syntax = "proto3";

package sources;

option go_package = "github.com/trufflesecurity/trufflehog/v3/pkg/pb/sourcespb";

import "validate/validate.proto";
import "credentials.proto";
import "google/protobuf/any.proto";
import "google/protobuf/duration.proto";
import "google/protobuf/timestamp.proto";


enum SourceType {
  SOURCE_TYPE_AZURE_STORAGE = 0;
  SOURCE_TYPE_BITBUCKET = 1;
  SOURCE_TYPE_CIRCLECI = 2;
  SOURCE_TYPE_CONFLUENCE = 3;
  SOURCE_TYPE_DOCKER = 4;
  SOURCE_TYPE_ECR = 5;
  SOURCE_TYPE_GCS = 6;
  SOURCE_TYPE_GITHUB = 7;
  SOURCE_TYPE_PUBLIC_GIT = 8;
  SOURCE_TYPE_GITLAB = 9;
  SOURCE_TYPE_JIRA = 10;
  SOURCE_TYPE_NPM_UNAUTHD_PACKAGES = 11;
  SOURCE_TYPE_PYPI_UNAUTHD_PACKAGES = 12;
  SOURCE_TYPE_S3 = 13;
  SOURCE_TYPE_SLACK = 14;
  SOURCE_TYPE_FILESYSTEM = 15;
  SOURCE_TYPE_GIT = 16;
  SOURCE_TYPE_TEST = 17;
  SOURCE_TYPE_S3_UNAUTHED = 18;
  SOURCE_TYPE_GITHUB_UNAUTHENTICATED_ORG = 19;
  SOURCE_TYPE_BUILDKITE = 20;
  SOURCE_TYPE_GERRIT = 21;
  SOURCE_TYPE_JENKINS = 22;
  SOURCE_TYPE_TEAMS = 23;
  SOURCE_TYPE_JFROG_ARTIFACTORY = 24;
  SOURCE_TYPE_SYSLOG = 25;
  SOURCE_TYPE_PUBLIC_EVENT_MONITORING = 26;
  SOURCE_TYPE_SLACK_REALTIME = 27;
  SOURCE_TYPE_GOOGLE_DRIVE = 28;
  SOURCE_TYPE_SHAREPOINT = 29;
  SOURCE_TYPE_GCS_UNAUTHED = 30;
  SOURCE_TYPE_AZURE_REPOS = 31;
  SOURCE_TYPE_TRAVISCI = 32;
  SOURCE_TYPE_POSTMAN = 33;
  SOURCE_TYPE_WEBHOOK = 34;
  SOURCE_TYPE_ELASTICSEARCH = 35;
  SOURCE_TYPE_HUGGINGFACE = 36;
  SOURCE_TYPE_GITHUB_EXPERIMENTAL = 37;
  SOURCE_TYPE_SENTRY = 38;
  SOURCE_TYPE_GITHUB_REALTIME = 39;
  SOURCE_TYPE_STDIN = 40;
  SOURCE_TYPE_SLACK_CONTINUOUS = 41;
  SOURCE_TYPE_JSON_ENUMERATOR = 42;
  SOURCE_TYPE_AWS_SECRETS_MANAGER = 43;
}

message LocalSource {
  string type = 1 ;
  string name = 2 ;
  // DEPRECATED: scan_interval is deprecated and can be removed when we no longer depend on the name.
  // Deprecating in favor of scan_period due to the fact that scan_interval is a duration
  // which is a fixed-length span of time represented as a count of seconds and fractions of seconds
  // at nanosecond resolution. Most of the time, we want to be able to specify a scan interval in
  // human-readable format (e.g. 45s, 30m, 12h, etc.) which is not possible with a duration.
  // https://protobuf.dev/reference/protobuf/google.protobuf/#duration
  google.protobuf.Duration scan_interval = 3 [deprecated = true];
  bool verify = 4;
  google.protobuf.Any connection = 5;
  string scan_period = 6;
}

// https://www.jfrog.com/confluence/display/JFROG/Artifactory+REST+API#ArtifactoryRESTAPI-RetrieveFolderorRepositoryArchive
message Artifactory {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.BasicAuth basic_auth = 2;
    string access_token = 3;
    credentials.Unauthenticated unauthenticated = 7;
  }
  repeated string repositories = 4;
  repeated string include_paths = 5;
  repeated string ignore_paths = 6;
}

message AzureStorage {
  oneof credential {
    string connection_string = 1;
    credentials.BasicAuth basic_auth = 2;
    string client_certificate = 3;
    credentials.Unauthenticated unauthenticated = 4;
  }
  repeated string storage_containers = 5;
}

enum BitbucketInstallationType {
  AUTODETECT = 0;
  CLOUD = 1;
  DATA_CENTER = 2;
}

message Bitbucket {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.Oauth2 oauth = 3;
    credentials.BasicAuth basic_auth = 4;
    credentials.BitbucketApiToken api_token = 14;
  }
  repeated string repositories = 5;
  repeated string ignore_repos = 6;
  bool skip_binaries = 7;
  bool skip_archives = 8;
  BitbucketInstallationType installation_type = 9;
  string oauth_authorization_endpoint = 10; // endpoint for OAuth authorization flow in Bitbucket Data Center application links
  string oauth_token_endpoint = 11; // endpoint for getting OAuth access/refresh tokens in Bitbucket Data Center application links
  repeated string oauth_scopes = 12; // list of OAuth scopes that the access token is provided
  bool allow_secrets_manager_write = 13; // flag to explicitly allow TruffleHog to write to the user's secrets manager
}

message CircleCI {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
  }
}

message TravisCI {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
  }
}

message Confluence {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.Unauthenticated unauthenticated = 2;
    credentials.BasicAuth basic_auth = 3;
    string token = 4;
  }

  enum GetAllSpacesScope {
    ALL = 0;
    GLOBAL = 1;
    PERSONAL = 2;
  }

  GetAllSpacesScope spaces_scope = 5;
  bool insecure_skip_verify_tls = 6;
  repeated string spaces = 7;
  repeated string ignore_spaces = 8;
  bool include_attachments = 9;
  bool skip_history = 10;
  bool include_comments = 11;
}

message Docker {
  oneof credential {
    credentials.Unauthenticated unauthenticated = 1;
    credentials.BasicAuth basic_auth = 2;
    string bearer_token = 3;
    bool docker_keychain = 4;
  }
  repeated string images = 5;
  repeated string exclude_paths = 6;
  string namespace = 7;
  string registry_token = 8;
}

message ECR {
  oneof credential {
    credentials.KeySecret access_key = 1;
  }
  repeated string registries = 2;
}

message Filesystem {
  // DEPRECATED: directories is deprecated and can be removed / renamed to
  // paths when we no longer depend on the name in enterprise configs.
  repeated string directories = 1;
  repeated string paths = 2;
  string include_paths_file = 3; // path to file containing newline separated list of paths
  string exclude_paths_file = 4; // path to file containing newline separated list of paths
  bool skip_binaries = 5; // allows skipping binary files from the scan
  int32 max_symlink_depth = 6; // allows following symlinks upto specified max depth
}

message GCS {
  oneof credential {
    string json_service_account = 1;
    string api_key = 2;
    credentials.Unauthenticated unauthenticated = 3;
    credentials.CloudEnvironment adc = 4;
    string service_account_file = 11;
    credentials.Oauth2 oauth = 12;
  }
  string project_id = 5;
  repeated string include_buckets = 6;
  repeated string exclude_buckets = 7;
  repeated string include_objects = 8;
  repeated string exclude_objects = 9;
  int64 max_object_size = 10;
}

message Git {
  oneof credential {
    credentials.BasicAuth basic_auth = 1;
    credentials.Unauthenticated unauthenticated = 2;
    credentials.SSHAuth ssh_auth = 5;
  }
  repeated string directories = 3;
  repeated string repositories = 4;
  string head = 6;
  string base = 7;
  bool bare = 8;
  string include_paths_file = 9; // path to file containing newline separated list of paths
  string exclude_paths_file = 10; // path to file containing newline separated list of paths
  string exclude_globs = 11; // comma separated list of globs
  int64  max_depth = 12;
  // This field is generally used by the CLI or within CI/CD systems to specify a single repository,
  // whereas the repositories field is used by the enterprise config to specify multiple repositories.
  // Passing a single repository via the uri field also allows for additional options to be specified
  // like head, base, bare, etc.
  string uri = 13; // repository URL. https://, file://, or ssh://
  bool skip_binaries = 14;
  bool skip_archives = 15;
  string clone_path = 16;
  bool no_cleanup = 17;
  bool print_legacy_json = 18;
  bool trust_local_git_config = 19; // Trust local git configurations
}

message GitLab {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.Oauth2 oauth = 3;
    credentials.BasicAuth basic_auth = 4;
  }
  repeated string repositories = 5;
  repeated string ignore_repos = 6;
  bool skip_binaries = 7;
  bool skip_archives = 8;
  repeated string include_repos = 9;
  bool exclude_projects_shared_into_groups = 10;
  bool remove_auth_in_url = 11;
  repeated string group_ids = 12;
  string clone_path = 13;
  bool no_cleanup = 14;
  bool print_legacy_json = 15;
}

message GitHub {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.GitHubApp github_app = 2;
    string token = 3;
    credentials.Unauthenticated unauthenticated = 4;
    credentials.BasicAuth basic_auth = 13;
  }
  repeated string repositories = 5;
  repeated string organizations = 6;
  bool scanUsers = 7;
  bool includeForks = 8;
  string head = 9;
  string base = 10;
  repeated string ignore_repos = 11;
  repeated string include_repos = 12;
  bool include_pull_request_comments = 14;
  bool include_issue_comments = 15;
  bool include_gist_comments = 16;
  bool skip_binaries = 17;
  bool skip_archives = 18;
  bool include_wikis = 19;
  uint32 comments_timeframe_days = 20;
  bool remove_auth_in_url = 21;
  string clone_path = 22;
  bool no_cleanup = 23;
  bool ignore_gists = 24;
  bool print_legacy_json = 25;
  // When using GitHub App credentials, enumerate repositories from every app installation.
  bool scan_all_installations = 26;
  bool exclude_archived = 27;
}

message GitHubExperimental {
  string repository = 1;
  oneof credential {
    string token = 2;
  }
  bool object_discovery = 3;
  int64 collision_threshold = 4;
  bool delete_cached_data = 5;
}

message GitHubRealtime {
  uint32 listener_port = 1;
  string webhook_secret = 2;
  string endpoint = 3 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.GitHubApp github_app = 4;
    string token = 5;
    credentials.Unauthenticated unauthenticated = 6;
    credentials.BasicAuth basic_auth = 7;
  }
}

message GoogleDrive {
  oneof credential {
    string refresh_token = 1;
    bool use_token_service = 2;
    credentials.Oauth2 oauth = 3; // on-prem scanning
    credentials.GoogleDriveDWD dwd = 4; // Domain-Wide Delegation
  }
}

message Huggingface {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.Unauthenticated unauthenticated = 3;
  }
  repeated string models = 4;
  repeated string spaces = 5;
  repeated string datasets = 12;
  repeated string organizations = 6;
  repeated string users = 15;
  repeated string ignore_models = 7;
  repeated string include_models = 8;
  repeated string ignore_spaces = 9;
  repeated string include_spaces = 10;
  repeated string ignore_datasets = 13;
  repeated string include_datasets = 14;
  bool  skip_all_models = 16;
  bool  skip_all_spaces = 17;
  bool  skip_all_datasets = 18;
  bool include_discussions = 11;
  bool include_prs = 19;
  repeated string buckets = 20;
  repeated string ignore_buckets = 21;
  repeated string include_buckets = 22;
  bool skip_all_buckets = 23;
}

enum JiraInstallationType {
  JIRA_INSTALLATION_TYPE_AUTODETECT = 0;
  JIRA_INSTALLATION_TYPE_CLOUD = 1;
  JIRA_INSTALLATION_TYPE_DATA_CENTER = 2;
}

message JIRA {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.BasicAuth basic_auth = 2;
    credentials.Unauthenticated unauthenticated = 3;
    credentials.Oauth2 oauth = 4;
    string token = 6;
  }
  repeated string projects = 5;
  repeated string ignore_projects = 7;
  bool insecure_skip_verify_tls = 8;
  
  // Override autodetection of Jira installation type.
  // Defaults to AUTODETECT which checks for cloud-only endpoints.
  JiraInstallationType installation_type = 9;
}

message NPMUnauthenticatedPackage {
  oneof credential {
    credentials.Unauthenticated unauthenticated = 1;
  }
}

message PyPIUnauthenticatedPackage {
  oneof credential {
    credentials.Unauthenticated unauthenticated = 1;
  }
}

message S3 {
  oneof credential {
    credentials.KeySecret access_key = 1;
    credentials.Unauthenticated unauthenticated = 2;
    credentials.CloudEnvironment cloud_environment = 4;
    credentials.AWSSessionTokenSecret session_token = 5;
  }
  repeated string buckets = 3;
  int64 max_object_size = 6;
  repeated string roles = 7;
  repeated string ignore_buckets = 8;
  bool enable_resumption = 9 [deprecated = true];
  // endpoint targets an S3-compatible service other than AWS S3. When set,
  // per-bucket region discovery is skipped because it is an AWS-only API.
  string endpoint = 10 [(validate.rules).string.uri_ref = true];
  // region is the region used to sign requests. Defaults to us-east-1.
  string region = 11;
  // include_prefixes limits the scan to objects whose key matches one of these
  // key prefixes. Leave empty to consider every object in the bucket.
  repeated string include_prefixes = 12;
  // exclude_prefixes skips objects whose key matches one of these key prefixes.
  repeated string exclude_prefixes = 13;
  // include_extensions limits the scan to objects whose key ends in one of these
  // file extensions, written without a leading dot (e.g. "tf", "yaml"). Leave
  // empty to consider every extension. Cannot be combined with
  // exclude_extensions.
  repeated string include_extensions = 14;
  // exclude_extensions skips objects whose key ends in one of these file
  // extensions, written without a leading dot (e.g. "zip", "mp4"). Cannot be
  // combined with include_extensions.
  repeated string exclude_extensions = 15;
}

message Slack {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.SlackTokens tokens = 5;
  }
  repeated string channels = 3;
  repeated string ignore_list = 4;
}

message Test{}

message Buildkite {
  oneof credential {
    string token = 1;
  }
}

message Gerrit {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.BasicAuth basic_auth = 2;
    credentials.Unauthenticated unauthenticated = 3;
  }
  repeated string projects = 4;
  bool skip_binaries = 5;
  bool skip_archives = 6;
}

message Jenkins {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    credentials.BasicAuth basic_auth = 2;
    credentials.Header header = 3;
    credentials.Unauthenticated unauthenticated = 5;
  }
  bool insecure_skip_verify_tls = 4;
}

message Teams {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.ClientCredentials authenticated = 3;
    credentials.Oauth2 oauth = 7;
  }
  repeated string channels = 4;
  repeated string ignore_list = 5;
  repeated string team_ids = 6;
  string tenant_id = 8;
}

message Syslog {
  string protocol = 1;
  string listen_address = 2;
  string tlsCert = 3;
  string tlsKey = 4;
  string format = 5;
}

message Forager {
  oneof credential {
    credentials.Unauthenticated unauthenticated = 1;
  }
  repeated string domains = 2;
  int64 max_depth = 3;
  google.protobuf.Timestamp since = 4;
}

message SlackRealtime {
  oneof credential {
    credentials.SlackTokens tokens = 1;
  }
}

message Sharepoint {
  oneof credential {
    credentials.Oauth2 oauth = 1;
  }
  string site_url = 2;
  bool include_versions = 3;
  bool include_comments = 4;
  bool include_attachments = 5;
}

message AzureRepos {
  string endpoint = 1 [(validate.rules).string.uri_ref = true];
  oneof credential {
    string token = 2;
    credentials.Oauth2 oauth = 3;
  }
  repeated string repositories = 4;
  repeated string organizations = 5;
  repeated string projects = 6;
  bool include_forks = 7;
  repeated string ignore_repos = 8;
  repeated string include_repos = 9;
  repeated string include_projects = 10;
  repeated string ignore_projects = 11;
  bool skip_binaries = 12;
  bool skip_archives = 13;
}

message Postman {
  oneof credential {
    credentials.Unauthenticated unauthenticated = 1;
    string token = 2;
  }
  repeated string workspaces = 3;
  repeated string collections = 4;
  repeated string environments = 5;
  repeated string exclude_collections = 6;
  repeated string exclude_environments = 7;
  repeated string include_collections = 8;
  repeated string include_environments = 9;
  repeated string detector_keywords = 10;
  repeated string workspace_paths = 11;
  repeated string collection_paths = 12;
  repeated string environment_paths = 13;
}

message Webhook {
  string listen_address = 1 [(validate.rules).string.hostname = true];
  oneof credential {
    credentials.Header header = 2;
  }
  oneof variant {
    Vector vector = 4;
  }
}

message Vector {
  string locator_field = 1;
  string link_format = 2;
}

message Elasticsearch {
  repeated string nodes = 1;
  string username = 2;
  string password = 3;
  string cloud_id = 4;
  string api_key = 5;
  string service_token = 6;
  string index_pattern = 7;
  string query_json = 8;
  string since_timestamp = 9;
  bool best_effort_scan = 10;
}

message Sentry {
  string endpoint = 1;
  oneof credential {
    string auth_token = 2;
    string dsn_key = 3;
    string api_key = 4;
  }
  bool insecure_skip_verify_tls = 5;
  string projects = 6;
}

message Stdin {}

message SlackContinuous {
  string namespace = 1;
  string project_id = 2;
}

message JSONEnumerator {
  repeated string paths = 1;
}