recce-nightly 0.62.0.20250417__py3-none-any.whl → 1.30.0.20251221__py3-none-any.whl

recce/util/perf_tracking.py

@@ -0,0 +1,85 @@
+import time
+from dataclasses import dataclass
+
+
+@dataclass
+class LineagePerfTracker:
+    lineage_start = None
+    lineage_elapsed = None
+    column_lineage_start = None
+    column_lineage_elapsed = None
+
+    total_nodes = None
+    init_nodes = None
+    cll_nodes = 0
+    change_analysis_nodes = 0
+    anchor_nodes = None
+
+    params = None
+
+    def start_lineage(self):
+        self.lineage_start = time.perf_counter_ns()
+
+    def end_lineage(self):
+        if self.lineage_start is None:
+            return
+        self.lineage_elapsed = (time.perf_counter_ns() - self.lineage_start) / 1000000
+
+    def start_column_lineage(self):
+        self.column_lineage_start = time.perf_counter_ns()
+
+    def end_column_lineage(self):
+        if self.column_lineage_start is None:
+            return
+        self.column_lineage_elapsed = (time.perf_counter_ns() - self.column_lineage_start) / 1000000
+
+    def set_total_nodes(self, total_nodes):
+        self.total_nodes = total_nodes
+
+    def set_init_nodes(self, init_nodes):
+        self.init_nodes = init_nodes
+
+    def set_anchor_nodes(self, anchor_nodes):
+        self.anchor_nodes = anchor_nodes
+
+    def increment_cll_nodes(self):
+        self.cll_nodes += 1
+
+    def increment_change_analysis_nodes(self):
+        self.change_analysis_nodes += 1
+
+    def set_params(self, has_node, has_column, change_analysis, no_cll, no_upstream, no_downstream):
+        self.params = {
+            "has_node": has_node,
+            "has_column": has_column,
+            "change_analysis": change_analysis,
+            "no_cll": no_cll,
+            "no_upstream": no_upstream,
+            "no_downstream": no_downstream,
+        }
+
+    def to_dict(self):
+        return {
+            "lineage_elapsed_ms": self.lineage_elapsed,
+            "column_lineage_elapsed_ms": self.column_lineage_elapsed,
+            "total_nodes": self.total_nodes,
+            "init_nodes": self.init_nodes,
+            "cll_nodes": self.cll_nodes,
+            "change_analysis_nodes": self.change_analysis_nodes,
+            "anchor_nodes": self.anchor_nodes,
+            "params": self.params,
+        }
+
+    def reset(self):
+        self.lineage_start = None
+        self.lineage_elapsed = None
+        self.column_lineage_start = None
+        self.column_lineage_elapsed = None
+
+        self.total_nodes = None
+        self.init_nodes = None
+        self.change_analysis_nodes = 0
+        self.cll_nodes = 0
+        self.anchor_nodes = 0
+
+        self.params = None

recce/util/recce_cloud.py

@@ -1,20 +1,30 @@
 import json
 import logging
 import os
-from typing import Dict, IO
+import typing
+from typing import IO, Dict
 
 import requests
 
+from recce import get_version
+from recce.event import get_user_id, is_anonymous_tracking
 from recce.pull_request import PullRequestInfo
 
-RECCE_CLOUD_API_HOST = os.environ.get('RECCE_CLOUD_API_HOST', 'https://cloud.datarecce.io')
+if typing.TYPE_CHECKING:
+    from recce.util.cloud import ChecksCloud
 
-logger = logging.getLogger('uvicorn')
+RECCE_CLOUD_API_HOST = os.environ.get("RECCE_CLOUD_API_HOST", "https://cloud.datarecce.io")
+RECCE_CLOUD_BASE_URL = os.environ.get("RECCE_CLOUD_BASE_URL", RECCE_CLOUD_API_HOST)
+
+DOCKER_INTERNAL_URL_PREFIX = "http://host.docker.internal"
+LOCALHOST_URL_PREFIX = "http://localhost"
+
+logger = logging.getLogger("uvicorn")
 
 
 class PresignedUrlMethod:
-    UPLOAD = 'upload'
-    DOWNLOAD = 'download'
+    UPLOAD = "upload"
+    DOWNLOAD = "download"
 
 
 class RecceCloudException(Exception):
@@ -23,7 +33,7 @@ class RecceCloudException(Exception):
         self.status_code = status_code
 
         try:
-            reason = json.loads(reason).get('detail', '')
+            reason = json.loads(reason).get("detail", "")
         except json.JSONDecodeError:
             pass
         self.reason = reason
@@ -31,147 +41,412 @@ class RecceCloudException(Exception):
 
 class RecceCloud:
     def __init__(self, token: str):
+        if token is None:
+            raise ValueError("Token cannot be None.")
         self.token = token
-        self.base_url = f'{RECCE_CLOUD_API_HOST}/api/v1'
+        self.token_type = "github_token" if token.startswith(("ghp_", "gho_", "ghu_", "ghs_", "ghr_")) else "api_token"
+        self.base_url = f"{RECCE_CLOUD_API_HOST}/api/v1"
+        self.base_url_v2 = f"{RECCE_CLOUD_API_HOST}/api/v2"
+
+        # Initialize modular clients
+        self._checks_client = None
+
+    @property
+    def checks(self) -> "ChecksCloud":
+        """
+        Get the checks client for check operations.
+
+        Returns:
+            ChecksCloud instance for check operations
+
+        Example:
+            >>> cloud = RecceCloud(token="your-token")
+            >>> checks = cloud.checks.list_checks("org", "proj", "sess")
+        """
+        if self._checks_client is None:
+            from recce.util.cloud import ChecksCloud
 
-    def _request(self, method, url, **kwargs):
+            self._checks_client = ChecksCloud(self.token)
+        return self._checks_client
+
+    def _request(self, method, url, headers: Dict = None, **kwargs):
         headers = {
-            'Authorization': f'Bearer {self.token}'
+            **(headers or {}),
+            "Authorization": f"Bearer {self.token}",
         }
         return requests.request(method, url, headers=headers, **kwargs)
 
-    def get_presigned_url(self,
-                          method: PresignedUrlMethod,
-                          repository: str,
-                          artifact_name: str,
-                          metadata: dict = None,
-                          pr_id: int = None,
-                          branch: str = None) -> str:
+    def verify_token(self) -> bool:
+        if self.token_type == "github_token":
+            return True
+        # Verify the Recce Cloud API token
+        api_url = f"{self.base_url}/verify-token"
+        try:
+            headers: Dict = None
+            if is_anonymous_tracking():
+                headers = {
+                    "X-Recce-Oss-User-Id": get_user_id(),
+                    "X-Recce-Oss-Version": get_version(),
+                }
+            response = self._request("GET", api_url, headers=headers)
+            if response.status_code == 200:
+                return True
+        except Exception:
+            pass
+        return False
+
+    def get_presigned_url_by_github_repo(
+        self,
+        method: PresignedUrlMethod,
+        repository: str,
+        artifact_name: str,
+        metadata: dict = None,
+        pr_id: int = None,
+        branch: str = None,
+    ) -> str:
         response = self._fetch_presigned_url(method, repository, artifact_name, metadata, pr_id, branch)
-        return response.get('presigned_url')
+        return response.get("presigned_url")
 
-    def get_download_presigned_url_with_tags(self,
-                                             repository: str,
-                                             artifact_name: str,
-                                             branch: str = None) -> (str, dict):
+    def _replace_localhost_with_docker_internal(self, url: str) -> str:
+        if url is None:
+            return None
+        if (
+            os.environ.get("RECCE_SHARE_INSTANCE_ENV") == "docker"
+            or os.environ.get("RECCE_TASK_INSTANCE_ENV") == "docker"
+            or os.environ.get("RECCE_INSTANCE_ENV") == "docker"
+        ):
+            # For local development, convert the presigned URL from localhost to host.docker.internal
+            if url.startswith(LOCALHOST_URL_PREFIX):
+                return url.replace(LOCALHOST_URL_PREFIX, DOCKER_INTERNAL_URL_PREFIX)
+        return url
+
+    def get_presigned_url_by_share_id(
+        self,
+        method: PresignedUrlMethod,
+        share_id: str,
+        metadata: dict = None,
+    ) -> str:
+        response = self._fetch_presigned_url_by_share_id(method, share_id, metadata=metadata)
+        presigned_url = response.get("presigned_url")
+        if not presigned_url:
+            raise RecceCloudException(
+                message="Failed to get presigned URL from Recce Cloud.",
+                reason="No presigned URL returned from the server.",
+                status_code=404,
+            )
+        presigned_url = self._replace_localhost_with_docker_internal(presigned_url)
+        return presigned_url
+
+    def get_download_presigned_url_by_github_repo_with_tags(
+        self, repository: str, artifact_name: str, branch: str = None
+    ) -> (str, dict):
         response = self._fetch_presigned_url(PresignedUrlMethod.DOWNLOAD, repository, artifact_name, branch=branch)
-        return response.get('presigned_url'), response.get('tags', {})
-
-    def _fetch_presigned_url(self,
-                             method: PresignedUrlMethod,
-                             repository: str,
-                             artifact_name: str,
-                             metadata: dict = None,
-                             pr_id: int = None,
-                             branch: str = None) -> str:
+        return response.get("presigned_url"), response.get("tags", {})
+
+    def _fetch_presigned_url(
+        self,
+        method: PresignedUrlMethod,
+        repository: str,
+        artifact_name: str,
+        metadata: dict = None,
+        pr_id: int = None,
+        branch: str = None,
+    ) -> str:
         if pr_id is not None:
-            api_url = f'{self.base_url}/{repository}/pulls/{pr_id}/artifacts/{method}?artifact_name={artifact_name}&enable_ssec=true'
+            api_url = f"{self.base_url}/{repository}/pulls/{pr_id}/artifacts/{method}?artifact_name={artifact_name}&enable_ssec=true"
         elif branch is not None:
-            api_url = f'{self.base_url}/{repository}/commits/{branch}/artifacts/{method}?artifact_name={artifact_name}&enable_ssec=true'
+            api_url = f"{self.base_url}/{repository}/commits/{branch}/artifacts/{method}?artifact_name={artifact_name}&enable_ssec=true"
         else:
-            raise ValueError('Either pr_id or sha must be provided.')
-        response = self._request('POST', api_url, json=metadata)
+            raise ValueError("Either pr_id or sha must be provided.")
+        response = self._request("POST", api_url, json=metadata)
         if response.status_code != 200:
             raise RecceCloudException(
-                message='Failed to {method} artifact {preposition} Recce Cloud.'.format(
-                    method=method,
-                    preposition='from' if method == PresignedUrlMethod.DOWNLOAD else 'to'
+                message="Failed to {method} artifact {preposition} Recce Cloud.".format(
+                    method=method, preposition="from" if method == PresignedUrlMethod.DOWNLOAD else "to"
                 ),
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
+            )
+        return response.json()
+
+    def _fetch_presigned_url_by_share_id(
+        self,
+        method: PresignedUrlMethod,
+        share_id: str,
+        metadata: dict = None,
+    ):
+        api_url = f"{self.base_url}/shares/{share_id}/presigned/{method}"
+        data = None
+        # Only provide metadata for upload requests
+        if method == PresignedUrlMethod.UPLOAD:
+            # Covert metadata values to strings to ensure JSON serializability
+            data = {"metadata": {key: str(value) for key, value in metadata.items()}} if metadata else None
+        response = self._request(
+            "POST",
+            api_url,
+            json=data,
+        )
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to {method} artifact {preposition} Recce Cloud.".format(
+                    method=method, preposition="from" if method == PresignedUrlMethod.DOWNLOAD else "to"
+                ),
+                reason=response.text,
+                status_code=response.status_code,
             )
         return response.json()
 
     def get_artifact_metadata(self, pr_info: PullRequestInfo) -> dict:
-        api_url = f'{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/metadata'
-        response = self._request('GET', api_url)
+        api_url = f"{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/metadata"
+        response = self._request("GET", api_url)
         if response.status_code == 204:
             return None
         if response.status_code != 200:
             raise RecceCloudException(
-                message='Failed to get artifact metadata from Recce Cloud.',
+                message="Failed to get artifact metadata from Recce Cloud.",
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
             )
         return response.json()
 
-    def purge_artifacts(self, pr_info: PullRequestInfo):
-        api_url = f'{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/artifacts'
-        response = self._request('DELETE', api_url)
+    def purge_artifacts(self, repository: str, pr_id: int = None, branch: str = None):
+        if pr_id is not None:
+            api_url = f"{self.base_url}/{repository}/pulls/{pr_id}/artifacts"
+            error_message = "Failed to purge artifacts from Recce Cloud."
+        elif branch is not None:
+            api_url = f"{self.base_url}/{repository}/commits/{branch}/artifacts"
+            error_message = "Failed to delete artifacts from Recce Cloud."
+        else:
+            raise ValueError(
+                "Please either run this command from within a pull request context "
+                "or specify a branch using the --branch option."
+            )
+        response = self._request("DELETE", api_url)
         if response.status_code != 204:
             raise RecceCloudException(
-                message='Failed to purge artifacts from Recce Cloud.',
+                message=error_message,
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
             )
 
     def check_artifacts_exists(self, pr_info: PullRequestInfo) -> bool:
-        api_url = f'{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/metadata'
-        response = self._request('GET', api_url)
+        api_url = f"{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/metadata"
+        response = self._request("GET", api_url)
         if response.status_code == 200:
             return True
         elif response.status_code == 204:
             return False
         else:
             raise RecceCloudException(
-                message='Failed to check if artifacts exist in Recce Cloud.',
+                message="Failed to check if artifacts exist in Recce Cloud.",
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
             )
 
     def share_state(self, file_name: str, file_io: IO):
-        api_url = f'{self.base_url}/recce-state/upload'
-        files = {'file': (file_name, file_io, 'application/json')}
-        response = self._request('POST', api_url, files=files)
+        api_url = f"{self.base_url}/recce-state/upload"
+        files = {"file": (file_name, file_io, "application/json")}
+        response = self._request("POST", api_url, files=files)
         if response.status_code == 403:
-            return {'status': 'error', 'message': response.json().get('detail')}
+            return {"status": "error", "message": response.json().get("detail")}
         if response.status_code != 200:
             raise RecceCloudException(
-                message='Failed to share Recce state.',
-                reason=response.text,
-                status_code=response.status_code
+                message="Failed to share Recce state.", reason=response.text, status_code=response.status_code
             )
         return response.json()
 
     def update_github_pull_request_check(self, pr_info: PullRequestInfo, metadata: dict = None):
-        api_url = f'{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/github/checks'
+        api_url = f"{self.base_url}/{pr_info.repository}/pulls/{pr_info.id}/github/checks"
         try:
-            self._request('POST', api_url, json=metadata)
+            self._request("POST", api_url, json=metadata)
         except Exception as e:
             # We don't care the response of this request, so we don't need to raise any exception.
-            logger.debug(f'Failed to update the GitHub PR check. Reason: {str(e)}')
+            logger.debug(f"Failed to update the GitHub PR check. Reason: {str(e)}")
 
     def get_user_info(self) -> Dict:
-        api_url = f'{self.base_url}/users'
-        response = self._request('GET', api_url)
+        api_url = f"{self.base_url}/users"
+        response = self._request("GET", api_url)
         if response.status_code != 200:
             raise RecceCloudException(
-                message='Failed to get user info from Recce Cloud.',
+                message="Failed to get user info from Recce Cloud.",
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
             )
-        return response.json().get('user')
+        return response.json().get("user")
 
     def set_onboarding_state(self, state: str):
-        api_url = f'{self.base_url}/users/onboarding-state'
-        response = self._request('PUT', api_url, json={'state': state})
+        api_url = f"{self.base_url}/users/onboarding-state"
+        try:
+            response = self._request("PUT", api_url, json={"state": state})
+            response.raise_for_status()
+        except requests.exceptions.HTTPError as e:
+            # Don't Raise an exception if setting onboarding_state fails
+            logger.warning(f"Failed to set Onboarding State in Recce Cloud. Reason: {str(e)}")
+        return
+
+    def get_session(self, session_id: str):
+        api_url = f"{self.base_url_v2}/sessions/{session_id}"
+        response = self._request("GET", api_url)
+        if response.status_code == 403:
+            return {"status": "error", "message": response.json().get("detail")}
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to get session from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        if data["success"] is not True:
+            raise RecceCloudException(
+                message="Failed to get session from Recce Cloud.",
+                reason=data.get("message", "Unknown error"),
+                status_code=response.status_code,
+            )
+        return data["session"]
+
+    def update_session(self, org_id: str, project_id: str, session_id: str, adapter_type: str):
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/sessions/{session_id}"
+        data = {"adapter_type": adapter_type}
+        response = self._request("PATCH", api_url, json=data)
+        if response.status_code == 403:
+            return {"status": "error", "message": response.json().get("detail")}
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to update session in Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        return response.json()
+
+    def get_download_urls_by_session_id(self, org_id: str, project_id: str, session_id: str) -> dict[str, str]:
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/sessions/{session_id}/download-url"
+        response = self._request("GET", api_url)
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to download session from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        if data["presigned_urls"] is None:
+            raise RecceCloudException(
+                message="No presigned URLs returned from the server.",
+                reason="",
+                status_code=404,
+            )
+
+        presigned_urls = data["presigned_urls"]
+        for key, url in presigned_urls.items():
+            presigned_urls[key] = self._replace_localhost_with_docker_internal(url)
+        return presigned_urls
+
+    def get_base_session_download_urls(self, org_id: str, project_id: str) -> dict[str, str]:
+        """Get download URLs for the base session of a project."""
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/base-session/download-url"
+        response = self._request("GET", api_url)
         if response.status_code != 200:
             raise RecceCloudException(
-                message='Failed to update onboarding state in Recce Cloud.',
+                message="Failed to download base session from Recce Cloud.",
                 reason=response.text,
-                status_code=response.status_code
+                status_code=response.status_code,
+            )
+        data = response.json()
+        if data["presigned_urls"] is None:
+            raise RecceCloudException(
+                message="No presigned URLs returned from the server.",
+                reason="",
+                status_code=404,
             )
 
+        presigned_urls = data["presigned_urls"]
+        for key, url in presigned_urls.items():
+            presigned_urls[key] = self._replace_localhost_with_docker_internal(url)
+        return presigned_urls
+
+    def get_upload_urls_by_session_id(self, org_id: str, project_id: str, session_id: str) -> dict[str, str]:
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/sessions/{session_id}/upload-url"
+        response = self._request("GET", api_url)
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to get upload URLs for session from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        if data["presigned_urls"] is None:
+            raise RecceCloudException(
+                message="No presigned URLs returned from the server.",
+                reason="",
+                status_code=404,
+            )
+
+        presigned_urls = data["presigned_urls"]
+        for key, url in presigned_urls.items():
+            presigned_urls[key] = self._replace_localhost_with_docker_internal(url)
+        return presigned_urls
+
+    def post_recce_state_uploaded_by_session_id(self, org_id: str, project_id: str, session_id: str):
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/sessions/{session_id}/recce-state-uploaded"
+        response = self._request("POST", api_url)
+        if response.status_code != 204:
+            raise RecceCloudException(
+                message="Failed to notify state uploaded for session in Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+
+    def list_organizations(self) -> list:
+        """List all organizations the user has access to."""
+        api_url = f"{self.base_url_v2}/organizations"
+        response = self._request("GET", api_url)
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to list organizations from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        return data.get("organizations", [])
+
+    def list_projects(self, org_id: str) -> list:
+        """List all projects in an organization."""
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects"
+        response = self._request("GET", api_url)
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to list projects from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        return data.get("projects", [])
+
+    def list_sessions(self, org_id: str, project_id: str) -> list:
+        """List all sessions in a project."""
+        api_url = f"{self.base_url_v2}/organizations/{org_id}/projects/{project_id}/sessions"
+        response = self._request("GET", api_url)
+        if response.status_code != 200:
+            raise RecceCloudException(
+                message="Failed to list sessions from Recce Cloud.",
+                reason=response.text,
+                status_code=response.status_code,
+            )
+        data = response.json()
+        return data.get("sessions", [])
+
 
 def get_recce_cloud_onboarding_state(token: str) -> str:
+    if token and token.startswith("rct-"):
+        return "undefined"
+
     try:
         recce_cloud = RecceCloud(token)
         user_info = recce_cloud.get_user_info()
         if user_info:
-            return user_info.get('onboarding_state')
+            return user_info.get("onboarding_state")
     except Exception as e:
         logger.debug(str(e))
-    return 'undefined'
+    return "undefined"
 
 
 def set_recce_cloud_onboarding_state(token: str, new_state: str):

recce/util/singleton.py

@@ -1,9 +1,9 @@
 class SingletonMeta(type):
     """
-        The Singleton class can be implemented in different ways in Python. Some
-        possible methods include: base class, decorator, metaclass. We will use the
-        metaclass because it is best suited for this purpose.
-        """
+    The Singleton class can be implemented in different ways in Python. Some
+    possible methods include: base class, decorator, metaclass. We will use the
+    metaclass because it is best suited for this purpose.
+    """
 
     _instances = {}