[FIX] RC3 - Adding support to NoOp, LLMW-v2 and Llama parse (#131)

* Exception handling for Prompt Service

* Remote storage-Support for NoOp and LLMW-v2

* Fixing mkdir method signature

* Passing param with arg

* Adding method params

* Version bump and passing extra_info param

Author: harini-venkataraman

src/unstract/sdk/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.54.0rc2"
+__version__ = "0.54.0rc3"
 
 
 def get_sdk_version():

src/unstract/sdk/adapters/x2text/llama_parse/src/llama_parse.py

@@ -69,12 +69,23 @@ def _call_parser(
                     input_file_path = ".".join(
                         (input_file_path_copy, input_file_extension)
                     )
+                    text_content = fs.read(
+                        path=input_file_path_copy, mode="rb", encoding="utf-8"
+                    )
+                    fs.write(
+                        path=input_file_path,
+                        data=text_content,
+                        mode="w",
+                        encoding="utf-8",
+                    )
                 except OSError as os_err:
                     logger.error("Exception raised while handling input file.")
                     raise AdapterError(str(os_err))
 
             file_bytes = fs.read(path=input_file_path, mode="rb")
-            documents = parser.load_data(file_bytes)
+            documents = parser.load_data(
+                file_bytes, extra_info={"file_name": input_file_path}
+            )
 
         except ConnectError as connec_err:
             logger.error(f"Invalid Base URL given. : {connec_err}")

src/unstract/sdk/adapters/x2text/llm_whisperer_v2/src/helper.py

@@ -306,6 +306,7 @@ def extract_text_from_response(
         output_file_path: Optional[str],
         response_dict: dict[str, Any],
         response: Response,
+        fs: FileStorage = FileStorage(provider=FileStorageProvider.LOCAL),
     ) -> str:
         output_json = {}
         if response.status_code == 200:
@@ -321,11 +322,16 @@ def extract_text_from_response(
             LLMWhispererHelper.write_output_to_file(
                 output_json=output_json,
                 output_file_path=Path(output_file_path),
+                fs=fs,
             )
         return output_json.get("result_text", "")
 
     @staticmethod
-    def write_output_to_file(output_json: dict, output_file_path: Path) -> None:
+    def write_output_to_file(
+        output_json: dict,
+        output_file_path: Path,
+        fs: FileStorage = FileStorage(provider=FileStorageProvider.LOCAL),
+    ) -> None:
         """Writes the extracted text and metadata to the specified output file
         and metadata file.
 
@@ -341,7 +347,9 @@ def write_output_to_file(output_json: dict, output_file_path: Path) -> None:
         try:
             text_output = output_json.get("result_text", "")
             logger.info(f"Writing output to {output_file_path}")
-            output_file_path.write_text(text_output, encoding="utf-8")
+            fs.write(
+                path=output_file_path, mode="w", data=text_output, encoding="utf-8"
+            )
         except Exception as e:
             logger.error(f"Error while writing {output_file_path}: {e}")
             raise ExtractorError(str(e))
@@ -352,13 +360,15 @@ def write_output_to_file(output_json: dict, output_file_path: Path) -> None:
             metadata_file_name = output_file_path.with_suffix(".json").name
             metadata_file_path = metadata_dir / metadata_file_name
             # Ensure the metadata directory exists
-            metadata_dir.mkdir(parents=True, exist_ok=True)
+            fs.mkdir(create_parents=True, path=metadata_dir)
             # Remove the "result_text" key from the metadata
             metadata = {
                 key: value for key, value in output_json.items() if key != "result_text"
             }
             metadata_json = json.dumps(metadata, ensure_ascii=False, indent=4)
             logger.info(f"Writing metadata to {metadata_file_path}")
-            metadata_file_path.write_text(metadata_json, encoding="utf-8")
+            fs.write(
+                path=metadata_file_path, mode="w", data=metadata_json, encoding="utf-8"
+            )
         except Exception as e:
             logger.warn(f"Error while writing metadata to {metadata_file_path}: {e}")

src/unstract/sdk/adapters/x2text/llm_whisperer_v2/src/llm_whisperer_v2.py

@@ -16,6 +16,8 @@
 )
 from unstract.sdk.adapters.x2text.llm_whisperer_v2.src.helper import LLMWhispererHelper
 from unstract.sdk.adapters.x2text.x2text_adapter import X2TextAdapter
+from unstract.sdk.file_storage.fs_impl import FileStorage
+from unstract.sdk.file_storage.fs_provider import FileStorageProvider
 
 logger = logging.getLogger(__name__)
 
@@ -60,6 +62,7 @@ def process(
         self,
         input_file_path: str,
         output_file_path: Optional[str] = None,
+        fs: FileStorage = FileStorage(provider=FileStorageProvider.LOCAL),
         **kwargs: dict[Any, Any],
     ) -> TextExtractionResult:
         """Used to extract text from documents.
@@ -75,7 +78,7 @@ def process(
         """
 
         response: requests.Response = LLMWhispererHelper.send_whisper_request(
-            input_file_path, self.config
+            input_file_path, self.config, fs=fs
         )
         response_text = response.text
         reponse_dict = json.loads(response_text)
@@ -85,7 +88,7 @@ def process(
 
         return TextExtractionResult(
             extracted_text=LLMWhispererHelper.extract_text_from_response(
-                self.config, output_file_path, reponse_dict, response
+                self.config, output_file_path, reponse_dict, response, fs=fs
             ),
             extraction_metadata=metadata,
         )

src/unstract/sdk/adapters/x2text/no_op/src/no_op_x2text.py

@@ -5,6 +5,8 @@
 
 from unstract.sdk.adapters.x2text.dto import TextExtractionResult
 from unstract.sdk.adapters.x2text.x2text_adapter import X2TextAdapter
+from unstract.sdk.file_storage.fs_impl import FileStorage
+from unstract.sdk.file_storage.fs_provider import FileStorageProvider
 
 logger = logging.getLogger(__name__)
 
@@ -41,6 +43,7 @@ def process(
         self,
         input_file_path: str,
         output_file_path: Optional[str] = None,
+        fs: FileStorage = FileStorage(provider=FileStorageProvider.LOCAL),
         **kwargs: dict[Any, Any],
     ) -> TextExtractionResult:
         extracted_text: str = (
@@ -49,8 +52,9 @@ def process(
         )
         time.sleep(self.config.get("wait_time"))
         if output_file_path:
-            with open(output_file_path, "w", encoding="utf-8") as f:
-                f.write(extracted_text)
+            fs.write(
+                path=output_file_path, mode="w", data=extracted_text, encoding="utf-8"
+            )
         return TextExtractionResult(extracted_text=extracted_text)
 
     def test_connection(self) -> bool: