groundx 2.6.0__py3-none-any.whl → 2.7.8__py3-none-any.whl

groundx/__init__.py

@@ -52,6 +52,7 @@ from .types import (
     WebsiteSource,
     WorkflowApplyRequest,
     WorkflowDetail,
+    WorkflowDetailChunkStrategy,
     WorkflowDetailRelationships,
     WorkflowEngine,
     WorkflowEngineReasoningEffort,
@@ -59,25 +60,22 @@ from .types import (
     WorkflowPrompt,
     WorkflowPromptGroup,
     WorkflowPromptRole,
+    WorkflowRequest,
+    WorkflowRequestChunkStrategy,
     WorkflowResponse,
     WorkflowStep,
+    WorkflowStepConfig,
+    WorkflowStepConfigField,
     WorkflowSteps,
-    WorkflowStepsChunkInstruct,
-    WorkflowStepsChunkSummary,
-    WorkflowStepsDocKeys,
-    WorkflowStepsDocSummary,
-    WorkflowStepsSearchQuery,
-    WorkflowStepsSectInstruct,
-    WorkflowStepsSectSummary,
     WorkflowsResponse,
 )
 from .errors import BadRequestError, UnauthorizedError
-from . import buckets, customer, documents, groups, health, search, workflow
+from . import buckets, customer, documents, groups, health, search, workflows
 from .environment import GroundXEnvironment
 from .ingest import AsyncGroundX, GroundX
 from .search import SearchContentRequestId
 from .version import __version__
-from .workflow import WorkflowGetRequestId
+from .workflows import WorkflowsGetRequestId
 
 __all__ = [
     "AsyncGroundX",
@@ -135,24 +133,22 @@ __all__ = [
     "WebsiteSource",
     "WorkflowApplyRequest",
     "WorkflowDetail",
+    "WorkflowDetailChunkStrategy",
     "WorkflowDetailRelationships",
     "WorkflowEngine",
     "WorkflowEngineReasoningEffort",
     "WorkflowEngineService",
-    "WorkflowGetRequestId",
     "WorkflowPrompt",
     "WorkflowPromptGroup",
     "WorkflowPromptRole",
+    "WorkflowRequest",
+    "WorkflowRequestChunkStrategy",
     "WorkflowResponse",
     "WorkflowStep",
+    "WorkflowStepConfig",
+    "WorkflowStepConfigField",
     "WorkflowSteps",
-    "WorkflowStepsChunkInstruct",
-    "WorkflowStepsChunkSummary",
-    "WorkflowStepsDocKeys",
-    "WorkflowStepsDocSummary",
-    "WorkflowStepsSearchQuery",
-    "WorkflowStepsSectInstruct",
-    "WorkflowStepsSectSummary",
+    "WorkflowsGetRequestId",
     "WorkflowsResponse",
     "__version__",
     "buckets",
@@ -161,5 +157,5 @@ __all__ = [
     "groups",
     "health",
     "search",
-    "workflow",
+    "workflows",
 ]

groundx/client.py

@@ -11,7 +11,7 @@ from .environment import GroundXEnvironment
 from .groups.client import AsyncGroupsClient, GroupsClient
 from .health.client import AsyncHealthClient, HealthClient
 from .search.client import AsyncSearchClient, SearchClient
-from .workflow.client import AsyncWorkflowClient, WorkflowClient
+from .workflows.client import AsyncWorkflowsClient, WorkflowsClient
 
 
 class GroundXBase:
@@ -78,7 +78,7 @@ class GroundXBase:
         self.search = SearchClient(client_wrapper=self._client_wrapper)
         self.buckets = BucketsClient(client_wrapper=self._client_wrapper)
         self.groups = GroupsClient(client_wrapper=self._client_wrapper)
-        self.workflow = WorkflowClient(client_wrapper=self._client_wrapper)
+        self.workflows = WorkflowsClient(client_wrapper=self._client_wrapper)
         self.customer = CustomerClient(client_wrapper=self._client_wrapper)
         self.health = HealthClient(client_wrapper=self._client_wrapper)
 
@@ -147,7 +147,7 @@ class AsyncGroundXBase:
         self.search = AsyncSearchClient(client_wrapper=self._client_wrapper)
         self.buckets = AsyncBucketsClient(client_wrapper=self._client_wrapper)
         self.groups = AsyncGroupsClient(client_wrapper=self._client_wrapper)
-        self.workflow = AsyncWorkflowClient(client_wrapper=self._client_wrapper)
+        self.workflows = AsyncWorkflowsClient(client_wrapper=self._client_wrapper)
         self.customer = AsyncCustomerClient(client_wrapper=self._client_wrapper)
         self.health = AsyncHealthClient(client_wrapper=self._client_wrapper)
 

groundx/core/client_wrapper.py

@@ -14,10 +14,10 @@ class BaseClientWrapper:
 
     def get_headers(self) -> typing.Dict[str, str]:
         headers: typing.Dict[str, str] = {
-            "User-Agent": "groundx/2.6.0",
+            "User-Agent": "groundx/2.7.8",
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "groundx",
-            "X-Fern-SDK-Version": "2.6.0",
+            "X-Fern-SDK-Version": "2.7.8",
         }
         headers["X-API-Key"] = self.api_key
         return headers

groundx/extract/classes/document.py

@@ -4,9 +4,11 @@ from io import BytesIO
 from pathlib import Path
 from PIL import Image
 from pydantic import BaseModel, ConfigDict, Field, PrivateAttr
+from urllib.parse import urlparse
 
 from .groundx import GroundXDocument
 from ..services.logger import Logger
+from ..services.upload import Upload
 from ..utility.classes import clean_json
 
 
@@ -44,6 +46,7 @@ class Document(BaseModel):
         base_url: str,
         cache_dir: Path,
         req: "DocumentRequest",
+        upload: typing.Optional[Upload] = None,
         **data: typing.Any,
     ) -> DocT:
         st = cls(**data)
@@ -56,7 +59,7 @@ class Document(BaseModel):
             base_url=base_url,
             documentID=req.document_id,
             taskID=req.task_id,
-        ).xray(cache_dir=cache_dir, clear_cache=req.clear_cache)
+        ).xray(upload=upload, cache_dir=cache_dir, clear_cache=req.clear_cache)
 
         for page in xray_doc.documentPages:
             st.page_images.append(page.pageUrl)
@@ -250,6 +253,7 @@ class DocumentRequest(BaseModel):
     def load_images(
         self,
         imgs: typing.List[str],
+        upload: typing.Optional[Upload] = None,
         attempt: int = 0,
         should_sleep: bool = True,
     ) -> typing.List[Image.Image]:
@@ -261,26 +265,39 @@ class DocumentRequest(BaseModel):
                     f"[{attempt}] loading cached [{self.page_image_dict[page]}] [{page}]",
                 )
                 pageImages.append(self.page_images[self.page_image_dict[page]])
-            else:
-                try:
-                    self.print("WARN", f"[{attempt}] downloading [{page}]")
-                    resp = requests.get(page)
-                    resp.raise_for_status()
-                    img = Image.open(BytesIO(resp.content))
+                continue
+
+            if upload:
+                parsed = urlparse(page)
+                path = parsed.path + ("?" + parsed.query if parsed.query else "")
+                ru = upload.get_object(path)
+                if ru:
+                    img = Image.open(BytesIO(ru))
                     if img:
                         self.page_image_dict[page] = len(self.page_images)
                         self.page_images.append(img)
                         pageImages.append(img)
-                except Exception as e:
-                    self.print(
-                        "ERROR", f"[{attempt}] Failed to load image from {page}: {e}"
+                        continue
+
+            try:
+                self.print("WARN", f"[{attempt}] downloading [{page}]")
+                resp = requests.get(page)
+                resp.raise_for_status()
+                img = Image.open(BytesIO(resp.content))
+                if img:
+                    self.page_image_dict[page] = len(self.page_images)
+                    self.page_images.append(img)
+                    pageImages.append(img)
+            except Exception as e:
+                self.print(
+                    "ERROR", f"[{attempt}] Failed to load image from {page}: {e}"
+                )
+                if attempt < 2:
+                    if should_sleep:
+                        time.sleep(2 * attempt + 1)
+                    return self.load_images(
+                        imgs, upload, attempt + 1, should_sleep=should_sleep
                     )
-                    if attempt < 2:
-                        if should_sleep:
-                            time.sleep(2 * attempt + 1)
-                        return self.load_images(
-                            imgs, attempt + 1, should_sleep=should_sleep
-                        )
 
         return pageImages
 

groundx/extract/classes/groundx.py

@@ -3,6 +3,8 @@ from pathlib import Path
 
 from pydantic import BaseModel, ConfigDict, Field
 
+from ..services.upload import Upload
+
 
 class GroundXDocument(BaseModel):
     model_config = ConfigDict(populate_by_name=True)
@@ -10,6 +12,9 @@ class GroundXDocument(BaseModel):
     document_id: str = Field(alias="documentID")
     task_id: str = Field(alias="taskID")
 
+    def xray_path(self) -> str:
+        return f"layout/processed/{self.task_id}/{self.document_id}-xray.json"
+
     def xray_url(self, base: typing.Optional[str] = None) -> str:
         if not base:
             base = self.base_url
@@ -20,6 +25,7 @@ class GroundXDocument(BaseModel):
     def xray(
         self,
         cache_dir: Path,
+        upload: typing.Optional[Upload] = None,
         clear_cache: bool = False,
         is_test: bool = False,
         base: typing.Optional[str] = None,
@@ -27,9 +33,10 @@ class GroundXDocument(BaseModel):
         return XRayDocument.download(
             self,
             cache_dir=cache_dir,
-            base=base,
+            upload=upload,
             clear_cache=clear_cache,
             is_test=is_test,
+            base=base,
         )
 
 
@@ -87,6 +94,7 @@ class XRayDocument(BaseModel):
         cls,
         gx_doc: GroundXDocument,
         cache_dir: Path,
+        upload: typing.Optional[Upload] = None,
         clear_cache: bool = False,
         is_test: bool = False,
         base: typing.Optional[str] = None,
@@ -99,30 +107,41 @@ class XRayDocument(BaseModel):
                 with cache_file.open("r", encoding="utf-8") as f:
                     payload = json.load(f)
 
+                return cls(**payload)
             except Exception as e:
                 raise RuntimeError(
                     f"Error loading cached X-ray JSON from {cache_file}: {e}"
                 )
-        else:
-            url = gx_doc.xray_url(base=base)
-            try:
-                resp = requests.get(url)
-                resp.raise_for_status()
-            except requests.RequestException as e:
-                raise RuntimeError(f"Error fetching X-ray JSON from {url}: {e}")
-
-            try:
-                payload = resp.json()
-            except ValueError as e:
-                raise RuntimeError(f"Invalid JSON returned from {url}: {e}")
 
-            if is_test is False:
+        if upload:
+            path = gx_doc.xray_path()
+            ru = upload.get_object(path)
+            if ru:
                 try:
-                    with cache_file.open("w", encoding="utf-8") as f:
-                        json.dump(payload, f)
+                    payload = json.loads(ru.decode("utf-8"))
+                    return cls(**payload)
                 except Exception as e:
-                    print(
-                        f"Warning: failed to write X-ray JSON cache to {cache_file}: {e}"
+                    raise RuntimeError(
+                        f"Error decoding X-ray JSON bytes from {path}: {e}"
                     )
 
+        url = gx_doc.xray_url(base=base)
+        try:
+            resp = requests.get(url)
+            resp.raise_for_status()
+        except requests.RequestException as e:
+            raise RuntimeError(f"Error fetching X-ray JSON from {url}: {e}")
+
+        try:
+            payload = resp.json()
+        except ValueError as e:
+            raise RuntimeError(f"Invalid JSON returned from {url}: {e}")
+
+        if is_test is False:
+            try:
+                with cache_file.open("w", encoding="utf-8") as f:
+                    json.dump(payload, f)
+            except Exception as e:
+                print(f"Warning: failed to write X-ray JSON cache to {cache_file}: {e}")
+
         return cls(**payload)

groundx/extract/services/logging_cfg.py

@@ -2,8 +2,6 @@ import typing
 
 
 def logging_config(name: str, level: str) -> typing.Dict[str, typing.Any]:
-    print(level)
-
     return {
         "version": 1,
         "disable_existing_loggers": False,

groundx/extract/services/upload.py

@@ -46,13 +46,8 @@ class Upload:
         else:
             raise Exception(f"unsupported upload.type [{self.settings.upload.type}]")
 
-    def get_file(self, url: str) -> bytes:
-        return bytes()
-
     def get_object(self, url: str) -> typing.Optional[bytes]:
-        self.client.get_object(url)
-
-        return None
+        return self.client.get_object(url)
 
     def put_object(
         self,

groundx/extract/services/upload_s3.py

@@ -25,12 +25,19 @@ class S3Client:
 
     def get_object(self, url: str) -> typing.Optional[bytes]:
         if not self.client:
+            print("get_object no client")
             return None
 
         try:
-            s3_uri_parts = url.replace("s3://", "").split("/")
-            s3_bucket = s3_uri_parts[0]
-            s3_key = "/".join(s3_uri_parts[1:])
+            if url.startswith("s3://"):
+                s3_uri_parts = url.replace("s3://", "").split("/")
+                s3_bucket = s3_uri_parts[0]
+                s3_key = "/".join(s3_uri_parts[1:])
+            else:
+                s3_bucket = self.settings.upload.bucket
+                s3_key = url
+                if url.startswith("/"):
+                    s3_key = url[1:]
 
             response = self.client.get_object(Bucket=s3_bucket, Key=s3_key)
 

groundx/extract/settings/settings.py

@@ -17,6 +17,8 @@ GX_DEFAULT_REGION: str = "GROUNDX_DEFAULT_REGION"
 GX_SECRET: str = "GROUNDX_SECRET_ACCESS_KEY"
 GX_TOKEN: str = "GROUNDX_SESSION_TOKEN"
 VALID_KEYS: str = "GROUNDX_VALID_API_KEYS"
+GX_ADMIN_API_KEY: str = "GROUNDX_ADMIN_API_KEY"
+GX_ADMIN_USERNAME: str = "GROUNDX_ADMIN_USERNAME"
 
 
 class AgentSettings(BaseModel):
@@ -77,22 +79,54 @@ class ContainerSettings(BaseModel):
         if key:
             return key
 
+        key = os.environ.get(GX_ADMIN_USERNAME)
+        if key:
+            return key
+
+        key = os.environ.get(GX_ADMIN_API_KEY)
+        if key:
+            return key
+
+        key = os.environ.get(GX_API_KEY)
+        if key:
+            return key
+
         raise Exception(f"you must set a callback_api_key")
 
     def get_valid_api_keys(self) -> typing.List[str]:
+        keys: typing.List[str] = []
+
         if self.valid_api_keys:
-            return self.valid_api_keys
+            keys = self.valid_api_keys
 
-        keys: typing.Optional[str] = os.environ.get(VALID_KEYS)
-        if not keys:
-            raise Exception(f"you must set an array of valid_api_keys")
+        env_keys: typing.Optional[str] = os.environ.get(VALID_KEYS)
+        if env_keys:
+            try:
+                data: typing.List[str] = json.loads(env_keys)
+                keys.extend(data)
+            except Exception as e:
+                raise Exception(f"you must set an array of valid_api_keys: {e}")
 
-        try:
-            data: typing.List[str] = json.loads(keys)
-        except Exception as e:
-            raise Exception(f"you must set an array of valid_api_keys: {e}")
+        key = os.environ.get(CALLBACK_KEY)
+        if key:
+            keys.append(key)
 
-        return data
+        key = os.environ.get(GX_ADMIN_API_KEY)
+        if key:
+            keys.append(key)
+
+        key = os.environ.get(GX_ADMIN_USERNAME)
+        if key:
+            keys.append(key)
+
+        key = os.environ.get(GX_API_KEY)
+        if key:
+            keys.append(key)
+
+        if len(keys) < 1:
+            raise Exception(f"you must set an array of valid_api_keys")
+
+        return keys
 
     def loglevel(self) -> str:
         return self.log_level.upper()
@@ -163,8 +197,16 @@ class GroundXSettings(BaseModel):
         if self.api_key:
             return self.api_key
 
+        key = os.environ.get(GX_ADMIN_USERNAME)
+        if key:
+            return key
+
         key = os.environ.get(GX_API_KEY)
         if key:
             return key
 
+        key = os.environ.get(GX_ADMIN_API_KEY)
+        if key:
+            return key
+
         raise Exception(f"you must set a valid GroundX api_key")

groundx/extract/settings/test_settings.py

@@ -45,7 +45,6 @@ class TestAgentSettings(unittest.TestCase):
                 "expect": {
                     "api_base": "http://test.com",
                     "api_key": "mykey",
-                    "api_key_env": "myenv",
                     "max_steps": 4,
                     "model_id": "gpt-5",
                 },
@@ -452,10 +451,8 @@ class TestGroundXSettings(unittest.TestCase):
     def test(self) -> None:
         tsts: typing.List[typing.Dict[str, typing.Any]] = [
             {
-                "api_key_env": "",
                 "expect": {
                     "api_key": Exception,
-                    "api_key_env": "",
                     "base_url": None,
                     "upload_url": "https://upload.eyelevel.ai",
                 },