groundx 2.5.6__tar.gz → 2.5.8__tar.gz

{groundx-2.5.6 → groundx-2.5.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: groundx
-Version: 2.5.6
+Version: 2.5.8
 Summary: 
 License: MIT
 Requires-Python: >=3.8,<4.0
@@ -33,6 +33,7 @@ Requires-Dist: google-auth-stubs ; extra == "extract"
 Requires-Dist: gspread ; extra == "extract"
 Requires-Dist: httpx (>=0.21.2)
 Requires-Dist: minio ; extra == "extract"
+Requires-Dist: pillow ; extra == "extract"
 Requires-Dist: pydantic (>=1.9.2)
 Requires-Dist: pydantic-core (>=2.18.2,<3.0.0)
 Requires-Dist: pytest ; extra == "extract"

{groundx-2.5.6 → groundx-2.5.8}/pyproject.toml

@@ -3,7 +3,7 @@ name = "groundx"
 
 [tool.poetry]
 name = "groundx"
-version = "2.5.6"
+version = "2.5.8"
 description = ""
 readme = "README.md"
 authors = []
@@ -48,6 +48,7 @@ google-auth-stubs = { version = "*", optional = true}
 gspread = { version = "*", optional = true}
 httpx = ">=0.21.2"
 minio = { version = "*", optional = true}
+pillow = { version = "*", optional = true}
 pydantic = ">= 1.9.2"
 pydantic-core = "^2.18.2"
 pytest = { version = "*", optional = true}
@@ -104,7 +105,7 @@ requires = ["poetry-core"]
 build-backend = "poetry.core.masonry.api"
 
 [tool.poetry.extras]
-extract=["boto3", "celery", "celery-types", "dateparser", "fastapi", "google-api-python-client", "google-api-python-client-stubs", "google-auth-stubs", "gspread", "minio", "pandas", "pandas-stubs", "pillow", "pytest", "redis", "types-boto3", "types-dateparser"]
+extract=["boto3", "celery", "celery-types", "dateparser", "fastapi", "google-api-python-client", "google-api-python-client-stubs", "google-auth-stubs", "gspread", "minio", "pillow", "pytest", "redis", "types-boto3", "types-dateparser"]
 
 [[tool.mypy.overrides]]
 module = ["dateparser.*"]

{groundx-2.5.6 → groundx-2.5.8}/src/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/2.5.6",
+            "User-Agent": "groundx/2.5.8",
             "X-Fern-Language": "Python",
             "X-Fern-SDK-Name": "groundx",
-            "X-Fern-SDK-Version": "2.5.6",
+            "X-Fern-SDK-Version": "2.5.8",
         }
         headers["X-API-Key"] = self.api_key
         return headers

{groundx-2.5.6 → groundx-2.5.8}/src/groundx/extract/classes/document.py

@@ -1,6 +1,7 @@
 import json, os, shutil, requests, time, typing
 from datetime import datetime, timezone
 from io import BytesIO
+from pathlib import Path
 from PIL import Image
 from pydantic import BaseModel, ConfigDict, Field, PrivateAttr
 
@@ -41,6 +42,7 @@ class Document(BaseModel):
     def from_request(
         cls: typing.Type[DocT],
         base_url: str,
+        cache_dir: Path,
         req: "DocumentRequest",
         **data: typing.Any,
     ) -> DocT:
@@ -54,7 +56,7 @@ class Document(BaseModel):
             base_url=base_url,
             documentID=req.document_id,
             taskID=req.task_id,
-        ).xray(clear_cache=req.clear_cache)
+        ).xray(cache_dir=cache_dir, clear_cache=req.clear_cache)
 
         for page in xray_doc.documentPages:
             st.page_images.append(page.pageUrl)

{groundx-2.5.6 → groundx-2.5.8}/src/groundx/extract/classes/groundx.py

@@ -19,12 +19,17 @@ class GroundXDocument(BaseModel):
 
     def xray(
         self,
+        cache_dir: Path,
         clear_cache: bool = False,
         is_test: bool = False,
         base: typing.Optional[str] = None,
     ) -> "XRayDocument":
         return XRayDocument.download(
-            self, base=base, clear_cache=clear_cache, is_test=is_test
+            self,
+            cache_dir=cache_dir,
+            base=base,
+            clear_cache=clear_cache,
+            is_test=is_test,
         )
 
 
@@ -81,11 +86,11 @@ class XRayDocument(BaseModel):
     def download(
         cls,
         gx_doc: GroundXDocument,
+        cache_dir: Path,
         clear_cache: bool = False,
         is_test: bool = False,
         base: typing.Optional[str] = None,
     ) -> "XRayDocument":
-        cache_dir = Path(__file__).resolve().parent.parent / "cache"
         cache_dir.mkdir(parents=True, exist_ok=True)
         cache_file = cache_dir / f"{gx_doc.document_id}-xray.json"
 

{groundx-2.5.6 → groundx-2.5.8}/src/groundx/extract/classes/test_document.py

@@ -3,6 +3,7 @@ import pytest, typing, unittest
 pytest.importorskip("PIL")
 
 from io import BytesIO
+from pathlib import Path
 from PIL import Image
 from unittest.mock import patch
 
@@ -14,6 +15,14 @@ def DR(**data: typing.Any) -> DocumentRequest:
     return DocumentRequest.model_validate(data)
 
 
+def test_doc() -> Document:
+    return Document.from_request(
+        cache_dir=Path("./cache"),
+        base_url="",
+        req=test_request(),
+    )
+
+
 def test_request() -> DocumentRequest:
     return DR(documentID="D", fileName="F", modelID=1, processorID=1, taskID="T")
 
@@ -28,12 +37,10 @@ class TestDocument(unittest.TestCase):
         self.mock_xray.return_value = TestXRay("http://test.co", [])
 
     def test_init_name(self) -> None:
-        st1: Document = Document.from_request(
-            base_url="",
-            req=test_request(),
-        )
+        st1: Document = test_doc()
         self.assertEqual(st1.file_name, "F")
         st2: Document = Document.from_request(
+            cache_dir=Path("./cache"),
             base_url="",
             req=DR(
                 documentID="D", fileName="F.pdf", modelID=1, processorID=1, taskID="T"
@@ -41,6 +48,7 @@ class TestDocument(unittest.TestCase):
         )
         self.assertEqual(st2.file_name, "F.pdf")
         st3: Document = Document.from_request(
+            cache_dir=Path("./cache"),
             base_url="",
             req=DR(documentID="D", fileName="F.", modelID=1, processorID=1, taskID="T"),
         )

{groundx-2.5.6 → groundx-2.5.8}/src/groundx/extract/classes/test_groundx.py

@@ -1,4 +1,5 @@
 import requests, typing, unittest
+from pathlib import Path
 from unittest.mock import patch
 
 from pydantic import ValidationError
@@ -42,6 +43,12 @@ def GD(**data: typing.Any) -> GroundXDocument:
     return GroundXDocument.model_validate(data)
 
 
+def test_xray(gx: GroundXDocument) -> XRayDocument:
+    return XRayDocument.download(
+        gx, cache_dir=Path("./cache"), base="https://upload.test", is_test=True
+    )
+
+
 class TestGroundX(unittest.TestCase):
     def make_dummy_response(
         self,
@@ -75,7 +82,7 @@ class TestGroundX(unittest.TestCase):
         dummy = self.make_dummy_response(payload=payload, status_ok=True)
         with patch("requests.get", return_value=dummy):
             gx = GD(base_url="", documentID="D", taskID="T")
-            xdoc = XRayDocument.download(gx, base="https://upload.test", is_test=True)
+            xdoc = test_xray(gx)
             self.assertIsInstance(xdoc, XRayDocument)
             self.assertEqual(xdoc.chunks, [])
             self.assertEqual(xdoc.documentPages, [])
@@ -85,7 +92,7 @@ class TestGroundX(unittest.TestCase):
         with patch("requests.get", side_effect=requests.RequestException("no network")):
             gx = GD(base_url="", documentID="D", taskID="T")
             with self.assertRaises(RuntimeError) as cm:
-                XRayDocument.download(gx, base="https://upload.test", is_test=True)
+                test_xray(gx)
             self.assertIn("Error fetching X-ray JSON", str(cm.exception))
 
     def test_download_http_error(self):
@@ -93,7 +100,7 @@ class TestGroundX(unittest.TestCase):
         with patch("requests.get", return_value=dummy):
             gx = GD(base_url="", documentID="D", taskID="T")
             with self.assertRaises(RuntimeError) as cm:
-                XRayDocument.download(gx, base="https://upload.test", is_test=True)
+                test_xray(gx)
             self.assertIn("HTTP error!", str(cm.exception))
 
     def test_download_json_error(self):
@@ -101,7 +108,7 @@ class TestGroundX(unittest.TestCase):
         with patch("requests.get", return_value=dummy):
             gx = GD(base_url="", documentID="D", taskID="T")
             with self.assertRaises(RuntimeError) as cm:
-                XRayDocument.download(gx, base="https://upload.test", is_test=True)
+                test_xray(gx)
             self.assertIn("Invalid JSON returned", str(cm.exception))
 
     def test_validation_error_on_missing_required_fields(self) -> None:
@@ -113,7 +120,7 @@ class TestGroundX(unittest.TestCase):
         with patch("requests.get", return_value=dummy):
             gx = GD(base_url="", documentID="D", taskID="T")
             with self.assertRaises(ValidationError) as cm:
-                XRayDocument.download(gx, base="https://upload.test", is_test=True)
+                test_xray(gx)
             self.assertIn("Field required", str(cm.exception))
 
     def test_xray_method_delegates_to_download(self) -> None:
@@ -121,7 +128,9 @@ class TestGroundX(unittest.TestCase):
 
         sentinel = object()
         with patch.object(XRayDocument, "download", return_value=sentinel):
-            result = gx.xray(base="https://upload.test", is_test=True)
+            result = gx.xray(
+                cache_dir=Path("./cache"), base="https://upload.test", is_test=True
+            )
             self.assertIs(result, sentinel)
 
     def test_chunk_json_alias(self) -> None:
@@ -186,7 +195,7 @@ class TestGroundX(unittest.TestCase):
         dummy = self.make_dummy_response(payload=payload, status_ok=True)
         with patch("requests.get", return_value=dummy):
             gx = GD(base_url="", documentID="D", taskID="T")
-            xdoc = XRayDocument.download(gx, base="https://upload.test", is_test=True)
+            xdoc = test_xray(gx)
 
             self.assertEqual(xdoc.fileType, "pdf")
             self.assertEqual(xdoc.fileName, "file.pdf")

{groundx-2.5.6 → groundx-2.5.8}/src/groundx/extract/settings/settings.py

@@ -54,6 +54,7 @@ class AgentSettings(BaseModel):
 
 class ContainerSettings(BaseModel):
     broker: str
+    cache_dir: str = "./cache"
     cache_to: int = 300
     google_sheets_drive_id: typing.Optional[str] = None
     google_sheets_template_id: typing.Optional[str] = None