Skip to content

Commit

Permalink
Update Adding_Detectors_external.md
Browse files Browse the repository at this point in the history
  • Loading branch information
zricethezav authored Sep 26, 2023
1 parent 1e42dae commit 0a37f69
Showing 1 changed file with 51 additions and 31 deletions.
82 changes: 51 additions & 31 deletions hack/docs/Adding_Detectors_external.md
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,7 @@ Note: Be sure to update the tests to reference the new secret values in GSM, or

3. Proceed from step 3 of [Creating a new Secret Scanner](#creating-a-new-secret-scanner)

### Creating a new Secret Scanner
### Creating a new Secret Detector

1. Add a new Secret Detector enum to the [`DetectorType` list here](/proto/detectors.proto).

Expand All @@ -69,40 +69,19 @@ Note: Be sure to update the tests to reference the new secret values in GSM, or
Add the secret scanner to the [`pkg/engine/defaults.go`](https://github.com/trufflesecurity/trufflehog/blob/main/pkg/engine/defaults.go) file like [`github.com/trufflesecurity/trufflehog/v3/pkg/detectors/<detector_name>`](https://github.com/trufflesecurity/trufflehog/blob/b71ea27a696bdf1c3141f637fda4ee4936c2f2d6/pkg/engine/defaults.go#L9) and
[`<detector_name>.Scanner{}`](https://github.com/trufflesecurity/trufflehog/blob/b71ea27a696bdf1c3141f637fda4ee4936c2f2d6/pkg/engine/defaults.go#L1546)
5. Complete the secret detector.
5. Complete the Secret Detector.
The previous step templated a boilerplate + some example code as a package in the `pkg/detectors` folder for you to work on.
The secret detector can be completed with these general steps:
The Secret Detector can be completed with these general steps:
1. Create a [test secrets file, and export the variable](#using-a-test-secret-file)
2. Update the pattern regex and keywords. Try iterating with [regex101.com](http://regex101.com/).
3. Update the verifier code to use a non-destructive API call that can determine whether the secret is valid or not.
1. Update the pattern regex and keywords. Try iterating with [regex101.com](http://regex101.com/).
2. Update the verifier code to use a non-destructive API call that can determine whether the secret is valid or not.
* Make sure you understand [verification indeterminacy](#verification-indeterminacy).
4. Update the tests with these test cases at minimum:
1. Found and verified (using a credential loaded from GCP Secrets)
2. Found and unverified (determinately, i.e. the secret is invalid)
3. Found and unverified (indeterminately due to timeout)
4. Found and unverified (indeterminately due to an unexpected API response)
5. Not found
6. Any false positive cases that you come across
5. Create a pull request for review.
3. Create a [test for the detector](#testing-the-detector)
4. Create a pull request for review.
## Addendum
### Verification indeterminacy
There are two types of reasons that secret verification can fail:
* The candidate secret is not actually a valid secret.
* Something went wrong in the process unrelated to the candidate secret, such as a transient network error or an unexpected API response.
In Trufflehog parlance, the first type of verification response is called _determinate_ and the second type is called _indeterminate_. Verification code should distinguish between the two by returning an error object in the result struct **only** for indeterminate failures. In general, a verifier should return an error (indicating an indeterminate failure) in all cases that haven't been explicitly identified as determinate failure states.

For example, consider a hypothetical authentication endpoint that returns `200 OK` for valid credentials and `403 Forbidden` for invalid credentials. The verifier for this endpoint could make an HTTP request and use the response status code to decide what to return:
* A `200` response would indicate that verification succeeded. (Or maybe any `2xx` response.)
* A `403` response would indicate that verification failed **determinately** and no error object should be returned.
* Any other response would indicate that verification failed **indeterminately** and an error object should be returned.

### Using a test secret file
### Testing the Detector
To ensure the quality of your PR, make sure your tests are passing with verified credentials.
1. Create a file called `.env` with this env file format:
Expand All @@ -116,8 +95,49 @@ For example, consider a hypothetical authentication endpoint that returns `200 O
```bash
export TEST_SECRET_FILE=".env"
```
The `.env` file should be in the new detector's directory like this:
```
├── tailscale
│   ├── .env
│   ├── tailscale.go
│   └── tailscale_test.go
```
Now that a `.env` file is present, the test file can load secrets locally.
Next, Update the tests with these test cases at minimum:
1. Found and verified (using a credential loaded from the .env file)
2. Found and unverified (determinately, i.e. the secret is invalid)
3. Found and unverified (indeterminately due to timeout)
4. Found and unverified (indeterminately due to an unexpected API response)
5. Not found
[Here is an exemplary test file for a detector which covers all 5 test cases](https://github.com/trufflesecurity/trufflehog/blob/6f9065b0aae981133a7fa3431c17a5c6213be226/pkg/detectors/browserstack/browserstack_test.go)
Now run the tests!
```bash
go test ./pkg/detectors/<detector> -tags=detectors
```

Now, the detector test should attempt to load the given env key from that file.
If the tests are passing, feel free to open a PR!




## Addendum

### Verification indeterminacy

There are two types of reasons that secret verification can fail:
* The candidate secret is not actually a valid secret.
* Something went wrong in the process unrelated to the candidate secret, such as a transient network error or an unexpected API response.

In Trufflehog parlance, the first type of verification response is called _determinate_ and the second type is called _indeterminate_. Verification code should distinguish between the two by returning an error object in the result struct **only** for indeterminate failures. In general, a verifier should return an error (indicating an indeterminate failure) in all cases that haven't been explicitly identified as determinate failure states.

For example, consider a hypothetical authentication endpoint that returns `200 OK` for valid credentials and `403 Forbidden` for invalid credentials. The verifier for this endpoint could make an HTTP request and use the response status code to decide what to return:
* A `200` response would indicate that verification succeeded. (Or maybe any `2xx` response.)
* A `403` response would indicate that verification failed **determinately** and no error object should be returned.
* Any other response would indicate that verification failed **indeterminately** and an error object should be returned.

### Adding Protos in Windows

Expand Down

0 comments on commit 0a37f69

Please sign in to comment.