chore(release): rilascia versione 1.2.2

Author: strawberry-code

CHANGELOG.md

@@ -7,6 +7,14 @@ and the project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.
 
 ## [Unreleased]
 
+## [1.2.2] - 2025-12-03
+
+### Added
+- Chunking parameters now configurable via environment variables: `CHUNK_SIZE` (default 400) and `CHUNK_MAX_TOKENS` (default 1500)
+
+### Fixed
+- Reduced default chunk size to prevent Ollama panic with nomic-embed-text model (2048 token context limit)
+
 ## [1.2.1] - 2025-12-02
 
 ### Fixed

README.md

@@ -92,6 +92,8 @@ docker compose up -d
 | `BASE_URL` | `http://localhost:8080` | Public URL for OAuth callbacks |
 | `API_PORT` | `8080` | API and Web UI port |
 | `OLLAMA_MODEL` | `nomic-embed-text` | Embedding model |
+| `CHUNK_SIZE` | `400` | Target chunk size in tokens |
+| `CHUNK_MAX_TOKENS` | `1500` | Maximum chunk size (safe margin for nomic-embed-text 2048 limit) |
 
 ## Features
 

lib/chunking.py

@@ -5,11 +5,18 @@
 """
 
 import logging
+import os
 from typing import Optional, TypedDict, List
 import tiktoken
 
 logger = logging.getLogger(__name__)
 
+# Chunking configuration from environment variables
+# CHUNK_SIZE: Target chunk size in tokens (default: 400)
+# CHUNK_MAX_TOKENS: Maximum chunk size before re-chunking (default: 1500, safe for nomic-embed-text 2048 limit)
+CHUNK_SIZE = int(os.getenv('CHUNK_SIZE', '400'))
+CHUNK_MAX_TOKENS = int(os.getenv('CHUNK_MAX_TOKENS', '1500'))
+
 # Custom exception for chunking failures
 class ChunkingError(RuntimeError):
     """Raised when semantic chunking cannot be performed."""
@@ -212,27 +219,29 @@ def _fallback_chunk(
     return final_chunks
 
 
-def validate_chunk_size(chunk_text: str, max_tokens: int = 2048) -> bool:
+def validate_chunk_size(chunk_text: str, max_tokens: int = None) -> bool:
     """
     Validate that chunk doesn't exceed embedding model's token limit.
 
     Args:
         chunk_text: Chunk text to validate
-        max_tokens: Maximum allowed tokens (nomic-embed-text: 2048)
+        max_tokens: Maximum allowed tokens (default: CHUNK_MAX_TOKENS env var or 1500)
         
     Returns:
         True if chunk is within limits
     """
+    if max_tokens is None:
+        max_tokens = CHUNK_MAX_TOKENS
     token_count = count_tokens(chunk_text)
     return token_count <= max_tokens
 
 
 def create_chunks(
     text: str,
-    chunk_size: int = 500,
+    chunk_size: int = None,
     chunk_overlap: int = 50,
     min_tokens: int = 0,
-    max_tokens: int = 2048
+    max_tokens: int = None
 ) -> list[dict]:
     """
     Create chunks from text using two-level semantic chunking (chonkie + semchunk).
@@ -244,14 +253,20 @@ def create_chunks(
 
     Args:
         text: Text to chunk
-        chunk_size: Target chunk size in tokens (default: 500)
+        chunk_size: Target chunk size in tokens (default: CHUNK_SIZE env var or 400)
         chunk_overlap: Overlap between chunks in tokens (default: 50)
-        min_tokens: Minimum chunk size to keep (default: 50)
-        max_tokens: Maximum chunk size before re-chunking (default: 2048)
+        min_tokens: Minimum chunk size to keep (default: 0)
+        max_tokens: Maximum chunk size before re-chunking (default: CHUNK_MAX_TOKENS env var or 1500)
 
     Returns:
         List of chunk dictionaries with text and metadata
     """
+    # Apply defaults from environment variables
+    if chunk_size is None:
+        chunk_size = CHUNK_SIZE
+    if max_tokens is None:
+        max_tokens = CHUNK_MAX_TOKENS
+
     if not text or len(text.strip()) == 0:
         return []
 
@@ -295,21 +310,24 @@ def create_chunks(
 def filter_chunks(
     chunks: list[dict],
     min_tokens: int = 0,
-    max_tokens: int = 8192
+    max_tokens: int = None
 ) -> list[dict]:
     """
     Filter chunks by token count.
 
     Args:
         chunks: List of chunk dictionaries
         min_tokens: Minimum token count - 0 = keep all (default)
-        max_tokens: Maximum token count (needs re-chunking)
+        max_tokens: Maximum token count (default: CHUNK_MAX_TOKENS env var or 1500)
         
     Returns:
         Filtered list of valid chunks
     """
+    if max_tokens is None:
+        max_tokens = CHUNK_MAX_TOKENS
+
     valid_chunks = []
-    
+
     for chunk in chunks:
         token_count = chunk.get('token_count', count_tokens(chunk['text']))