Spaces:
Sleeping
Sleeping
File size: 10,059 Bytes
0a4529c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 | # DEPENDENCIES
from typing import List
from typing import Optional
from config.models import DocumentChunk
from config.settings import get_settings
from config.models import DocumentMetadata
from config.models import ChunkingStrategy
from config.logging_config import get_logger
from chunking.base_chunker import BaseChunker
from chunking.base_chunker import ChunkerConfig
from chunking.token_counter import TokenCounter
from chunking.semantic_chunker import SemanticChunker
from llama_index.core.node_parser import SentenceSplitter
from llama_index.core.node_parser import TokenTextSplitter
from llama_index.core.schema import Document as LlamaDocument
from llama_index.embeddings.huggingface import HuggingFaceEmbedding
from llama_index.core.node_parser import SemanticSplitterNodeParser
# Setup Settings and Logging
logger = get_logger(__name__)
settings = get_settings()
class LlamaIndexChunker(BaseChunker):
"""
LlamaIndex-based semantic chunking strategy:
- Uses LlamaIndex's advanced semantic splitting algorithms
- Provides superior boundary detection using embeddings
- Supports multiple LlamaIndex splitter types
Best for:
- Documents requiring sophisticated semantic analysis
- When LlamaIndex ecosystem integration is needed
- Advanced chunking with embedding-based boundaries
"""
def __init__(self, chunk_size: int = None, overlap: int = None, splitter_type: str = "semantic", min_chunk_size: int = 100):
"""
Initialize LlamaIndex chunker
Arguments:
----------
chunk_size { int } : Target tokens per chunk
overlap { int } : Overlap tokens between chunks
splitter_type { str } : Type of LlamaIndex splitter ("semantic", "sentence", "token")
min_chunk_size { int } : Minimum chunk size in tokens
"""
# Use SEMANTIC since it's semantic-based
super().__init__(ChunkingStrategy.SEMANTIC)
self.chunk_size = chunk_size or settings.FIXED_CHUNK_SIZE
self.overlap = overlap or settings.FIXED_CHUNK_OVERLAP
self.splitter_type = splitter_type
self.min_chunk_size = min_chunk_size
# Initialize token counter
self.token_counter = TokenCounter()
# Initialize LlamaIndex components
self._splitter = None
self._initialized = False
self._initialize_llamaindex()
self.logger.info(f"Initialized LlamaIndexChunker: chunk_size={self.chunk_size}, overlap={self.overlap}, splitter_type={self.splitter_type}")
def _initialize_llamaindex(self):
"""
Initialize LlamaIndex splitter with proper error handling
"""
try:
# Initialize embedding model
embed_model = HuggingFaceEmbedding(model_name = settings.EMBEDDING_MODEL)
# Initialize appropriate splitter based on type
if (self.splitter_type == "semantic"):
self._splitter = SemanticSplitterNodeParser(buffer_size = 1,
breakpoint_percentile_threshold = 95,
embed_model = embed_model,
)
elif (self.splitter_type == "sentence"):
self._splitter = SentenceSplitter(chunk_size = self.chunk_size,
chunk_overlap = self.overlap,
)
elif (self.splitter_type == "token"):
self._splitter = TokenTextSplitter(chunk_size = self.chunk_size,
chunk_overlap = self.overlap,
)
else:
self.logger.warning(f"Unknown splitter type: {self.splitter_type}, using semantic")
self._splitter = SemanticSplitterNodeParser(buffer_size = 1,
breakpoint_percentile_threshold = 95,
embed_model = embed_model,
)
self._initialized = True
self.logger.info(f"Successfully initialized LlamaIndex {self.splitter_type} splitter")
except ImportError as e:
self.logger.error(f"LlamaIndex not available: {repr(e)}")
self._initialized = False
except Exception as e:
self.logger.error(f"Failed to initialize LlamaIndex: {repr(e)}")
self._initialized = False
def chunk_text(self, text: str, metadata: Optional[DocumentMetadata] = None) -> List[DocumentChunk]:
"""
Chunk text using LlamaIndex semantic splitting
Arguments:
----------
text { str } : Input text
metadata { DocumentMetaData } : Document metadata
Returns:
--------
{ list } : List of DocumentChunk objects
"""
if not text or not text.strip():
return []
# Fallback if LlamaIndex not available
if not self._initialized:
self.logger.warning("LlamaIndex not available, falling back to simple semantic chunking")
return self._fallback_chunking(text = text,
metadata = metadata,
)
document_id = metadata.document_id if metadata else "unknown"
try:
# Create LlamaIndex document
llama_doc = LlamaDocument(text = text)
# Get nodes from splitter
nodes = self._splitter.get_nodes_from_documents([llama_doc])
# Convert nodes to our DocumentChunk format
chunks = list()
start_pos = 0
for i, node in enumerate(nodes):
chunk_text = node.text
# Create chunk
chunk = self._create_chunk(text = self._clean_chunk_text(chunk_text),
chunk_index = i,
document_id = document_id,
start_char = start_pos,
end_char = start_pos + len(chunk_text),
metadata = {"llamaindex_splitter" : self.splitter_type,
"node_id" : node.node_id,
"chunk_type" : "llamaindex_semantic",
}
)
chunks.append(chunk)
start_pos += len(chunk_text)
# Filter out chunks that are too small
chunks = [c for c in chunks if (c.token_count >= self.min_chunk_size)]
self.logger.debug(f"Created {len(chunks)} chunks using LlamaIndex {self.splitter_type} splitter")
return chunks
except Exception as e:
self.logger.error(f"LlamaIndex chunking failed: {repr(e)}")
return self._fallback_chunking(text = text,
metadata = metadata,
)
def _fallback_chunking(self, text: str, metadata: Optional[DocumentMetadata] = None) -> List[DocumentChunk]:
"""
Fallback to basic semantic chunking when LlamaIndex fails
Arguments:
----------
text { str } : Input text
metadata { DocumentMetaData } : Document metadata
Returns:
--------
{ list } : List of chunks
"""
fallback_chunker = SemanticChunker(chunk_size = self.chunk_size,
overlap = self.overlap,
similarity_threshold = 0.95,
min_chunk_size = self.min_chunk_size,
)
return fallback_chunker.chunk_text(text, metadata)
def get_splitter_info(self) -> dict:
"""
Get information about the LlamaIndex splitter configuration
Returns:
--------
{ dict } : Splitter information
"""
return {"splitter_type" : self.splitter_type,
"chunk_size" : self.chunk_size,
"overlap" : self.overlap,
"initialized" : self._initialized,
"min_chunk_size" : self.min_chunk_size,
}
@classmethod
def from_config(cls, config: ChunkerConfig) -> 'LlamaIndexChunker':
"""
Create LlamaIndexChunker from configuration
Arguments:
----------
config { ChunkerConfig } : ChunkerConfig object
Returns:
--------
{ LlamaIndexChunker } : LlamaIndexChunker instance
"""
return cls(chunk_size = config.chunk_size,
overlap = config.overlap,
splitter_type = config.extra.get('llamaindex_splitter', 'semantic'),
min_chunk_size = config.min_chunk_size,
) |