airbyte-cdk 6.13.1.dev4106__py3-none-any.whl → 6.13.1.dev4108__py3-none-any.whl

airbyte_cdk/sources/file_based/file_types/unstructured_parser.py

@@ -42,16 +42,32 @@ from airbyte_cdk.utils.traced_exception import AirbyteTracedException
 unstructured_partition_pdf = None
 unstructured_partition_docx = None
 unstructured_partition_pptx = None
-nltk_data_dir = "/tmp/nltk_data"
+
+
+def get_ntlk_temp_folder() -> str:
+    """
+    For non-root connectors /tmp is not currently writable, but we should allow it in the future.
+    It's safe to use /airbyte for now. Fallback to /tmp for local development.
+    """
+    try:
+        nltk_data_dir = "/airbyte/nltk_data"
+        os.makedirs(nltk_data_dir, exist_ok=True)
+    except OSError:
+        nltk_data_dir = "/tmp/nltk_data"
+        os.makedirs(nltk_data_dir, exist_ok=True)
+    return nltk_data_dir
+
 
 try:
-    os.makedirs(nltk_data_dir, exist_ok=True)
+    nltk_data_dir = get_ntlk_temp_folder()
     nltk.data.path.append(nltk_data_dir)
     nltk.data.find("tokenizers/punkt.zip")
     nltk.data.find("tokenizers/punkt_tab.zip")
+    nltk.data.find("tokenizers/averaged_perceptron_tagger_eng.zip")
 except LookupError:
     nltk.download("punkt", download_dir=nltk_data_dir)
     nltk.download("punkt_tab", download_dir=nltk_data_dir)
+    nltk.download("averaged_perceptron_tagger_eng", download_dir=nltk_data_dir)
 
 
 def optional_decode(contents: Union[str, bytes]) -> str:

{airbyte_cdk-6.13.1.dev4106.dist-info → airbyte_cdk-6.13.1.dev4108.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: airbyte-cdk
-Version: 6.13.1.dev4106
+Version: 6.13.1.dev4108
 Summary: A framework for writing Airbyte Connectors.
 License: MIT
 Keywords: airbyte,connector-development-kit,cdk

{airbyte_cdk-6.13.1.dev4106.dist-info → airbyte_cdk-6.13.1.dev4108.dist-info}/RECORD

@@ -217,7 +217,7 @@ airbyte_cdk/sources/file_based/file_types/file_transfer.py,sha256=HyGRihJxcb_lEs
 airbyte_cdk/sources/file_based/file_types/file_type_parser.py,sha256=JgpH21PrbRqwK92BJklZWvh2TndA6xZ-eP1LPMo44oQ,2832
 airbyte_cdk/sources/file_based/file_types/jsonl_parser.py,sha256=GwyNyxmST4RX-XpXy7xVH0D-znYWWBmGv_pVAu95oHQ,5886
 airbyte_cdk/sources/file_based/file_types/parquet_parser.py,sha256=XenFg5sJ-UBnIkSmsiNJRou11NO0zZXx-RXgPHMT2NA,10487
-airbyte_cdk/sources/file_based/file_types/unstructured_parser.py,sha256=vypnocjZkMIkc84XeYtTMO0Vy70wXwHZ0Ug3B4sD_dM,18788
+airbyte_cdk/sources/file_based/file_types/unstructured_parser.py,sha256=FOZL-RiXc7sndfdYZtLVSR93c_eHlKaS_nv2KrqFu2E,19371
 airbyte_cdk/sources/file_based/remote_file.py,sha256=yqRz93vPe8PBXLIMJ5W5u2JRlZRhg6sBrAjn3pPjJ8A,315
 airbyte_cdk/sources/file_based/schema_helpers.py,sha256=Cf8FH1bDFP0qCDDfEYir_WjP4exXUnikz8hZ40y1Ek0,9601
 airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py,sha256=FkByIyEy56x2_awYnxGPqGaOp7zAzpAoRkPZHKySI9M,536
@@ -342,8 +342,8 @@ airbyte_cdk/utils/slice_hasher.py,sha256=-pHexlNYoWYPnXNH-M7HEbjmeJe9Zk7SJijdQ7d
 airbyte_cdk/utils/spec_schema_transformations.py,sha256=-5HTuNsnDBAhj-oLeQXwpTGA0HdcjFOf2zTEMUTTg_Y,816
 airbyte_cdk/utils/stream_status_utils.py,sha256=ZmBoiy5HVbUEHAMrUONxZvxnvfV9CesmQJLDTAIWnWw,1171
 airbyte_cdk/utils/traced_exception.py,sha256=C8uIBuCL_E4WnBAOPSxBicD06JAldoN9fGsQDp463OY,6292
-airbyte_cdk-6.13.1.dev4106.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
-airbyte_cdk-6.13.1.dev4106.dist-info/METADATA,sha256=zMtiITJXfqZ91Kxu5yHIvi1S4rXhdN_ktOCOi7-pC30,6008
-airbyte_cdk-6.13.1.dev4106.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
-airbyte_cdk-6.13.1.dev4106.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
-airbyte_cdk-6.13.1.dev4106.dist-info/RECORD,,
+airbyte_cdk-6.13.1.dev4108.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
+airbyte_cdk-6.13.1.dev4108.dist-info/METADATA,sha256=4MmYepPLObB4KNT5D7ttKKhwRdVaXdJnZMcOLewYXOU,6008
+airbyte_cdk-6.13.1.dev4108.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
+airbyte_cdk-6.13.1.dev4108.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
+airbyte_cdk-6.13.1.dev4108.dist-info/RECORD,,