classifyre-cli 0.4.68__tar.gz → 0.4.70__tar.gz

{classifyre_cli-0.4.68 → classifyre_cli-0.4.70}/.turbo/turbo-build.log

@@ -1,3 +1,3 @@
 $ uv sync
-Resolved 300 packages in 269ms
+Resolved 302 packages in 146ms
 Checked 51 packages in 1ms

{classifyre_cli-0.4.68 → classifyre_cli-0.4.70}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: classifyre-cli
-Version: 0.4.68
+Version: 0.4.70
 Summary: Classifyre CLI — scan and classify unstructured data sources
 License: MIT
 Keywords: data,ingestion,metadata,pii,secrets,unstructured

{classifyre_cli-0.4.68 → classifyre_cli-0.4.70}/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@classifyre/cli",
-  "version": "0.4.68",
+  "version": "0.4.70",
   "private": true,
   "scripts": {
     "build": "uv sync",

{classifyre_cli-0.4.68 → classifyre_cli-0.4.70}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "classifyre-cli"
-version = "0.4.68"
+version = "0.4.70"
 description = "Classifyre CLI — scan and classify unstructured data sources"
 readme = "README.md"
 requires-python = ">=3.12"
@@ -227,6 +227,9 @@ google-workspace = [
 google-cloud-storage = [
     "google-cloud-storage>=3.10.1,<4.0.0",
 ]
+dropbox = [
+    "dropbox>=12.2.1,<13.0.0",
+]
 email = [
     "imap-tools>=1.10.0,<2.0.0",
 ]
@@ -377,6 +380,7 @@ module = [
     "confluent_kafka.*",
     "boto3.*",
     "botocore.*",
+    "dropbox.*",
 ]
 ignore_missing_imports = true
 

{classifyre_cli-0.4.68 → classifyre_cli-0.4.70}/src/models/generated_input.py

@@ -54,6 +54,7 @@ class AssetType(StrEnum):
     LOCAL_FOLDER = 'LOCAL_FOLDER'
     MICROSOFT_365 = 'MICROSOFT_365'
     GOOGLE_WORKSPACE = 'GOOGLE_WORKSPACE'
+    DROPBOX = 'DROPBOX'
     SANDBOX = 'SANDBOX'
 
 
@@ -354,6 +355,7 @@ class Type(StrEnum):
     LOCAL_FOLDER = 'LOCAL_FOLDER'
     MICROSOFT_365 = 'MICROSOFT_365'
     GOOGLE_WORKSPACE = 'GOOGLE_WORKSPACE'
+    DROPBOX = 'DROPBOX'
     SANDBOX = 'SANDBOX'
 
 
@@ -3067,6 +3069,7 @@ class Type21(StrEnum):
     LOCAL_FOLDER = 'LOCAL_FOLDER'
     MICROSOFT_365 = 'MICROSOFT_365'
     GOOGLE_WORKSPACE = 'GOOGLE_WORKSPACE'
+    DROPBOX = 'DROPBOX'
     SANDBOX = 'SANDBOX'
 
 
@@ -3333,7 +3336,7 @@ class NotionInput(CoreInput):
 
 class KafkaSecurityProtocol(StrEnum):
     """
-    Kafka client security protocol
+    Kafka client security protocol. Leave unset to derive it from the authentication mode (SASL → SASL_SSL, Client Certificate → SSL, No Authentication → PLAINTEXT); set it only to override, e.g. SASL_PLAINTEXT for a SASL broker without TLS.
     """
 
     PLAINTEXT = 'PLAINTEXT'
@@ -3398,19 +3401,21 @@ class NoAuthentication(BaseModel):
         extra='forbid',
     )
     auth_mode: Literal['NONE']
-    bootstrap_servers: str = Field(
-        ..., description='Comma-separated Kafka bootstrap servers (host:port)'
+    host: str = Field(
+        ..., description='Kafka broker host name or IP (e.g. kafka.example.com)'
     )
+    port: int = Field(..., description='Kafka broker port', ge=1, le=65535)
 
 
-class SASL(BaseModel):
+class SASLUsernamePassword(BaseModel):
     model_config = ConfigDict(
         extra='forbid',
     )
     auth_mode: Literal['SASL']
-    bootstrap_servers: str = Field(
-        ..., description='Comma-separated Kafka bootstrap servers (host:port)'
+    host: str = Field(
+        ..., description='Kafka broker host name or IP (e.g. kafka.example.com)'
     )
+    port: int = Field(..., description='Kafka broker port', ge=1, le=65535)
 
 
 class ClientCertificateMTLS(BaseModel):
@@ -3418,9 +3423,26 @@ class ClientCertificateMTLS(BaseModel):
         extra='forbid',
     )
     auth_mode: Literal['CLIENT_CERT']
-    bootstrap_servers: str = Field(
-        ..., description='Comma-separated Kafka bootstrap servers (host:port)'
+    host: str = Field(
+        ..., description='Kafka broker host name or IP (e.g. kafka.example.com)'
     )
+    port: int = Field(..., description='Kafka broker port', ge=1, le=65535)
+
+
+class KafkaRESTProxyUsernamePassword(BaseModel):
+    """
+    Connect over the Kafka REST Proxy HTTP API (Confluent REST Proxy, Karapace) instead of the native broker protocol.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    auth_mode: Literal['REST']
+    host: str = Field(
+        ...,
+        description='Kafka REST proxy host name or IP (e.g. kafka-rest.example.com)',
+    )
+    port: int = Field(..., description='Kafka REST proxy port', ge=1, le=65535)
 
 
 class NoAuthentication1(BaseModel):
@@ -3429,9 +3451,9 @@ class NoAuthentication1(BaseModel):
     )
 
 
-class SASL1(BaseModel):
+class SASLUsernamePassword1(BaseModel):
     """
-    SASL username/password credentials.
+    SASL credentials. Set the SASL mechanism under connection options (PLAIN by default).
     """
 
     model_config = ConfigDict(
@@ -3439,24 +3461,42 @@ class SASL1(BaseModel):
     )
     sasl_username: str = Field(..., description='SASL username')
     sasl_password: str = Field(..., description='SASL password')
+    ssl_ca: str | None = Field(
+        None,
+        description="PEM-encoded CA certificate used to verify the broker's TLS certificate. Required for private or self-signed CAs; leave empty to trust the system CA bundle.",
+    )
 
 
 class ClientCertificateMTLS1(BaseModel):
     """
-    mTLS client certificate credentials.
+    mTLS credentials: the client certificate and its private key, plus the CA certificate when the broker uses a private CA.
     """
 
     model_config = ConfigDict(
         extra='forbid',
     )
-    ssl_certfile: str = Field(
-        ..., description='PEM-encoded client certificate (access cert)'
-    )
+    ssl_certfile: str = Field(..., description='PEM-encoded client access certificate')
     ssl_keyfile: str = Field(
-        ..., description='PEM-encoded client private key (access key)'
+        ..., description='PEM-encoded client access key (private key)'
+    )
+    ssl_ca: str | None = Field(
+        None,
+        description="PEM-encoded CA certificate used to verify the broker's TLS certificate. Required for private or self-signed CAs; leave empty to trust the system CA bundle.",
     )
 
 
+class KafkaRESTProxyUsernamePassword1(BaseModel):
+    """
+    HTTP basic auth credentials for the REST proxy.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    username: str = Field(..., description='REST proxy username')
+    password: str = Field(..., description='REST proxy password')
+
+
 class KafkaOptionalConnection(BaseModel):
     """
     Kafka client connection and security options.
@@ -3465,11 +3505,11 @@ class KafkaOptionalConnection(BaseModel):
     model_config = ConfigDict(
         extra='forbid',
     )
-    security_protocol: KafkaSecurityProtocol | None = 'PLAINTEXT'
+    security_protocol: KafkaSecurityProtocol | None = None
     sasl_mechanism: KafkaSaslMechanism | None = 'PLAIN'
-    ssl_ca: str | None = Field(
-        None,
-        description="PEM-encoded CA certificate for TLS verification (optional for client-certificate auth; validates the broker's certificate)",
+    rest_use_tls: bool | None = Field(
+        True,
+        description='REST proxy only: call the proxy over HTTPS. Turn off for a plain-HTTP proxy (e.g. a local Karapace).',
     )
     request_timeout_ms: int | None = Field(
         30000, description='Client request timeout in milliseconds', ge=1000
@@ -3508,12 +3548,19 @@ class KafkaInput(CoreInput):
     type: Literal['KAFKA'] | None = Field(
         None, description='Type of the asset or source'
     )
-    required: NoAuthentication | SASL | ClientCertificateMTLS = Field(
-        ..., title='KafkaRequired'
-    )
-    masked: NoAuthentication1 | SASL1 | ClientCertificateMTLS1 | None = Field(
-        None, title='KafkaMasked'
-    )
+    required: (
+        NoAuthentication
+        | SASLUsernamePassword
+        | ClientCertificateMTLS
+        | KafkaRESTProxyUsernamePassword
+    ) = Field(..., title='KafkaRequired')
+    masked: (
+        NoAuthentication1
+        | SASLUsernamePassword1
+        | ClientCertificateMTLS1
+        | KafkaRESTProxyUsernamePassword1
+        | None
+    ) = Field(None, title='KafkaMasked')
     optional: KafkaOptional | None = None
     detectors: list[Detector] | None = Field(
         None, description='Detectors to run on ingested content'
@@ -4174,6 +4221,193 @@ class GoogleWorkspaceInput(CoreInput):
     )
 
 
+class DropboxRequiredAccessToken(BaseModel):
+    """
+    Access token issued from the Dropbox App Console. Simplest to set up; short-lived tokens expire after 4 hours.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    auth_method: Literal['access_token']
+
+
+class DropboxRequiredOAuth(BaseModel):
+    """
+    OAuth 2.0 app credentials with a refresh token. Recommended for unattended scans — the client mints a fresh access token on every run.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    auth_method: Literal['oauth']
+    app_key: str = Field(
+        ...,
+        description='Dropbox app key (client ID) of the OAuth app that issued the refresh token',
+    )
+
+
+class DropboxMaskedAccessToken(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    access_token: str = Field(
+        ..., description='Dropbox OAuth 2.0 access token (sl.… for short-lived tokens)'
+    )
+
+
+class DropboxMaskedOAuth(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    app_secret: str = Field(
+        ...,
+        description='Dropbox app secret of the OAuth app that issued the refresh token',
+    )
+    refresh_token: str = Field(
+        ...,
+        description='Dropbox OAuth 2.0 refresh token (obtained with token_access_type=offline)',
+    )
+
+
+class DropboxOptionalScope(BaseModel):
+    """
+    Folder selection and content filtering.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    folder_path: str | None = Field(
+        None,
+        description='Folder to scan, for example /Finance/Exports. Leave empty to scan the whole account root.',
+    )
+    recursive: bool | None = Field(
+        True, description='Descend into subfolders of the selected folder'
+    )
+    include_extensions: list[str] | None = Field(
+        None,
+        description='Optional extension allowlist (for example, .pdf, .csv, .docx)',
+    )
+    exclude_extensions: list[str] | None = Field(
+        None, description='Optional extension denylist'
+    )
+    include_mounted_folders: bool | None = Field(
+        True, description='Include shared folders mounted into the account'
+    )
+    include_empty_objects: bool | None = Field(
+        False, description='Include zero-byte files in extraction results'
+    )
+    include_object_metadata: bool | None = Field(
+        True,
+        description='Attach Dropbox metadata (content hash, size, revision, timestamps) to asset checksums',
+    )
+    include_content_preview: bool | None = Field(
+        True,
+        description='Download file bytes to infer MIME and extract detector-ready text previews',
+    )
+    export_non_downloadable: bool | None = Field(
+        True,
+        description='Export files that have no direct download (Dropbox Paper docs) to Markdown so their text can be scanned',
+    )
+
+
+class DropboxOptionalConnection(BaseModel):
+    """
+    Network and pagination controls for Dropbox API requests.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    max_entries_per_page: int | None = Field(
+        500, description='Maximum entries requested per list_folder call', ge=1, le=2000
+    )
+    max_object_bytes: int | None = Field(
+        5242880,
+        description='Maximum bytes downloaded per file for MIME detection and text extraction',
+        ge=1024,
+        le=52428800,
+    )
+    request_timeout_seconds: float | None = Field(
+        60,
+        description='Network timeout in seconds for list/download operations',
+        ge=1.0,
+        le=300.0,
+    )
+    max_retries: int | None = Field(
+        4,
+        description='Maximum retries on transient Dropbox errors (5xx and rate limits)',
+        ge=0,
+        le=10,
+    )
+
+
+class DropboxOptionalTeam(BaseModel):
+    """
+    Dropbox Business team scanning. The credential shape is the same as for an individual account, but the Dropbox app must be created with a team access type ('Team member file access' plus 'Team information'). Leave disabled for individual accounts.
+    """
+
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    enabled: bool | None = Field(
+        False,
+        description='Scan the whole Dropbox Business team instead of a single account',
+    )
+    member_emails: list[str] | None = Field(
+        None, description='Only scan these team members (empty = every active member)'
+    )
+    include_member_folders: bool | None = Field(
+        True, description="Scan each selected member's own Dropbox folder"
+    )
+    include_team_space: bool | None = Field(
+        True, description='Scan the shared team folders in the team space'
+    )
+    include_suspended_members: bool | None = Field(
+        False, description='Also scan members whose account is suspended'
+    )
+    as_admin: bool | None = Field(
+        False,
+        description="Traverse team-owned folders with a team administrator's privileges instead of acting as each member. Reaches content a member cannot list themselves and does not mark files as viewed by them. Requires at least one team admin on the team.",
+    )
+
+
+class DropboxOptional(BaseModel):
+    model_config = ConfigDict(
+        extra='forbid',
+    )
+    scope: DropboxOptionalScope | None = None
+    team: DropboxOptionalTeam | None = None
+    connection: DropboxOptionalConnection | None = None
+
+
+class DropboxInput(CoreInput):
+    type: Literal['DROPBOX'] | None = Field(
+        None, description='Type of the asset or source'
+    )
+    required: DropboxRequiredAccessToken | DropboxRequiredOAuth = Field(
+        ..., title='DropboxRequired'
+    )
+    masked: DropboxMaskedAccessToken | DropboxMaskedOAuth = Field(
+        ..., title='DropboxMasked'
+    )
+    optional: DropboxOptional | None = None
+    detectors: list[Detector] | None = Field(
+        None, description='Detectors to run on ingested content'
+    )
+    custom_detectors: list[CustomDetectorSelection] | None = Field(
+        None,
+        description='Reusable custom detector IDs selected from the custom detector catalog.',
+    )
+    sampling: SamplingConfig
+    resources: ResourceOverrides | None = None
+    cleanup_removed_detector_findings: bool | None = Field(
+        True,
+        description='When enabled (default), findings produced by detectors that are no longer configured on this source (removed or disabled) are automatically resolved at the start of the next run, keeping the findings list in step with the current detector set.',
+    )
+
+
 class YouTubeInput(CoreInput):
     type: Literal['YOUTUBE'] | None = Field(
         None, description='Type of the asset or source'
@@ -4291,6 +4525,7 @@ class SourceInput(
         | MeilisearchInput
         | Microsoft365Input
         | GoogleWorkspaceInput
+        | DropboxInput
     ]
 ):
     root: (
@@ -4327,6 +4562,7 @@ class SourceInput(
         | MeilisearchInput
         | Microsoft365Input
         | GoogleWorkspaceInput
+        | DropboxInput
     ) = Field(
         ...,
         description='Merged configuration schema with all source types and common definitions',

classifyre_cli-0.4.70/src/sources/dropbox/__init__.py

@@ -0,0 +1,3 @@
+from .source import DropboxSource
+
+__all__ = ["DropboxSource"]