chunklet-py 2.3.0__tar.gz → 2.3.1__tar.gz

{chunklet_py-2.3.0/src/chunklet_py.egg-info → chunklet_py-2.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: chunklet-py
-Version: 2.3.0
+Version: 2.3.1
 Summary: High-fidelity context-aware chunking and interactive visualization for RAG. Advanced segmentation for code and documents, because your LLM is only as smart as the fragments you feed it.
 Author-email: speedyk_005 <speedy40115719@gmail.com>
 License-Expression: MIT
@@ -28,8 +28,8 @@ License-File: LICENSE
 Requires-Dist: loguru<1.0,>=0.7.3
 Requires-Dist: pysbd<1.0,>=0.3.4
 Requires-Dist: sentsplit<2.0,>=1.0.8
-Requires-Dist: sentencex<2.0,>=1.0.0; platform_system != "Android"
-Requires-Dist: sentencex<=0.6.1; platform_system == "Android"
+Requires-Dist: sentencex<=0.6.1; sys_platform == "linux" and (platform_machine == "aarch64" or platform_machine == "armv7l")
+Requires-Dist: sentencex<2.0,>=1.0.0; sys_platform != "linux" or (platform_machine != "aarch64" and platform_machine != "armv7l")
 Requires-Dist: indic-nlp-library<1.0,>=0.92
 Requires-Dist: py3langid<1.0,>=0.3.0
 Requires-Dist: mpire<3.0,>=2.10.2

{chunklet_py-2.3.0 → chunklet_py-2.3.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "chunklet-py"
-version = "2.3.0"
+version = "2.3.1"
 description = "High-fidelity context-aware chunking and interactive visualization for RAG. Advanced segmentation for code and documents, because your LLM is only as smart as the fragments you feed it."
 authors = [
   { name = "speedyk_005", email = "speedy40115719@gmail.com" }
@@ -38,9 +38,9 @@ dependencies = [
   "pysbd>=0.3.4,<1.0",
   "sentsplit>=1.0.8,<2.0",
 
-  # Rust binding (1.0+) for faster performance on supported platforms, legacy for Android.
-  "sentencex>=1.0.0,<2.0; platform_system != 'Android'",
-  "sentencex<=0.6.1; platform_system == 'Android'",
+  # Rust binding (1.0+) for faster performance on supported platforms, legacy for Android temporarily.
+  "sentencex<=0.6.1; sys_platform == 'linux' and (platform_machine == 'aarch64' or platform_machine == 'armv7l')",
+  "sentencex>=1.0.0,<2.0; sys_platform != 'linux' or (platform_machine != 'aarch64' and platform_machine != 'armv7l')",
 
   "indic-nlp-library>=0.92,<1.0",
   "py3langid>=0.3.0,<1.0",

{chunklet_py-2.3.0 → chunklet_py-2.3.1}/src/chunklet/document_chunker/_plain_text_chunker.py

@@ -120,7 +120,7 @@ class PlainTextChunker:
         """
         chunks_out = []
         for i, chunk_str in enumerate(chunks, start=1):
-            chunk = DotDict()
+            chunk = DotDict({})
             chunk.content = chunk_str.strip()
             chunk.metadata = copy.deepcopy(base_metadata)
             chunk.metadata["chunk_num"] = i

{chunklet_py-2.3.0 → chunklet_py-2.3.1}/src/chunklet/document_chunker/document_chunker.py

@@ -607,7 +607,7 @@ class DocumentChunker(BaseChunker):
             sections_per_path[curr_path] -= 1
 
     @deprecated_callable(
-        use_instead="chunk_file", deprecated_in="2.2.0", removed_in="3.0.0"
+        use_instead="chunk_file or chunk_text", deprecated_in="2.2.0", removed_in="3.0.0"
     )
     def chunk(  # pragma: no cover
         self,
@@ -632,7 +632,7 @@ class DocumentChunker(BaseChunker):
         return self.chunk_file(**params)
 
     @deprecated_callable(
-        use_instead="chunk_files", deprecated_in="2.2.0", removed_in="3.0.0"
+        use_instead="chunk_files or chunk_texts", deprecated_in="2.2.0", removed_in="3.0.0"
     )
     def batch_chunk(  # pragma: no cover
         self,

{chunklet_py-2.3.0 → chunklet_py-2.3.1}/src/chunklet/sentence_splitter/_universal_splitter.py

@@ -36,6 +36,11 @@ class UniversalSplitter:
         self.numbered_list_pattern = re.compile(r"[\n:]\s*\p{N}\.")
 
         # Core sentence split regex
+        # NOTE: Acronyms like "U.S.A" are protected primarily by the lookahead (?=\s+...).
+        # Since "U.S.A," has no space after it (just punctuation), the lookahead fails
+        # and no split occurs. The negative lookbehind handles other abbreviations like "Dr."
+        # This means acronym protection is *not* dependent on masking—it's explicit in the
+        # lookahead requirement for whitespace or newline before the next uppercase letter.
         self.sentence_end_pattern = re.compile(
             rf"""
             (?<!\b(\p{{Lu}}\p{{Ll}}{{1, 4}}\.)*)   # Latin-only abbreviation

{chunklet_py-2.3.0 → chunklet_py-2.3.1/src/chunklet_py.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: chunklet-py
-Version: 2.3.0
+Version: 2.3.1
 Summary: High-fidelity context-aware chunking and interactive visualization for RAG. Advanced segmentation for code and documents, because your LLM is only as smart as the fragments you feed it.
 Author-email: speedyk_005 <speedy40115719@gmail.com>
 License-Expression: MIT
@@ -28,8 +28,8 @@ License-File: LICENSE
 Requires-Dist: loguru<1.0,>=0.7.3
 Requires-Dist: pysbd<1.0,>=0.3.4
 Requires-Dist: sentsplit<2.0,>=1.0.8
-Requires-Dist: sentencex<2.0,>=1.0.0; platform_system != "Android"
-Requires-Dist: sentencex<=0.6.1; platform_system == "Android"
+Requires-Dist: sentencex<=0.6.1; sys_platform == "linux" and (platform_machine == "aarch64" or platform_machine == "armv7l")
+Requires-Dist: sentencex<2.0,>=1.0.0; sys_platform != "linux" or (platform_machine != "aarch64" and platform_machine != "armv7l")
 Requires-Dist: indic-nlp-library<1.0,>=0.92
 Requires-Dist: py3langid<1.0,>=0.3.0
 Requires-Dist: mpire<3.0,>=2.10.2

{chunklet_py-2.3.0 → chunklet_py-2.3.1}/src/chunklet_py.egg-info/requires.txt

@@ -12,10 +12,10 @@ typer<1.0,>=0.19.0
 charset-normalizer<4.0,>=3.4.2
 setuptools<81
 
-[:platform_system != "Android"]
+[:sys_platform != "linux" or (platform_machine != "aarch64" and platform_machine != "armv7l")]
 sentencex<2.0,>=1.0.0
 
-[:platform_system == "Android"]
+[:sys_platform == "linux" and (platform_machine == "aarch64" or platform_machine == "armv7l")]
 sentencex<=0.6.1
 
 [all]