Spaces:
Sleeping
Sleeping
File size: 14,791 Bytes
c0f31c1 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 | """
ChunkingManager.py
A manager class that orchestrates document chunking using different strategies.
"""
from typing import Dict, List, Optional, Union
from pathlib import Path
from sentence_transformers import SentenceTransformer
from langchain_core.documents import Document
# Import chunker strategies
from core.BaseChunker import BaseChunker
from core.PageChunker import PageChunker
from core.ParagraphChunker import ParagraphChunker
from core.SemanticChunker import SemanticChunker
from core.HierarchicalChunker import HierarchicalChunker
from core.TokenChunker import TokenChunker
import logging
# Configure logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
class ChunkingStrategy:
"""Enumeration of available chunking strategies."""
PAGE = "page"
PARAGRAPH = "paragraph"
SEMANTIC = "semantic"
HIERARCHICAL = "hierarchical"
TOKEN = "token"
class ChunkingManager:
"""Manager class for document chunking strategies."""
def __init__(
self,
embedding_model_name: str = "all-mpnet-base-v2",
token_model_name: Optional[str] = None
):
"""
Initialize chunking manager.
Args:
embedding_model_name: Name of the sentence transformer model
token_model_name: Name of the token counting model
"""
self.token_model_name = token_model_name
self.embedding_model_name = embedding_model_name
self._embedding_model = None
self._chunkers = {}
@property
def embedding_model(self):
"""Lazy-load the embedding model."""
if self._embedding_model is None:
try:
# Only try to load as SentenceTransformer if it's a known SentenceTransformer model
if self.embedding_model_name and not any(x in self.embedding_model_name.lower() for x in ["gpt", "text-embedding", "openai"]):
logger.info(f"Loading embedding model: {self.embedding_model_name}")
self._embedding_model = SentenceTransformer(self.embedding_model_name)
else:
# Return a dummy embedding model that returns None
logger.info("Using dummy embedding model for tokenization only")
class DummyEmbedder:
def encode(self, text, **kwargs):
return [0.0] * 384 # Return dummy vector
self._embedding_model = DummyEmbedder()
except Exception as e:
logger.error(f"Error loading embedding model: {e}")
# Return a dummy embedding model that returns None
class DummyEmbedder:
def encode(self, text, **kwargs):
return [0.0] * 384 # Return dummy vector
self._embedding_model = DummyEmbedder()
return self._embedding_model
def _get_chunker(self, strategy: str) -> BaseChunker:
"""Get or create chunker for the specified strategy."""
strategy = strategy.lower()
if strategy not in self._chunkers:
if strategy == ChunkingStrategy.PAGE:
self._chunkers[strategy] = PageChunker(
model_name=self.token_model_name,
embedding_model=self.embedding_model
)
elif strategy == ChunkingStrategy.PARAGRAPH:
self._chunkers[strategy] = ParagraphChunker(
model_name=self.token_model_name,
embedding_model=self.embedding_model
)
elif strategy == ChunkingStrategy.SEMANTIC:
self._chunkers[strategy] = SemanticChunker(
embedding_model=self.embedding_model,
model_name=self.token_model_name
)
elif strategy == ChunkingStrategy.HIERARCHICAL:
self._chunkers[strategy] = HierarchicalChunker(
model_name=self.token_model_name,
embedding_model=self.embedding_model
)
elif strategy == ChunkingStrategy.TOKEN:
self._chunkers[strategy] = TokenChunker(
model_name=self.token_model_name,
embedding_model=self.embedding_model,
chunk_size=256, # Default values, could be made configurable
chunk_overlap=50
)
else:
raise ValueError(f"Unknown chunking strategy: {strategy}")
return self._chunkers[strategy]
def process_document(
self,
file_path: str,
strategy: str = ChunkingStrategy.PARAGRAPH,
preprocess: bool = True
) -> Union[List[Document], Dict[str, List[Document]]]:
"""
Process document using specified chunking strategy.
Args:
file_path: Path to document file
strategy: Chunking strategy to use
preprocess: Whether to preprocess text
Returns:
Chunked document(s) according to strategy
"""
# Validate file exists
path = Path(file_path)
if not path.exists():
raise FileNotFoundError(f"File not found: {file_path}")
# Determine file type
file_extension = path.suffix.lower()
# Process based on file type
if file_extension == '.csv':
return self._process_csv(file_path, strategy)
elif file_extension == '.txt':
return self._process_txt(file_path, strategy, preprocess)
elif file_extension == '.pdf':
# Get appropriate chunker and process document
chunker = self._get_chunker(strategy)
logger.info(f"Processing document using {strategy} chunking strategy")
if strategy == ChunkingStrategy.PAGE:
return chunker.page_process_document(file_path, preprocess)
elif strategy == ChunkingStrategy.PARAGRAPH:
return chunker.paragraph_process_document(file_path, preprocess)
elif strategy == ChunkingStrategy.SEMANTIC:
return chunker.semantic_process_document(file_path, preprocess)
elif strategy == ChunkingStrategy.HIERARCHICAL:
return chunker.hierarchical_process_document(file_path, preprocess)
elif strategy == ChunkingStrategy.TOKEN:
return chunker.token_process_document(file_path, preprocess)
else:
raise ValueError(f"Unknown chunking strategy: {strategy}")
else:
raise ValueError(f"Unsupported file type: {file_extension}. Supported types: .pdf, .csv, .txt")
def process_directory(
self,
dir_path: str,
strategy: str = ChunkingStrategy.PARAGRAPH,
preprocess: bool = True
) -> Dict[str, Union[List[Document], Dict[str, List[Document]]]]:
"""
Process all supported documents in a directory.
Args:
dir_path: Directory containing files
strategy: Chunking strategy to use
preprocess: Whether to preprocess text
Returns:
Dictionary mapping filenames to their processed documents
"""
path = Path(dir_path)
if not path.is_dir():
raise NotADirectoryError(f"Not a directory: {dir_path}")
results = {}
# Find supported files (PDFs, CSVs, and TXT files)
pdf_files = list(path.glob("**/*.pdf"))
csv_files = list(path.glob("**/*.csv"))
txt_files = list(path.glob("**/*.txt"))
all_files = pdf_files + csv_files + txt_files
logger.info(f"Found {len(pdf_files)} PDF files, {len(csv_files)} CSV files, and {len(txt_files)} TXT files in {dir_path}")
for file in all_files:
try:
logger.info(f"Processing {file.name}")
result = self.process_document(
str(file),
strategy=strategy,
preprocess=preprocess
)
results[file.name] = result
except Exception as e:
logger.error(f"Error processing {file.name}: {e}")
results[file.name] = {"error": str(e)}
return results
def _process_txt(self, file_path: str, strategy: str, preprocess: bool) -> List[Document]:
"""Process a TXT file into document chunks."""
logger.info(f"Processing TXT file: {file_path}")
# Validate strategy for TXT files
if strategy not in [ChunkingStrategy.PARAGRAPH, ChunkingStrategy.TOKEN]:
raise ValueError(f"TXT files only support paragraph and token chunking strategies. Got: {strategy}")
# Get appropriate chunker
chunker = self._get_chunker(strategy)
# Process based on strategy
if strategy == ChunkingStrategy.PARAGRAPH:
return chunker.process_text_file(file_path, preprocess)
elif strategy == ChunkingStrategy.TOKEN:
return chunker.process_text_file(file_path, preprocess)
else:
raise ValueError(f"Unsupported chunking strategy for TXT: {strategy}")
def _process_txt(self, file_path: str, strategy: str, preprocess: bool) -> List[Document]:
"""Process a TXT file into document chunks."""
logger.info(f"Processing TXT file: {file_path}")
# Validate strategy for TXT files
if strategy not in [ChunkingStrategy.PARAGRAPH, ChunkingStrategy.TOKEN]:
raise ValueError(f"TXT files only support paragraph and token chunking strategies. Got: {strategy}")
# Get appropriate chunker
chunker = self._get_chunker(strategy)
# Process based on strategy
if strategy == ChunkingStrategy.PARAGRAPH:
return chunker.process_text_file(file_path, preprocess)
elif strategy == ChunkingStrategy.TOKEN:
return chunker.process_text_file(file_path, preprocess)
else:
raise ValueError(f"Unsupported chunking strategy for TXT: {strategy}")
def _process_csv(self, file_path: str, strategy: str) -> List[Document]:
"""Process a CSV file into document chunks."""
import pandas as pd
logger.info(f"Loading CSV file: {file_path}")
# Read the CSV file
df = pd.read_csv(file_path)
# Determine the chunking approach based on strategy
if strategy == ChunkingStrategy.PARAGRAPH:
# For these strategies, we treat each row as a separate document
# with columns combined into a structured text format
return self._chunk_csv_by_row(df, file_path)
elif strategy == ChunkingStrategy.PAGE:
# For page strategy, we create larger chunks with multiple rows
return self._chunk_csv_by_page(df, file_path)
elif strategy == ChunkingStrategy.HIERARCHICAL:
# For hierarchical, create documents with metadata structure
return {"chunks": self._chunk_csv_by_row(df, file_path)}
else:
raise ValueError(f"Unsupported chunking strategy for CSV: {strategy}")
def _chunk_csv_by_row(self, df, file_path: str) -> List[Document]:
"""Convert each CSV row to a document chunk."""
chunks = []
file_name = Path(file_path).name
# Get column names
columns = df.columns.tolist()
# Process each row
for i, row in df.iterrows():
# Convert row to formatted text
content = "\n".join([f"{col}: {row[col]}" for col in columns])
# Create metadata
metadata = {
"source": file_path,
"file_name": file_name,
"file_type": "csv",
"row_index": i,
"chunk_type": "csv_row",
}
# Add columns as additional metadata
for col in columns:
# Convert to string to ensure compatibility
metadata[f"csv_{col}"] = str(row[col])
# Create document
doc = Document(page_content=content, metadata=metadata)
chunks.append(doc)
logger.info(f"Created {len(chunks)} chunks from CSV (row-based)")
return chunks
def _chunk_csv_by_page(self, df, file_path: str, rows_per_chunk: int = 20) -> List[Document]:
"""Convert CSV into larger chunks with multiple rows per chunk."""
chunks = []
file_name = Path(file_path).name
columns = df.columns.tolist()
# Calculate number of chunks
total_rows = len(df)
chunk_count = (total_rows + rows_per_chunk - 1) // rows_per_chunk # Ceiling division
# Generate chunks
for chunk_idx in range(chunk_count):
start_row = chunk_idx * rows_per_chunk
end_row = min(start_row + rows_per_chunk, total_rows)
chunk_df = df.iloc[start_row:end_row]
# Format the chunk content
content = f"CSV Data (Rows {start_row+1}-{end_row}):\n\n"
# Add header row
content += " | ".join(columns) + "\n"
content += "-" * (sum(len(col) for col in columns) + 3 * (len(columns) - 1)) + "\n"
# Add data rows
for _, row in chunk_df.iterrows():
content += " | ".join(str(row[col]) for col in columns) + "\n"
# Create metadata
metadata = {
"source": file_path,
"file_name": file_name,
"file_type": "csv",
"chunk_type": "csv_page",
"start_row": start_row,
"end_row": end_row - 1,
"row_count": end_row - start_row,
}
# Create document
doc = Document(page_content=content, metadata=metadata)
chunks.append(doc)
logger.info(f"Created {len(chunks)} chunks from CSV (page-based)")
return chunks |