pangea-sdk 3.8.0b1__py3-none-any.whl → 5.3.0__py3-none-any.whl

pangea/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "3.8.0beta1"
+__version__ = "5.3.0"
 
 from pangea.asyncio.request import PangeaRequestAsync
 from pangea.config import PangeaConfig

pangea/asyncio/file_uploader.py

@@ -28,7 +28,7 @@ class FileUploaderAsync:
     ) -> None:
         if transfer_method == TransferMethod.PUT_URL:
             files = [("file", ("filename", file, "application/octet-stream"))]
-            await self._request.put_presigned_url(url=url, files=files)  # type: ignore[arg-type]
+            await self._request.put_presigned_url(url=url, files=files)
         elif transfer_method == TransferMethod.POST_URL:
             files = [("file", ("filename", file, "application/octet-stream"))]
             await self._request.post_presigned_url(url=url, data=file_details, files=files)  # type: ignore[arg-type]

pangea/asyncio/request.py

@@ -1,20 +1,24 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+from __future__ import annotations
 
 import asyncio
 import json
-import os
 import time
-from typing import Dict, List, Optional, Tuple, Type, Union
+from typing import Any, Dict, List, Optional, Sequence, Tuple, Type, Union
 
 import aiohttp
 from aiohttp import FormData
+from pydantic import BaseModel
+from typing_extensions import Any, TypeVar
 
 import pangea.exceptions as pe
 from pangea.request import MultipartResponse, PangeaRequestBase
 from pangea.response import AttachedFile, PangeaResponse, PangeaResponseResult, ResponseStatus, TransferMethod
 from pangea.utils import default_encoder
 
+TResult = TypeVar("TResult", bound=PangeaResponseResult)
+
 
 class PangeaRequestAsync(PangeaRequestBase):
     """An object that makes direct calls to Pangea Service APIs.
@@ -28,12 +32,12 @@ class PangeaRequestAsync(PangeaRequestBase):
     async def post(
         self,
         endpoint: str,
-        result_class: Type[PangeaResponseResult],
-        data: Union[str, Dict] = {},
-        files: List[Tuple] = [],
+        result_class: Type[TResult],
+        data: str | BaseModel | dict[str, Any] | None = None,
+        files: Optional[List[Tuple]] = None,
         poll_result: bool = True,
         url: Optional[str] = None,
-    ) -> PangeaResponse:
+    ) -> PangeaResponse[TResult]:
         """Makes the POST call to a Pangea Service endpoint.
 
         Args:
@@ -44,6 +48,13 @@ class PangeaRequestAsync(PangeaRequestBase):
             PangeaResponse which contains the response in its entirety and
                various properties to retrieve individual fields
         """
+
+        if isinstance(data, BaseModel):
+            data = data.model_dump(exclude_none=True)
+
+        if data is None:
+            data = {}
+
         if url is None:
             url = self._url(endpoint)
 
@@ -91,9 +102,7 @@ class PangeaRequestAsync(PangeaRequestBase):
 
         return self._check_response(pangea_response)
 
-    async def get(
-        self, path: str, result_class: Type[PangeaResponseResult], check_response: bool = True
-    ) -> PangeaResponse[Type[PangeaResponseResult]]:
+    async def get(self, path: str, result_class: Type[TResult], check_response: bool = True) -> PangeaResponse[TResult]:
         """Makes the GET call to a Pangea Service endpoint.
 
         Args:
@@ -110,7 +119,7 @@ class PangeaRequestAsync(PangeaRequestBase):
 
         async with self.session.get(url, headers=self._headers()) as requests_response:
             await self._check_http_errors(requests_response)
-            pangea_response = PangeaResponse(  # type: ignore[var-annotated]
+            pangea_response = PangeaResponse(
                 requests_response, result_class=result_class, json=await requests_response.json()
             )
 
@@ -131,11 +140,11 @@ class PangeaRequestAsync(PangeaRequestBase):
             raise pe.ServiceTemporarilyUnavailable(await resp.json())
 
     async def poll_result_by_id(
-        self, request_id: str, result_class: Union[Type[PangeaResponseResult], Type[dict]], check_response: bool = True
-    ):
+        self, request_id: str, result_class: Type[TResult], check_response: bool = True
+    ) -> PangeaResponse[TResult]:
         path = self._get_poll_path(request_id)
         self.logger.debug(json.dumps({"service": self.service, "action": "poll_result_once", "url": path}))
-        return await self.get(path, result_class, check_response=check_response)  # type: ignore[arg-type]
+        return await self.get(path, result_class, check_response=check_response)
 
     async def poll_result_once(self, response: PangeaResponse, check_response: bool = True):
         request_id = response.request_id
@@ -160,7 +169,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         if resp.status < 200 or resp.status >= 300:
             raise pe.PresignedUploadError(f"presigned POST failure: {resp.status}", await resp.text())
 
-    async def put_presigned_url(self, url: str, files: List[Tuple]):
+    async def put_presigned_url(self, url: str, files: Sequence[Tuple]):
         # Send put request with file as body
         resp = await self._http_put(url=url, files=files)
         self.logger.debug(
@@ -173,7 +182,20 @@ class PangeaRequestAsync(PangeaRequestBase):
         if resp.status < 200 or resp.status >= 300:
             raise pe.PresignedUploadError(f"presigned PUT failure: {resp.status}", await resp.text())
 
-    async def download_file(self, url: str, filename: Optional[str] = None) -> AttachedFile:
+    async def download_file(self, url: str, filename: str | None = None) -> AttachedFile:
+        """
+        Download file
+
+        Download a file from the specified URL and save it with the given
+        filename.
+
+        Args:
+            url: URL of the file to download
+            filename: Name to save the downloaded file as. If not provided, the
+              filename will be determined from the Content-Disposition header or
+              the URL.
+        """
+
         self.logger.debug(
             json.dumps(
                 {
@@ -209,16 +231,16 @@ class PangeaRequestAsync(PangeaRequestBase):
                 )
 
                 return AttachedFile(filename=filename, file=await response.read(), content_type=content_type)
-            else:
-                raise pe.DownloadFileError(f"Failed to download file. Status: {response.status}", await response.text())
+            raise pe.DownloadFileError(f"Failed to download file. Status: {response.status}", await response.text())
 
-    async def _get_pangea_json(self, reader: aiohttp.MultipartReader) -> Optional[Dict]:
+    async def _get_pangea_json(self, reader: aiohttp.multipart.MultipartResponseWrapper) -> Optional[Dict[str, Any]]:
         # Iterate through parts
         async for part in reader:
-            return await part.json()
+            if isinstance(part, aiohttp.BodyPartReader):
+                return await part.json()
         return None
 
-    async def _get_attached_files(self, reader: aiohttp.MultipartReader) -> List[AttachedFile]:
+    async def _get_attached_files(self, reader: aiohttp.multipart.MultipartResponseWrapper) -> List[AttachedFile]:
         files = []
         i = 0
 
@@ -229,7 +251,7 @@ class PangeaRequestAsync(PangeaRequestBase):
             if name is None:
                 name = f"default_file_name_{i}"
                 i += 1
-            files.append(AttachedFile(name, await part.read(), content_type))
+            files.append(AttachedFile(name, await part.read(), content_type))  # type: ignore[union-attr]
 
         return files
 
@@ -237,13 +259,12 @@ class PangeaRequestAsync(PangeaRequestBase):
         # Parse the multipart response
         multipart_reader = aiohttp.MultipartReader.from_response(resp)
 
-        pangea_json = await self._get_pangea_json(multipart_reader)  # type: ignore[arg-type]
+        pangea_json = await self._get_pangea_json(multipart_reader)
         self.logger.debug(
             json.dumps({"service": self.service, "action": "multipart response", "response": pangea_json})
         )
 
-        multipart_reader = multipart_reader.__aiter__()
-        attached_files = await self._get_attached_files(multipart_reader)  # type: ignore[arg-type]
+        attached_files = await self._get_attached_files(multipart_reader)
         return MultipartResponse(pangea_json, attached_files)  # type: ignore[arg-type]
 
     async def _http_post(
@@ -251,7 +272,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         url: str,
         headers: Dict = {},
         data: Union[str, Dict] = {},
-        files: List[Tuple] = [],
+        files: Optional[List[Tuple]] = [],
         presigned_url_post: bool = False,
     ) -> aiohttp.ClientResponse:
         if files:
@@ -276,7 +297,7 @@ class PangeaRequestAsync(PangeaRequestBase):
     async def _http_put(
         self,
         url: str,
-        files: List[Tuple],
+        files: Sequence[Tuple],
         headers: Dict = {},
     ) -> aiohttp.ClientResponse:
         self.logger.debug(
@@ -328,9 +349,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         # Receive 202
         return await self._poll_presigned_url(accepted_exception.response)
 
-    async def _poll_presigned_url(
-        self, response: PangeaResponse[Type[PangeaResponseResult]]
-    ) -> PangeaResponse[Type[PangeaResponseResult]]:
+    async def _poll_presigned_url(self, response: PangeaResponse[TResult]) -> PangeaResponse[TResult]:
         if response.http_status != 202:
             raise AttributeError("Response should be 202")
 
@@ -373,8 +392,7 @@ class PangeaRequestAsync(PangeaRequestBase):
 
         if loop_resp.accepted_result is not None and not loop_resp.accepted_result.has_upload_url:
             return loop_resp
-        else:
-            raise loop_exc
+        raise loop_exc
 
     async def _handle_queued_result(self, response: PangeaResponse) -> PangeaResponse:
         if self._queued_retry_enabled and response.http_status == 202:

pangea/asyncio/services/__init__.py

@@ -1,8 +1,10 @@
 from .audit import AuditAsync
 from .authn import AuthNAsync
+from .authz import AuthZAsync
 from .embargo import EmbargoAsync
 from .file_scan import FileScanAsync
 from .intel import DomainIntelAsync, FileIntelAsync, IpIntelAsync, UrlIntelAsync, UserIntelAsync
 from .redact import RedactAsync
+from .sanitize import SanitizeAsync
 from .share import ShareAsync
 from .vault import VaultAsync

pangea/asyncio/services/audit.py

@@ -1,11 +1,14 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+from __future__ import annotations
+
 import datetime
-from typing import Any, Dict, List, Optional, Union
+from typing import Any, Dict, Iterable, List, Optional, Sequence, Union
 
 import pangea.exceptions as pexc
 from pangea.asyncio.services.base import ServiceBaseAsync
-from pangea.response import PangeaResponse
+from pangea.config import PangeaConfig
+from pangea.response import PangeaResponse, PangeaResponseResult
 from pangea.services.audit.audit import AuditBase
 from pangea.services.audit.exceptions import AuditException
 from pangea.services.audit.models import (
@@ -13,6 +16,7 @@ from pangea.services.audit.models import (
     DownloadRequest,
     DownloadResult,
     Event,
+    ExportRequest,
     LogBulkResult,
     LogResult,
     PublishedRoot,
@@ -56,14 +60,33 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
 
     def __init__(
         self,
-        token,
-        config=None,
+        token: str,
+        config: PangeaConfig | None = None,
         private_key_file: str = "",
-        public_key_info: Dict[str, str] = {},
-        tenant_id: Optional[str] = None,
-        logger_name="pangea",
-        config_id: Optional[str] = None,
-    ):
+        public_key_info: dict[str, str] = {},
+        tenant_id: str | None = None,
+        logger_name: str = "pangea",
+        config_id: str | None = None,
+    ) -> None:
+        """
+        Audit client
+
+        Initializes a new Audit client.
+
+        Args:
+            token: Pangea API token.
+            config: Configuration.
+            private_key_file: Private key filepath.
+            public_key_info: Public key information.
+            tenant_id: Tenant ID.
+            logger_name: Logger name.
+            config_id: Configuration ID.
+
+        Examples:
+             config = PangeaConfig(domain="pangea_domain")
+             audit = AuditAsync(token="pangea_token", config=config)
+        """
+
         # FIXME: Temporary check to deprecate config_id from PangeaConfig.
         # Delete it when deprecate PangeaConfig.config_id
         if config_id and config is not None and config.config_id is not None:
@@ -151,14 +174,16 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         verbose: Optional[bool] = None,
     ) -> PangeaResponse[LogResult]:
         """
-        Log an entry
+        Log an event
 
         Create a log entry in the Secure Audit Log.
+
         Args:
             event (dict[str, Any]): event to be logged
             verify (bool, optional): True to verify logs consistency after response.
             sign_local (bool, optional): True to sign event with local key.
             verbose (bool, optional): True to get a more verbose response.
+
         Raises:
             AuditException: If an audit based api exception happens
             PangeaAPIException: If an API Error happens
@@ -169,18 +194,12 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
                 Available response fields can be found in our [API documentation](https://pangea.cloud/docs/api/audit#log-an-entry).
 
         Examples:
-            try:
-                log_response = audit.log({"message"="Hello world"}, verbose=False)
-                print(f"Response. Hash: {log_response.result.hash}")
-            except pe.PangeaAPIException as e:
-                print(f"Request Error: {e.response.summary}")
-                for err in e.errors:
-                    print(f"\\t{err.detail} \\n")
+            response = await audit.log_event({"message": "hello world"}, verbose=True)
         """
 
         input = self._get_log_request(event, sign_local=sign_local, verify=verify, verbose=verbose)
         response: PangeaResponse[LogResult] = await self.request.post(
-            "v1/log", LogResult, data=input.dict(exclude_none=True)
+            "v1/log", LogResult, data=input.model_dump(exclude_none=True)
         )
         if response.success and response.result is not None:
             self._process_log_result(response.result, verify=verify)
@@ -216,7 +235,7 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
 
         input = self._get_log_request(events, sign_local=sign_local, verify=False, verbose=verbose)
         response: PangeaResponse[LogBulkResult] = await self.request.post(
-            "v2/log", LogBulkResult, data=input.dict(exclude_none=True)
+            "v2/log", LogBulkResult, data=input.model_dump(exclude_none=True)
         )
         if response.success and response.result is not None:
             for result in response.result.results:
@@ -254,7 +273,7 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         input = self._get_log_request(events, sign_local=sign_local, verify=False, verbose=verbose)
         try:
             response: PangeaResponse[LogBulkResult] = await self.request.post(
-                "v2/log_async", LogBulkResult, data=input.dict(exclude_none=True), poll_result=False
+                "v2/log_async", LogBulkResult, data=input.model_dump(exclude_none=True), poll_result=False
             )
         except pexc.AcceptedRequestException as e:
             return e.response
@@ -272,10 +291,11 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         end: Optional[Union[datetime.datetime, str]] = None,
         limit: Optional[int] = None,
         max_results: Optional[int] = None,
-        search_restriction: Optional[dict] = None,
+        search_restriction: Optional[Dict[str, Sequence[str]]] = None,
         verbose: Optional[bool] = None,
         verify_consistency: bool = False,
         verify_events: bool = True,
+        return_context: Optional[bool] = None,
     ) -> PangeaResponse[SearchOutput]:
         """
         Search the log
@@ -301,10 +321,11 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             end (datetime, optional): An RFC-3339 formatted timestamp, or relative time adjustment from the current time.
             limit (int, optional): Optional[int] = None,
             max_results (int, optional): Maximum number of results to return.
-            search_restriction (dict, optional): A list of keys to restrict the search results to. Useful for partitioning data available to the query string.
+            search_restriction (Dict[str, Sequence[str]], optional): A list of keys to restrict the search results to. Useful for partitioning data available to the query string.
             verbose (bool, optional): If true, response include root and membership and consistency proofs.
             verify_consistency (bool): True to verify logs consistency
             verify_events (bool): True to verify hash events and signatures
+            return_context (bool): Return the context data needed to decrypt secure audit events that have been redacted with format preserving encryption.
 
         Raises:
             AuditException: If an audit based api exception happens
@@ -332,10 +353,11 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             max_results=max_results,
             search_restriction=search_restriction,
             verbose=verbose,
+            return_context=return_context,
         )
 
         response: PangeaResponse[SearchOutput] = await self.request.post(
-            "v1/search", SearchOutput, data=input.dict(exclude_none=True)
+            "v1/search", SearchOutput, data=input.model_dump(exclude_none=True)
         )
         if verify_consistency:
             await self.update_published_roots(response.result)  # type: ignore[arg-type]
@@ -347,8 +369,10 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         id: str,
         limit: Optional[int] = 20,
         offset: Optional[int] = 0,
+        assert_search_restriction: Optional[Dict[str, Sequence[str]]] = None,
         verify_consistency: bool = False,
         verify_events: bool = True,
+        return_context: Optional[bool] = None,
     ) -> PangeaResponse[SearchResultOutput]:
         """
         Results of a search
@@ -361,8 +385,11 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             id (string): the id of a search action, found in `response.result.id`
             limit (integer, optional): the maximum number of results to return, default is 20
             offset (integer, optional): the position of the first result to return, default is 0
+            assert_search_restriction (Dict[str, Sequence[str]], optional): Assert the requested search results were queried with the exact same search restrictions, to ensure the results comply to the expected restrictions.
             verify_consistency (bool): True to verify logs consistency
             verify_events (bool): True to verify hash events and signatures
+            return_context (bool): Return the context data needed to decrypt secure audit events that have been redacted with format preserving encryption.
+
         Raises:
             AuditException: If an audit based api exception happens
             PangeaAPIException: If an API Error happens
@@ -378,7 +405,8 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             result_res: PangeaResponse[SearchResultsOutput] = audit.results(
                 id=search_res.result.id,
                 limit=10,
-                offset=0)
+                offset=0,
+                assert_search_restriction={'source': ["monitor"]})
         """
 
         if limit <= 0:  # type: ignore[operator]
@@ -391,13 +419,116 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             id=id,
             limit=limit,
             offset=offset,
+            assert_search_restriction=assert_search_restriction,
+            return_context=return_context,
         )
-        response = await self.request.post("v1/results", SearchResultOutput, data=input.dict(exclude_none=True))
+        response = await self.request.post("v1/results", SearchResultOutput, data=input.model_dump(exclude_none=True))
         if verify_consistency and response.result is not None:
             await self.update_published_roots(response.result)
 
         return self.handle_results_response(response, verify_consistency, verify_events)
 
+    async def export(
+        self,
+        *,
+        format: DownloadFormat = DownloadFormat.CSV,
+        start: Optional[datetime.datetime] = None,
+        end: Optional[datetime.datetime] = None,
+        order: Optional[SearchOrder] = None,
+        order_by: Optional[str] = None,
+        verbose: bool = True,
+    ) -> PangeaResponse[PangeaResponseResult]:
+        """
+        Export from the audit log
+
+        Bulk export of data from the Secure Audit Log, with optional filtering.
+
+        OperationId: audit_post_v1_export
+
+        Args:
+            format: Format for the records.
+            start: The start of the time range to perform the search on.
+            end: The end of the time range to perform the search on. If omitted,
+                then all records up to the latest will be searched.
+            order: Specify the sort order of the response.
+            order_by: Name of column to sort the results by.
+            verbose: Whether or not to include the root hash of the tree and the
+                membership proof for each record.
+
+        Raises:
+            AuditException: If an audit based api exception happens
+            PangeaAPIException: If an API Error happens
+
+        Examples:
+            export_res = await audit.export(verbose=False)
+
+            # Export may take several dozens of minutes, so polling for the result
+            # should be done in a loop. That is omitted here for brevity's sake.
+            try:
+                await audit.poll_result(request_id=export_res.request_id)
+            except AcceptedRequestException:
+                # Retry later.
+
+            # Download the result when it's ready.
+            download_res = await audit.download_results(request_id=export_res.request_id)
+            download_res.result.dest_url
+            # => https://pangea-runtime.s3.amazonaws.com/audit/xxxxx/search_results_[...]
+        """
+        input = ExportRequest(
+            format=format,
+            start=start,
+            end=end,
+            order=order,
+            order_by=order_by,
+            verbose=verbose,
+        )
+        try:
+            return await self.request.post(
+                "v1/export", PangeaResponseResult, data=input.model_dump(exclude_none=True), poll_result=False
+            )
+        except pexc.AcceptedRequestException as e:
+            return e.response
+
+    async def log_stream(self, data: dict) -> PangeaResponse[PangeaResponseResult]:
+        """
+        Log streaming endpoint
+
+        This API allows 3rd party vendors (like Auth0) to stream events to this
+        endpoint where the structure of the payload varies across different
+        vendors.
+
+        OperationId: audit_post_v1_log_stream
+
+        Args:
+            data: Event data. The exact schema of this will vary by vendor.
+
+        Raises:
+            AuditException: If an audit based api exception happens
+            PangeaAPIException: If an API Error happens
+
+        Examples:
+            data = {
+                "logs": [
+                    {
+                        "log_id": "some log ID",
+                        "data": {
+                            "date": "2024-03-29T17:26:50.193Z",
+                            "type": "sapi",
+                            "description": "Create a log stream",
+                            "client_id": "some client ID",
+                            "ip": "127.0.0.1",
+                            "user_agent": "AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0",
+                            "user_id": "some user ID",
+                        },
+                    }
+                    # ...
+                ]
+            }
+
+            response = await audit.log_stream(data)
+        """
+        return await self.request.post("v1/log_stream", PangeaResponseResult, data=data)
+
     async def root(self, tree_size: Optional[int] = None) -> PangeaResponse[RootResult]:
         """
         Tamperproof verification
@@ -420,10 +551,14 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             response = audit.root(tree_size=7)
         """
         input = RootRequest(tree_size=tree_size)
-        return await self.request.post("v1/root", RootResult, data=input.dict(exclude_none=True))
+        return await self.request.post("v1/root", RootResult, data=input.model_dump(exclude_none=True))
 
     async def download_results(
-        self, result_id: str, format: Optional[DownloadFormat] = None
+        self,
+        result_id: Optional[str] = None,
+        format: DownloadFormat = DownloadFormat.CSV,
+        request_id: Optional[str] = None,
+        return_context: Optional[bool] = None,
     ) -> PangeaResponse[DownloadResult]:
         """
         Download search results
@@ -435,6 +570,8 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         Args:
             result_id: ID returned by the search API.
             format: Format for the records.
+            request_id: ID returned by the export API.
+            return_context (bool): Return the context data needed to decrypt secure audit events that have been redacted with format preserving encryption.
 
         Returns:
             URL where search results can be downloaded.
@@ -450,8 +587,13 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
             )
         """
 
-        input = DownloadRequest(result_id=result_id, format=format)
-        return await self.request.post("v1/download_results", DownloadResult, data=input.dict(exclude_none=True))
+        if request_id is None and result_id is None:
+            raise ValueError("must pass one of `request_id` or `result_id`")
+
+        input = DownloadRequest(
+            request_id=request_id, result_id=result_id, format=format, return_context=return_context
+        )
+        return await self.request.post("v1/download_results", DownloadResult, data=input.model_dump(exclude_none=True))
 
     async def update_published_roots(self, result: SearchResultOutput):
         """Fetches series of published root hashes from Arweave
@@ -476,12 +618,31 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         for tree_size in tree_sizes:
             pub_root = None
             if tree_size in arweave_roots:
-                pub_root = PublishedRoot(**arweave_roots[tree_size].dict(exclude_none=True))
+                pub_root = PublishedRoot(**arweave_roots[tree_size].model_dump(exclude_none=True))
                 pub_root.source = RootSource.ARWEAVE
             elif self.allow_server_roots:
                 resp = await self.root(tree_size=tree_size)
                 if resp.success and resp.result is not None:
-                    pub_root = PublishedRoot(**resp.result.data.dict(exclude_none=True))
+                    pub_root = PublishedRoot(**resp.result.data.model_dump(exclude_none=True))
                     pub_root.source = RootSource.PANGEA
             if pub_root is not None:
                 self.pub_roots[tree_size] = pub_root
+
+        await self._fix_consistency_proofs(tree_sizes)
+
+    async def _fix_consistency_proofs(self, tree_sizes: Iterable[int]) -> None:
+        # on very rare occasions, the consistency proof in Arweave may be wrong
+        # override it with the proof from pangea (not the root hash, just the proof)
+        for tree_size in tree_sizes:
+            if tree_size not in self.pub_roots or tree_size - 1 not in self.pub_roots:
+                continue
+
+            if self.pub_roots[tree_size].source == RootSource.PANGEA:
+                continue
+
+            if self.verify_consistency_proof(tree_size):
+                continue
+
+            resp = await self.root(tree_size=tree_size)
+            if resp.success and resp.result is not None and resp.result.data is not None:
+                self.pub_roots[tree_size].consistency_proof = resp.result.data.consistency_proof