hie-rag 0.2.2__py3-none-any.whl → 0.2.3__py3-none-any.whl

hie_rag/hie_rag.py

@@ -16,7 +16,7 @@ class HieRag:
     def process_and_save_index_stream(self, file_name: str, uploaded_file: bytes, min_chunk_size, max_chunk_size):
         yield {"status": "🔍 Extracting text..."}
         print(f"Extracting text from {file_name}")
-        extracted_text = self.utils.extract_text(uploaded_file)
+        extracted_text = self.utils.extract_text(uploaded_file, file_name=file_name)
 
         yield {"status": "✂️ Splitting into chunks..."}
         print(f"Splitting text into chunks with min size {min_chunk_size} and max size {max_chunk_size}")

hie_rag/utils.py

@@ -16,36 +16,26 @@ class Utils:
         # self.client = OpenAI(api_key=api_key)
         self.client = AiClient(base_url=base_url)
 
-    def extract_text(self, uploaded_file: bytes):
+    def extract_text(self, uploaded_bytes: bytes, file_name: str):
         """Extract text from an uploaded file using MarkItDown."""
-        # md = MarkItDown(llm_client=self.client, llm_model="gpt-4o")
         md = MarkItDown()
 
-        # Accept both raw bytes and file-like objects with `.read()`
-        if isinstance(uploaded_file, bytes):
-            file_bytes = uploaded_file
-            suffix = ".bin"  # fallback generic extension
-        elif hasattr(uploaded_file, "read"):
-            file_bytes = uploaded_file.read()
-            filename = getattr(uploaded_file, "name", None) or getattr(uploaded_file, "filename", None)
-            suffix = os.path.splitext(filename)[-1] if filename else ".bin"
-        else:
-            raise TypeError("Unsupported file type: must be bytes or file-like object")
-
-        # Write to temp file for MarkItDown to process
-        with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as temp_file:
-            temp_file_path = temp_file.name
-            temp_file.write(file_bytes)
+        # derive a real suffix from the filename
+        suffix = os.path.splitext(file_name)[1].lower() or ".txt"
+
+        # write to temp file
+        with tempfile.NamedTemporaryFile(delete=False, suffix=suffix) as tmp:
+            tmp.write(uploaded_bytes)
+            tmp_path = tmp.name
 
         try:
-            # Redirect stderr to suppress native print warnings like "CropBox missing"
             with contextlib.redirect_stderr(io.StringIO()):
-                extracted_text = md.convert(temp_file_path)
+                result = md.convert(tmp_path)
         finally:
-            # Clean up the temporary file
-            os.remove(temp_file_path)
+            os.remove(tmp_path)
 
-        return extracted_text.text_content
+        # depending on MarkItDown version this may return a str or an object
+        return getattr(result, "text_content", result)
 
     def count_tokens(self, text: str, encoding="cl100k_base") -> int:
         """Count tokens in text using tiktoken"""

{hie_rag-0.2.2.dist-info → hie_rag-0.2.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hie_rag
-Version: 0.2.2
+Version: 0.2.3
 Summary: A hierarchical RAG framework for chunks retrieval.
 License: Apache License
         Version 2.0, January 2004

{hie_rag-0.2.2.dist-info → hie_rag-0.2.3.dist-info}/RECORD

@@ -1,14 +1,14 @@
 hie_rag/__init__.py,sha256=p2glSTkCqGvMlcivcuKBStFh2C5adojaC9aGmF6nbhY,358
 hie_rag/ai_client.py,sha256=VbGQ0e3vZNn8W2YoR15Vvq2r-MUs-TBRNLGiImT4QxU,1000
-hie_rag/hie_rag.py,sha256=Nl_1WZM9IWhpNyZMvPzsae_u_xaCWEwrJgorZV-hp20,2741
+hie_rag/hie_rag.py,sha256=6WG5TyX3w_2jm1lIWFgCn-J8lElPOmjqNfUKe7akAI0,2762
 hie_rag/process.py,sha256=Z4qpNmxSsxUJgnqJtw8cYWJTS6SxhRR7F7eX_akyVCU,2427
 hie_rag/split.py,sha256=gEQVt57xWruT5e1psgSOnwuBrQngzri3S4H6ZvKzsw4,5082
 hie_rag/split_and_process.py,sha256=PkFlnOF7nW4Zs47JTsGF4AY9VDOXz1AtxG9Die8_mQk,572
 hie_rag/tree_index.py,sha256=iTa25ohMv5O0HYc5JtzIzVAIhNdVklYiAIJvqyE8sbM,2722
-hie_rag/utils.py,sha256=GwGiQj-zc8-U9UXOFHTKkjHWHx8YTYquR27gsXJgzCE,4687
+hie_rag/utils.py,sha256=_4TGiHuJ-Xo4JEEEdOjp4d1zxw6dNVsxROcom-vr7uU,4059
 hie_rag/vectordb.py,sha256=iI73ujrONjDaHU66RNdHnD2PZWSppnjm0isIHPJEGAY,11068
-hie_rag-0.2.2.dist-info/licenses/LICENSE,sha256=IwAxruLb1UG8F0KZtfnV6MJq10FRAxWM-XOTWkWsJt4,632
-hie_rag-0.2.2.dist-info/METADATA,sha256=3vTI_zyvJxOOq8VrrchOAn0a7m8hwQPISnlholFi3u0,1698
-hie_rag-0.2.2.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-hie_rag-0.2.2.dist-info/top_level.txt,sha256=tN2S3VpMUl6oLWL9sN4xIh4o2na_zjnW8rHiwPFf0T8,8
-hie_rag-0.2.2.dist-info/RECORD,,
+hie_rag-0.2.3.dist-info/licenses/LICENSE,sha256=IwAxruLb1UG8F0KZtfnV6MJq10FRAxWM-XOTWkWsJt4,632
+hie_rag-0.2.3.dist-info/METADATA,sha256=QkUjUo8zbEa7Ijm6mWOkmo-hoLzcYPDOQqzeqR7iL8g,1698
+hie_rag-0.2.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+hie_rag-0.2.3.dist-info/top_level.txt,sha256=tN2S3VpMUl6oLWL9sN4xIh4o2na_zjnW8rHiwPFf0T8,8
+hie_rag-0.2.3.dist-info/RECORD,,