otherwise it is None |
| version | Yes | String | The version part of the status line of the response (ie 'HTTP/1.1') |
| raw_headers | No | String | A text version of the headers of the response |
| status_line | Yes | String | The status line of the response |
| raw_data | Yes | String | The data portion of the response |
| full_response | No | String | The full text version of the response including headers and data |
Like I said, these interfaces are prone to change and will probably crash when you use them. If you get a traceback, send me an email so I can fix it.
There are also a few functions which could be useful for creating requests in macros. It's worth pointing out that `request_by_id` is useful for passing request objects as arguments. For example, here is a macro that lets you resubmit a request with the Google Bot user agent:
```
## macro_googlebot.py
from pappyproxy.http import Request, get_request, post_request, request_by_id
| get_request(url, url_params={}) | Returns a Request object that contains a GET request to the given url with the given url params |
| post_request(url, post_params={}, url_params={}) | Returns a Request object that contains a POST request to the given url with the given url and post params |
| main_context_ids() | Returns a list of the IDs that are in the current context. Use this for macros that need to act on every in-context request. For example, it can be used in a macro to resubmit a set of requests. |
Intercepting macros let you mangle requests as they pass through the proxy. Similarly to normal macros, an intercepting macro is any python script with an "int" prefix. For example, `int_name.py` would be a valid intercepting macro name. They are also loaded with the `lma` command. An intercepting macro can define two functions: `mangle_request` or `mangle_response`. Both requests only take a `Request` object as a parameter. `mangle_request` returns either a new, modified Request object to change it, or it can return the original object to not mangle it. The `mange_response` must return a `Response` (not request!) object. The request passed in to `mangle_response` will have an associated response with it. If you want to modify the response, copy `request.response`, make modifications, then return it. If you would like to pass it through untouched, just return `request.response`.
Note, that due to twisted funkyness, *you cannot save requests from intercepting macros*. Technically you **can**, but to do that you'll have to define `async_mangle_request` (or response) instead of `mangle_request` (or response) then use `Request.async_deep_save` which generates a deferred, then generate a deferred from `async_mangle_requests` (inline callbacks work too). If you've never used twisted before, please don't try. Twisted is hard. Plus the mangled request will be saved before it is submitted anyways.
Confusing? Here are some example intercepting macros:
In addition, you can use an `init(args)` function to get arguments from the command line. If no arguments are passed, args will be an empty list. Here is an example macro that does a search and replace:
You can use this macro to do any search and replace that you want. For example, if you wanted to replace "Google" with "Skynet", you can run the macro like this:
You can use the following commands to start/stop intercepting macros
| Command | Aliases | Description |
|:--------|:--------|:------------|
| `lma [dir]` | `load_macros`, `lma` | Load macros from a directory. If `dir` is not given, use the current directory (the project directory) |
| `rim <macro name>` | `run_int_macro`, `rim` | Run an intercepting macro. Similarly to normal macros you can use the name, short name, or file name of the macro. |
| `sim <macro name> [args]` | `stop_int_macro`, `sim` | Stop an intercepting macro. If arguments are given, they will be passed to the macro's `init(args)` function if it exists. |
| `lim` | `list_int_macros`, `lsim` | List all enabled/disabled intercepting macros |
Pappy also includes some other templates for generating macros. They can be generated with the `gtma` command. You can then modify the generated macros to do what you want. For example, you could modify the resubmit macro to get a new session token before submitting each request. Using a template can save you from writing boilerplate for commonly created macros.
Examples:
```
# The same as gma foo 1,2,3
pappy> gtma foo macro 1,2,3
Wrote script to macro_foo.py
# Generate a macro that resubmits all in-context requests
pappy> gtma suball resubmit
Wrote script to macro_suball.py
# Generate an intercepting macro that modifies headers as they pass through the proxy
| `gtma <name> <template name> [template arguments]` | `generate_template_macro`, `gtma` | Generate a macro using a template. |
Available macro templates:
| Name | Arguments | Description |
|:-----|:----------|:------------|
| `macro` | `[reqids]` | The template used to generate macros from request IDs. |
| `intmacro` | None | The template used to generate an intercepting macro. |
| `modheader` | None | Create an intercepting macro that modifies a header in the request or response. |
| `resubmit` | None | Create a macro that resubmits all in-context requests. Includes commented out code to maintain session state using a cookie jar. |
Resubmitting Groups of Requests
-------------------------------
You can use the `submit` request to resubmit requests. It is suggested that you use this command with a heavy use of filters and using the wildcard (`*`) to submit all in-context requests. Be careful submitting everything in context, remember, if you have to Ctl-C out you will close Pappy and lose all in-memory requests!
| Command | Aliases | Description |
|:--------|:--------|:------------|
| `submit reqids [-m] [-u] [-p] [-c [COOKIES [COOKIES ...]]] [-d [HEADERS [HEADERS ...]]]` | `submit` | Submit a given set of requests. Request IDs must be passed in as the first argument. The wildcard (`*`) selector can be very useful. Resubmitted requests are given a `resubmitted` tag. See the arguments section for information on the arguments. |
### Useful Filters For Selecting Requests to Resubmit
*`before` and `after` to select requests in a time range. You can use the `after` filter on the most recent request, browse the site, then use the `before` filter to select a continuous browsing session.
*`verb` if you only want to select GET requests
*`path ct logout` to avoid logging out
### Arguments
There are a few simple parameters you can pass to the command to modify requests. These behave like normal command parameters in the terminal. If you need something more complex (ie getting CSRF tokens, refreshing the session token, reacting to Set-Cookie headers, etc.) you should consider writing a macro and using the `main_context_ids` function to get in-context IDs then iterating over them and handling them however you want.
| Argument | Description |
|:---------|:------------|
| `-c <cookie>=<val>` | Modify a cookie on each request before submitting. Can pass more than one pair to the flag to modify more than one cookie. Does not encode the cookie values in any way. |
| `-d <header>=<val>` | Modify a header on each request before submitting. Can pass more than one pair to the flag to modify more than one header. |
| `-m` | Store requests in memory instead of saving to the data file. |
| `-u` | Only submit one request per endpoint. Will count requests with the same path but different url params as *different* endpoints. |
| `-p` | Only submit one request per endpoint. Will count requests with the same path but different url params as *the same* endpoints. |
# Resubmit all in-context requests with the User-Agent header set to "Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)" then store them in memory
You can watch in real-time what requests are going through the proxy. Verbosisty defaults to 1 which just states when connections are made/lost and some information on what is happening. If verbosity is set to 3, it includes all the data which is sent through the proxy and processed. It will print the raw response from the server, what it decodes it to, etc. Even if you don't run this command, all the information is stored in the dubug directory (the directory is cleared every start though!)
| `log [verbosity]` | View the log at the given verbosity. Default verbosity is 1 which just shows connections being made/lost and some other info, verbosity 3 shows full requests/responses as they pass through and are processed by the proxy |
This is a list of other random stuff you can do that isn't categorized under anything else. These are mostly commands that I found that I needed while doing a test and just added. They likely don't do a ton of error checking.
| Command | Aliases | Description |
|:--------|:--------|:------------|
| `dump_response <reqid> [filename]` | `dump_response` | Dumps the data from the response to the given filename (useful for images, .swf, etc). If no filename is given, it uses the name given in the path. |
| `export <req|rsp> <reqid>` | `export` | Writes either the full request or response to a file in the current directory. |
If you don't have any intercepting macros running, Pappy will forward data to the browser as it gets it. However, if you're trying to mangle messages/responses, Pappy will need to download the entire message first.
You can view responses in your browser by visiting `http://pappy/rsp/<rspid>` (NOT pappy.com) in your browser while connected to the proxy. For example, if you want to view the response to request 123, you can visit `http://pappy/rsp/123` to view the response. Pappy will return a response with the same body as the original response and will not make a request to the server. The response will not have the same headers as the original response (aside from the Content-Type header). In addition, Pappy doesn't modify any URLs in the page which means your browser will still fetch external resources like images, JavaScript etc from external servers.
While information on websockets is scattered throughout the readme, this section will collect it all into one place. Websockets can be intercepted/modified, viewed, etc just like regular requests and responses.
### Viewing a WebSocket Session
Websocket sessions are referred to by the request that was used to initiate their handshake. In the following example, we look at all our websocket connections to an echo websocket server. In both connections, we send "Hello" from the browser. In the second connection, we sent "Hello" twice and used the interceptor to change the second "Hello" message to say "HelloFool" instead.
```
pappy> f ws
pappy> ls
ID Verb Host Path S-Code Req Len Rsp Len Time Mngl
6 GET localhost:9000 /?compressed=true 101 Switching... 0 0 0.34 --
1 GET localhost:9000 /?compressed=true 101 Switching... 0 0 0.32 --
pappy> vfq 1
Websocket session handshake
GET /?compressed=true HTTP/1.1
Host: localhost:9000
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:49.0) Gecko/20100101 Firefox/49.0
You can intercept websocket messages by running `ic` with a `ws` argument:
```
pappy> ic ws
```
### Intercepting Macros
You can automatically mangle websocket messages similarly to requests and responses. Just define a `mangle_ws` function or an `async_mangle_ws` (see the [Intercepting Macros](#intercepting-macros) section for more details).
Note that this section is a very quick overview of plugins. For a full description of how to write them, please see [the official docs](https://roglew.github.io/pappy-proxy/pappyplugins.html).
It is also possible to write plugins which are reusable across projects. Plugins are simply Python scripts located in `~/.pappy/plugins`. Plugins are able to create new console commands and maintain state throughout a Pappy session. They can access the same API as macros, but the plugin system is designed to allow you to create general purpose commands as compared to macros which are meant to be project-specific scripts. Still, it may not be a bad idea to try building a macro to do something in a quick and dirty way before writing a plugin since plugins are more complicated to write.
A simple hello world plugin could be something like:
```
## hello.py
import shlex
def hello_world(line):
if line:
args = shlex.split(line)
print 'Hello, %s!' % (', '.join(args))
else:
print "Hello, world!"
###############
## Plugin hooks
def load_cmds(cmd):
cmd.set_cmds({
'hello': (hello_world, None),
})
cmd.add_aliases([
('hello', 'hlo'),
('hello', 'ho'),
])
```
You can also create commands which support autocomplete:
return [n for n in _AUTOCOMPLETE_NAMES if n.startswith(text)]
###############
## Plugin hooks
def load_cmds(cmd):
cmd.set_cmds({
'hello': (hello_world, complete_hello_world),
})
cmd.add_aliases([
('hello', 'hlo'),
])
```
Then when you run Pappy you can use the ``hello`` command:
```
$ pappy -l
Temporary datafile is /tmp/tmpBOXyJ3
Proxy is listening on port 8000
pappy> ho
Hello, world!
pappy> ho foo bar baz
Hello, foo, bar, baz!
pappy> ho foo bar "baz lihtyur"
Hello, foo, bar, baz lihtyur!
pappy>
```
### Should I Write a Plugin or a Macro?
A lot of the time, you can get away with writing a macro. However, you may consider writing a plugin if:
* You find yourself copying one macro to multiple projects
* You want to write a general tool that can be applied to any website
* You need to maintain state during the Pappy session
My guess is that if you need one quick thing for a project, you're better off writing a macro first and seeing if you end up using it in future projects. Then if you find yourself needing it a lot, write a plugin for it. You may also consider keeping a `mine.py` plugin where you can write out commands that you use regularly but may not be worth creating a dedicated plugin for.
There are some settings that apply to Pappy as a whole and are stored in `~/.pappy/global_config.json`. These settings are generally for tuning performance or modifying behavior on a system-wide level. No information about projects is put in here since it is world readable. You can technically add settings in here for plugins that you write, but if it's at all possible, please keep settings in the normal project config.
Settings included in `~/.pappy/global_config.json`:
| Setting | Description |
|:--------|:------------|
| cache_size | The number of requests from history that will be included in memory at any given time. Set to -1 to keep everything in memory. See the request cache section for more info. |
Pappy allows you to use an upstream HTTP proxy. You can do this by adding an `http_proxy` value to config.json. You can use the following for anonymous access to the proxy:
At the moment, only basic auth is supported. Anything in-scope that passes through any of the active listeners will use the proxy. Out of scope requests will not be sent through the proxy.
Using a SOCKS Proxy
-------------------
Pappy allows you to use an upstream SOCKS proxy. You can do this by adding a `socks_proxy` value to config.json. You can use the following for anonymous access to the proxy:
Sometimes you get a frustrating thick client that doesn’t let you mess with proxy settings to get it to go through a proxy. However, if you can redirect where it sends its traffic to localhost, you can get Pappy to take that traffic and redirect it to go where it should.
It takes root permissions to listen on low numbered ports. As a result, we’ll need to do some root stuff to listen on ports 80 and 443 and get the data to Pappy. There are two ways to get the traffic to Pappy. The first is to set up port forwarding as root to send traffic from localhost:80 to localhost:8080 and localhost:443 to localhost:8443 (since we can listen on 8080 and 8443 without root). Or you can YOLO, run Pappy as root and just have it listen on 80 and 443.
According to Google you can use the following command to forward port 80 on localhost to 8080 on Linux:
This configuration will cause Pappy to open a port on 8080 that will accept connections normally and a port on 8443 which will accept SSL connections. The forward_host setting tells Pappy to redirect any requests sent to the port to the given host. It will also update the request’s host header. forward_host_ssl does the same thing, but it listens for SSL connections and forces the connection to use SSL.
Or if you’re going to YOLO it do the same thing then listen on port 80/443 directly. I do not suggest you do this.
Pappy includes some basic features for automatically compressing and encrypting your project directory with a password. However, before I go into details on how to do this, I need to make one thing clear.
**Don't rely on Pappy to encrypt confidential information. Use a dedicated encryption product to encrypt your project directory instead.**
Other commercial and large open source crypto projects have had a much larger number of people look at their crypto implementations and are less likely to have errors in their implementation. However, for cases where you don't need enterprise level security or if you just want your project stored in a single password-protected file instead of a directory, Pappy's got you covered.
Here is how Pappy's project encryption works:
* Open a project by running Pappy with the `-c` flag
* Pappy creates a `crypt/` directory in the current directory and changes the working directory into it
* Do work as normal. You can use other tools in the created `crypt/` directory
* When you quit Pappy, the file is compressed and encrypted with the provided password
* The project directory is deleted
Unfortunately, if Pappy hard crashes the files will not be cleaned up. However, if you start Pappy and it notices a `crypt/` directory, it will attempt to use it as the project directory and create a new encrypted project file upon exiting.
### Text just appeared over my prompt! What do I do?!
Unfortunately I've been a bit lazy when it comes to printing errors to the terminal. A lot of stuff is just printed to stdout using `print`. This has the side effect of printing over your input. I'm very sorry and I'm trying to work on better solutions, but for now:
* Hit Ctl-L to clear the terminal. Your input will be saved.
You can't do anything with a request/response until it is decoded and saved to disk. In between the time when a request is decoded and when it's saved to disk, it will have an ID of `--`. So just wait a little bit and it will get an ID you can use.
I do some stuff to try and keep speed and memory usage to reasonable levels. Unfortunately, things might seem slow in some areas. This is where I try and explain why those exist. Honestly, you probably don't care about this, but I'd rather have it written down and have nobody read it than just leave people in the dark.
### Request Cache / Memory usage
For performance reasons, Pappy by default will not store every request in memory. The cache will store a certain number of the most recently accessed requests in memory. This means that if you go through all of history, it could be slow (for example running `ls a` or `sm`). If you have enough RAM to keep everything in memory, you can set the request cache size to -1 to just keep everything in memory. However, even if the cache size is unlimited, it still won't load a request into memory untill you access it. So if you want to load everything in memory, run `ls a`.
By default, Pappy will cache 2000 requests. This is kind of heavy, but it's assumed you're doing testing on a reasonably specced laptop. Personally, I live on the edge and use -1 until I run into memory issues.
